# IMPACT OF CANCER PLASTICITY ON DRUG RESISTANCE AND TREATMENT IN SOLID TUMORS

EDITED BY : Dong-Hua Yang, Pascale Cohen and Chang Zou PUBLISHED IN : Frontiers in Oncology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88966-275-3 DOI 10.3389/978-2-88966-275-3

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# IMPACT OF CANCER PLASTICITY ON DRUG RESISTANCE AND TREATMENT IN SOLID TUMORS

Topic Editors: Dong-Hua Yang, St. John's University, United States Pascale Cohen, Université Claude Bernard Lyon 1, France Chang Zou, Jinan University, China

Citation: Yang, D.-H., Cohen, P., Zou, C., eds. (2020). Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-275-3

# Table of Contents

*06 Editorial: Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors* Pascale A. Cohen, Chang Zou and Dong-Hua Yang *09 SIRT5 Promotes Cisplatin Resistance in Ovarian Cancer by Suppressing DNA Damage in a ROS-Dependent Manner via Regulation of the Nrf2/HO-1 Pathway* Xiaodan Sun, Shouhan Wang, Junda Gai, Jingqian Guan, Ji Li, Yizhuo Li, Jinming Zhao, Chen Zhao, Lin Fu and Qingchang Li *25 MTA3-SOX2 Module Regulates Cancer Stemness and Contributes to Clinical Outcomes of Tongue Carcinoma* Zhimeng Yao, Liang Du, Min Xu, Kai Li, Haipeng Guo, Guodong Ye, Dianzheng Zhang, Robert P. Coppes and Hao Zhang *37 An Engineered Fusion Protein Anti-CD19(Fab)-LDM Effectively Inhibits ADR-Resistant B Cell Lymphoma* Dongmei Fan, Linlin Jiang, Yuewen Song, Shiqi Bao, Yuanyuan Yang, Xiangfei Yuan, Yongsu Zhen, Ming Yang and Dongsheng Xiong *48 Current Advance of Therapeutic Agents in Clinical Trials Potentially Targeting Tumor Plasticity* Xiao-Guang Yang, Lan-Cao Zhu, Yan-Jun Wang, Yan-Yu Li and Dun Wang *58 MicroRNA and mRNA Interaction Network Regulates the Malignant Transformation of Human Bronchial Epithelial Cells Induced by Cigarette Smoke* Jin Wang, Xiao-fan Yu, Nan Ouyang, Shiyu Zhao, Haiping Yao, Xifei Guan, Jian Tong, Tao Chen and Jian-xiang Li *73 MiR-199a Inhibits Tumor Growth and Attenuates Chemoresistance by Targeting K-RAS via AKT and ERK Signalings* Wei Li, Lin Wang, Xiang-Bo Ji, Li-Hong Wang, Xin Ge, Wei-Tao Liu, Ling Chen, Zhong Zheng, Zhu-Mei Shi, Ling-Zhi Liu, Marie C. Lin, Jie-Yu Chen and Bing-Hua Jiang *83 LncRNA AFAP1-AS1 Supresses miR-139-5p and Promotes Cell Proliferation and Chemotherapy Resistance of Non-small Cell Lung Cancer by Competitively Upregulating RRM2*

Na Huang, Wei Guo, Ke Ren, Wancheng Li, Yi Jiang, Jian Sun, Wenjing Dai and Wei Zhao

*100 Chronic BDE-47 Exposure Aggravates Malignant Phenotypes and Chemoresistance by Activating ERK Through ER*a *and GPR30 in Endometrial Carcinoma*

Fan Zhang, Lin Peng, Yiteng Huang, Xueqiong Lin, Li Zhou and Jiongyu Chen

*112 The Tumor Suppressor Role of Zinc Finger Protein 671 (*ZNF671*) in Multiple Tumors Based on Cancer Single-Cell Sequencing* Jian Zhang, Jianli Luo, Huali Jiang, Tao Xie, Jieling Zheng, Yunhong Tian, Rong Li, Baiyao Wang, Jie Lin, Anan Xu, Xiaoting Huang and Yawei Yuan


Meng Gao, Chengyuan Li, Han Xiao, Hang Dong, Siyi Jiang, Yunfeng Fu and Liying Gong

*140 Canmei Formula Reduces Colitis-Associated Colorectal Carcinogenesis in Mice by Modulating the Composition of Gut Microbiota*

Huayue Zhang, Dengcheng Hui, Yuan Li, Guangsu Xiong and Xiaoling Fu


Jinan Guo, Pan Zhao, Zengqin Liu, Zaishang Li, Yeqing Yuan, Xueqi Zhang, Zhou Yu, Jiequn Fang and Kefeng Xiao


Li-Na Yu, Zhen Liu, Yan Tian, Pei-Pei Zhao and Xing Hua


Kaisheng Liu, Juan Chen, Fang Yang, Zhifan Zhou, Ying Liu, Yaomin Guo, Hong Hu, Hengyuan Gao, Haili Li, Wenbin Zhou, Bo Qin and Yifei Wang


De-Rong Tang, Cheng-Lin Li, Ke-Ping Xu, Qing-Quan Wu, Qi-You Chen, Jun-Jie Lv, Jian Ji, Bao Zang, Chen Chen, Biao Gu and Jian-Qiang Zhao

*235 Knockdown of Thymidine Kinase 1 Suppresses Cell Proliferation, Invasion, Migration, and Epithelial–Mesenchymal Transition in Thyroid Carcinoma Cells*

Chang Liu, Jian Wang, Li Zhao, Hui He, Pan Zhao, Zheng Peng, Feiyuan Liu, Juan Chen, Weiqing Wu, Guangsuo Wang and Fajin Dong

*244 miR-936 Suppresses Cell Proliferation, Invasion, and Drug Resistance of Laryngeal Squamous Cell Carcinoma and Targets GPR78*

Xi-Jun Lin, Hui Liu, Pei Li, Hai-Feng Wang, An-Kui Yang, Jin-Ming Di, Qi-Wei Jiang, Yang Yang, Jia-Rong Huang, Meng-Ling Yuan, Zi-Hao Xing, Meng-Ning Wei, Yao Li, Zhi Shi and Jin Ye

*252 7-Methoxy-1-Tetralone Induces Apoptosis, Suppresses Cell Proliferation and Migration in Hepatocellular Carcinoma via Regulating c-Met, p-AKT, NF-*k*B, MMP2, and MMP9 Expression*

Ying Wen, Xiaoyan Cai, Shaolian Chen, Wei Fu, Dong Chai, Huainian Zhang and Yongli Zhang


Zhuo-Xun Wu, Xingduo Dong, Leli Zeng, Linguo Zhao, Dong-Hua Yang and Zhe-Sheng Chen


Giuseppe Nicolò Fanelli, Antonio Giuseppe Naccarato and Cristian Scatena

*321 Long-Term Exposure of Early-Transformed Human Mammary Cells to Low Doses of Benzo[a]pyrene and/or Bisphenol A Enhances Their Cancerous Phenotype via an AhR/GPR30 Interplay*

Caterina F. Donini, Myriam El Helou, Anne Wierinckx, Balázs Győrffy, Sophie Aires, Aurélie Escande, Séverine Croze, Philippe Clezardin, Joël Lachuer, Mona Diab-Assaf, Sandra E. Ghayad, Béatrice Fervers, Vincent Cavaillès, Véronique Maguer-Satta and Pascale A. Cohen


Assil Fahs, Farah Ramadan, Farah Ghamloush, Abeer J. Ayoub, Fatima Ali Ahmad, Firas Kobeissy, Yehia Mechref, Jingfu Zhao, Rui Zhu, Nader Hussein, Raya Saab and Sandra E. Ghayad

# Editorial: Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors

#### Pascale A. Cohen1,2,3 \*, Chang Zou<sup>4</sup> and Dong-Hua Yang<sup>5</sup> \*

<sup>1</sup> Université Lyon 1, Lyon, France, <sup>2</sup> INSERM, UMR1033 LYOS, Lyon, France, <sup>3</sup> CRCL-Centre de Recherche en Cancérologie de Lyon-Inserm U1052-CNRS U5286, Lyon, France, <sup>4</sup> Clinical Medical Research Center, Shenzhen People's Hospital, Jinan University, Shenzhen, China, <sup>5</sup> Department of Pharmaceutical Sciences, College of Pharmacy and Health Sciences, St. John's University, Queens, NY, United States

Keywords: cancer plasticity, solid tumor, drug resistance, biomarker, therapeutic

#### **Editorial on the Research Topic**

#### **Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors**

The Research Topic "Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors" consists of 32 articles contributed by more than 270 authors in the field of oncology, pharmacology, and translational research. Our aim was to provide a collaborative discussion on molecular and cellular regulators of cancer cell plasticity contributing to tumor progression and drug resistance for the future direction of biomarker discovery and therapeutic strategies.

Cancer stem cells, tumor microenvironment, stroma/cancer cells interactions, changes in metabolism and epithelial-mesenchymal transition offer explanation for tumor plasticity. The current state of art in this era was elegantly reviewed by Fanelli et al., Yang et al., and Lin X. et al., who discussed the clinical relevance of cancer cell plasticity, the novel approaches for monitoring tumor plasticity and the current advances for therapeutic targeting. Yu et al. found that the FAP-a+GOLPH3<sup>+</sup> immunophenotype, combining the expression of both the fibroblast activation protein-alpha and the oncogenic Golgi phosphoprotein 3 protein predict the recurrence and progression of ductal carcinoma in-situ (DCIS) into invasive breast cancer. Yao et al. demonstrated in mouse experiments that the levels of MTA3 and SOX2 decreased and increased, respectively, during the progression of tongue squamous cell cancer (TSCC), and that MTA3low/SOX2high can serve as an independent prognostic factor for TSCC patients. Chen et al. confirmed that overexpression of PD-L1 occurred predominantly in highly aggressive glioma cells, and Akt binding/activation prevented autophagic cytoskeleton collapse, thus facilitating glioma cell invasion upon starvation stress. Sun et al. showed that SIRT5, a mitochondrial class III NAD-dependent deacetylase, contributes to cisplatin resistance in ovarian cancer by suppressing cisplatin-induced DNA damage in a reactive oxygen species (ROS)-dependent manner, via the regulation of the nuclear factor erythroid 2-related factor 2 (Nrf2)/heme oxygenase 1 (HO-1) pathway. The study from Tang et al. suggested that the Pigment epithelium-derived factor (PEDF) participates the carcinogenesis of human esophageal squamous cell carcinoma and might be a candidate therapeutic target. Finally, analyses conducted by Zhang J. et al. on single-cell sequencing datasets of several human cancers indicated a tumor suppression function of the ZNF671 transcription factor. Fahs et al. demonstrated that the PAX3-FOXO1 fusion protein modulates exosome cargo to confer a protective effect on recipient cells against oxidative stress and to promote plasticity and survival, potentially contributing to the known aggressive phenotype of the fusion gene-positive subtype of Rhabdomyosarcoma. Guo T. et al. reported a clinical case showing change of pathological type to metaplastic squamous cell carcinoma of the breast during disease recurrence.

Edited and reviewed by: Paolo Pinton, University of Ferrara, Italy

#### \*Correspondence:

Pascale A. Cohen pascale.cohen@univ-lyon1.fr Dong-Hua Yang yangd1@stjohns.edu

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 20 August 2020 Accepted: 16 September 2020 Published: 23 October 2020

#### Citation:

Cohen PA, Zou C and Yang D-H (2020) Editorial: Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors. Front. Oncol. 10:596963. doi: 10.3389/fonc.2020.596963

Epigenetic reprogramming favors cancer plasticity. The discovery of non-coding RNA such as microRNA (miR), Long non-coding RNA (LncRNA) and circular RNA (circ-RNA) is propelling the future advancement of biomarker development and offers opportunities to understand their role in the hallmarks of cancer, including signaling pathways involved in cell proliferation, cell invasion, metabolic plasticity and drug resistance. Wan et al. deciphered the functional domains of the channel-kinase transient receptor potential ion channel subfamily M, member 7 (TRPM7) involved in glioma cell growth or migration/invasion. TRPM7 was found to regulate miR-28-5p expression, which suppresses cell proliferation and invasion in glioma cells by targeting the Rap1b signaling. Guo J. et al. demonstrated that miR-204-3p, whose down-regulation was significantly associated with poor prognosis in bladder cancer patients, negatively modulated the proliferation of bladder cancer cells via targeting the lactate dehydrogenase (LDHA) mediated glycolysis. Huang et al. elegantly provided evidence that LncRNA AFAP1-S1 up-regulates the RRM2 protein levels by sponging miR-139-5, then activating an RRM2/EGFR/Akt axis that promotes chemoresistance in non-small cell lung cancer. Supportive in vivo experiments further demonstrated that knockdown of AFAP1-AS1 significantly suppressed tumor growth and chemoresistance. Li W. et al. proved that miR-199a, by directly regulating K-RAS and thus the downstream AKT and ERK signaling, inhibits glioma cell proliferation in vitro, tumor growth in vivo and increases sensitivity to telozomide, a drug used in first line treatment of glioma. Lin X.-J. et al. highlighted the role of miR-936 in sensitizing laryngeal squamous cancer cells to doxorubicin and cisplatin. Liu C. et al. experiments suggested that miR-34a-5p, by directly targeting thymidine kinase 1 (TKI), may be part of the mechanisms negatively regulating TKI-driven thyroid carcinoma cell aggressiveness. Growing body of evidence indicate that circRNAs play a role in disease progression, partly by sponging miRNA, and may be used as biomarkers. Gao et al. identified a candidate circRNA associated with poor prognosis in multiple myeloma. Finally, the review by Guo Q. et al. elegantly depicted and discussed the role of exosomal miRNA as a regulators and biomarkers in cancer drug resistance.

It is of utmost importance to decipher how chronic exposure to environmental carcinogens contribute to cell plasticity and tumor progression. The identification of such molecular mechanisms may help in the discovery of human biomarkers of environmental carcinogen exposure and the development of candidate preventive strategies. Using an in vitro model for malignant transformation of normal lung cells upon longterm exposure to cigarette smoke, Wang et al. deciphered complex miRNA-mRNA networks associated with cancerrelated signaling pathways, in particular those governing the metastasis-associated epithelial-mesenchymal transition and the PI3K/Akt/mTOR survival pathway. Donini et al. demonstrated a functional interplay between the aryl hydrocarbon receptor (AhR) and the G protein-coupled receptor 30 (GPR30) by which chronic and low-dose exposure of the genotoxic Benzo[a]pyrene and/or the endocrine disruptor Bisphenol A fosters the progression of early-transformed human mammary cells into a more aggressive stage. Zhang F. et al. proved that the chronic exposure of endometrial carcinoma cells to the polybrominated diphenyl ether endocrine disruptor BDE-47 triggers phenotypic plasticity, promotes progression and chemoresistance to cisplatin or paclitaxel, at least in part, via ERα/GPR30 and EGFR/ERK signaling pathways.

Antineoplastic drugs can induce cancer cells resistant to treatment that makes the therapeutic effect reduced. Circumventing drug resistance or delineating novel biomarkers of drug efficacy thus represent a great challenge, in particular in the era of precision medicine. To overcome multidrug resistance (MDR) mediated by overexpression of ATP binding cassette (ABC), Wang et al. provided a promising strategy, by nicely highlighting that NVP-TAE684, a novel ALK inhibitor, could reduce the chemo-resistance of MDR cells via the reversion of ABCG2-mediated efflux activity. Zhou et al. found that Erastin inhibits the drug efflux activity of ABCB1 and reverses ABCB1 mediated docetaxel resistance in ovarian cancer. Altogether, these two studies reveal that combination of NVP-TAE684 or Erastin with classical chemotherapy may offer potential effective combinations for the treatment of MDR cancers. Fan et al. explored novel methods to circumvent MDR in B cell lymphoma and demonstrated that an engineered anti-CD19(Fab)-lidamycin cytotoxic fusion protein could effectively inhibit, both in vitro and in vivo, the growth of adriamycinresistant B cell lymphoma cells. Finally, to clarify the correlation between drug efficacy and mutations in circulating tumor DNA (ctDNA), Cao et al. monitored the mutational changes and therapeutic response of late-stage colorectal cancers following chemotherapy combined with bevacizumab and/or cetuximab, and confirmed that dynamic changes in drug resistance can be sensitively monitored by gene variation status in ctDNA.

Elucidating the mechanisms of cancer plasticity offers new opportunities for the development of therapeutic targeting. Among future candidate targets is the microbiota, demonstrated to be involved in both tumor initiation and progression. Zhang H. et al. found that Canmei formula, a classical traditional Chinese herbal formulation, reduced in vivo colitis-associated colorectal carcinogenesis by modulating inflammation and the composition of the gut microbiota. Huang et al. explored in vitro the anticancerous activity the Methyl-cantharidimide drug, originally discovered in insects, and proved that it inhibits growth of human hepatocellular carcinoma cells by inducing cell cycle arrest and promoting apoptosis. Nan et al. proposed that ROS-mediated proteasome-dependent pathway can be exploited to overcome apoptosis resistance triggered by aberrant expression of the anti-apoptotic survivin protein in cancers, by demonstrating that the alkaloid/amide Piperlongumine extracted from peppers efficiently induced the proteasome-dependent degradation of the survivin in vitro, downregulated survivin in vivo and inhibition of ovarian cancer cells xenograft tumor growth. Finally, an elegant study from Liu K. et al. identified, by proteomics investigation of breast cancer specimen, increased protein levels of Hsp90 proteins associated with poor prognosis. BJ-B11, an Hsp90 inhibitor, was shown to hamper in vitro the cancerous properties of triple negative breast cancer cells and to inhibit tumor growth in xenograft model. Wen et al. explored the antitumor effects of 7-methoxy-1-tetralone in hepatocellular carcinoma and found that this compound induces apoptosis, suppresses cell proliferation and migration via regulating c-Met, p-AKT, NF-κB, MMP2, and MMP9 expression. Finally, Yan and Wang nicely reviewed brain cancer-associated alterations of proteoglycans making these latter as putative biomarkers or therapeutic targets.

In conclusion, the "Impact of Cancer Plasticity on Drug Resistance and Treatment in Solid Tumors" Research Topic highlights the complex phenomenon of cancer cell plasticity. The recent insights into the role of plasticity in cancer progression implies the need to continue improving our understanding into the fundamental mechanisms governing tumor progression and drug resistance, with the aim to identify relevant new biomarkers and to develop innovative therapies.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# FUNDING

PC was funded by the Region Auvergne Rhone-Alpes, France (Pack Ambition International 2020) and the Agence Nationale de la Recherche, France (2011 ANR-CESA-018-01).

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Cohen, Zou and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# SIRT5 Promotes Cisplatin Resistance in Ovarian Cancer by Suppressing DNA Damage in a ROS-Dependent Manner via Regulation of the Nrf2/HO-1 Pathway

Xiaodan Sun<sup>1</sup> , Shouhan Wang<sup>2</sup> , Junda Gai <sup>1</sup> , Jingqian Guan<sup>1</sup> , Ji Li <sup>1</sup> , Yizhuo Li <sup>1</sup> , Jinming Zhao<sup>1</sup> , Chen Zhao<sup>3</sup> , Lin Fu1,4 and Qingchang Li 1,4 \*

*<sup>1</sup> Department of Pathology, College of Basic Medical Sciences, China Medical University, Shenyang, China, <sup>2</sup> Department of Hepatopancreatobiliary Surgery, Jilin Province Cancer Hospital, Changchun, China, <sup>3</sup> Department of Pathology, Carver College of Medicine, University of Iowa, Iowa City, IA, United States, <sup>4</sup> Department of Pathology, The First Affiliated Hospital, China Medical University, Shenyang, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Monica Montopoli, University of Padova, Italy Daniele Vergara, University of Salento, Italy*

\*Correspondence:

*Qingchang Li qcli@cmu.edu.cn*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

> Received: *03 June 2019* Accepted: *26 July 2019* Published: *13 August 2019*

#### Citation:

*Sun X, Wang S, Gai J, Guan J, Li J, Li Y, Zhao J, Zhao C, Fu L and Li Q (2019) SIRT5 Promotes Cisplatin Resistance in Ovarian Cancer by Suppressing DNA Damage in a ROS-Dependent Manner via Regulation of the Nrf2/HO-1 Pathway. Front. Oncol. 9:754. doi: 10.3389/fonc.2019.00754* Sirtuin 5 (SIRT5), a mitochondrial class III NAD-dependent deacetylase, plays controversial roles in tumorigenesis and chemoresistance. Accordingly, its role in ovarian cancer development and drug resistance is not fully understood. Here, we demonstrate that SIRT5 is increased in ovarian cancer tissues compared to its expression in normal tissues and this predicts a poor response to chemotherapy. SIRT5 levels were also found to be higher in cisplatin-resistant SKOV-3 and CAOV-3 ovarian cancer cells than in cisplatin-sensitive A2780 cells. Furthermore, this protein was revealed to facilitate ovarian cancer cell growth and cisplatin-resistance *in vitro*. Mechanistically, we show that SIRT5 contributes to cisplatin resistance in ovarian cancer by suppressing cisplatin-induced DNA damage in a reactive oxygen species (ROS)-dependent manner via regulation of the nuclear factor erythroid 2-related factor 2 (Nrf2)/heme oxygenase 1 (HO-1) pathway.

Keywords: ovarian cancer, SIRT5, Nrf2/HO-1, reactive oxygen species, drug resistance

# INTRODUCTION

SIRT5 is a unique member of the Sirtuin family (Sirt1–7), which possesses multiple enzymatic activities including NAD-dependent histone deacetylase (1), potent lysine demalonylase, lysine desuccinylase (2), and lysine glutarylase (3) activities. These specific enzymatic activities indicate that SIRT5 plays a crucial role in regulating multiple cellular metabolic processes such as glycolysis, the tricarboxylic acid cycle, fatty acid oxidation, nitrogen metabolism, and the pentose phosphate pathway (4, 5). In addition, certain aspects of cancer biology, such as stress responses (6, 7), apoptosis (8, 9), and autophagy (10, 11), can be regulated by SIRT5. Moreover, altered cellular metabolism has been recently identified as a hallmark of malignancy and emerging literature suggests that SIRT5 is involved in oncogenesis. For example, either the mRNA or protein expression of SIRT5 was found to be increased in non-small cell lung cancer (NSCLC) (12, 13), hepatocellular carcinoma (HCC) (14), colorectal cancer (CRC) (15), Waldenstrom macroglobulinemia (16), and breast cancer (9, 17), compared to the levels in matched normal tissues. Janus-faced roles of SIRT5 in cancer have also been described. Specifically, the downregulation of SIRT5 was observed

**9**

in head and neck squamous cell carcinoma (18), liver cancer (19), and endometrial carcinoma (20), which highlight the tumor-suppressive role of SIRT5. Furthermore, controversial roles for SIRT5 in drug resistance have been reported. SIRT5 facilitates NSCLC resistance to cisplatin, 5-fluorouracil (5- FU), and bleomycin (13). Moreover, SIRT5-positive cells in wild-type Kras CRCs were found to be resistant to either chemotherapeutic agents or cetuximab (21). However, a positive association between SIRT5 expression and complete response to neoadjuvant chemotherapy in triple-negative breast cancer patients was previously shown by analyzing data from the Gene Expression Omnibus (GEO) DataSet. In addition, by analyzing the ONCOMINE online database, SIRT5 expression levels were found to be higher in chemotherapy-responders than in non-responders (17). Considering these discrepant findings, further studies are needed to explore the functions of SIRT5 in tumorigenesis and chemoresistance.

Globally, ovarian cancer ranked eighth in incidence and seventh in mortality among all cancers in females in 2018 (WHO, http://gco.iarc.fr/today/home). Standard treatment for this disease involves surgery combined with chemotherapy. However, chemoresistance has become a major reason for poor outcomes in ovarian cancer. Although emerging targeted therapies have improved the survival of chemoresistant ovarian cancer patients, their quality of life and overall survival are still limited due to the side effects associated with such drugs, and the 5-year survival rate for advanced-stage ovarian cancer is only 28.9% (22). Therefore, better curative outcomes for ovarian cancer therapeutics are required. Considering the diverse roles of SIRT5 in cancer biology, we speculated that SIRT5 is a therapeutic target for ovarian cancer. To date, the role of this protein in ovarian cancer has not been elucidated. Therefore, in this context, we investigated the expression pattern of SIRT5 in both human ovarian cancer tissues and ovarian cancer cells. Furthermore, we reveal a potential role for SIRT5 in ovarian cancer cell growth and chemoresistance.

# MATERIALS AND METHODS

#### Antibodies and Reagents

The following antibodies were purchased: primary antibodies to SIRT5 (Cell Signaling Technology, Danvers, MA), H2A histone family member X phosphorylated on S139 (γ-H2AX, Cell Signaling Technology), nuclear factor erythroid-2 related factor 2 (Nrf2; Proteintech, Wuhan, China), heme oxygenase 1 (HO-1; Proteintech), manganese-dependent superoxide dismutase (MnSOD)/SOD2 (Proteintech), breast cancer gene 1 (BRCA1, Cell Signaling Technology), histone H3 (Cell Signaling Technology), and β-actin (Cell Signaling Technology); and secondary antibodies, specifically goat antirabbit IgG (Proteintech), goat anti-mouse IgG (Proteintech), and tetramethylrhodamine (TRITC)-conjugated secondary antibody (Proteintech).

Cisplatin and ML385 (an Nrf2-specific inhibitor) were purchased from MCE, China. N-acetyl-L-cysteine (NAC), a reactive oxygen species (ROS) scavenger, was purchased from Selleck, China.

#### Immunohistochemical Staining

SIRT5 staining was performed on an ovarian cancer and normal tissue microarray (Biomax, USA), which contained 90 tumor tissues and 10 normal ovarian tissues. Immunostaining was performed based on the ABC immunostaining protocol (MaiXin, Fuzhou, China). Two independent investigators scored the microarray by evaluating the staining intensity and percentage of stained cells blindly and randomly. The staining intensity was scored from 0 to 3 as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The percentage of positively stained cells was scored from 0 to 4 as follows: 0 (0%), 1 (1–25%), 2 (26–50%), 3 (51–75%), and 4 (76–100%). The final multiplied scores ranged from 0 to 12 and SIRT5 expression was regarded as positive if the final score was ≥ 6.

# Cell Culture and Transfection

A2780, SKOV-3, and CAOV-3 human ovarian cancer cell lines were used in this study. A2780 and CAOV-3 cells were cultured in DMEM supplemented with 10% fetal bovine serum. SKOV-3 cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum. These cells were grown at 37◦C in a humidified atmosphere with 5% CO2.

Transient transfection was carried out using Lipofectamine 3000 reagent (Invitrogen) according to the manufacturer's instructions. A SIRT5 expression plasmid (OriGene, Rockville, MD, USA) and the corresponding empty plasmid (OriGene) were used for SIRT5 overexpression and as a negative control, respectively. Cells were transfected with SIRT5-siRNA and negative control siRNA (GenePharma, Shanghai, China) for SIRT5 knockdown experiments.

The cells were pretreated with 5 mM NAC for 2 h to inhibit ROS generation. ML385 was added to cells at a concentration of 5µM for 48 h in the presence of cisplatin to block the Nrf2 pathway.

#### Cell Proliferation and Cell Viability Assays

Cell Counting Kit-8 (CCK-8) assays were performed to assess cell proliferation and cell viability in vitro. For cell proliferation, the cells were seeded in 96-well plates at a density of 3,000 cells/well and incubated for 5 days. For cell viability, the cells were seeded in 96-well plates at a density of 5,000 cells/well and treated with the indicated concentration of cisplatin for 48 h or 5 days, changing the cisplatin-containing medium every 2 days. Next, 10 µl of CCK-8 reagent (Beyotime, Shanghai, China) was added to each well, and the cells were incubated for 2 h. The absorbance value (OD) of each well was measured at 450 nm and the cell viability was calculated as follows: cell viability (%) = experimental group OD value / control group OD value × 100%. IC<sup>50</sup> values (50% inhibition of surviving fraction) were calculated using GraphPad Prism 7.0 software.

#### Colony Formation Assays

Cells were plated in six-well culture plates at 500 cells/well. After incubation for 14 days at 37◦C in a humidified atmosphere at 5% CO2, the cells were washed three times with PBS and stained with Giemsa solution. The number of colonies containing ≥ 50 cells was then counted under a microscope. Colony formation efficiency was calculated as colony numbers / 500 × 100%.

#### ROS Detection and Measurement of Intracellular Glutathione (GSH)

The ROS levels induced by cisplatin in ovarian cancer cells were detected using the probe 2′ ,7′ -dichlorodihydrofluorescein diacetate (DCFH-DA, Beyotime), which can be oxidized by intracellular oxygen to dichlorofluorescein, a highly fluorescent compound. After exposure to the indicated concentration of cisplatin for 2, 6, 24, or 48 h, the cells were incubated with a final concentration of 10µM DCFH-DA in the dark for 20 min at 37◦C in a humidified atmosphere at 5% CO2, after which the cells were washed three times with cold PBS to remove excess fluorescent probe. The cells were then resuspended in 300 µl of PBS and assessed for fluorescence intensity using a flow cytometer (LSRFortessa, BD Biosciences). The data were analyzed using FlowJo X 10.0.7 Software.

Intracellular GSH levels were measured using a Total Glutathione Assay Kit (Beyotime) according to the manufacturer's instructions. Briefly, the ovarian cancer cells were harvested and lysed in the protein removal solution S provided in the kit. After incubation for 5 min at 4◦C, the samples were centrifuged at 10,000 g for 10 min at 4◦C. The supernatant was treated with assay solution for 5 min at 25◦C and the absorbance at 412 nm was measured using a microplate

FIGURE 1 | SIRT5 expression is increased in ovarian cancer tissues and high SIRT5 level predicts poor chemo-response. (A) The mRNA expression of *Sirtuins* in patients with ovarian cancer (GEPIA). (B) Immunohistochemical staining results of SIRT5 expression in 90 cases ovarian cancer tissues and 10 cases normal tissues microarrays. (C) Representative immunohistochemical staining images of SIRT5 expression (positive or negative) in ovarian cancer and normal tissues. (Magnification, × 400; Scale bar = 20µm). (D) High SIRT5 level predicts poor progression-free survival (PFS) by online Kaplan-Meier analysis. \*\**P* < 0.01.

TABLE 1 | Association of SIRT5 expression with clinicopathological characteristics of ovarian cancer tissue microarrays.


*The bold fonts represent the significant values.*

reader (SpectraMax i3x, Molecular Devices, Sunnyvale, CA). Intracellular GSH levels were quantified by interpolation on standard curves and relative GSH levels were calculated by normalization to the values obtained from A2780 cells.

#### Immunofluorescence Staining

Cells were plated in 20-mm culture plates, pretreated with the indicated concentrations of cisplatin for 24 h to observe γ-H2AX foci formation, washed with PBS three times, fixed with 4% paraformaldehyde for 15 min, and permeabilized in 0.1% Triton X-100 for 5 min. After blocking with 5% bovine serum albumin for 1 h at room temperature, the cells were incubated with primary antibodies against γ-H2AX (dilution 1:200), SIRT5 (dilution 1:200), or Nrf2 (dilution 1:200) overnight at 4◦C. Then, TRITC-conjugated secondary antibody (dilution 1:1,000) was incubated with the cells for 2 h in the dark at room temperature, and the cells were stained with 4′ ,6-diamidino-2 phenylindole (DAPI) for 5 min to visualize their nuclei. Images were captured using a fluorescence microscope or an Olympus FV1000 confocal laser-scanning microscope (Olympus, Tokyo, Japan). For quantification of γ-H2AX foci, 5 random fields of cells from each slide were quantified by ImageJ software and foci containing ≥5 cells were considered positive.

#### Western Blotting and Isolation of Cytosolic and Nuclear Cell Fractions

Total protein was isolated from SIRT5 overexpressing or knockdown ovarian cancer cells and their corresponding controls, with or without cisplatin treatment. The cells were washed with ice-cold PBS three times and lysed in lysis buffer supplemented with a cocktail of proteinase inhibitors (MCE). Equal amounts of protein (60 µg) from cell extracts were separated by 10% SDS-PAGE and transferred to 0.45 or 0.22-µm (the latter for γ-H2AX) polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA, USA). Following blocking with 5% fat-free milk for 2 h at room temperature, the membranes were incubated with primary antibodies (anti-SIRT5, anti-actin, anti-Nrf2, anti-HO-1, anti-MnSOD/SOD2, anti-BRCA1, anti-histone H3, or anti-γ-H2AX; dilution, 1:1,000) in blocking buffer overnight at 4◦C. Then, the membranes were washed three times in Tris-buffered saline with Tween 20 and incubated with secondary antibodies at a dilution of 1:2,500 for 2 h at 37◦C. Immunoreactivity was detected using ECL (Thermo Fisher Scientific, Waltham, MA, USA) with a BioImaging System (UVP Inc., Upland, CA, USA). ImageJ software was used to evaluate the gray value of each band.

For cytosolic and nuclear isolation, a Nuclear and Cytoplasmic Protein Extraction kit (Beyotime) was used, following the manufacturer's instructions. Briefly, the collected cells were suspended in ice-cold hypotonic buffer and incubated on ice for 20 min. The extracts were then centrifuged at 12,000 g for 5 min, and the supernatants were collected as cytosolic fractions. The pellets were washed with ice-cold PBS and resuspended in lysis buffer, followed by vortexing at the highest speed. These extracts were centrifuged at 12,000 g for 10 min, and the supernatants were collected as nuclear fractions.

#### Gene Expression Profiling Interactive Analysis (GEPIA) Database and Statistical Analysis

The GEPIA database (http://gepia.cancer-pku.cn/), a newly developed web-based tool, provides tumor vs. normal differential expression analysis, correlation analysis based on the Cancer Genome Atlas, and genotype-tissue expression data. It was used to analyze the expression of SIRT5, Nrf2, and HO-1 in ovarian cancer and normal tissues and the correlation between SIRT5 and Nrf2, MnSOD, and BRCA1. All experiments were repeated at least three times and the data were expressed as the means ± standard deviations. Statistical analysis was performed

FIGURE 2 | SIRT5 expression is involved in cisplatin-resistant of ovarian cancer cells. (A) SIRT5 expression in three ovarian cancer cell lines analyzed by immunofluorescence staining using TRITC-labeled antibodies; nuclei were stained with DAPI. (B) Differences in cisplatin sensitivity of three ovarian cancer cell lines assessed by CCK-8 assay. (C) Calculated IC50 values after cells were exposed to the indicated dose of cisplatin for 48 h. (D) SIRT5 protein levels in three ovarian cancer cell lines assessed by western blot and (E) relative protein expression was quantified by relative gray value of bands with ImageJ software. (F,G) SIRT5 protein levels were upregulated during exposed to indicated dose of cisplatin in three ovarian cancer cell lines for 48 h. DDP, cisplatin. Data are presented as the mean ± SD of three independent experiments. Scale bar = 20 µm, \**P* < 0.05, \*\**P* < 0.01, compared with A2780 or control cells.

FIGURE 3 | treated with indicated dose of cisplatin, the relative cell numbers were evaluated with CCK-8 assay at the indicated time points. (D) A significant increase in colony number was observed in A2780 cells with SIRT5 overexpression and an opposite trend was observed in SKOV-3 and CAOV-3 cells with knockdown of SIRT5. (E) Colony numbers were counted under microscope and the colony formation efficiency was calculated. (F) IC50 of cisplatin was increased in A2780 cells with SIRT5 overexpression and decreased in SKOV-3 and CAOV-3 cells with knockdown of SIRT5. (G) The transfected cells were treated with IC50 dose of cisplatin for consecutive 5 days respectively. Cell viability was assessed by CCK-8 assay. Data are presented as the mean ± SD of three independent experiments. EV, empty vector. NC, negative control. DDP, cisplatin. \**P* < 0.05, \*\**P* < 0.01, compared with the EV or NC cells.

with GraphPad Prism 7.0 software. Statistical significance was determined based on a Student's t-test or one-way ANOVA. The χ 2 test was used to determine the correlation between SIRT5 expression and clinicopathologic characteristics. P < 0.05 was considered to denote a statistically significant difference.

#### RESULTS

# SIRT5 Expression Is Increased in Ovarian Cancer Tissues and High SIRT5 Levels Predict Poor Chemotherapy Response

First, the GEPIA database was used to compare the mRNA expression of seven Sirtuin members between ovarian cancer and normal tissues. Only the mRNA level of SIRT5 was higher, while that of other isoforms was lower, in tumors than it was in normal tissues (**Figure 1A**). Then, immunohistochemistry was performed to verify this result. SIRT5 was more highly expressed in ovarian cancer tissues than in normal tissues and was mainly localized to the cytoplasm (**Figures 1B,C**). In addition, as shown in **Table 1**, higher SIRT5 levels were positively correlated with advanced International Federation of Gynecology and Obstetrics (FIGO) stage (P < 0.0001) and lymph node metastasis (P = 0.0019), but negatively correlated with a low grade of differentiation (P = 0.0179). The prognostic value of SIRT5 was then determined by Kaplan–Meier analysis using the KM plotter online software (http://kmplot.com/analysis/) based on 614 ovarian cancer patients who received chemotherapy. To investigate whether SIRT5 is relevant to chemoresistance, progression-free survival (PFS) was chosen as the primary endpoint, and patients with high SIRT5 expression exhibited significantly shorter PFS than those with low expression (P = 0.0018; **Figure 1D**). In summary, SIRT5 expression is increased in ovarian cancer tissues and high SIRT5 levels predict poor chemotherapy response.

#### SIRT5 Expression Is Increased in Cisplatin-Resistant Ovarian Cancer Cells and Cisplatin Upregulates SIRT5 Levels

Based on the aforementioned results, the expression pattern of SIRT5 in ovarian cancer cells was assessed. First, immunofluorescence staining indicated that this marker was localized mainly to the cytoplasm of A2780, SKOV-3, and CAOV-3 cells (**Figure 2A**), in agreement with the immunohistochemistry staining of tissues. Western blotting of cytosolic and nuclear fractions verified this result (**Figure S1A**). Then, the sensitivities to cisplatin among the three ovarian cancer cell lines were compared. The cells were treated with different concentrations of cisplatin (0–80µg/ml) for 24 h and different degrees of sensitivity to cisplatin were noted. SKOV-3 (IC<sup>50</sup> = 39.743 ± 4.756µg/ml) and CAOV-3 (IC<sup>50</sup> = 80.813 ± 7.058µg/ml) cell lines were less sensitive to cisplatin than A2780 cells (IC<sup>50</sup> = 9.362 ± 0.489µg/ml; **Figures 2B,C**). These results are consistent with previous descriptions of these three ovarian cancer cell lines (23–25).

SIRT5 expression was higher in SKOV-3 and CAOV-3 cisplatin-resistant cells than in A2780 cisplatin-sensitive cells (**Figures 2D,E**). Then, whether cisplatin could affect the protein levels of SIRT5 was investigated. The three ovarian cancer cell lines were incubated without cisplatin, with 50% of the IC<sup>50</sup> dose, and with the IC<sup>50</sup> dose of cisplatin for 24 h. SIRT5 expression was significantly upregulated in the cisplatin-treated cells in a concentration-dependent manner compared to control cells (**Figures 2F,G**). Taken together, these results reveal a causal relationship between SIRT5 expression and cisplatin resistance in ovarian cancer cells.

#### SIRT5 Promotes Cell Proliferation and Cisplatin Resistance in Ovarian Cancer Cells

As SIRT5 was hypothesized to play a role in ovarian cancer development and chemoresistance, its effect on the biological behaviors of the three cell lines was investigated. Based on its basal protein levels, SIRT5 was either overexpressed or silenced. A2780 cells were transfected with a SIRT5 overexpression plasmid or empty vector, while knockdown of SIRT5 in SKOV-3 and CAOV-3 cells was achieved by transfecting them with SIRT5 siRNA or negative control siRNA (**Figures 3A,B**). The results indicated that SIRT5 overexpression could significantly promote A2780 cell proliferation and colony formation, while the opposite effect was observed in SKOV-3 and CAOV-3 cells upon SIRT5 knockdown (**Figures 3C–E**).

Next, cell viability assays were performed to investigate the relationship between SIRT5 and cisplatin resistance. The transfected cell lines were treated with increasing concentrations of cisplatin for 24 h to observe the effect on cisplatin IC<sup>50</sup> values. Moreover, these transfected cells were treated with their respective IC<sup>50</sup> dose of cisplatin for 5 consecutive days. The results showed that the IC<sup>50</sup> values and cell viability were elevated after overexpression of SIRT5 in A2780 cells, whereas both values were decreased upon SIRT5 downregulation in SKOV-3 and CAOV-3 cells (**Figures 3F,G**). These findings suggest that SIRT5 promotes proliferation and cisplatin resistance in ovarian cancer cells.

cells assessed by western blot and (B) the relative gray values were shown in histogram. (C) The transfected cells were treated with or without IC50 dose of cisplatin for 24 h. The γ-H2AX protein levels were assessed by western blot and (D) the relative gray values were shown in histogram. (E) The γ-H2AX foci formation was *(Continued)* FIGURE 4 | observed by immunofluorescence staining. The mock cells were untreated and the transfected cells were treated with IC50 dose of cisplatin for 24 h respectively. (F) The quantification results of γ-H2AX foci in three ovarian cancer cells. Data are presented as the mean ± SD of three independent experiments. EV, empty vector. NC, negative control. DDP, cisplatin. Scale bar = 10µm, \**P* < 0.05, \*\**P* < 0.01.

### SIRT5 Suppresses Cisplatin-Induced DNA Damage

As mentioned above, SIRT5 was found to promote resistance to cisplatin in ovarian cancer cells, raising the question of how SIRT5 participates in the induction of chemoresistance. First, we confirmed that cisplatin can lead to the accumulation of γ-H2AX protein, which is a marker of DNA double-strand breaks, in a concentration-dependent manner (**Figures 4A,B**). Moreover, cisplatin-induced DNA damage could be suppressed by the overexpression of SIRT5 in A2780 cells, whereas γ-H2AX protein levels were increased upon SIRT5 downregulation in SKOV-3 and CAOV-3 cells (**Figures 4C,D**). In addition, immunofluorescence validated that SIRT5 overexpression could inhibit the formation of γ-H2AX foci, whereas this suppressive effect was abrogated upon SIRT5 knockdown (**Figures 4E,F**). Therefore, SIRT5 was shown to promote cisplatin resistance in ovarian cancer cells by inhibiting DNA damage.

#### SIRT5 Eliminates Cisplatin-Induced ROS to Reduce DNA Damage

Next, the mechanism by which SIRT5 suppresses cisplatininduced DNA damage was investigated. As SIRT5 has been reported to regulate ROS (26) and excessive ROS induced by cisplatin can lead to DNA damage (27–29), we hypothesized that SIRT5 suppresses cisplatin-induced DNA damage by eliminating ROS. First, we confirmed that the levels of intracellular ROS induced by cisplatin were increased in both a concentrationand time-dependent manner. The levels of ROS peaked after 24 h of exposure to cisplatin and decreased by 48 h in all three ovarian cancer cell lines (**Figures 5A,B**). Therefore, we observed ROS levels in these cell lines at 24 h after cisplatin treatment for the subsequent experiments. Cisplatin-induced ROS levels were significantly inhibited upon overexpression of SIRT5 in A2780 cells, whereas the levels were increased after SIRT5 was silenced in SKOV-3 and CAOV-3 cells (**Figures 5C,D**). As expected, the inhibition of cisplatin-induced DNA damage by SIRT5 was reversed by the administration of NAC, a ROS scavenger (**Figures 5E–G**). These results indicate that SIRT5 suppresses cisplatin-induced DNA damage in a ROS-dependent manner in ovarian cancer cells.

In addition, using the GEPIA database, SIRT5 expression was found to have a positive relationship with BRCA1 expression, which is a well-known DNA damage repair gene (P = 6.6e-7, **Figure S1B**). In the present study, the levels of γ-H2AX were suppressed when BRCA1 was upregulated upon overexpression of SIRT5, and they were increased after downregulation of BRCA1 upon knockdown of SIRT5 (**Figure 5H**). These results suggest that SIRT5 also suppresses γ-H2AX expression by positively regulating BRCA1 expression, but the specific mechanism needs to be further explored.

# SIRT5 Inhibits ROS by Positively Regulating the Nrf2/HO-1 Pathway

As SIRT5 is closely related to the mitochondria, the expression of MnSOD/SOD2, which is a well-known antioxidant enzyme, was measured initially. However, the level of this enzyme was not significantly changed after upregulation or downregulation of SIRT5 and the expression of SOD2 mRNA had no significant relationship with SIRT5 mRNA expression in ovarian cancer based on the GEPIA database (**Figures S1C,D**). Next, whether SIRT5 could inhibit ROS by regulating the Nrf2/HO-1 pathway, which is also known to eliminate ROS (30), was investigated. The mRNA expression levels of Nrf2 and HO-1 were lower in ovarian cancer than in normal tissues based on the GEPIA database (**Figure 6A**). Interestingly, the protein levels of Nrf2 and HO-1 were higher in SKOV-3 and CAOV-3 cells than in A2780 cells (**Figure 6B**). Moreover, a significant, positive correlation was identified between Nrf2 and SIRT5 mRNA expression in ovarian cancer based on the GEPIA database (P = 1.5e-10, R = 0.3) (**Figure 6C**). Further, Nrf2 and HO-1 proteins were upregulated upon SIRT5 overexpression in A2780 cells and were downregulated upon SIRT5 silencing in SKOV-3 and CAOV-3 cells (**Figure 6D**). In addition, overexpression of SIRT5 facilitated the nuclear translocation of Nrf2 by immunofluorescence staining, which was corroborated by western blotting of cytosolic and nuclear fractions (**Figures 6E,F**). These observations suggest that SIRT5 can enhance the expression of Nrf2 and its target gene HO-1 in ovarian cancer.

To provide further support, ML385, a specific inhibitor of Nrf2, was utilized to treat A2780 cells for 24 or 48 h. As shown in **Figure 6G**, Nrf2 was significantly suppressed at 48 h. ROS levels in A2780 cells overexpressing SIRT5 were then measured after pretreating the cells with or without ML385 for 48 h and cisplatin for 24 h. The results show that the inhibition of ROS by SIRT5 was reversed upon ML385 treatment (**Figures 6H,I**). Taken together, SIRT5 can inhibit ROS by positively regulating the Nrf2/HO-1 pathway.

#### DISCUSSION

In this study, among the Sirtuin family, SIRT5 was highly expressed in ovarian cancer compared to its expression in normal tissues, based on the GEPIA database and this result was verified by immunohistochemistry. In addition, high levels of SIRT5 predicted shorter PFS and were positively associated with clinicopathologic characteristics of ovarian cancer, such as advanced FIGO stage and lymph node metastasis, but were negatively correlated with differentiation. These results reveal a potential prognostic role for SIRT5 in ovarian cancer patients. Consistent with our results, SIRT5 was shown to be overexpressed in human NSCLC (13), triple-negative breast cancer, breast cancer with BRCA1 mutation subtypes (17), CRC

FIGURE 5 | cytometry and 0 h served as a control group. (C) Overexpression of SIRT5 inhibited ROS production and knockdown of SIRT5 attenuated the inhibition effect. (D) Representative flow cytometry results analyzed by FlowJo software. (E,F) NAC, a ROS scavenger, reversed the inhibition of cisplatin induced DNA damage in A2780 cells with overexpression SIRT5 by assessing γ-H2AX foci formation and (G) γ-H2AX protein levels. Cells were pretreated with 5 mM NAC for 2 h to inhibit ROS generation and then exposure to cisplatin for 24 h. (H) The levels of γ-H2AX was suppressed when BRCA1 was upregulated upon overexpression of SIRT5, and increased after downregulated of BRCA1 upon knockdown of SIRT5. Data are presented as the mean ± SD of three independent experiments. EV, empty vector; NC, negative control; DDP, cisplatin; NS, not significant. Scale bar = 10 µm, \**P* < 0.05, \*\**P* <0.01, \*\*\**P* <0.001.

(15), and HCC (14). Further statistical analysis showed that higher SIRT5 was significantly associated with malignant tumor characteristics such as larger tumor size, lymph node metastasis, advanced TNM stage, and poor survival. Besides, a shorter time to post-therapeutic recurrence in wild-type Kras CRC patients was found to correlate with high expression of SIRT5 (21). However, upregulated SIRT5 in liver cancer tissues was found to be associated with favorable prognosis (19). Regarding mRNA or protein levels of SIRT5, Janus-faced expression has been identified. For example, SIRT5 was overexpressed in the aforementioned tumors and B cell malignancies (16), but was decreased in endometrial carcinoma (20) and head and neck squamous cell carcinoma (18).To summarize, SIRT5 might act as either a tumor promoter or suppressor, in a contextspecific manner.

Although SIRT5 is predominantly a mitochondrial matrix protein, possessing a well-defined mitochondrial localization sequence (31), it was localized to both the cytoplasm and nucleus in three ovarian cancer cell lines, but the majority of SIRT5 was found in the cytoplasm in our study, which is consistent with other reports (32–34). Interestingly, nuclear and cytosolic SIRT5 in cerebellar granule neurons exerted a protective effect for cells, whereas mitochondrial SIRT5 promoted neuronal death (34). However, the mechanism by which SIRT5 localization contributes to these effects remains unknown.

Additionally, we revealed that SIRT5 expression was higher in SKOV-3 and CAOV-3 cells, which were shown to be resistant to cisplatin in our study, than in A2780 cells, which were confirmed to be sensitive to cisplatin (23–25). Moreover, SIRT5 expression was upregulated by cisplatin in a concentrationdependent manner. Similarly, the significant enrichment of endogenous SIRT5 protein after exposure to chemotherapeutic agents or cetuximab was confirmed previously in two wildtype Kras CRC cell lines (21). Consequently, we speculated that SIRT5 is involved in the progression and chemoresistance of ovarian cancer. As expected, the overexpression of SIRT5 significantly promoted cell proliferation and cisplatin resistance in vitro, while an inhibitory effect was observed upon SIRT5 downregulation. In agreement with our results, SIRT5 was found to drive HEK293 cancer cell proliferation via desuccinylation and activation of the serine hydroxymethyltransferase SHMT2 (35) and promote cell proliferation in HCC by targeting E2F transcription factor 1 (14). In contrast, SIRT5 was found to inhibit gastric cancer cell proliferation and tumor formation by inhibiting aerobic glycolysis (36). Interestingly, SIRT5 is not necessary for BrafV600E-mediated cutaneous melanoma initiation and growth in vivo (37). With respect to its function in chemoresistance, SIRT5 knockdown sensitizes NSCLC A549 cells to multiple chemotherapeutics including cisplatin (13). Moreover, SIRT5 was determined to have a partial inhibitory effect on the tumor suppressor SUN2, which was found to increase the sensitivity of lung cancer to cisplatin by inducing apoptosis (12). In wild-type Kras CRCs, SIRT5-positive cells were also shown to be resistant to either chemotherapeutic agents, such as 5-FU and oxaliplatin, or cetuximab (21). In contrast, an analysis of data from the GEO DataSet revealed high levels of SIRT5 in triple-negative breast cancer patients who showed complete response to neoadjuvant chemotherapy. Analysis of the ONCOMINE database also suggested that SIRT5 expression levels were higher in chemotherapy responders than in nonresponders (17). In conclusion, further study is needed to explore the role of SIRT5 in tumorigenesis and chemoresistance.

Cisplatin, one of the most widely applied chemotherapeutic agents for multiple solid tumors including ovarian cancer, was initially described as a DNA intrastrand cross-linker that interacts with DNA to form DNA adducts, resulting in the activation of several signal transduction pathways including those involved in DNA damage repair and apoptosis (38). Recent reports have suggested that a novel cytotoxic effect for most genotoxic drugs including cisplatin is to promote ROS-dependent apoptosis (39, 40) or DNA damage (27–29). Therefore, the activation of a ROS-scavenging mechanism in cancer cells confers resistance to chemotherapy (21). In this study, we found that SIRT5 potently inhibits cisplatin-induced DNA damage, as indicated by γ-H2AX protein levels in western blotting and foci formation in immunofluorescence staining. Similarly, after SIRT5 knockdown, the levels of γ-H2AX were previously shown to be significantly upregulated in both CRCs and HCC (15, 19). Then, we confirmed that cisplatin can induce ROS production in a concentration- and time-dependent manner, and that ROS levels peaked after exposure to an IC<sup>50</sup> treatment of cisplatin for 24 h. ROS levels induced by cisplatin were significantly reduced upon SIRT5 overexpression but were increased after SIRT5 knockdown. Furthermore, the inhibitory effect of SIRT5 on DNA damage was attenuated when NAC, a ROS scavenger, was applied. These results verified our hypothesis that SIRT5 promotes cisplatin resistance in ovarian cancer by suppressing cisplatin-induced DNA damage in a ROS-dependent manner. In agreement with our results, SIRT5 was found to demalonylate and inactivate succinate dehydrogenase complex subunit A (SDHA) in CRCs, resulting in activation of the ROS-scavenging enzyme, thioredoxin reductase 2 (TrxR2) and finally leading to chemotherapy resistance (21). Likewise, other mechanisms underlying SIRT5-mediated ROS detoxification have been reported. For instance, SIRT5 was identified as a safeguard against oxidative stress-induced apoptosis in cardiomyocytes and neuroblastoma cells (6, 7). In NSCLC, SIRT5 was found to desuccinylate and activate Cu/Zn superoxide

FIGURE 6 | upregulated Nrf2 and HO-1 protein levels and knockdown of SIRT5 reduced their expression assessed by western blot and the relative gray values were shown in histogram. (E,F) Overexpression of SIRT5 facilitated the nuclear translocation of Nrf2 by immunofluorescence staining and western blot after nuclear and cytoplasm isolation. (G) ML385, a specific inhibitor of Nrf2, was added into A2780 cells with terminal concentration of 5µM for 24 or 48 h and the inhibition efficiency was measured by western blot. (H) The inhibition effect of ROS by overexpression SIRT5 in A2780 cells was reversed when ML385 was applied. (I) Representative flow cytometry results analyzed by FlowJo software. Data are presented as the mean ± SD of three independent experiments. EV, empty vector; NC, negative control; DDP, cisplatin; TPM, transcripts per million. Scale bar = 20µm, \**P* < 0.05, \*\**P*<0.01, \*\*\**P*<0.001, compared with A2780, the EV or NC cells.

dismutase (SOD1) to eliminate ROS when the proteins were co-expressed (41). Moreover, the desuccinylation of isocitrate dehydrogenase 2 (IDH2) or pyruvate kinase M2 (PKM2) and the deglutarylation of glucose-6-phosphate dehydrogenase (G6PD) by SIRT5 leads to the production of sufficient NADPH, a major intracellular reductant, to attenuate cellular ROS levels (42, 43). Interestingly, the expression of MnSOD/SOD2, a known antioxidant enzyme, was not significantly changed after upregulation or downregulation of SIRT5 in our study. Collectively, these studies highlight the fact that SIRT5 promotes ROS detoxification via the post-translational modification of multiple, vital antioxidant enzymes.

Another mechanism of cisplatin resistance is the enhancement of DNA damage repair. BRCA1, which is a key gene in the DNA damage repair pathway, was identified as having a positive correlation with SIRT5 expression in ovarian cancer, based on the GEPIA database. Further, the levels of γ-H2AX were suppressed when BRCA1 was upregulated upon overexpression of SIRT5, and they were increased when BRCA1 was downregulated upon knockdown of SIRT5. These results suggested that SIRT5 also contributes to cisplatin resistance by positively regulating BRCA1 expression, but the exact mechanism needs to be explored further.

In addition to the abovementioned enzymes, transcription factors that promote the expression of antioxidant defense genes, such as forkhead box O3 (FOXO3), could be deacetylated at critical lysine residues by SIRT5, promoting their nuclear localization and leading to decreased ROS levels (44). In our study, another transcription factor, Nrf2, which promotes the expression of antioxidant and detoxifying genes, and HO-1, a key target gene of Nrf2, were overexpressed in cisplatinresistant SKOV-3 and CAOV-3 cells, in agreement with reports of high levels of Nrf2 and HO-1 in drug-resistant tumor cells (45–47). Interestingly, the mRNA expressions of Nrf2 and HO-1 were lower in ovarian cancer than in normal tissues based on the GEPIA database, in contrast to reports of their upregulated protein levels in tumors such as ovarian cancer (48, 49). Moreover, in our study, the Nrf2/HO-1 pathway was found to be positively regulated by SIRT5 and overexpression of SIRT5 facilitated Nrf2 nuclear translocation, consistent with a report that the mRNA levels of Nrf2 and its downstream target genes are reduced upon SIRT5 knockdown in NSCLC (13). In addition, in the present study, the downregulation of ROS by SIRT5 was reversed when a specific Nrf2 inhibitor, ML385, was utilized. These results provide evidence that SIRT5 inhibits ROS production via positive regulation of the Nrf2/HO-1 pathway. GSH, a known ROS scavenger, was identified previously as mediating resistance to cisplatin (50–53). It was reported that Nrf2 is involved in the regulation of GSH abundance and that Nrf2 activation can result in high GSH dependency of the affected cells (54). Consistent with these reports, in our study, higher levels of GSH were found in SKOV-3 and CAOV-3 cisplatinresistant cell lines relative to the levels in A2780 sensitive cells (**Figure S1E**). To summarize, SIRT5 may also function as a ROS inhibitor by activating Nrf2, leading to increased GSH levels, but this hypothesis should be confirmed by future experiments.

The limitation of our study is that the exact mechanism by which SIRT5 functions as a mitochondrial enzyme to regulate the Nrf2/HO-1 pathway was not clarified. However, one possibility is that SIRT5 inhibits autophagic flux via the deacetylation of lactate dehydrogenase B (LDHB) (10), leading to an increase in p62, which competes with Nrf2 for Kelch-like ECH-associated protein 1 (KEAP1) binding (55). This results in Nrf2 dissociation from KEAP1 and prolonged activation of Nrf2 (56, 57). Another possibility is that SIRT5 competes with SIRT2 to interact with Nrf2, which blocks the deacetylation of Nrf2 by SIRT2, leading to an increase in nuclear Nrf2 levels (58, 59). In addition, it was also reported that BRCA1 could interact with Nrf2 and promote its stability and activation (60–62), we hypothesize that SIRT5 actives Nrf2 pathway by positively regulating BRCA1. However, these hypotheses need to be tested further.

In conclusion, our study implies that SIRT5 expression is increased in ovarian cancer tissues and its high level predicts a poor chemotherapy response. We also reveal a potential function for SIRT5 in ovarian cancer proliferation and chemoresistance. Specifically, SIRT5 contributes to cisplatin resistance by suppressing cisplatin-induced DNA damage in a ROS-dependent manner via the regulation of Nrf2/HO-1 signaling (**Figure 7**). Therefore, SIRT5 might serve as a prognostic factor for ovarian cancer and SIRT5 inhibitors combined with chemotherapy could represent a novel therapeutic strategy for ovarian cancer patients.

# DATA AVAILABILITY

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher. Publicly available datasets were also analyzed in this study. This data can be found here: http://gepia. cancer-pku.cn/.

# AUTHOR CONTRIBUTIONS

QL, LF, CZ, SW, and XS contributed conception and design of the study. XS, SW, JGa, JGu, JL, YL, and JZ performed experiment and the statistical analysis. XS and SW wrote the first draft of the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.

# FUNDING

This work was supported by the National Natural Science Foundation of China (grant numbers 81672964, 81874214, and 81702269).

# ACKNOWLEDGMENTS

We would like to thank Editage [www.editage.cn] for English language editing and Mr. Yifu Song for technological support. XS would like to thank her families for helping her take care of her little daughter.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2019.00754/full#supplementary-material

Figure S1 | (A) The localization of SIRT5 in three ovarian cancer cell lines was validated by western blot after nuclear and cytoplasm isolation. (B) *SIRT5* had a positive relationship with *BRCA1* based on GEPIA database. (C) MnSOD/SOD2 protein levels were not significantly changed after upregulation or downregulation of SIRT5 levels. (D) *SIRT5* had no significant relationship with *MnSOD/SOD2* based on GEPIA database. (E) The relative levels of glutathione (GSH) in three ovarian cancer cells. TPM, transcripts per million. <sup>∗</sup>*P* < 0.05.

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Sun, Wang, Gai, Guan, Li, Li, Zhao, Zhao, Fu and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# MTA3-SOX2 Module Regulates Cancer Stemness and Contributes to Clinical Outcomes of Tongue Carcinoma

Zhimeng Yao1†, Liang Du1,2†, Min Xu<sup>3</sup> , Kai Li <sup>1</sup> , Haipeng Guo<sup>3</sup> , Guodong Ye<sup>4</sup> , Dianzheng Zhang<sup>5</sup> , Robert P. Coppes <sup>2</sup> and Hao Zhang4,6 \*

*<sup>1</sup> Cancer Research Center, Shantou University Medical College, Shantou, China, <sup>2</sup> Department of Biomedical Sciences of Cells and Systems, Section Molecular Cell Biology and Radiation Oncology, University Medical Center Groningen, University of Groningen, Groningen, Netherlands, <sup>3</sup> Department of Head and Neck Surgery, Cancer Hospital of Shantou University Medical College, Shantou, China, <sup>4</sup> Institute of Precision Cancer Medicine and Pathology, Jinan University Medical College, Guangzhou, China, <sup>5</sup> Department of Bio-Medical Sciences, Philadelphia College of Osteopathic Medicine, Philadelphia, PA, United States, <sup>6</sup> Research Centre of Translational Medicine, The Second Affiliated Hospital of Shantou University Medical College, Shantou, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Anca Maria Cimpean, Victor Babes University of Medicine and Pharmacy, Romania Jun Yan, Nanjing University, China*

> \*Correspondence: *Hao Zhang haozhang@jnu.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

> Received: *04 June 2019* Accepted: *09 August 2019* Published: *27 August 2019*

#### Citation:

*Yao Z, Du L, Xu M, Li K, Guo H, Ye G, Zhang D, Coppes RP and Zhang H (2019) MTA3-SOX2 Module Regulates Cancer Stemness and Contributes to Clinical Outcomes of Tongue Carcinoma. Front. Oncol. 9:816. doi: 10.3389/fonc.2019.00816* Cancer cell plasticity plays critical roles in both tumorigenesis and tumor progression. Metastasis-associated protein 3 (MTA3), a component of the nucleosome remodeling and histone deacetylase (NuRD) complex and multi-effect coregulator, can serve as a tumor suppressor in many cancer types. However, the role of MTA3 in tongue squamous cell cancer (TSCC) remains unclear although it is the most prevalent head and neck cancer and often with poor prognosis. By analyzing both published datasets and clinical specimens, we found that the level of MTA3 was lower in TSCC compared to normal tongue tissues. Data from gene set enrichment analysis (GSEA) also indicated that MTA3 was inversely correlated with cancer stemness. In addition, the levels of MTA3 in both samples from TSCC patients and TSCC cell lines were negatively correlated with SOX2, a key regulator of the plasticity of cancer stem cells (CSCs). We also found that SOX2 played an indispensable role in MTA3-mediated CSC repression. Using the mouse model mimicking human TSCC we demonstrated that the levels of MTA3 and SOX2 decreased and increased, respectively, during the process of tumorigenesis and progression. Finally, we showed that the patients in the MTA3low/SOX2high group had the worst prognosis suggesting that MTA3low/SOX2high can serve as an independent prognostic factor for TSCC patients. Altogether, our data suggest that MTA3 is capable of repressing TSCC CSC properties and tumor growth through downregulating SOX2 and MTA3low/SOX2high might be a potential prognostic factor for TSCC patients.

Keywords: tongue squamous cell carcinoma, MTA3, SOX2, cancer stem cell, prognosis, proliferation, progression

# INTRODUCTION

More than 500,000 new cases of oral and pharyngeal cancers are diagnosed yearly worldwide (World Cancer report 2014, https://www.who.int/cancer/publications/WRC\_2014/en/). Oral cancer is malignant neoplasia which arises on the lip or oral cavity. Although progress has been made in cancer treatments, the oral cancer survival rate has not been improved significantly

**25**

for decades (1, 2). Since ∼90% of these cancers are histologically shown to be originated from squamous cells (3), this subtype of cancers is traditionally defined as oral squamous cell carcinoma (OSCC). OSCC may show different levels of differentiation and has a propensity to coincide with lymph node metastasis (4). Among OSCC, tongue squamous cell cancer (TSCC) has the highest incidence and is usually associated with a poor survival rate. Therefore, TSCC is one of the most lethal types of cancers in the head and neck region (5). Thus, a better understanding of the underlying mechanisms in TSCC development will provide not only a more reliable biomarkers for diagnosis and prognosis but also potential therapeutic targets for the treatment of this cancer.

Cellular plasticity plays critical roles in tumor initiation, progression, and metastasis. It is now well-established that stem cell-like cancer cells or cancer stem cells (CSCs) are responsible for both cell plasticity and treatment (6–10). CSCs are a small subset of cancer cells and multiple lines of evidence indicate that CSCs are responsible for tumor initiation, indefinite, and progression (7, 11, 12). Accumulating data also indicate that plasticity of CSCs closely correlates with recurrences and metastasis (13–16) with poor prognosis in a wide variety of cancers, including tongue cancer (17–20).

Metastasis-associated protein 3 (MTA3) is a multi-effect coregulatory factor and plays indispensable roles in cell proliferation, tumorigenesis, and metastasis (21–26). Compelling evidence suggests that MTA3 is a tumor suppressor in many cancer types (26–28) by serving as an integral subunit of the nucleosome remodeling and histone deacetylase (NuRD) complex (21, 25, 27). As a transcriptional corepressor (29), MTA3 either directly or indirectly regulates the expression and activity of EMT-associated genes such as Snail and Ecadherin (25, 27). Dysregulation of MTA3 has been observed in many different human tumors (26–28). Reduced levels of MTA3 lead to the upregulation of Snail and subsequently enhance the process of epithelial-mesenchymal transition (EMT) (25, 27, 30). Consistently, the dysfunctional MTA3 reduces cell-cell adhesion and promotes cancer invasion and metastasis (23, 26). Moreover, reduced expression of MTA3 in tumor specimens has been associated with poor survival and therefore the expression of MTA3 has been suggested as an independent predictor of patient prognosis in uterine non-endometrioid carcinomas, gastroesophageal junction adenocarcinoma, glioma, and colorectal cancer (27, 28, 31, 32). However, the role of MTA3 and the underlying mechanism of MTA3's function in TSCC remain largely unknown.

In this study, we found that reduced levels of MTA3 in the patient specimens correlated with poorer clinical outcomes with concurrently increased cancer stemness. We also showed that MTA3 was capable of repressing cancer cell proliferation through inhibiting SOX2 expression. Using a chemical-induced mouse model of TSCC, we demonstrated that MTA3 and SOX2 decreased and increased, respectively, during the process of carcinogenesis and progression. Finally, our findings suggested that MTA3low/SOX2high could potentially serve as an independent prognostic factor for TSCC patients.

# MATERIALS AND METHODS

#### Patient Tissue Samples

A total of 119 patients with TSCC were recruited at the Affiliated Tumor Hospital of Shantou University Medical College from 2009 to 2011 and their TSCC were clinically diagnosed and histologically confirmed. The primary TSCC specimens and their matched non-cancerous tissues were paraffin-embedded. Samples from patients who underwent preoperative radiotherapy or chemotherapy for TSCC were excluded. Clinical research protocols of this study were reviewed and approved by the Ethics Committee of Shantou University Medical College.

### Immunohistochemistry (IHC)

Tissue sections (4µm) from the formalin-fixed paraffinembedded clinical specimens or 4NQO-induced tongue tumor tissues were processed and immune-stained with antibodies against MTA3 (Catalog No. A300-160A, Bethyl, 1: 600), SOX2 (Catalog No. 23064, Cell Signaling, 1: 200), each with at least two cores of the primary tumor as well as two cores of normal tongue tissue. Sections immune-labeled with rabbit IgG or mouse IgG as the primary antibody were used as negative controls, known MTA3 and SOX2 positive slides were used as a positive control.

# IHC Evaluation

The percentage of positively stained cells were scored using the following scales: 0, no staining in any field; 1, ≤ 10; 2, 11–50; 3, 51–75; 4, > 75%. The intensity of staining was scored using the following scales: 1+, weak staining; 2+, moderate staining; 3+, strong staining. Percentage (P) and intensity (I) of nuclear, cytoplasm or membrane expression were multiplied to generate a numerical score (S = P • I).

The tissue sections were scored by two pathologists blind to the clinical outcomes. Receiver operating characteristic (ROC) curves were employed to define an optical cut-off score, which was closest to the point with maximum sensitivity and specificity. The cases with scores lower than or equal to the cut-off value were designated as low expression group and those with higher scores were categorized as high expression group.

# Histological Analysis

For histological analysis, tissues were fixed in 4% neutral buffer formalin, embedded in paraffin, sectioned (4µm) and stained with hematoxylin and eosin.

#### Gene Set Enrichment Analyses

Microarray data (accession no. GSE78060) were obtained from the Gene Expression Omnibus of NCBI (http://www.ncbi.nlm. nih.gov/geo/) and subjected to Gene set enrichment analysis (GSEA) using GSEA software (version 2.0.13) (http://www. broadinstitute.org/gsea/index.jsp).

#### Immunofluorescence Staining

FFPE tissue sections were deparaffinized and dehydrated in xylene and graded ethanol solutions in preparation for MTA3 and SOX2 double immunofluorescence (IF) staining. All slides were subjected to heat-induced epitope retrieval in Citrate Buffer [pH = 6.0]. Endogenous tissue peroxidases were blocked by incubating the slides in 3% hydrogen peroxide solution and blocking buffer before incubation with MTA3 (Catalog A300-160A, Bethyl, 1: 600), SOX2 (Catalog No. #23064, Cell Signaling, 1: 200) as primary antibodies. And HRP-conjugated streptavidin as the secondary antibody. The signal in IF labeled slides were visualized with AlexaFluor 488 and AlexaFluor 594 Tyramide Super Boost kits (Invitrogen, Carlsbad, CA), and nuclei were visualized with Prolong Diamond Antifade Reagent with 4',6-diamidino-2-phenylindole (DAPI; Invitrogen, Carlsbad, CA). Primary and secondary antibodies were stripped using Citrate Buffer [pH = 6.0] in the microwave. Known MTA3 and SOX2 positive slides were used as a positive control. Immunofluorescence staining was analyzed using the PerkinElmer Vectra analysis platform to estimate the cell numbers. The percentage of positive cells was estimated by two pathologists.

#### Cell Culture

Human TSCC cell lines (SCC-25 and SCC-4) were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The cells were cultured in Dulbecco's Modified Eagle Media (DMEM, Gibco/Invitrogen) supplemented with 10% FBS (Gibco/Invitrogen) at 37◦C in a humidified atmosphere containing 5% CO2.

# Virus Production and Transduction

The full-length cDNA of MTA3 was PCR amplified from SCC-25 cells and cloned into the pCDNA3.1-flag plasmid. The shRNA targeting human MTA3 (target sequence: GAGGATACCTTCTTCTACTCA) was cloned into pBabe/U6 plasmid. The SOX2 overexpression plasmid and SOX2 short hairpin RNA (shSOX2) plasmid (shSOX2 target sequence: GGTTGACACCGTTGGTAATTT) were obtained from GeneCopoeia. Transfection of plasmid was performed using Lipofectamine 3000 (Thermo Fisher Scientific, catalog no. L3000015) according to the manufacturer's instructions. Stable cells were selected by culturing the cells in the medial with puromycin for 2 weeks.

#### RNA Isolation and Quantitative Real-Time PCR

Total RNA was extracted from cells using TRIzol (Invitrogen) according to the manufacturer's instruction and 2 µg RNA was reversely transcribed using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). The cDNA was amplified and quantified in ABI-7500 system (Applied Biosystems) using SYBR Green Master (Roche). The cDNA was subjected to quantitative real-time PCR (qPCR) with the following primers: MTA3 forward: 5′ -AAGCCTGGT GCTGTGAAT-3′ and reverse: 5′ -AGGGTCCTCTGTAGTTGG-3 ′ ; SOX2 forward: 5′ -CATCACCCACAGCAAATGACA-3′ and reverse: 5′ -GCTCCTACCGTACCACTAGAACTT-3′ ; GADPH forward: 5′ -TCCTCCTGTTTCATCCAAGC-3′ and reverse: 5′ - TAGTAGCCGGGCCCTACTTT-3′ .

# Western Blot Analysis

Whole-cell lysates were prepared by lysing the cells in lysis buffer. Cell lysates with an equal amount of proteins were separated on 10% SDS-PAGE and transferred to PVDF membranes. The membranes were incubated with primary antibodies (MTA3, Catalog No. A300-160A, Bethyl, 1: 2,000; SOX2, Catalog No.23064, Cell Signaling, 1: 1,000; GAPDH, Catalog No. ab8245, Abcam, 1:3,000) followed by HRP-conjugated secondary antibodies as previously described (33). Blotted proteins were visualized by incubating in SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) followed by exposure to X-ray film (Eastman Kodak, Rochester, NY, USA) (33).

# ALDEFLUOR Assay

ALDEFLUOR assay kit (Stem Cell TechnologiesTM, Vancouver, BC, Canada) was used to determine ALDH1 activity according to the manufacturer's protocol. Cells were suspended in ALDEFLUOR assay buffer containing 1µM per 1 × 10<sup>6</sup> cells of the ALDH substrate, boron-dipyrromethene-aminoacetaldehyde (BAAA), and incubated for 50 min at 37◦C. Each sample was treated with 50 mM of an ALDH-specific inhibitor, and diethylaminobenzaldehyde (DEAB) as a negative control. Stained cells were analyzed by BD FACSAriaTM II (BD Biosciences, San Jose, CA, USA). To evaluate cell viability the cells were stained with 1 mg/ml of propidium iodide prior to analysis.

# Proliferation and Survival Assays

Real-time cell analysis (RTCA) was performed to estimate cell proliferation using the xCELLigence DP device (ACEA Biosciences) as described in the supplier's instructions. In brief, 3,000 cells were seeded in E-plates, and the plates were locked into the RTCA DP device supplied with humidified air with 5% CO<sup>2</sup> at 37◦C. The proliferative ability was monitored by the xCELLigence RTCA Analyzer (Roche Applied Science, Mannheim, Germany) (34, 35).

# Animals and Carcinogen Treatment

Male and female wild-type C57BL/six mice were supplied by Beijing Vital Laboratory Animal Technology (Beijing, China). To induce tumorigenesis in the tongue, 4NQO (Sigma-Aldrich, St. Louise, MO) was added to the drinking water (100 mg/mL) for the 6-week-old young adult mice for 16 weeks. The mice were sacrificed when the bodyweight loss >1/3, otherwise they were sacrificed at the indicated time. After sacrifice, the tongue surface was photographed. The tissues were resected for histopathological examination and immunohistochemistry (IHC) analyses. Animals were housed in pathogen-free conditions at the Animal Center of Shantou University Medical College in compliance with Institutional Animal Care and Use Committee (IACUC) regulations (SUMC2014-148). All animal experiments were performed according to protocols approved by the Animal Care and Use Committee of the Medical College of Shantou University.

#### Gaussia Luciferase Assay

The pEZX-PG04 plasmid carrying double-expression cassette for Gaussia luciferase under the control of the SOX2 promoter (+270 to −1038), and secreted Alkaline Phosphatase (SeAP) under the control of the CMV promoter was obtained from GeneCopoeia (Catalog No. HPRM15202). Cells were seeded in 24-well plates, and transiently transfected with the above plasmid using Lipofectamine 3000 (Thermo Fisher Scientific, catalog no. L3000015) according to the manufacturer's instructions. After 72 h of transfection, the culture medium was collected for analysis of Gaussia luciferase and secreted Alkaline Phosphatase (SeAP) activities using a Secrete-PairTM Dual Luminescence Assay Kit (GeneCopoeia, SPDA-D010) according to the manufacturer's instructions. Gaussia luciferase activity was normalized on the basis of seAP activity.

#### Statistical Analyses

All statistical analyses except for microarray data were carried out using the statistical software package SPSS 17.0 (SPSS, Inc., Chicago, IL, USA). The comparisons between two groups were performed with Student's t-test. The correlation between MTA3 expression and clinicopathological data of patients was analyzed with Pearson χ2 test. Survival curves were plotted with the Kaplan-Meier method and compared by log-rank test. Survival data were evaluated by univariate and multivariate Cox regression analyses. The correlations of the histoscore between MTA3 and SOX2 was determined by Spearman's rank test. Two-way ANOVA followed by a Tukey–Kramer post hoc test was performed to compare the difference of proliferation affected by MTA3 and SOX2 among four groups. All data were presented as the mean ± SEM. The P < 0.05 was considered statistically significant.

### RESULTS

#### MTA3 Is Reduced in Human TSCC

To estimate the expression MTA3, we first assessed the mRNA levels of MTA3 in OSCC from GEO database (https://www. ncbi.nlm.nih.gov/geo/) GSE30784 (36) and GSE25099 (37). We found that the MTA3 mRNA levels were significantly lower in OSCC when compared with the normal controls (P < 0.001 and 0.01, respectively; **Figure 1A** and **Supplementary Figure 1A**). Since TSCC is the highest incidence of all oral squamous cell cancers (5), we focused on the role of MTA3 in TSCC. Data from both datasets GSE78060 (38) and GSE34105 (39) revealed higher MTA3 mRNA levels in normal tongue tissues than in TSCC tissues (P = 0.014 and 0.003, respectively; **Figure 1B** and **Supplementary Figure 1B**). Next, we examined the MTA3 expression at protein levels in TSCC of 119 patient specimens using immunohistochemistry (IHC). Representative

FIGURE 1 | MTA3 is downregulated in human TSCC. (A) Analysis of *MTA3* mRNA expression was performed in an OSCC dataset from GEO (GSE30784). (B) *MTA3* mRNA expression was analyzed in a TSCC dataset from GEO (GSE78060). (C) MTA3 expression in 119 human TSCC tissues and paired adjacent normal tissues (ANT) was monitored by immunohistochemistry (IHC) (left panel). The immunohistochemistry score of MTA3 in TSCC (filled bar) and paired normal adjacent (open bar) tissues were plotted (right panel). Shown are the mean values or representative data from at least three independent experiments. Error bars indicate SEM. \**P* < 0.05, \*\*\**P* < 0.001 using student's *t*-test.

photomicrographs for MTA3 IHC scores of level 0, 4, 6, 9, and 12 are shown in **Figure 1C** (left panel). TSCC showed significantly (P < 0.001, n = 119) lower levels of MTA3 protein in the primary tumors compared to the corresponding normal tissue (**Figure 1C**, Right panel). These findings demonstrate that MTA3 is downregulated in TSCC tissues compared to normal controls.

#### Downregulation of MTA3 Correlates With Clinical Outcomes in TSCC Patients

We next assessed the prognostic impact of MTA3 on the outcome of TSCC patients. An optimal cutoff value was identified using Receiver operator characteristic (ROC) analysis which categorized 51.3% (61/119) of the patient cohort into a high MTA3 group and the remainder into a low MTA3 group (**Figure 2A**). Then Kaplan-Meier survival analyses were performed and showed that patients with low MTA3 were associated with shorter overall survival than those with high MTA3 (P = 0.002, **Figure 2B**).

Univariate analyses found that MTA3 expression, pTNM stage, pN status, and tumor depth were significantly related to TSCC patient outcome (**Table 1**). However, after multivariate Cox regression analysis only MTA3 expression (HR 0.420; 95% CI 0.218–0.810; P = 0.010) and pTNM stage (HR 3.029; 95% CI 1.075–8.538; P = 0.036) were independently significant with overall survival (**Table 1**). These results reveal that reduced expression of MTA3 may be an independent prognostic factor for the overall survival of patients with TSCC.

#### MTA3 Inhibits Key TSCC Plasticity Regulator SOX2

The plasticity of CSC plays an important role in oncogenesis and progression (7, 11, 12, 40). Therefore, we next explored the relationship between MTA3 level and cancer cell stemness using gene set enrichment analysis (GSEA) of published human TSCC expression profiles (GSE78060) and found that a cancer stemness related gene signature (BOQUEST\_STEM\_CELL\_UP) was significantly enriched in TSCC with low MTA3 expression (P < 0.001, **Figure 3A**). Interestingly, SOX2, a key regulator in the plasticity of cancer stemness, was also closely associated with poor prognosis in TSCC patients (20). To study the relation between MTA3 and SOX2, we measured the (co-)expression



*HR, hazard ratio; CI, confidence interval.*

*High in this analysis is based on an MTA3 level* > *5.5; the remaining individuals were classified as low.*

FIGURE 2 | Downregulation of MTA3 correlates with poor prognosis in human TSCC. (A) Receiver operating characteristic (ROC) curve analysis was performed to determine the cut-off score for the low expression of MTA3. (B) Kaplan–Meier curves compared the overall survival in TSCC patients with high and low protein levels of MTA3.

depletion (Left panel) and SCC-4 cells with MTA3 overexpression (Right panel).

of MTA3 and SOX2 in adjacent non-tumor tissues (ANT) and TSCC tissues from the 119 patients. Indeed, a highly significant inverse correlation between the MTA3 and SOX2 levels (r <sup>2</sup> = 0.534, P < 0.001 and r <sup>2</sup> = 0.624, P < 0.001, respectively; **Figures 3B,C**) was observed. Moreover, a low level of SOX2 was accompanied by high expression of MTA3, and vice versa, as indicated by double immunofluorescent staining (**Figure 3D**). In addition, two TSCC cell lines, SSC-25 and SSC-4, were transfected with shMTA3 or the plasmid overexpressing MTA3, respectively. The mRNA and protein levels of MTA3 were then assessed by qRT-PCR and western blot analysis, respectively. When compared to the controls, the expression level of MTA3 was obviously enhanced in cells transfected with pcMTA3 (P < 0.001), while it was significantly decreased in cells transfected with shMTA3 (P < 0.001) (**Figures 3E,F**). Knockdown MTA3 dramatically increased, whereas overexpression of MTA3 decreased, the expression levels of SOX2 (**Figures 3E,F**), suggesting that the MTA3 inhibits SOX2 expression in TSCC cells.

#### MTA3 Suppresses TSCC Cell Stemness and Proliferation Via SOX2

To determine whether SOX2 plays a key role in MTA3 mediated plasticity of CSC and cell growth, we first established TSCC cells with knockdown of MTA3 or SOX2 alone or in combination using specific shRNA (**Figure 4A**) or stably

overexpressing MTA3 or SOX2 alone or in combination (**Figure 4B**). As expected, knockdown MTA3 significantly promoted the percentage of cells expressing ALDH1, a major CSC marker (**Figure 4C**). MTA3-depletion-mediated increase of the number of ALDH1-positive cells were significantly reduced when SOX2 is removed suggesting SOX2 plays a crucial role in MTA3-repressed CSC properties (**Figure 4C**). Next, the effects of overexpressed MTA3 individually or in combination with SOX2 were examined. **Figure 4D** showed that overexpression of MTA3 and SOX2 reduced and increased the number of ALDH1-positive cells, respectively. However, overexpressed SOX2 counteracted MTA3-repressed the number of ALDH1-positive cells (**Figure 4D**). In addition, real-time proliferation assays were conducted to assess changes in cell dynamics. Silencing SOX2 inhibited MTA3-depletion-induced cell proliferation (**Figure 4E**) and forced SOX2 expression was capable of counteracting MTA3-repressed cell proliferation (**Figure 4F**). These data altogether demonstrate that in TSCC cells MTA3 represses CSC property and proliferation by targeting SOX2.

#### Dysregulated MTA3/SOX2 Axis Is Associated With Tumor Progression

The results above indicate a role of MTA3/SOX2 in TSCC progression. Therefore, we next investigated the role of MTA3/SOX2 in a mouse model of tongue tumorigenesis induced by 4 nitroquinoline 1-oxide (4NQO). Exposure to 4NQO caused a temporal progression from hyperplasia to invasive carcinoma in the murine tongue, resembling human tongue carcinogenesis and development (41, 42). Mice were exposed to 4NQO in daily drinking water for 16 weeks followed by 4NQO-free drinking water for 12 additional weeks (**Figure 5A**), which resulted in tongue carcinogenesis and progression (**Figure 5B**) similar to what has been published previously (41, 42). Next, we studied the expression of MTA3/SOX2 in normal tongue tissue samples, hyperplasia, carcinoma in situ, early invasive carcinoma and invasive carcinoma by immunohistochemistry. As shown in **Figure 5C**, the expression of MTA3 gradually

staining image against MTA3 and SOX2 of mice tongue neoplastic tissues. DAPI = blue, MTA3 = green, SOX2 = red.

decreased during the process of carcinogenesis and progression, in contrast to SOX2. This is indicative of an inverse relation between MTA3 and SOX2 associated with cancer occurrence and development, which was confirmed in mouse TSCC by double immunofluorescent staining (**Figure 5D**). Thus, the levels of MTA3 and SOX2 may change dynamically during tongue carcinogenesis and progression.

#### Dysregulated MTA3/SOX2 Axis Associated Strongly With Poor Prognosis

To validate our findings in a clinical setting, we assessed the levels of MTA3 and SOX2 in 119 TSCC samples and related this to the overall survival rate. Receiver operator characteristic (ROC) analysis identified an optimal cutoff value and categorized 39.5% (47/119) of the patients into a high SOX2 group and the remainder into a low SOX2 group (**Figure 6A**). Indeed, patients with low levels of MTA3 but high levels of SOX2 (MTA3low/SOX2high) had significantly shorter overall survival rate (P < 0.001; **Figure 6B**) than those with high levels of MTA3. Moreover, low levels of SOX2 (MTA3high/SOX2low), were associated with an even shorter overall survival compared to TSCC patients with any other expression pattern (MTA3high/SOX2high , MTA3low/SOX2high, MTA3low/SOX2low) (P < 0.001; **Figure 6C**). These data altogether suggest that combined MTA3low-SOX2high expression had stronger correlation with worst patient prognosis than that of the individual components. Indeed, univariate and multivariate Cox regression analysis (P < 0.001 and P < 0.001, HR = 4.044, 95 %CI = 1.925–8.495, respectively) indicated that combined MTA3low/SOX2high expression is an independent prognostic factor of TSCC as it was significantly associated with prognosis (**Table 2**). Overall, these data indicate that dysregulated MTA3/SOX2 axis may contribute to patient outcomes and could be of value as a predictive biomarker for TSCC prognosis.

#### DISCUSSION

Here we demonstrated that MTA3 was a potential independent prognosis factor for TSCC patients. Moreover, we established the negative regulation of SOX2, a key regulator in the plasticity of



*HR, hazard ratio; CI, confidence interval.*

cancer stemness, by MTA3 repressed the number of ALDH1 positive cells and cell proliferation in TSCC cells. We identified the expression of MTA3 and SOX2 as important markers in a 4NQO-induced TSCC mouse model, human TSCC progression and clinical outcomes.

The biological and clinical significance of the MTA family members in malignancies has been well-established. The levels of these factors have even been proposed as potential diagnostic parameters and targeting each one of the family members could be potential treatments for different cancers (26, 43–45). However, unlike MTA1 and MTA2 which were mainly involved in cancer progression and metastasis, MTA3 possesses both tumor-suppressing and tumor-promoting properties depending on specific cancer types (21, 43). We found that MTA3 was negatively associated with overall survival and could act as an independent prognosis factor in TSCC. To the best of our knowledge, this is the first study to show the expression of MTA3 in TSCC.

The plasticity of CSCs is key mechanisms in oncogenesis and progression of cancers (7, 11, 12, 16) and a predictive factor of poor prognosis in a wide variety of cancers, including tongue cancer (17–20). Accumulating evidence suggests that transcription factor SOX2 is a key regulator in the plasticity of cancer stemness (46–48). Moreover, overexpression of SOX2 is often associated with increased cancer aggressiveness, resistance to chemoradiation therapy and decreased survival rate, which has been reported in various cancer types (49, 50), including TSCC (51). Previous studies revealed that SOX2 is vital in the regulation of TSCC motility, invasion, tumorigenicity, and upregulated SOX2 is significantly associated with the progression of TSCC (51). SOX2 is detectable in oral pre-invasive lesions, suggesting that SOX2 upregulation may be an early event in TSCC carcinogenesis. We found that MTA3 can inhibit CSC properties and cell proliferation via downregulating SOX2 in TSCC cells. In addition, luciferase reporter assays showed that knockdown or overexpression MTA3 had no significant effect on the luciferase activity (**Supplementary Figure 2**), suggesting that MTA3 does not regulate the expression of SOX2 by interacting with its proximal promoter. However, these results do not necessarily rule out the possibility that MTA3 is directly involved in the regulation of SOX2 expression by interacting with its enhancer region or the sequence other than the proximal promoter. Further studies will better the understanding about the mechanisms in MTA3-regulated SOX2 expression in tongue cancer.

Given the fact that alterations of MTA3 and SOX2 highly correlate with clinical outcomes and the predictability

#### REFERENCES


of prognosis, the levels of MTA3 and SOX2, especially MTA3low/SOX2high, could be used as diagnostic parameters. In addition, targeting MTA3 and/or SOX2 could be potential therapeutic strategies in TSCC treatment. Finally, 4NQO-induce TSCC mice model has been widely used in the study of TSCC (41, 42). And we used this model to further elaborate the dynamic changes of MTA3 and SOX2 in the occurrence and developmental progression of TSCC.

In summary, consistent with the fact that MTA3 is often silenced in the process of carcinogenesis and development, we found that MTA3 was capable of inhibiting CSC properties and cell proliferation by negatively regulating SOX2. Additionally, we found that TSCC patients in the MTA3low/SOX2high group had a poor prognosis. Future research may involve translational research to clinically evaluate the efficacy of this therapeutic strategy using IHC evaluation of low MTA3 and high SOX2 as a companion diagnostic for patient selection.

#### DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

#### AUTHOR CONTRIBUTIONS

HZ designed the experiments. ZY and LD performed the in vivo experiments. ZY, KL, and LD assayed and analyzed of patient samples. LD and KL performed bioinformatics analyses. LD, MX, KL, ZY, and GY performed the in vitro experiments. LD, MX, KL, and ZY performed the statistical analysis. MX and HG collected the clinical samples. DZ and HZ analyzed data. HZ, LD, RC, and DZ prepared the manuscript. All the authors read and approved the final manuscript.

#### FUNDING

This work was supported in part by funding from the National Natural Science Foundation of China (Grant Nos. 81773087, 81071736, and 81572876 to HZ and 81871977 to GY).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2019.00816/full#supplementary-material


of gastroesophageal junction adenocarcinoma. PLoS ONE. (2013) 8:e62986. doi: 10.1371/journal.pone.0062986


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Yao, Du, Xu, Li, Guo, Ye, Zhang, Coppes and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# An Engineered Fusion Protein Anti-CD19(Fab)-LDM Effectively Inhibits ADR-Resistant B Cell Lymphoma

Dongmei Fan1†, Linlin Jiang2†, Yuewen Song<sup>1</sup> , Shiqi Bao<sup>1</sup> , Yuanyuan Yang<sup>1</sup> , Xiangfei Yuan<sup>1</sup> , Yongsu Zhen<sup>3</sup> , Ming Yang<sup>1</sup> \* and Dongsheng Xiong<sup>1</sup> \*

<sup>1</sup> State Key Laboratory of Experimental Hematology, Institute of Hematology and Hospital of Blood Diseases, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin, China, <sup>2</sup> School of Life Sciences, Ludong University, Yantai, China, <sup>3</sup> Department of Oncology, Institute of Medicinal Biotechnology (IMB), Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China

#### Edited by:

Dong-Hua Yang, St. John's University, United States

#### Reviewed by:

Pranav Gupta, Massachusetts General Hospital and Harvard Medical School, United States Hamid Morjani, Université de Reims Champagne-Ardenne, France

#### \*Correspondence:

Ming Yang my1970@163.com Dongsheng Xiong dsxiong9070@163.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 24 June 2019 Accepted: 20 August 2019 Published: 04 September 2019

#### Citation:

Fan D, Jiang L, Song Y, Bao S, Yang Y, Yuan X, Zhen Y, Yang M and Xiong D (2019) An Engineered Fusion Protein Anti-CD19(Fab)-LDM Effectively Inhibits ADR-Resistant B Cell Lymphoma. Front. Oncol. 9:861. doi: 10.3389/fonc.2019.00861 The 5-year survival rate of patients with B cell lymphoma is about 50% after initial diagnosis, mainly because of resistance to chemotherapy. Hence, it is necessary to understand the mechanism of chemo-resistance and to explore novel methods to circumvent multidrug resistance. Previously, we showed that an engineered cytotoxic fusion protein anti-CD19(Fab)-LDM (lidamycin), can induce apoptosis of B-lymphoma cells. Herein, we successfully established an adriamycin (ADR)-resistant B cell lymphoma cell line BJAB/ADR. The mRNA and protein level of ATP-binding cassette subfamily B member 1 (ABCB1) were significantly overexpressed in BJAB/ADR cells. Increased efflux function of ABCB1 was observed by analyzing intracellular accumulation and efflux of Rhodamine 123. The efflux of Rhodamine 123 could be significantly ameliorated by verapamil. Treatment with anti-CD19(Fab)-LDM at different concentrations induced cytotoxic response of BJAB/ADR cells similar to that of the sensitive cells. In vivo studies showed that anti-CD19(Fab)-LDM had better antitumor effect in BJAB and BJAB/ADR cell lymphoma xenografts compared with ADR or LDM treatment alone. Taken together, anti-CD19(Fab)-LDM can effectively inhibit the growth of BJAB/ADR cells both in vitro and in vivo. Anti-CD19(Fab)-LDM could be a promising molecule for the treatment of drug resistant cancers.

Keywords: adriamycin, BJAB cell line, BJAB/ADR cell line, engineered fusion protein, anti-CD19(Fab)-LDM

# INTRODUCTION

Lymphomas are a common heterogeneic group of hematologic diseases, among which B cell origin lymphoma represents the largest proportion (1, 2). At present, chemotherapy or chemoimmunotherapy remains the most effective therapeutic modality in the multifaceted treatment of lymphomas (3). Most patients who experience remission for more than 5 years have benefitted from the overall improvements in the treatment of B cell lymphomas. However, a significant portion of patients still show unfavorable response toward drug treatment. Currently, the clinical approaches to relapsed B lymphomas mainly involve in administering high-dose chemotherapeutic agents, using inhibitors to reverse drug resistance toward chemotherapy (4), or

**37**

finding novel therapeutic strategies such as targeting CD20 or using Chimeric Antigen Receptor T-Cell Immunotherapy (CAR-T) (5–8). Multidrug resistance (MDR) or acquired chemo-drug resistance is a major contributor to the failure of chemotherapy as well as one of the major reasons for tumor relapse and metastasis (9–11). To investigate the mechanisms involved in the acquisition of chemotherapy resistance and subsequent poor prognosis, it is necessary to establish a proper resistant cell model derived from a drug-sensitive human lymphoma cell line. Adriamycin (ADR; generic name: doxorubicin, DOX) is a chemotherapeutic drug frequently used in multiple clinical protocols of chemotherapy and is also a critical drug in the treatment of lymphoma (12). Unfortunately, some lymphomas have shown ADR resistance with continued treatment (13, 14). Therefore, establishing an ADR-resistant lymphoma cell model is useful for studying the mechanism of resistance in B cell lymphoma and for searching solutions regarding ADR resistance.

Lidamycin (LDM), originally named C-1027, is a member of the enediyne antibiotic family with strong cytotoxic effect toward human cancer cells and its mechanism of action is related to DNA damage. Importantly, LDM molecule is composed of a highly active group enediyne chromophore (AE) and a protective group apoprotein (LDP) (15, 16). The non-covalent bond between AE and LDP can be dissociated and re-associated, leading to rebuilting a molecule that exhibits similar activity as that of natural LDM. Taking advantage of the LDP genetic reassortment and the specific targeting capability of antibody fragments, different types of engineered fusion proteins were created (17–20). In short, lidamycin can be linked with another component, such as antibodies, due to its unique structure. As a result, lidamycin can target a specific site with its cytotoxicity. CD19 is a biomarker that is expressed on virtually all neoplastic cells of the B-cell lineage (21, 22). Previous studies demonstrated that the engineered fusion protein anti-CD19(Fab)-LDM, which comprises the chemo-drug lidamycin and anti-CD19(Fab) antibody, showed targeted cytotoxicity against lymphoma cells both in vitro and in vivo (23).

In this article, to verify the anticancer activity of the engineered fusion protein anti-CD19(Fab)-LDM on multidrugresistant cells, we established an ADR resistant lymphoma cell line BJAB/ADR. Furthermore, we showed that anti-CD19(Fab)- LDM engineered fusion proteins could target the cell surface marker CD19 and exert the same cytotoxicity effect on ADRresistant BJAB cells as on BJAB-sensitive cells. Our study indicates that anti-CD19(Fab)-LDM has anticancer effects on ADR-resistant B cell lymphoma. This result sheds light on the therapeutic effect of this fusion protein and provides a promising solution for MDR, especially ADR-resistant B cell lymphoma.

# MATERIALS AND METHODS

#### Chemicals and Reagents

Adriamycin (ADR), propidium iodide (PI), verapamil and RNase A were obtained from Sigma-Aldrich Trading Co, Ltd (St. Louis, MO, USA). The phospho-glycoprotein (P-gp, MDR1) mouse monoclonal antibody conjugated with Alexa Fluor 594 (sc-390883), ABCG2 mouse monoclonal antibody conjugated with Alexa Fluor 488 (sc-18841) and MRP1 mouse monoclonal antibody conjugated with Alexa Fluor 488 (sc-53130) were obtained from Santa Cruz Biotechnology, Inc (Dallas, TX, USA). LDM was provided by the Institute of Medicinal Biotechnology of the Chinese Academy of Medical Sciences (Beijing, China). Antitumor agents were prepared fresh in PBS (phosphatebuffered saline) immediately prior to use.

# Cells and Cell Culture

Cell culture supplies, including Dulbecco's modified Eagle's Medium (DMEM), fetal bovine serum (FBS), penicillin/streptomycin and 0.25% trypsin, were purchased from Corning Incorporated (Corning, NY, USA). The BJAB cell line was obtained from Cell Resource Center, Institute of Hematology and Hospital of Blood Diseases, Peking Union Medical College (PUMC) (Beijing, China). The cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin, and they are maintained in an incubator containing 37◦C humidified air with 5% CO2.

# Establishment of an ADR-Resistant BJAB Cell Line

The ADR-resistant cell line was created from the BJAB parental cell line via intermittent exposure to increasing concentrations of ADR for 6 months. Briefly, BJAB/ADR cells were treated with ADR with the concentrations ranging from 37 nM to 294 nM in a stepwise increasing manner. At first, the majority of the cells died after being treated with low concentrations of ADR for 24 h. We used 0.01 mol/L PBS to wash the surviving cells and continued to culture them in ADR-free growth medium. When cells were in the logarithmic growth phase, they were exposed to a higher ADR concentration for 24 h. After this process was repeated in a stepwise manner, a single-cell-derived ADR-resistant subclone, designated as BJAB/ADR, was established. For the maintenance of MDR, BJAB/ADR cells were cultured with 147 nM ADR. Two weeks before the experiment, BJAB/ADR cells were maintained in drug-free culture medium and passaged at least 3 times.

# Cell Growth Assay

To investigate cell growth in both BJAB and BJAB/ADR cells, a cell proliferation assay was performed. Briefly, we seeded cells into 24-well culture plates at a density of 5 × 10<sup>3</sup> cells per well and cultured in complete RPMI 1640 culture medium for 8 days. Trypan blue exclusion-based methods were used to determine cell counts, and cells from triplicate wells were counted every 24 h for 8 days. All experiments were independently performed three times.

# Analysis of Cell Cycle Distribution

After BJAB and BJAB/ADR cells were treated with ADR, they were harvested, washed twice with ice-cold PBS (pH 7.2), centrifuged and resuspended in 500 µL ice-cold PBS, and adjusted to a density of 1 × 10<sup>6</sup> cells/mL. Then, the cells were fixed with 70% ethanol at −20◦C overnight. For the next step, the cells were incubated with 100 µL RNase (100µg/mL, Sigma) for half an hour and stained with 200 µL PI (50µg/mL) for 1 h. Data from 100,000 events/sample were collected via FACScan flow cytometer (Becton Dickinson, San Diego, CA) and analyzed using FlowJo software.

# Cell Viability Analysis (MTT Assay)

The MTT colorimetric assay was used to determine cell viability. Briefly, BJAB or BJAB/ADR cells (approximately 6,000 cells/well) were seeded into 96-well plates one day before drug treatment. After 72 h of drug treatment, 20 µL MTT solution (5 mg/mL thiazolyl blue powder in PBS) was added into each well and further incubated for 4 h at 37◦C in a humidified atmosphere with 5% CO2. At the end of the incubation period, the supernatant, including the medium and MTT solution, was removed from each well, and the forming formazan crystals were dissolved by adding 100 µL dimethyl sulphoxide (DMSO) solution and agitating the plate for 15 min. The spectrophotometric absorbance was measured at 570 nm. The percentage of viable cells was calculated compared with the untreated control group (assumed 100% viability). After treatment with ADR at different concentrations, the resistance fold was reflected by MTT colorimetric analysis. In a separate experiment, 10 µL verapamil (2.5 mg/mL) was used to treat BJAB/ADR cells to observe whether verapamil can reverse ADRinduced MDR. The cytotoxicity of anti-CD19(Fab)-LDM on cells was evaluated in the same way. All experiments were repeated three times independently.

# Primer Design

Primers were designed according to published sequences using web-based software. We used the "BLAST" program (http:// www.ncbi.nlm.nih.gov/blast) to determine the specificity of the primers. The primers used in this study were as follows: GAPDH-F: GAAGGTGAAGGTCGGAGTC, GAPDH-R: GAAGATGGTGATGGGATTTC; and MDR1-F: CCCATC ATTGCAATAGCAGG, MDR1-R: GTTCAAACTTCTGCTC CTGA.

# RNA Extraction

Total RNA was extracted from cells using a RNeasy Mini Kit (Qiagen) following the instructions of manufacturer. Then, cDNA was synthesized with an M-MLV Reverse Transcriptase Kit (InvitrogenTM) following the manufacturer's instructions.

# Quantitative Real-Time PCR (qRT-PCR)

Quantitative Real-time PCR (RT-PCR) was used to quantitatively detect mRNA of ABCB1, ABCC1, and ABCG2 in BJAB and BJAB/ADR cells. RT-PCR was performed using a SYBR <sup>R</sup> Green PCR Master Mix kit (Applied Biosystems <sup>R</sup> ) on the Applied Biosystems 7500 system. The thermal profile comprised 40 cycles as follows: 95◦C for 30 s, 55◦C for 60 s, and 72◦C for 30 s. The expression of each gene was normalized using the mean expression of the housekeeping gene. Linearized relative expression was obtained according to the 2−11CT method (24).

# Detection of MDR Protein Expression Level by Flow Cytometry

Control and ADR-resistant BJAB cells were harvested, washed twice with ice-cold PBS (pH 7.2) and placed on ice immediately after collection. Samples (50 µL) were stained at 4◦C for 20 min using predetermined saturating concentrations of phycoerythrin (PE)-labeled anti-P-gp monoclonal antibody, fluorescein (FITC) labeled anti-ABCG2 monoclonal antibody, or FITC-labeled anti-MRP1 monoclonal antibody, respectively. Cells were analyzed on a FACScan flow cytometer (Becton Dickinson, San Diego, CA). Positive and negative cell populations were determined by using unreactive isotype-matched mAbs (Coulter) as controls for background staining. Background levels of staining were delineated using gates established to include 98% of the control cells.

# Assessment of the Efflux Function of ABC Transporter in BJAB/ADR Cells

Because Rhodamine 123 (Rho 123) is a reference fluorescent substrate of ABCB1, we detected the fluorescence intensity to obtain efflux function of ABCB1 (25). Briefly, BJAB and BJAB/ADR cells were suspended at a density of 5 × 10<sup>5</sup> cells/mL in serum-free RPMI 1640 medium, and 200 µL of the cell suspension was put into 1.5 mL microcentrifuge tubes. BJAB cells were divided into two groups: negative control group (PBS) and positive control group (Rho 123). BJAB/ADR cells were divided into three groups: negative control group (PBS), positive control group (Rho 123) and experimental group (verapamil plus Rho 123). In the experimental group, 50 µmol/L verapamil was added to the tubes and incubated for 30 min at 37◦C. After incubation, Rho 123 (200 nmol/L) was added to each tube. The cells were incubated for 1 h and then washed twice with ice-cold PBS after the incubation period. Finally, the fluorescence intensity was determined by flow cytometry to measure intracellular accumulation and efflux of Rho 123. Data was obtained from FACScan flow cytometer (Becton Dickinson, San Diego, CA). All experiments were independently conducted three times.

#### In vivo Antitumor Activity in Subcutaneous Xenograft Tumor Models

All experiments on mice received humane care in compliance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals. The study protocol was approved by the Institutional Animal Care and Use Committee of the State Key Laboratory of Experimental Hematology (SKLEH).

BJAB and BJAB/ADR cells were harvested, suspended in PBS, and then subcutaneously injected into 5-week-old female BALB/c nude mice (1 × 10<sup>7</sup> cells/0.2 mL/mouse) to establish the BJAB and BJAB/ADR xenograft tumor models. When tumor volumes reached 60–80 mm<sup>3</sup> , mice were randomized into eight treatment groups (five mice per group). Group 1: animals received PBS; group 2: animals received 6 nmol/kg ADR; group 3–5: animals received 2 nmol/kg, 4 nmol/kg, and 6 nmol/kg LDM, respectively; and group 6–8: animals received 2 nmol/kg, 4 nmol/kg, and 6 nmol/kg anti-CD19(Fab)-LDM, respectively. Drugs were intravenously injected once. The body weights of the mice and the two perpendicular diameters of the tumors were recorded every third day, and tumor volumes were calculated by the following formula: tumor volume=1/2 × length × width<sup>2</sup> . Animals were sacrificed, and xenograft tumors were surgically dissected, weighed and measured 28 days after treatment initiation.

#### Statistical Analysis

All the data was shown as the mean ± SD obtaining from three independent experiments performed in triplicate. The results were analyzed with one-way analysis of variance (ANOVA). Comparisons are made between control groups and corresponding treatment groups and they were carried out via SPSS 10.0 software. A value of P < 0.05 was considered statistically significant.

#### RESULTS

#### Successful Establishment of the ADR Resistant BJAB/ADR Cell Line

The BJAB/ADR cell line was established after intermittent treatment with ADR at concentrations ranging from 37 to 294 nM in a stepwise increasing manner. Over 6 months, a clone of BJAB cells that resistant to ADR was successfully screened and named BJAB/ADR. BJAB/ADR cells could grow sufficiently even if cultured in RPMI 1640 medium with 294 nM ADR, and cells also maintained resistance to ADR after removal of the drug for at least 2 weeks. Normally, the ADR-resistant cells were maintained in complete culture medium with 147 nM ADR, which is the approximate IC<sup>50</sup> value (concentration that reduces viability to 50%) of BJAB/ADR cells to ADR. Moreover, BJAB/ADR cells can stably grow in drug-free RPMI 1640 medium for more than 2 weeks. These results suggest that an ADR-resistant cell line was successfully established.

In addition, the morphological characteristics of the established ADR-resistant BJAB/ADR cells were distinct from those of its parental cells under optical microscope. Although both types of cells exhibited suspension growth and had relatively consistent size and shape, the BJAB parental cells grew as a monolayer (**Figure 1A**), while the BJAB/ADR resistant cells tended to grow in clusters (**Figure 1B**).

#### BJAB/ADR Cells Had a Slower Growth Rate Than the Parental Cells and Were Arrested in G0/G1 Phase

The growth curves of BJAB and BJAB/ADR cells are shown in **Figure 2A**. The proliferation rate of both cell lines was not significantly different when the cells were cultured at low density. However, the growth rate of BJAB cells increased much more quickly as the density increased (P < 0.05). Specifically, the cell population doubling times for BJAB and BJAB/ADR cells were 31.66 ± 1.2 h and 35.19 ± 2.1 h, respectively (P < 0.05).

To investigate the effect of ADR on cell growth, a cell cycle assay was performed via flow cytometry. The results showed that the proportion of BJAB/ADR cells in the G0/G1 phase increased (P < 0.05, **Table 1**) and was accompanied by a decreased proportion in the S phase and G2/M phase. These results indicated that ADR could induce G0/G1 phase arrest in BJAB/ADR cells compared with the phase distribution of BJAB cells. However, the difference of phase distribution between BJAB and BJAB/ADR was not obvious, even they were repeatable and statistically significant. Hence, BJAB/ADR cell line may involve in other mechanisms of action resulting in ADR-resistant, which is needed to be addressed further.

#### BJAB/ADR Cells Exhibited a 43-fold Greater ADR Resistance Level Than the Parental Cells and Showed Cross-Resistance to Other Anticancer Drugs

After cells were treated with ADR, we performed the MTT assay to determine the drug resistance factor (RF). The IC<sup>50</sup> values of ADR for BJAB and BJAB/ADR cells were 57.156 ± 2.30 nM and 2,434 ± 111.476 nM (**Figure 2B**), respectively. As shown in the results, the ADR resistance level of BJAB/ADR cells was 43-fold

FIGURE 2 | Cell growth curve of BJAB and BJAB/ADR cells. (A) Growth curves of BJAB and BJAB/ADR cells. BJAB/ADR cells grew slower than BJAB cells (P < 0.05). (B) Cytotoxicity of adriamycin against BJAB and BJAB/ADR cells. Each data point was obtained from three independent experiments in triplicate. BJAB/ADR cells were resistant to adriamycin.

TABLE 1 | Cell cycle distribution of BJAB and BJAB/ADR cells.


a \*P < 0.05, the values are shown as the mean ± SD obtained from three independent assays.

<sup>b</sup>The increased proportion of resistant BJAB/ADR cells in G0/G1 phase was accompanied by a decreased proportion of cells in S and G2/M phases.

greater than that of the parental cells. This result verified that the BJAB/ADR cell line acquired ADR resistance.

Additionally, BJAB/ADR cells showed resistance to various structurally unrelated anticancer drugs other than ADR. The cross-resistance profile of BJAB/ADR was summarized in **Table 2**. BJAB/ADR cells showed strong cross-resistance to etoposide, daunorubicin, homoharringtonine, and mitoxantrone but not cisplatin. Interestingly, the ADR-resistant subclone was 40 times more resistant to daunorubicin than the parental cell line. Hence, the established BJAB/ADR cell line can also be used for MDR study of its substrate drugs.

#### The Expression of Both ABCB1 Gene and P-gp Protein Increased in BJAB/ADR Cells

Since it is reported that ADR is transported by ATP-binding cassette (ABC) transporters, especially ABCB1 and ABCG2 (4, 11), we hypothesized that the resistance mechanism of BJAB/ADR cells was associated with overexpression of ABC transporters. The ABCB1 gene is a member of the ABC transporter superfamily that encodes a 170-kDa plasma membrane ABCB1 (P-glycoprotein, P-gp). ABC transporter functions as a drug efflux pump, thus resulting in decreased intracellular concentrations of broad drugs, such as paclitaxel, doxorubicin and others (26).

To determine the underlying resistance mechanism of BJAB/ADR cells, qRT-PCR was performed to detect the expression of the ABCB1, ABCC1, and ABCG2 genes. We found TABLE 2 | Cross-resistance profile of BJAB/ADR cells to other anticancer drugs.


a IC<sup>50</sup> values are shown as the mean ± SD calculated from the results of at least three independent MTT assays.

<sup>b</sup>RF refers to the resistance factor, which was calculated by dividing the IC<sup>50</sup> values of the resistance cell line by the IC<sup>50</sup> values of the respective parental cell line.

that the ABCB1 gene was upregulated in BJAB/ADR cells (P < 0.05) (**Figure 3A**). Moreover, the expression of P-gp protein was evaluated by flow cytometry analysis. Compared with the BJAB-sensitive cells, BJAB/ADR cells showed higher protein expression level of P-gp (P < 0.05) (**Figure 3B**). The results were consistent with those results from the qRT-PCR assay. We also examined the mRNA level of ABCC1 and ABCG2, plus protein expression level of MRP1 (multidrug resistanceassociated protein 1) and BCRP (breast cancer resistance protein), as shown in **Figures 3C–F**. Both transporters showed increased mRNA and protein expression, but not increased as much as ABCB1 and P-gp expression. These results implicated that these two transporters were probably related to crossresistance or MDR, suggesting that the mechanism of drug resistance of BJAB/ADR cells might be due to the overexpression of ABCB1.

#### Overexpression of ABCB1 in BJAB/ADR Cells Increased Drug Efflux, and Verapamil Reversed the Chemoresistance of the Cells to Adriamycin

As shown above, ABCB1 was overexpressed in the ADRresistant cell line. To further understand the effects of

FIGURE 3 | Expression of the ABCB1, ABCC1, ABCG2 genes and P-gp, MRP1, BCRP proteins in BJAB/ADR cells. (A) The expression of the ABCB1 gene in the ADR-resistant BJAB/ADR cells. The mRNA level of MDR1 analyzed by qRT-PCR and normalized to the mRNA level of the housekeeping gene GAPDH. Compared with the sensitive cells BJAB, BJAB/ADR drug-resistant cells showed increased expression of ABCB1 mRNA. \*P < 0.05, compared with control group. (B) The protein expression of P-gp on BJAB/ADR cells. The expression of P-gp was evaluated via flow cytometry analysis. BJAB/ADR cells showed higher expression of the membrane protein P-gp compared with the BJAB sensitive cells. (C,E) The qRT-PCR on the gene expression of ABCC1 and ABCG2, respectively. These results showed higher expression levels of both genes, but the increased expression was not as high as that of ABCB1. \*P < 0.05, compared with control group. (D,F) The protein expression level of MRP1 and BCRP, separately. The results were obtained from flow cytometry analysis. These results indicated that higher expression levels of both protein, but the increased expression was not as high as that of P-gp.

ABCB1 overexpression on drug resistance, we performed an accumulation and efflux assay using Rho 123, a reference fluorescence substrate of ABCB1, via flow cytometry (27). As **Figure 4A**a,b shown, the mean values of the fluorescence intensity in BJAB and BJAB/ADR cells were 11,900 ± 312.05 and 165.67 ± 24.74, respectively, with statistical significance (P < 0.05). This result suggested that overexpression of ABCB1 in BJAB/ADR cells could decrease intracellular chemo-drug accumulation by increasing its efflux function.

Verapamil is a known reversal agent against drug resistance that can reverse MDR by blocking the efflux function of ABCB1 without changing its expression level (4, 11). After confirming the efflux function of ABCB1, we further treated BJAB/ADR cells with verapamil to observe the cells' sensitivity to ADR. When BJAB/ADR cells were pretreated with verapamil, the peak fluorescence intensity significantly shifted to the right, and the mean values of the fluorescence intensity increased to 4,890 ± 43.52 (**Figure 4A**c). As shown in **Figure 4B**, verapamil could sensitize the chemoresistance of BJAB/ADR cells to ADR and make BJAB/ADR cells more sensitive to ADR. These results suggested that the resistance mechanism of BJAB/ADR cells might be due to the increased efflux function of ABCB1, and verapamil could mitigate the efflux activity of ABCB1.

#### Anti-CD19(Fab)-LDM Had Similar Antitumor Activity in Both Resistant and Parental Cells

Previous experiments in our laboratory showed that the engineered fusion protein anti-CD19(Fab)-LDM exerted significant cytotoxic effects on BJAB cells (23). We performed the MTT assay to ascertain the cytotoxic effect of anti-CD19(Fab)-LDM toward BJAB/ADR cells. As shown in **Figure 5A**, the growth inhibition curves showed that two types of cells had similar drug sensitivity to anti-CD19(Fab)- LDM (P > 0.05). Additionally, the engineered fusion protein anti-CD19(Fab)-LDM showed a much stronger inhibitory effect than ADR in ADR-resistant cells (P < 0.01). More importantly, the fusion protein had a stronger cytotoxic effect than LDM alone (**Figure 5B**). These results suggested that anti-CD19(Fab)-LDM exerted cytotoxic effects on BJAB and BJAB/ADR cells and had a much stronger inhibitory function than either ADR or LDM alone in BJAB/ADR cells.

#### Anti-CD19(Fab)-LDM Inhibited Tumor Growth in Both BJAB and BJAB/ADR Xenograft Tumors in BALB/c Nude Mice

We previously demonstrated that anti-CD19(Fab)-LDM suppresses tumor growth in a human B-cell lymphoma xenograft model (23). To assess whether the observed anti-CD19(Fab)-LDM-mediated inhibition of cell growth of MDR cells in vitro would extend to animal models, we established BJAB and BJAB/ADR xenograft tumor mouse models to investigate the MDR phenomenon in vivo to investigate the therapeutic effect of anti-CD19(Fab)-LDM on the BJAB/ADR xenograft model.

We induced tumors by subcutaneously injecting BJAB or BJAB/ADR cells into the nude mice. When the tumor volume reached 60–80 mm<sup>3</sup> , we treated the mice with PBS (as a control), ADR (6 nmol/kg), LDM (2, 4, or 6 nmol/kg), or anti-CD19(Fab)- LDM (2, 4, or 6 nmol/kg). Tumor volume was measured every 3 days following inoculation. Compared with the LDM- and ADR-treated mice, mice treated with anti-CD19(Fab)-LDM at doses of 2, 4, and 6 nmol/kg showed a significant inhibition of tumor growth in a dose-dependent manner in both the BJAB and BJAB/ADR xenograft models (P < 0.05) as shown in **Figure 6A**. Specifically, the ratio of tumor volume of the anti-CD19(Fab)- LDM group (6 nmol/kg) compared to the PBS control group was 92.79% on day 30, while the inhibitory effect of ADR was 53.45%.

After treatment for 28 days, the tumor tissues were excised and weighed. In the ADR-resistant xenograft model, the antitumor activity of anti-CD19(Fab)-LDM was stronger than that of LDM or ADR alone in a concentration-dependent manner (P < 0.05) (**Figure 6B**). More importantly, anti-CD19(Fab)-LDM was well-tolerated in both the BJAB and

FIGURE 5 | Antitumor activity of anti-CD19(Fab)-LDM on ADR resistant cells. Antitumor activity of anti-CD19(Fab)-LDM on BJAB and BJAB/ADR cells assessed by MTT assay. (A) The anti-CD19(Fab)-LDM has a similar cytotoxic effect on resistant cells and their corresponding parental cells. (B) The anti-CD19(Fab)-LDM had a stronger inhibitory effect on BJAB/ADR cells than either adriamycin or LDM alone (P < 0.05).

BJAB/ADR xenograft models, as indicated by the absence of significant differences in body weight compared with that in the vehicle-treated animals (P > 0.05) (**Figure 6C**). These results suggested that anti-CD19(Fab)-LDM was able to inhibit the growth of ADR-resistant BJAB cells and was welltolerated. Therefore, anti-CD19(Fab)-LDM could be exploited as a potential drug used in the treatment of multidrugresistant tumors.

# DISCUSSION

B cell lymphoma is a hematopoietic malignant tumor, and its poor prognosis and short survival are mainly associated with multidrug resistance (MDR). Overcoming MDR and enhance the therapeutic effect of regimens for the treatment of B cell lymphoma is a major concern in clinical oncology (28–30). Thus, there is an immediate need to identify novel targets for the treatment of B cell leukemias and lymphomas. It was known that the poor response of lymphoma to chemotherapeutic drugs is mainly due to acquired MDR rather than innate resistance (31–33). Therefore, an appropriate experimental model is urgently needed for the study of MDR in B cell lymphoma. Since Bielder and Riehm first reported the MDR phenomenon of tumor cells in 1970, a series of multidrugresistant cell lines have been constructed. However, there is few report on the stable MDR cell line of B cell lymphoma (34). In this article, our laboratory successfully established a B lymphoma MDR cell line, named BJAB/ADR, with the first-line chemotherapeutic drug adriamycin (ADR). The resistance factor (RF) between the parental and resistant cell lines was 43-fold. In fact, the resistance fold is highly variable between cell lines. For example, Wattanawongdon established two gemcitabineresistant human cholangiocarcinoma cell lines with resistance indices of approximately 25- and 62-fold, respectively (35). In contrast, Iwasaki developed a cisplatin-resistant human neuroblastoma cell line with a resistance variant of approximately 1.1 (36). Generally, medium resistance is the most common type encountered in clinical practice. It is worth mentioning that BJAB/ADR cells could stably grow in drug-free medium for several weeks, and the morphological characteristics are consistent with those of parental cells, indicating a resistancemediated improvement in survival. However, BJAB/ADR cells prefer to grow in clusters and have a slower growth rate than its parental cells (**Figures 1**, **2**). More importantly, these cells exhibited cross-resistance to a variety of structurally and functionally unrelated antineoplastic agents, such as etoposide, daunorubicin, homoharringtonine and mitoxantrone (**Table 2**). This result provides important information for further clinical evaluation.

We firstly examined the cell cycle of BJAB and BJAB/ADR. The results showed that ADR could induce G0/G1 phase arrest in BJAB/ADR cells compared with that in BJAB cells (**Table 1**). Combined the results of growth rate in **Figure 2**, we postulated that ADR could induce longer proliferation time and poorer proliferation ability. However, the difference of phase distribution between BJAB and BJAB/ADR was not obvious. Hence, it is needed more further studies to figure out the exact resistance mechanism of ADR on BJAB/ADR cell line. Also, we hypothesized that ADR resistance is probably associated with the overexpression of ABC transporters. This hypothesis is supported by the cross-resistance to other structurally unrelated chemotherapeutic drugs, most of which are substrates of ABC transporters, in resistant cells. ABCB1 (P-gp), ABCG2 (BCRP), and ABCC1 (MRP1) are three ABC transporters that are broadly expressed in multidrug-resistant cell lines (37). Thus, we examined the expression of these three ABC transporters and found that the ABCB1 gene and P-gp protein expression was significantly upregulated in the BJAB/ADR cells. In contrast, the ABCC1 and ABCG2 mRNA and protein levels were only slightly increased compared to ABCB1 (**Figure 3**). This result is consistent with the previous reports that upregulated ABCB1 gene is the main response for MDR in B-cell lymphoma (38, 39). Moreover, the Rhodamine 123 (Rho 123) exclusion assay verified that overexpression of ABCB1 participated in MDR, and verapamil, a known ABCB1 inhibitor, could reverse this drug resistance, thus increasing the sensitivity of BJAB/ADR cells to ADR (**Figure 4**). According to the present results, we could conclude that ABCB1-overexpressing is responsible for chemoresistance in BJAB/ADR cell line and poor efficacy of chemotherapeutic agents.

About 80–90% of cases of non-Hodgkin lymphoma (NHL) are of B-cell origin (40). The current therapeutic approach for B cell lymphoma involves chemotherapy, radiotherapy and the incorporation of the anti-CD20 monoclonal antibody rituximab (41, 42). Chemotherapy is the most common treatment strategy, but the outcomes of patients are often very poor, because of the development of resistance to conventional chemotherapeutic strategies. To overcome this issue, chemoimmunotherapies using rituximab in combination with CHOP (refers to cyclophosphamide, doxorubicin, vincristine, and prednisone) (R-CHOP) have markedly improved the outcome of patients with B cell lymphoma in recent decades. Currently, there are some novel treatment regimes, such as bendamustine or valproate, in combination with R-CHOP for patients with different phases of lymphoma (43–45). Unfortunately, about 10–15% of patients fail to respond to R-CHOP treatment, and 20–25% of patients develop relapse (46, 47). Therefore, novel strategies are needed to improve patients' response rate. Lidamycin is a novel antibiotic with antitumor activity emerged in recent years. Its mechanism of antineoplastic action is to inhibit DNA synthesis and break down cellular DNA in carcinoma cells (16). Due to its unique structure, lidamycin is often reconstituted with antibodies to establish engineered fusion proteins to maintain both the target property of antibodies and the cytotoxic effect. This type of biopharmaceutical drug is called antibody-drug conjugate (ADC) (48). Specifically, anti-C19(Fab)-LDM is an engineered fusion protein previously established in our laboratory and has been reported to have high antineoplastic activity toward B cell lymphoma (23). The fusion protein anti-CD19(Fab)-LDM was developed as a targeted therapy for lymphoma and induces significant tumor-specific cytotoxicity. Thus, anti-CD19(Fab)-LDM can overcome the deficiencies of traditional chemotherapy agents and significantly decrease adverse effects in patients. Additionally, this study shed light on the solution of drug resistance in tumor treatment. With all of these advantages, the use of engineered fusion proteins can circumvent the clinical issue of chemotherapy in the treatment of lymphoma. More importantly, due to the strong cytotoxic effects of LDM, the antibody-drug conjugate anti-CD19(Fab)- LDM can be administered at a lower dose to achieve therapeutic effects. Thus, it is a novel strategy worth exploring to find out its promising potential in preclinical and clinical trials.

Considering the strong antitumor activity of LDM and the B cell-targeted property of the anti-C19(Fab) antibody, we postulated that the anti-C19(Fab)-LDM could exert cytotoxic effect on the resistant cells of B cell lymphoma. As expected, anti-C19(Fab)-LDM showed similar cytotoxic effects toward BJAB/ADR and BJAB cells and showed a greater effect than either LDM or ADR alone (**Figure 5**). From the in vivo results, anti-C19(Fab)-LDM exhibited more potent antitumor activities than LDM and ADR in the BJAB/ADR xenograft mouse model (**Figure 6**). The in vivo results were in consistent with the in vitro results. Importantly, the therapeutic effect of anti-C19(Fab)- LDM was better than that of LDM both in vitro and in vivo. Therefore, our current results indicated that anti-C19(Fab)-LDM could be a promising targeted therapy for patients with ADRresistant B cell lymphoma. Considering our in vitro results above, it is reasonable to postulate that anti-C19(Fab)-LDM may have inhibitory effect to pumped function of ABCB1, in turn probably increase the intracellular concentration of antineoplastic drugs. However, the exactly underlying re-sensitive mechanism of anti-C19(Fab)-LDM is needed to be addressed in the future.

In summary, we established an MDR B cell lymphoma cell line named BJAB/ADR, which could represent a drug-resistant cell model for lymphoma research. Additionally, our previously developed engineered fusion protein anti-C19(Fab)-LDM can be

#### REFERENCES


used to overcome MDR for the treatment of B cell lymphoma, especially in patients with acquired ADR resistance.

# DATA AVAILABILITY

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

# AUTHOR CONTRIBUTIONS

DF, LJ, MY, and DX designed the experiments. DF, LJ, YS, and SB performed experiments. YY, XY, and YZ provided technical and material support. DF and LJ wrote the first draft. DF, LJ, and MY revised the manuscript. All authors discussed the results and implications and developed the manuscript at all stages.

#### FUNDING

This work was supported by the Natural Science Foundation of China (Grant numbers 30873091, 30971291, and 81170512), the Natural Science Foundation of Tianjin (Grant number 05YFGZGX02800), and the National Science and Technology Major Project (Grant number 2012ZX09102301-015).


lymphoma xenografts with enhanced anticancer activity. J Drug Target. (2015) 24:47–5. doi: 10.3109/1061186X.2015.1055568


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Fan, Jiang, Song, Bao, Yang, Yuan, Zhen, Yang and Xiong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Current Advance of Therapeutic Agents in Clinical Trials Potentially Targeting Tumor Plasticity

#### Xiao-Guang Yang† , Lan-Cao Zhu† , Yan-Jun Wang, Yan-Yu Li and Dun Wang\*

*Key Laboratory of Structure-Based Drug Design & Discovery of Ministry of Education, Shenyang Pharmaceutical University, Shenyang, China*

Tumor plasticity refers to tumor cell's inherent property of transforming one type of cell to different types of cells. Tumor plasticity is the main cause of tumor relapse, metastasis and drug resistance. Cancer stem cell (CSC) model embodies the trait of tumor plasticity. During carcinoma progression, epithelial-mesenchymal transition (EMT) plays crucial role in the formation of CSCs and vasculogenic mimicry (VM) based on epithelial-mesenchymal plasticity. And the unique tumor microenvironment (TME) not only provides suitable niche for CSCs but promotes the building of CSCs and VM that nourishes tumor tissue together with neoplasm metabolism by affecting tumor plasticity. Therapeutic strategies targeting tumor plasticity are promising ways to treat malignant tumor. In this article, we discuss the recent developments of potential drug targets related to CSCs, EMT, TME, VM, and metabolic pathways and summarize drugs that target these areas in clinical trials.

Keywords: tumor plasticity, cancer stem cells, vasculogenic mimicry, extracellular matrix, tumor microenvironment, targeting

#### INTRODUCTION

The universal methods for cancer treatment include surgery, radiotherapy, and chemotherapy. Chemotherapy is the principle modality for the treatment of malignant tumor, especially tumors in the late stages. Despite significant improvement of cancer chemotherapy in clinical practice, there are still many obstacles that chemotherapeutic drugs must overcome: (1) lack of effective treatments for metastatic tumors; (2) ineffectiveness in killing drug-resistant tumor cells; and (3) lack of new targets based on the characteristics of neoplasm, such as tumor plasticity.

Tumor plasticity prompts tumor cells to differentiate into a variety of cell types to adapt to different environment (1). Emerging evidence suggested that tumor plasticity played critical roles in the emergence of drug resistance and the promotion of tumor growth, invasion and metastasis. Therefore, there is an urgent need to develop new therapeutic agents to target tumor plasticity.

The cancer stem cells (CSCs) model offers one explanation for tumor plasticity. The CSCs model revealed that only a minority of tumorigenic cells contribute to tumor growth and progression. However, there are many other aspects closely related to tumor plasticity. For example: (1) epithelial-mesenchymal transition (EMT), which contributes to the phenotypic plasticity and promotes cancer metastasis; (2) tumor microenvironment (TME), which contains extracellular matrix (ECM) and cells such as fibroblasts, endothelial and immune cells that are the primary source of signals to and from the tumors; (3) vasculogenic mimicry (VM), which is a microcirculation that is independent of angiogenesis in aggressive primary and metastatic tumors

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Aamir Ahmad, Mitchell Cancer Institute, United States Xiaojun Yang, Shantou University Medical College, China*

> \*Correspondence: *Dun Wang*

*wangduncn@hotmail.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *24 July 2019* Accepted: *27 August 2019* Published: *10 September 2019*

#### Citation:

*Yang X-G, Zhu L-C, Wang Y-J, Li Y-Y and Wang D (2019) Current Advance of Therapeutic Agents in Clinical Trials Potentially Targeting Tumor Plasticity. Front. Oncol. 9:887. doi: 10.3389/fonc.2019.00887*

**48**

and comprised of non-endothelial cell generated by tumor cells and ECM; and (4) neoplastic metabolic pathways, that mainly include glycolysis and oxidative phosphorylation (OXPHOS). Changes of metabolic pathways between glycolysis and OXPHOS in cancer cells is prevalent during tumorigenesis and metastasis. Hence, targeting glycolysis and OXPHOS is essential to wipe out metabolic plasticity in cancer cells. Here, the potential targets related to tumor plasticity was summarized in **Figure 1**. In this mini review, we summarize the recent advances in anticancer compounds targeting CSCs, ETM, TEM, VM formation, and metabolic pathways, which is associated with tumor plasticity.

# THERAPEUTIC TARGETING OF CSCS

The concept of CSCs was proposed several decades ago. The existence of CSCs has been confirmed by lineage tracing and cell ablation experiments in tumors (2–6). Similar to normal stem cells, a small subset of CSCs could proliferate and differentiate into a wide variety of cell types to sustain and promote tumor growth. The characteristic of tumor plasticity in CSCs is that CSCs could differentiate in different directions. The CSCs model provides a new explanation for the metastasis and recurrence of malignant tumors. CSCs have also been recognized as a major driver of tumor growth, metastasis and chemotherapeutic resistance. Therefore, CSCs has become crucial targets for cancer treatment. The ways to eliminate CSCs mainly consist of two aspects (7): (1) inhibition of key CSCs signaling pathways, including Wnt pathway, porcupine (PORCN) pathway and Hedgehog (Hh) pathway (8, 9); and (2) ablate CSCs by targeting CSC surface markers, such as CD133, CD44, (leucine-rich repeat containing G protein-coupled receptor 5) LGR5, (aldehyde dehydrogenase1) ALDH1, and breast cancer-resistant protein (BCRP; ABCG2). **Table 1** summarizes drugs that target CSCs in recent clinical trials.

Therapeutic agents targeting Wnt signaling pathway in clinical trials include porcupine (PORCN) inhibitors, β-catenin inhibitors and antibodies against Wnt signaling molecules (24). Among these, PORCN inhibitors gradually became research focus of antitumor drugs. **WNT-974**, an orally first-in-class PORCN inhibitor, is a pyridinyl acetamide derivative that target Wnt signaling to inhibit the expression of Wnt related genes and Wnt-dependent LRP6 phosphorylation. **WNT-974** showed significant growth inhibitory effect on Wnt-driven neoplasms, such as pancreatic cancer and head and neck squamous cell carcinoma. The pharmacokinetics (PK) and pharmacodynamics (PD) of **WNT974** were tested in patients with advanced cancers in phase I clinical trial, and the results showed rapid absorption (median Tmax 1–3 h) and appropriate elimination half-life of 5– 8 h. These clinical data demonstrated that **WNT-974** possesses favorable safety profile and potential antineoplastic activity in selected populations (25). Currently, **WNT-974** is being tested in a phase I study for the treatment of solid tumors including colorectal cancer and melanoma (10). In addition, PORCN inhibitor **ETC-159** is in phase I clinical trial for advanced solid


TABLE 1 | Potential drugs targeting CSCs in clinical trials.

tumors, **CGX-1321** is in phase I clinical trials for advanced gastrointestinal cancers and **RXC-004** is in phase I/II clinical trials for the treatment of solid tumors (11–14). Through the inhibition of β-catenin, **Tegavivint** (BC-2059), an anthraquinone derivative and **E-7386** are both being evaluated in phase I clinical trials to treat symptomatic or progressive unresectable desmoid tumors and solid tumors (15, 16).

The small-molecule inhibitors and macromolecule monoclonal antibodies (mAbs) including γ-secretase inhibitors and mAbs to NOTCH receptors have been tested in clinical trials. A small-molecule inhibitor of γ-secretase, which is a key enzyme in NOTCH signaling pathway, **AL-101** with favorable in vitro potency and oxidative metabolic stability, is in phase II clinical development for the treatment of adenoid cystic carcinoma bearing NOTCH activated mutations (17). On the other hand, among the therapeutic molecules targeting Hh pathway, smoothened (Smo) receptor antagonists are the most promising molecules (26). A novel small-molecule inhibitor or antagonist of Smo, **Sonidegib phosphate** was launched in 2015 for the treatment of advanced basal cell carcinoma (BCC). **Sonidegib phosphate** exhibited excellent therapeutic TABLE 2 | Potential drugs targeting CSC surface marker in clinical trials.


effect (roughly 35–60% response rates of patients) in patients with locally advanced, unresectable and metastatic BCC, with high disease control rates and clinical benefit (19, 27). Recent advances in the development of Hh signaling inhibitors include **Vismodegib** (18), which is launched in 2012 for the treatment of patients with advanced BCC; **Patidegib**, which is in phase III clinical trial for reducing the incidence of BCC (20, 21) and **Taladegib**, which is in phase I/II clinical trial) for the treatment of patients with recurrent, advanced solid tumors (22, 23).

Because of the highly plasticity of CSCs in tumors, the identification and eradication of CSCs are difficult. Generally, their identification depends on cell surface markers. CD34, CD44, and CD133 are common examples of CSC-specific surface markers (28). CSC surface markers can mediate adhesion of the cells. A cell surface membrane protein CD133, which was first discovered in hematopoietic stem and progenitor cells, is considered to be one of the common surface markers in multiple stem cells (29). Others like ALDH1 and ABCG2 also play significant roles in the regulation of CSCs (30, 31). Because CSCs drive cancer development, a number of agents targeting the biomarkers of CSCs have been developed (**Table 2**).

A novel mAb **P5**, which targets CD49e/CD29, is currently being tested in phase III clinical trials to evaluate its antitumor effect, but there are only a few reports about its progress of new clinical trials (32). As a FK506 binding protein like (FKBPL) peptide derivative, **ALM-201** can bind to CD44 and inhibit cancer related pathways, such as DLL4/NOTCH signal pathway as well as inhibit cell migration, tubule formation and angiogenesis. **ALM-201** showed an excellent safety profile and acceptable PK in patients with advanced solid tumors in a phase I dose-escalation study (39). This candidate is currently in phase I clinical trials for the treatment of patients with advanced ovarian cancer and other solid tumors (33). **RO-5429083** and **RG-7356**

are both humanized monoclonal antibodies against extracellular domain of CD44 which had been used in phase I clinical studies for the treatment of acute myeloid leukemia and solid tumors (34, 35). In addition, **AMC-303**, a high specific inhibitor of CD44v6, was evaluated as monotherapy to treat patients with advanced epithelial tumors. **AMC-303** was proved to be welltolerated with a favorable PK profile (t1/<sup>2</sup> of 4–7 h, CL of 40– 60 mL/h/kg) (40). At present, **AMC-303** is in phase I/II clinical trials to treat patients with advanced or metastatic malignant solid tumors of epithelial origin (36). A probody drug conjugate **CX-2009** against CD166 is in phase I/II clinical development for the treatment of adult patients with metastatic or locally advanced unresectable solid tumors (37). Furthermore, a recent research reported that **chrysin**, which is an ABCG2 inhibitor, could enhance sorafenib mediated inhibition of cell viability by sustained phosphorylation of ERK1/2 (41). And **chrysin** is being used in phase II clinical trials to treat CLL (38).

# THERAPEUTIC TARGETING OF EMT

The conversion of cells from epithelial phenotype into mesenchymal phenotype is a critical transformation for embryonic development and during cancer progression. Through EMT process, tumor cells can acquire the ability to disarm anti-tumor defenses in the body, resist apoptosis and antineoplastic drugs, spread through the body and expand the population of tumor cells (42). At the same time, EMT may play an important role in generating CSCs (43). Hence, EMT is an important target for inhibiting tumor metastasis and reducing drug resistance. Various approaches can be used to target the EMT process: (1) targeting the inducing signals in EMT process; (2) reversing EMT to reduce tumor cell aggressiveness; and (3) killing the cells in EMT-like state (44). As one of the key factors of tumor invasion, metastasis and drug resistance, EMT is a promising target for oncotherapy. The following summarized the progress of potential drugs targeting EMT-related signals (**Table 3**).

Modulators of transcription factors, such as nuclear factorkappa B (NF-κB) and signaling transducer and activator of transcription 3 (STAT3) have made progress in clinical trials (59, 60). **Denosumab**, which is a macromolecule of humanized mAbs to receptor activator of NF-κB ligand (RANKL), was originally approved to treat and prevent postmenopausal osteoporosis in 2010 (45). **Denosumab** prevents RANKL binding to RANK, and blocks the development of osteoclasts, leading to restraining the resorption of bone. So far, phase III clinical studies have been ongoing for evaluating its therapeutic effect on metastatic non-small cell lung cancer (NSCLC) together with other chemotherapeutics. **TK-006** is another anti-RANKL antibody in early clinical development for the treatment of patients with bone metastases caused by breast cancer through hypodermic injection (46). In addition, **WO-1066** is a JAK/STAT3 (the Janus kinase/signal transducer and activator of tran-ions 3) signaling pathway and programmed cell death-ligand 1 (PD-L1) inhibitor, which is derived from the JAK2 inhibitor AG490. In 2019, the compound was granted an orphan drug designation TABLE 3 | Potential drugs targeting EMT-related modulators in clinical trials.


in the U.S. for treating glioblastoma. Currently, the candidate is in phase I clinical trials for patients with melanoma or glioblastoma multiforme with brain metastases (47, 48). **DSP-0337**, **Danvatirsen** and **OPB-111077,** all inhibit STAT3 and are in phase I or II clinical trials to assess their therapeutic efficacy in solid tumors (49–52).

Hypoxia-inducible factor 1α (HIF1α) and β-catenin also regulate the expression of other transcription factors related to EMT (61, 62). **PEGPH20** (PEGylated recombinant human hyaluronidase PH20), which enzymatically degrades hyaluronic acid (HA), is currently being evaluated in phase II and III trials. It shows promising efficacy in preclinical and early clinical studies in the treatment of metastatic pancreatic carcinoma and other malignant tumors (55, 56). **CRLX-101** was proved to be a potent topoisomerase 1 and HIF1α inhibitor, which is a nanoparticle composed of CPT conjugated to a biocompatible copolymer of cyclodextrin and polyethylene glycol (PEG). Currently, CRLX101 is being evaluated in phase II clinical trials for several tumor types (58).

#### THERAPEUTIC TARGETING OF TME

Studies have shown that epigenetic changes of tumor cells caused by TME play a prominent role in tumor progression and invasion (1, 63). Tumor cells usually adapt to the changing external environment through changing the plasticity of tumor cells to meet the demand of tumor development. The research of relationship between TME and tumor plasticity is making progress in recent years (64). TME is composed of a complex Yang et al. Therapeutic Strategies Targeting Tumor Plasticity

mixture of ECM and various cells including cancer associated fibroblasts (CAFs) (65), cancer associated macrophages (CAMs) (66) and endothelial progenitor cells (EPCs) (67). Many components in ECM contribute to tumor growth. TME has become one of the key targets in tumor treatment due to its special pathophysiological characteristics and physicochemical properties (**Table 4**).

Tumor necrosis factor alpha (TNF-α) could promote tumor growth via a PKCa- and AP-1-dependent pathway (90).



**Avadomide** (**CC-122**) is a small molecule drug that inhibits both TNF-α and cereblon E3 ligase. The first-in-human phase I study, which evaluated the safety and clinical therapeutic effect of **avadomide** in patients with advanced solid tumors and others, showed acceptable safety and favorable pharmacokinetics (68). A**vadomide** is currently being evaluated in advanced melanoma in combination with Nivolumab. Transforming growth factorβ (TGF-β) signaling pathway is related to EMT in cancer cells (91). Therapeutic agents modulating the expression of TGF-β that are monoclonal antibodies include: **NIS-793** (a humanized anti-TGF-β monoclonal antibody), **AVID-200** (a recombinant inhibitor of TGF-β1 and TGF-β3), **SAR-439459** (targeting transforming TGF-β) and **fresolimumab** (a pan-specific human anti-TGF-β monoclonal antibody). Among these therapeutic agents, **fresolimumab** is able to neutralize all human isoforms of transforming TGF-β and being evaluated in phase I/II trials (72, 92).

Epidermal growth factor receptor (EGFR) regulates ECM and promotes cancer invasion (93). A small EGFR inhibitor **Simotinib** is used in phase I study to treat NSCLC (73). Platelet derived growth factor receptor alpha (PDGFRα), which contributes to fibroblast reprograming toward CAFs, plays a significant role in colorectal carcinogenesis (94). **Amcasertib**, a PDGFRα inhibitor and cancer stemness kinase inhibitor, is used to treat hepatocellular carcinoma and cholangiocarcinoma in phase II trials (74). Different from **Amcasertib**, **Lartruvo**(**R**) (**olaratumab**) is a fully humanized monoclonal antibody to neutralize PDGFRα. It was first launched in the U.S. for front-line treatment with doxorubicin in adults with soft tissue sarcoma in 2016 (75).

Some signaling pathways are also critical in cancer development. Janus kinase 1 (JAK1)/Rho kinase1 (ROCK1) signaling could promote fibroblast-dependent carcinoma cell invasion (95). **Cerdulatinib** is a small-molecule anti-cancer drug targeting JAK and syk kinase for the treatment of hematologic cancers (76). Liver kinase B1 (LKB1)/mammalian target of rapamycin (mTOR) signaling axis regulates ECM stiffness and participates in lung adenocarcinoma progression (96). Potential drugs such as **AZD-8055**, **BI-860585**, **DCBCI-0901**, **LXI-15029,** and **ABI-009** are in early clinical stage for various cancers (77–81). **Sapanisertib** is an orally and highly selective ATP-competitive inhibitor of mTORC1/2 and demonstrates satisfactory anticancer activity. The phase II study of **sapanisertib** in metastatic castration resistant prostate cancer was not entirely satisfactory likely because of dose reductions secondary to toxicity (82). In addition, abnormal expression of protein kinase B (PKB/Akt) is related to many cancers (97). **GSK-690693** (83), **ARQ-751** (84), and **TAS-117** (85) that can effectively treat solid tumors through inhibiting PKB/Akt are being evaluated in phase I and II clinical studies. **Ipatasertib** has been combined with other antitumor drugs to treat prostate cancer and breast cancer and is undergoing an investigation in a phase III clinical trial (86).

With the exception of targets above, interleukin-6 (IL-6) showed high expression in prostate cancer (98). **Siltuximab**, a chimeric monoclonal antibody, was first launched in 2014 to treat HIV-negative and Human Herpes Virus-8 negative multicentric Castleman's disease. Its tight binding to IL-6 inhibits IL-6 bioactivity and thus causes apoptosis of tumor cell. Recently, a phase II clinical trial of **siltuximab** was conducted for the treatment of multiple myeloma (87). Others like immunityrelated programmed cell death receptor-1 (PD-1) and PD-L1 inhibitors show satisfied antitumor effects by restoring antitumor immunity. **Sintilimab** is a fully human IgG4 mAb, which blocks the interaction of PD-1 with PD-L1 and PL-L2 (88). It was firstly approved in China to treat classical Hodgkin's lymphoma. **Avelumab**, an anti-PD-L1 antibody, was approved by the FDA in 2019 for first-line treatment of advanced renal cell carcinoma together with axitinib (89).

#### VM RELATED TARGETS AND THERAPEUTIC AGENTS

VM refers to a tumor microcirculation pattern that tumor cells aggregate, migrate and remodel to form a vascular-like structure based on the adhesion of ECM. VM differs from traditional endothelial tumor angiogenesis and plays a crucial role in tumor invasion and spreading. It is worth noting that there is an obvious increase of EMT-related regulators and transcription factors in VM, which indicates the crucial rule of EMT in VM formation (99). VM has been observed in a broad range of tumor types such as prostate cancer, malignant glioma, and melanoma (100). Currently, certain mechanism of VM formation remains matters of frenetic investigation and the mechanism of VM formation mainly include TME, EMT, tumor plasticity, RNA, and other regulators (100). Because VM is important for tumor progression, targeted therapies related to VM could also be a promising antitumor strategy to reducing tumor plasticity.

The major signaling molecules participating in VM formation and promising drugs are summarized in **Table 5**. Histone deacetylases inhibitor (HDACi) inhibits key molecule MMP-2 in PI3K-MMPs-Ln-5γ2 signaling pathway to block VM formation (112). **Panobinostat lactate**, which is lunched in 2015, is a firstline HDAC inhibitor applied in combination with bortezomib and dexamethasone to the treatment of multiple myeloma (113). **Panobinostat lactate** is not only a HDAC inhibitor but also a pan-deacetylase inhibitor. The pharmacokinetics of **panobinostat lactate** is affected by some factors such as hepatic impairment. HDAC inhibitor **romidepsin**, which is launched in 2010, could cause cell cycle arrest, differentiation and apoptosis in various cancer cells and is used for the treatment of cutaneous T-cell lymphoma (103). **OKI-179** and **remetinostat** are HDAC inhibitors in early clinical development (101, 102).

Phosphatidylinositide 3-kinases (PI3K) participate in VM formation by activating matrix metalloproteinases (MMPs) (114). The PI3Kα/δ inhibitor **copanlisib hydrochloride** was launched in 2017 as a treatment for relapsed follicular lymphoma in patients receiving two or more prior therapy regimens (110). **Copanlisib** characterizes low risk of PK-related pharmacological interaction due to reduced oxidation metabolism and unchanged excretion of copanlisib. Other PI3K inhibitors in clinical trials include **MEN-1611** (phase I for breast cancer), **HMPL-689** (phase I for B-cell lymphoma), **Gedatolisib** (phase II for acute myeloid leukemia and solid tumors), **GDC-0980** (phase II for prostate cancer) and **Buparlisib** (phase III in patients with head and neck squamous cell carcinoma, HNSCC) (105–109).

VE-cadherin mediates the activities of epithelial cell kinase (Eck/EphA2) to affect the formation of VM (115). EphA2 interacts with cell membrane surface ligands by phosphorylation and regulates the extracellular expression of protein kinases ERK and focal adhesion kinase FAK to activate PI3K (116, 117). **SiRNA-EphA2-DOPC** is a small interfering RNA targeting EphA2 loaded in neutral 1,2 dioleoyl-sn-glycero-3-phosphocholin (DOPC) liposomes (111). **SiRNA-EphA2-DOPC** reaches to tumor site by interacting with endothelial cells of tumor vasculature. As an EphA2 inhibitor, **siRNA-EphA2-DOPC** is in early clinical investigations to treat recurrent and advanced solid tumors.

#### THERAPEUTIC TARGETING OF NEOPLASM METABOLIC PATHWAYS

Cancer cells reprogram metabolic pathways by oncogenic mutations, result in enhanced demand of nutrient uptake to supply anabolic metabolism. Not only must energy production and consumption processes in cancer cells be balanced to

#### TABLE 5 | Potential drugs targeting VM in clinical trials.


TABLE 6 | Potential drugs targeting neoplasm metabolic pathways in clinical trials.


sustain tumor growth, but also cancer cells have to adapt to the changes in nutrition and oxygen supply caused by their rapid growth. Hence, malignant cells exhibit metabolic flexibility for them to exist and develop. Different from normal cells, cancer cells are more dependent on anaerobic glycolysis even in a sufficient oxygen supply environment, called Warburg effect (118). HIF-1α is crucial for anaerobic glycolysis under oxygen free conditions. Tumor suppressor liver kinase B1 (LKB1) regulates HIF-1α-dependent metabolic reprogramming (119). Recent studies have shown that Pyruvate kinase M2 (PKM2) plays a crucial part in the plasticity of cancer metabolism, and up regulation of PKM2 leads to oxidative metabolism (120). **Dimethylaminomicheliolide** (DMAMCL), a PKM2 activator, is a prodrug of micheliolide (MCL) that suppresses tumor growth and targets CSCs in the form of guaianolide sesquiterpene lactone. **Dimethylaminomicheliolide** could inhibit inflammation and tumor growth by releasing MCL into plasma. Early clinical trial using **Dimethylaminomicheliolide** for patients with solid tumors is being conducted (**Table 6**) (121).

In addition to this, oxidative phosphorylation plays an important role in cancer metabolism. Oxidative phosphorylation is mainly regulated by AMP-activated protein kinase (AMPK) (123). As an AMPK activator, **acadesine** increases the availability of adenosine in tissues under ischemic conditions and shows antitumor activity. **Acadesine** causes B cells apoptosis selectively in chronic lymphocytic leukemia (CLL) and phase I/II studies are being tested for sieving out the best methods for the treatment of resistant/refractory B-cell chronic lymphocytic leukemia (122).

#### REFERENCES


Metabolic plasticity of cancer triggers the adaptive "metabolic switch" needed for cancer development. Mechanism of metabolic switch provides insights into therapies, which could be used to target cancer development.

#### CONCLUSIONS

Tumor plasticity provides new explanation for the mechanisms of drug resistance, metastasis and recurrence of neoplasm. Interfering tumor plasticity is becoming strategies to treat malignant tumors. The drugs in clinical trials that targeting tumor plasticity are still on intense research. However, targeted therapy also has some limitations that most drugs could only be effective on a small part of tumors of genetic transformation and engender drug resistance after a period of time of taking drugs. How to find effective multi-targeted inhibitors or combine with traditional chemotherapeutic drugs and other therapeutics like photodynamic or photothermal therapy become particularly important. The quest for new therapeutic targets toward tumor plasticity continues to be a great impetus to promote cancer treatment.

# AUTHOR CONTRIBUTIONS

DW, X-GY, and L-CZ wrote the draft. Y-JW and Y-YL edited the manuscript. All authors read and approved the final version of manuscript.


patients (pts) with advanced solid tumours. Ann Oncol. (2017) 28:128. doi: 10.1093/annonc/mdx367.017


improves rectal cancer chemoradiotherapy by inhibiting DNA repair and HIF1α. Cancer Res. (2017) 77:112–22. doi: 10.1158/0008-5472.CAN-15-2951


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Yang, Zhu, Wang, Li and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# MicroRNA and mRNA Interaction Network Regulates the Malignant Transformation of Human Bronchial Epithelial Cells Induced by Cigarette Smoke

Jin Wang, Xiao-fan Yu, Nan Ouyang, Shiyu Zhao, Haiping Yao, Xifei Guan, Jian Tong, Tao Chen and Jian-xiang Li\*

*Department of Toxicology, School of Public Health, Medical College of Soochow University, Suzhou, China*

This study analyzes the correlation and interaction of miRNAs and mRNAs and their biological function in the malignant transformation of BEAS-2B cells induced by cigarette smoke (CS). Normal human bronchial epithelial cells (BEAS-2B) were continuously exposed to CS for 30 passages (S30) to establish an *in vitro* cell model of malignant transformation. The transformed cells were validated by scratch wound healing assay, transwell migration assay, colony formation and tumorigenicity assay. The miRNA and mRNA sequencing analysis were performed to identify differentially expressed miRNAs (DEMs) and differentially expressed genes (DEGs) between normal BEAS-2B and S30 cells. The miRNA-seq data of lung cancer with corresponding clinical data obtained from TCGA was used to further identify lung cancer-related DEMs and their correlations with smoking history. The target genes of these DEMs were predicted using the miRDB database, and their functions were analyzed using the online tool "Metascape." It was found that the migration ability, colony formation rate and tumorigenicity of S30 cells enhanced. A total of 42 miRNAs and 753 mRNAs were dysregulated in S30 cells. The change of expression of top five DEGs and DEMs were consistent with our sequencing results. Among these DEMs, eight miRNAs were found dysregulated in lung cancer tissues based on TCGA data. In these eight miRNAs, six of them including miR-96-5p, miR-93-5p, miR-106-5p, miR-190a-5p, miR-195-5p, and miR-1-3p, were found to be associated with smoking history. Several DEGs, including *THBS1*, *FN1*, *PIK3R1*, *CSF1*, *CORO2B*, and *PREX1*, were involved in many biological processes by enrichment analysis of miRNA and mRNA interaction. We identified the negatively regulated miRNA-mRNA pairs in the CS-induced lung cancer, which were implicated in several cancer-related (especially EMT-related) biological process and KEGG pathways in the malignant transformation progress of lung cells induced by CS. Our result demonstrated the dysregulation of miRNA-mRNA profiles in cigarette smoke-induced malignant transformed cells, suggesting that these miRNAs might contribute to cigarette smoke-induced lung cancer. These genes may serve as biomarkers for predicting lung cancer pathogenesis and progression. They can also be targets of novel anticancer drug development.

Keywords: cigarette smoke, BEAS-2B, miRNA-mRNA network, lung cancer, The Cancer Genome Atlas

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Chun Wu, Bristol Myers Squibb, United States Yi-Chao Zheng, Zhengzhou University, China*

> \*Correspondence: *Jian-xiang Li aljxcr@suda.edu.cn*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *25 July 2019* Accepted: *23 September 2019* Published: *09 October 2019*

#### Citation:

*Wang J, Yu X, Ouyang N, Zhao S, Yao H, Guan X, Tong J, Chen T and Li J (2019) MicroRNA and mRNA Interaction Network Regulates the Malignant Transformation of Human Bronchial Epithelial Cells Induced by Cigarette Smoke. Front. Oncol. 9:1029. doi: 10.3389/fonc.2019.01029*

**58**

# INTRODUCTION

Lung cancer is one of the most common carcinomas in men and women around the world. It is the first and second leading cause of cancer-related deaths in men and women, respectively (1, 2). There were 2.09 million new lung cancer cases and 1.76 million lung cancer deaths, which accounts for about 18.4% of all cancer deaths around the world in 2018 (3). The incidence of lung cancer is closely associated with cigarette smoking (2, 4). The risk of developing lung cancer in smokers is nearly ten times higher than that in non-smokers (5, 6). However, it is still not clear how normal lung epithelial cells become cancerous change in cigarette smokers.

It is well-known that the initiation and development of lung cancer are associated with abnormal expression of oncogenes and tumor suppressor genes. Numerous evidence suggested that the change in gene expression, which affects the occurrence and progression of tumors is closely related to epigenetic modification (7). Epigenetic modification could be DNA methylation, microRNAs (miRNAs), histone modifications, and nucleosome remodeling. These modifications are independent but could interact with each other to regulate gene expression (8). Epigenetic disruptions could promote the acquisition of a cancerous phenotype and aggressive behavior in lung cancer cells as well as primary or acquired resistance to treatment (9).

MiRNAs are highly conserved non-coding RNAs and consist of 18–24 nucleotides (nt) that are involved in the posttranscriptional regulation of gene (10). An individual miRNA is able to regulate many different transcripts. It is also believed that miRNAs can regulate more than one in three coding RNAs in the genome (11). MiRNAs participate in many vital biological processes through pairing with target mRNAs and regulating their expression (12, 13). The imbalance of miRNAs is usually associated with the progression and suppression of cancer, suggesting that miRNAs may play important roles as oncogenes or tumor suppressors (14).

The rapid development of high-throughput next-generation sequencing technology made it possible to identify changes in single bases in coding sequences of specific genes during lung tumorigenesis. A meticulous and thorough analysis of these data identified various vital genes and signaling pathways related to the tumor resulting in a better understanding of the mechanism of occurrence, development, and prognosis of lung cancer. Using novel technology and bioinformatics analysis, The Cancer Genome Atlas (TCGA, http://cancergenome.nih.gov/) project has previously identified panels of genetic mutations contributed to or were associated with the cause of a variety of cancers (15). Recently, the TCGA had shown studies on lung adenocarcinoma (LUAD) and squamous cell carcinomas (LUSC) at the molecular level (16, 17).

The aim of this study is to analyze the correlation and regulating mechanism of the regulatory network of miRNAs and mRNAs during carcinogenesis. An in vitro cell model of malignant transformation was established by exposing normal lung epithelial cells BEAS-2B to cigarette smoke (CS). Using high-throughput sequencing analysis, we analyzed the miRNA and mRNA expression profile in BEAS-2B cells with or without CS exposure. The differential expression miRNAs (DEMs) and differentially expressed genes (DEGs) were selected, and the integrative miRNA-mRNA network was analyzed. We identified some critical genes involved in carcinogenesis. This study will provide potential target candidates for novel drug development.

# METHODS

### Preparation of Malignantly Transformed Cells

The CS-exposed malignant transformed cell model was established as described previously. Briefly, exponentially growing BEAS-2B cells were plated onto transwell membrane (Corning, USA) with 1 × 10<sup>5</sup> in a single well (18). CS was produced using an automatic smoking machine, and the CS was pumped into an inhalation exposure chamber. Cells were directly exposed to CS for 10 min every day at a smoke concentration of 20%. This procedure was continued until the cells reached 30 passages and named S30 cells (18).

# Scratch Wound Healing Assay

The normal BEAS-2B cells and S30 cells (2 × 10<sup>5</sup> ) were seeded into 6 well plates and cultured at 37 ◦C. Cells were allowed to grow up to 100% confluence and a scratch was made in the plate using a P10 pipette. The cells were cultured in fresh serumfree DMEM medium. Images were collected at 0 and 24 h under an inverted microscope (Olympus, Germany) and quantitatively analyzed with NIH ImageJ software.

#### Transwell Migration Assay

The normal BEAS-2B cells and S30 cells (2 × 10<sup>5</sup> ) were seeded in the upper chambers (pore size, 8µm) of the 6 well plate (Corning, USA) in 1 ml serum-free medium. The lower chambers were filled with 2 ml complete medium with 10% FBS, and the plate was incubated under standard conditions for 24 h. At the end of incubation, after removing the cells in the upper surface of the membrane with a cotton swab, cells in the lower chamber were fixed with methanol and stained with 0.5% crystal violet solution. The images were taken with an inverted microscope (Olympus, Germany) and analyzed using NIH ImageJ software.

# Colony Formation Assay

1 × 10<sup>3</sup> normal BEAS-2B cells and S30 cells were plated in 0.35% agarose on top of a 0.7% agarose base supplemented with complete medium. The medium was renewed every 2–3 days. The colonies were stained with 0.5% crystal violet (Sigma, USA) for 20 min at room temperature. The colony formation rate was calculated by the following equation: colony formation rate = the number of colonies/number of seeded cells × 100%.

#### Tumorigenicity Assay

Five-week-old male BALB/c-nude mice of SPF grade were purchased from Beijing Vital River Laboratory Animal Technology Company Limited (Beijing, China). All nude mice were housed in the Laboratory Animal Center Soochow University. The animal experiment protocol was approved by the Laboratory Animal Ethics Committee of Experiment Animal Center of the Soochow University (Suzhou, China). Approximately 5 × 10<sup>6</sup> normal BEAS-2B cells or S30 cells were injected subcutaneously into the right flank of male BALB/c-nude mice (5 mice were used for BEAS-2B cells injection and 10 mice for S30 cells injection). Animals were euthanized 45 days after injection, and tumors were collected and photographed.

#### RNA Extraction and Sequencing

Total cellular RNA was extracted from S30 and normal BEAS-2B cells using TRIzol reagent (Invitrogen, USA) according to the manufacturer's protocol. Small RNA sequencing was performed on the Illumina Hiseq 2500 platform (Illumina, San Diego, CA). NEBNext <sup>R</sup> Multiplex Small RNA Library Prep Set for Illumina <sup>R</sup> (NEB, USA.) was used to prepare the small RNA sequencing library. To determine the known and novel miRNAs, unique clustered reads were aligned with the reference genome and database obtained from miRBase 20.0 (http://www.mirbase.org/). The miRDeep2 algorithm was used to predict novel miRNA precursors. The expression levels were estimated by transcript per million (TPM) and mRNA sequencing was performed on the Illumina HiSeq 4000 platform. The Illumina TruSeq RNA kit was used for preparing the mRNA sequencing library. The mRNAs with expression profiles that differed between the samples were normalized as fragments per kilobase of transcript per million mapped reads (FKPM). The DEGSeq package was used to analyze the differential expressed miRNAs (DEMs) or mRNAs (DEGs). P-value < 0.05 and |log2 (foldchange)| ≥ 1 were regarded as the threshold of significantly differential expression.

#### Data Source and Processing

The NSCLC (LUAD and LUSC) miRNA-Seq datasets and related clinicopathology information were obtained from the


*Reverse primers of miRNAs and U6 primers are provided by Mir-XTM miRNA First-Strand Synthesis Kit.*

Xena (https://xena.ucsc.edu). The LUAD miRNA expression data included a total of 493 samples consisting of 448 LUAD samples and 45 normal adjacent samples. The LUSC miRNA expression data included a total of 380 samples comprising 336 LUSC samples and 44 normal adjacent samples. The limma package was used to identify the differential expressed miRNAs in LUAD and LUSC when compared with corresponding normal adjacent samples. The differentially expressed miRNAs were defined by a threshold of p-value < 0.05 and |log2 (foldchange)| ≥ 1.

#### Real-Time Quantitative PCR

A total of 1.5 µg RNA isolated from each sample was reversely transcribed into complementary DNA (cDNA) using Revert Aid First Strand Complementary DNA Synthesis Kit (for mRNA detecting, Thermo, USA) or Mir-XTM miRNA First-Strand Synthesis Kit (for miRNA detecting, Clontech Laboratories, USA) according to the manufacturer's instructions. Quantitative PCR (qPCR) was performed using NovoScript <sup>R</sup> SYBR Two-Step qRT-PCR Kit (novoprotein, China) with a QuantStudioTM 6 Flex qRT-PCR system (USA). The internal control for the quantitive analysis of miRNA and mRNA were U6 and GAPDH, respectively. The primer used for qPCR were listed in **Table 1**.

#### Analysis of Gene Expression and Smoking History

To validate the correlation between expression of miRNAs and patients' smoking history, all valid LUAD samples in the TCGA database were divided into four groups according to smoking history, including (1) lifelong non-smoker (n = 66); (2) current smokers (n = 104); (3) Current reformed smoker for >15 years (n = 116); (4) current reformed smoker for ≤15 years (n = 144). The expression of miRNAs in lung adenocarcinoma tissues of each group was compared.

#### miRNA-mRNA Integrative Network

For identification of the candidate miRNA-mRNA network in smoking-induced malignant transformed cells, two separate steps were carried out. First, the target mRNAs of interest miRNAs were predicted through the miRDB database (http://mirdb.org/ miRDB/). Second, the intersection of differently expressed genes and target genes was taken to screen the potential target genes of miRNAs in smoking-induced malignant transformed cells. These different expression target genes and miRNAs were used to construct the miRNA-mRNA regulation network through the Cytoscape software (V3.7.1, https://cytoscape.org).

#### Enrichment Analyses

Metascape (http://metascape.org/gp/index.html) is an effective and efficient tool for experimental biologists to comprehensively analyze and interpret OMICs-based studies in the big data era (19). The database was used to perform the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, which is used to predict the potential biological functions of the overlapping genes of the DEGs and target genes.

FIGURE 1 | CS-induced malignant transformation in BEAS-2B cells *in vitro* and *in vivo*. (A) Representative photographs of the colony formation assay of the normal BEAS-2B cells and S30 cells. (B) Graph of soft agar colony forming rate of normal BEAS-2B cells and S30 cells. \*\**p* < 0.01 vs. BEAS-2B. (C) Photographs of tumors excised 45 days after injection of normal BEAS-2B cells and S30 cells into nude mice. (D) Representative HE staining histopathologic image of tumor tissues excised 45 days after injection of S30 cells into nude mice.

#### Statistical Analysis

SPSS 22.0 was used for statistical analysis. Values were presented as mean ± standard deviation (SD). Difference analysis between two groups was performed by using student t-test. A p < 0.05 was considered statistically significant. Correlation between differentially expressed miRNAs and predicted target mRNAs were calculated by Pearson's correlation. A p < 0.05 was regarded as statistically significant.

# RESULTS

#### CS-Induced Malignant Transformation in BEAS-2B Cells

To validate the malignant change of S30 cells, the normal BEAS-2B cells and S30 cells were seeded in soft agar. As shown in **Figure 1A**, cells formed significantly more and bigger colonies in S30 cells compared to the normal BEAS-2B cells. Besides, colony formation rate in S30 cells was remarkably higher than in the normal BEAS-2B cells (**Figure 1B**). Moreover, the normal BEAS-2B cells and S30 cells were used to generate xenograft tumors in nude mice. Among the ten mice injected with S30 cells, 3 developed tumor tissue (**Figures 1C,D**). While no tumor was found in the five mice injected with normal BEAS-2B cells.

# CS Promoted the Migratory Ability of BEAS-2B Cells

To investigate the effect of CS in cell migration, we performed scratch wound healing and transwell cell migration assays. Scratch wound healing assay indicated that the migratory ability was significantly increased in S30 cells compared to the normal BEAS-2B cells (**Figures 2A,B**). As shown in **Figures 2C,D**, further transwell migratory assay demonstrated that the migrated cells were significantly increased in S30 cells compared to the normal BEAS-2B cells. These results suggested that longterm exposure to CS could promote the migratory ability of BEAS-2B cells.

### Differentially Expressed miRNAs Between S30 Cells and Normal BEAS-2B Cells

To test whether CS affects the miRNA-mRNA regulatory network in BEAS-2B cells, the miRNAs in normal BEAS-2B cells and S30 cells were quantitatively analyzed using small RNA sequencing. Compared with the normal BEAS-2B cells, the S30 cells showed dysregulation of 42 miRNAs that had significantly different expression levels with 2-fold change as a cut off (**Figures 3A,B**, **Supplementary Table 1**). Of these 42 miRNAs, 28 were upregulated (67%), and 14 were downregulated

FIGURE 3 | Graphical representation of the 42 miRNAs differentially expressed between S30 cells and normal BEAS-2B cells. (A) Heatmap of the 42 differentially expressed miRNAs (DEMs) between the CS-induced malignant transformed cells (S30) and normal BEAS-2B cells. The colors in the heatmap represent the normalized expression values, with low expression values being colored in shades of green and high expression values in shades of red. (B) Volcano plots were generated to visualize the distribution of DEMs between normal BEAS-2B and S30 cells. The top five most significantly dysregulated miRNAs are marked. (C) Counts of miRNAs upregulated or downregulated in S30 cells.

(33%) in the S30 cells (**Figure 3C**). The top five most significantly aberrant expression miRNAs are marked in the scatter plot; miR-106b-5p, miR-589-5p, and miR-96-5p were upregulated, and miR-181a-5p and miR-361-3p were downregulated (**Figure 3B**). The qPCR results of the top five miRNAs showed the increased miR-106b-5p, miR-589-5p, and miR-96-5p and decreased miR-181a-5p and miR-361-3p expression in S30 cells compared to normal BEAS-2B cells (**Figure 4**).

# Differentially Expressed mRNAs Between S30 Cells and Normal BEAS-2B Cells

Next, we investigated the expression patterns of mRNAs using transcriptome resequencing. Compared with the normal BEAS-2B cells, the S30 cells showed dysregulation of 753 mRNAs that had significantly different expression levels with 2-fold change as a cut off (**Figures 5A,B**). Of these 753 mRNAs, 273 were upregulated (36%), and 480 were downregulated (64%) in the S30 cells (**Figure 5C**). The top five most significantly dysregulated mRNAs are marked in the scatter plot; IGFBP3 and KRT17 were upregulated, and FAM129A, FLNC, and TIE1 were downregulated (**Figure 5B**). The qPCR results of the top five mRNAs validated the increased IGFBP3 and KRT17 and decreased FAM129A, FLNC, and TIE1 expression in S30 cells compared to normal BEAS-2B cells (**Figure 6**).

#### Integrated Analysis of the DEMs in S30 Cells and Lung Cancer Samples

To explore whether the DEMs' expression is altered in lung cancer tissues, we analyzed the miRNAs sequencing data of lung cancer, including lung adenocarcinomas (LUAD) and squamous cell lung carcinomas (LUSC), in the TCGA database. A total of 8 miRNAs were found dysregulation in S30 cells, LUAD and LUSC samples with a similar tendency of change. Among these 8 miRNAs, 5 were upregulated (miR-96-5p, miR-93- 5p, miR-589-5p, miR-4661-5p, and miR-106b-5p) and 3 were downregulated (miR-190a-5p, miR-195-5p, and miR-1-3p) in the three datasets (**Figure 7**).

#### Association of miRNA Expression With Smoking History

Among the five up-regulated miRNAs, three miRNAs, including miR-96-5p, miR-93-5p, and miR-106-5p, showed a higher

expression in current smoking LUAD patients when compared with the lifelong non-smokers (**Table 2**). Three of the screened down-regulated miRNAs, including miR-190a-5p, miR-195-5p, and miR-1-3p, showed lower expression in current smoking LUAD patients when compared with the lifelong non-smokers (**Table 2**). Moreover, miR-96-5p and miR-106b-5p are overexpressed in the current reformed smoker for >15 years, while miR-190a-5p has lower expression in the current reformed smoker for >15 years when compared with the lifelong non-smoker (**Table 2**).

#### Integrative Analysis of Correlation of miRNA and mRNA in S30 Cells

To understand the potential functions of the smoking-related differentially expressed miRNAs, and to explore miRNA-mRNA interaction in S30 cells, we predicted the target genes of miRNAs and performed an intersection analysis with the gene expression data to identify genes that were inversely co-expressed with miRNAs. A total of 2,477 target genes of low-expressed miRNAs and 2,295 target genes of high-expressed miRNAs were screened by searching miRDB database. Consequently, 25 upregulated genes and 70 down-regulated genes were found to have at least one negatively regulated miRNA-mRNA pair for smoking-related differentially expressed miRNAs (**Figure 8A**, **Supplementary Table 2**). The miRNAs-DEGs network was generated by Cytoscape software, as showed in **Figure 8B**.

#### Enrichment Analysis of Correlation of miRNA and mRNA in S30 Cells

To further explore the function of the negatively correlated miRNA-mRNA pairs, 95 up-regulated or down-regulated target genes in S30 cells were selected for mapping into the Metascape database and subjected to functional enrichment analysis. As shown in **Figure 9A**, GO analysis demonstrated that these target genes are associated with several cancerrelated, especially tumor migration related GO terms, including "positive regulation of cell motility," "regulation of cell adhesion," "mononuclear cell migration," "cell junction organization," "extracellular structure organization" and so on. Among these enriched DEGs, several DEGs, including THBS1, FN1, PIK3R1, CSF1, CORO2B, and PREX1, were involved in many biologic processes which derived from enrichment analysis of negative miRNA-mRNA correlations (**Figure 9B**). Moreover, the KEGG pathway enrichment analysis suggested that these target genes were significantly correlated with "TNF signaling pathway," "Small cell lung cancer," "Rap1 signaling pathway," "PI3K-Akt signaling pathway," "mTOR signaling pathway," "FoxO signaling pathway," "Focal adhesion," "ECM-receptor interaction," and some other cancer-related pathways (**Figure 10A**). In particular, "Focal adhesion" and "ECM-receptor interaction" are closely related to cell migration. In addition, THBS1, FN1, PIK3R1, and IRS1, were involved in many KEGG pathways which derived from enrichment analysis of negative miRNA-mRNA correlations (**Figure 10B**).

# DISCUSSION

There are nearly 1.3 billion cigarette smokers in the world, which leads to 5 million cancer related deaths every year (20). Cigarettesmoking is a notable risk factor for multiple pathologies (21–23), among them lung cancer takes the lead with smokers having a much higher risk than non-smokers. Our previous studies have

FIGURE 7 | Identification of differential expressed miRNAs in S30 cells and lung cancer samples. DEM\_UP/DEM\_DOWN: up-regulated/ down-regulated miRNAs in S30 cells; LUAD\_UP/LUAD\_DOWN: up-regulated/ down-regulated miRNAs in LUAD samples; LUSC\_UP/LUSC\_DOWN: up-regulated/ down-regulated miRNAs in LUSC samples.


TABLE 2 | The expression of miRNAs in the LUAD patients with different smoking history.

*Lifelong non-smoker (*<*100 cigarettes smoked in Lifetime)* = *1; Current smoker (includes daily smokers and non-daily smokers or occasional smokers)* = *2; Current reformed smoker for* > *15 years (*>*15 years)* = *3; Current reformed smoker for* ≤*15 years (*≤*15 years)* = *4 (*\**p* < *0.05 vs. non-smoker;* \*\**p* < *0.01 vs. non-smoker).*

suggested that chronic exposure to CS could induce malignant transformation of the human bronchial epithelial cell line (BEAS-2B) and tumorigenesis (18, 24). In recent years, studies have indicated that miRNAs play an essential role in tumor initiation, development, and metastasis as well as the cellular response to stress by modulating the expression of their target genes (25–27). In this study, we investigate the effect of chronic exposure to CS on the expression of miRNA and mRNA in BEAS-2B cells and further examined the interaction of miRNA and mRNAs.

Based on our high throughput sequencing data and the TCGA database analysis, we found significant dysregulation of 6 smoking-related miRNAs in S30 cells compared with normal BEAS-2B cells. Among these miRNAs, miR-190a is found downregulated in aggressive neuroblastoma (NBL). Overexpression of miR-190a contributed to the inhibition of tumor growth and prolonged the dormancy period of fastgrowing tumors (28). A recent study showed that miR-190a could inhibit the metastasis of breast tumor by involving in estrogen receptor (ERα) signaling (29). miR-195 usually serves as a tumor suppressor in several cancer types, such as gastric cancer (30), renal cancer (31), cervical cancer (32), liver cancer (33), and osteosarcoma (34), and its downregulation was related to lymph node metastasis and advanced clinical stage (32). Similarly, miR-1 can regulate multiple behavior of the tumor cells, such as proliferation (35, 36), migration (37), apoptosis (38, 39), and metabolism (40). In addition, miR-106b and miR-93 are both the members of miR-106b∼25 cluster, which have regarded as significant oncogenic drivers as well as potential biomarkers and therapeutic targets in various tumors (41–44). Moreover, Several studies have demonstrated that miR-96 could act as an oncogene (45–47) or tumor suppressor (48, 49) depending on the different types of cancer. Although these miRNAs have extensively been reported to be associated with many other kinds of cancer, their roles in lung cancer has yet been demonstrated.

Numerous studies have established the regulatory relationships between miRNA and mRNA expression (50, 51). CS-induced DEMs can bind to 3′UTR regions of several genes and down-regulate their expression, indicating that these miRNAs may contribute to the pathogenesis of smoking-related diseases. It has been reported that negatively regulated miRNAmRNA pairs are significantly contributed to the initialization and development of different kinds of cancer (52–54). In order to further comprehend the roles of the miRNA-mRNA pairs in CS-induced lung cancer, we selected 95 dysregulated target mRNAs of the 6 CS-related miRNAs and found that they are involved in several cancer-related signaling pathways including TNF signaling pathway, Rap1 signaling pathway, PI3K-Akt signaling pathway, mTOR signaling pathway, FoxO signaling pathway, ECM-receptor interaction, and so on. Meanwhile, the GO enrichment analysis results indicated that these target genes were participated in a series of cell adhesion and migration biological processes, suggesting these miRNA-mRNA pairs related to the process of epithelial-mesenchymal transformation (EMT). EMT is considered to be an important regulator of metastasis by promoting the invasion and spread of tumor cells to distant organs (55). Among these enriched DEGs, IRS1, PIK3R1, THBS1, and FN1 are related to more than 4 KEGG pathways. As an adaptor of the insulin growth factor-1 receptor, IRS1 plays an essential role in cell growth and proliferation, primarily via the Akt pathway, and it was reported to be regulated by several miRNAs through direct or indirect action (56–59). Moreover, studies have demonstrated that PIK3R1 was directly targeted by miR-127 (60), miR-21 (61), miR-155 (62), and some other miRNAs in different kinds of cancers. It's reported that the activity of phosphoinositide 3- kinase (PI3K) is activated by many oncogenes and the PI3K family members are involved in a serious of biological processes and the genesis and progression of various tumors (63). Thrombospondin 1 (THBS1) is a secreted protein with multiple biological functions (64), including a potent anti-angiogenic activity and activation of latent transforming growth factor beta (TGF-β) (65, 66). A recent study suggested that the expression of THBS1 was modulated by multiple miRNAs (67). Moreover, it's reported that fibronectin 1 (FN1) is crucial to many cellular processes, including cell proliferation, adhesion, migration and differentiation (68, 69), and the expression of FN1 was regulated by miR-1271 (70), miR-9 (71), and miR-206 (72). Similar to previous studies, we identified the negatively regulated miRNA-mRNA pairs in the CS-induced lung cancer, which were implicated in several cancer-related (especially EMT-related) biological process and KEGG pathways in the malignant transformation progress of lung cells induced by CS. Further study will be needed to explore

pairs within the four datasets. (B) Regulatory network for 95 negatively correlated miRNA-mRNA, including 25 DEG\_UP/DEM\_DOWN\_Targets pairs and 70 DEG\_DOWN/DEM\_UP\_Targets pairs. DEG\_UP/DEG\_DOWN: up-regulated/down-regulated genes in S30 cells; DEM\_UP/DEM\_DOWN: up-regulated/ down-regulated miRNAs in S30 cells.

the targeting relationships of these miRNAs and their target mRNAs and their possible roles on cancer-related molecular mechanisms for the development of novel targeted therapy for CS-induced lung cancer.

In conclusion, our study demonstrated that the expression profiles of miRNA and mRNA were significantly dysregulated in BEAS-2B cells with long-term exposure to CS. Smoking induced miRNAs are associated with EMT and carcinogenesis.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study can be found in the Sequence Read Archive (SRA) database (https://trace.ncbi.nlm. nih.gov/Traces/sra/) with identifier SRP182926 and SRP181756. The LUAD and LUSC datasets analyzed for this study can be obtained from UCSC Xena (https://xenabrowser.net/datapages/).

#### ETHICS STATEMENT

The animal study was reviewed and approved by The Laboratory Animal Ethics Committee of Experiment Animal Center of the Soochow University. Written informed consent was obtained from the owners for the participation of their animals in this study.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

JL conceived and designed the study. JW performed and analyzed the experiments. XY, NO, SZ, HY, and XG assisted to collect and analyze the data. JW wrote the manuscript. JT and TC were of immense help in the modification of the manuscript. All authors read and approved the final manuscript.

#### FUNDING

This study was supported by the National Natural Science Foundation of China (81573178 and 81172707) and the Suzhou Science and Technology Project (SS201832). The study was also supported by Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases as well as the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2019.01029/full#supplementary-material


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wang, Yu, Ouyang, Zhao, Yao, Guan, Tong, Chen and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# MiR-199a Inhibits Tumor Growth and Attenuates Chemoresistance by Targeting K-RAS via AKT and ERK Signalings

Wei Li 1†, Lin Wang2†, Xiang-Bo Ji <sup>2</sup> , Li-Hong Wang<sup>2</sup> , Xin Ge<sup>3</sup> , Wei-Tao Liu<sup>3</sup> , Ling Chen<sup>1</sup> , Zhong Zheng<sup>1</sup> , Zhu-Mei Shi <sup>3</sup> , Ling-Zhi Liu<sup>4</sup> , Marie C. Lin<sup>2</sup> , Jie-Yu Chen<sup>1</sup> \* and Bing-Hua Jiang2,3,4 \*

<sup>1</sup> Department of Pathology, Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, China, <sup>2</sup> Institute of Medical and Pharmaceutical Sciences, The Academy of Medical Sciences, Zhengzhou University, Zhengzhou, China, <sup>3</sup> Key Laboratory of Human Functional Genomics of Jiangsu Province, Department of Pathology, Nanjing Medical University, Nanjing, China, <sup>4</sup> Department of Pathology, Carver College of Medicine, The University of Iowa, Iowa, IA, United States

#### Edited by:

Pascale Cohen, Université Claude Bernard Lyon 1, France

Reviewed by:

Jessica Dal Col, University of Salerno, Italy Vishwa Jeet Amatya, Hiroshima University, Japan

#### \*Correspondence:

Jie-Yu Chen crucianfish@hotmail.com Bing-Hua Jiang binghjiang@yahoo.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 09 July 2019 Accepted: 30 September 2019 Published: 15 October 2019

#### Citation:

Li W, Wang L, Ji X-B, Wang L-H, Ge X, Liu W-T, Chen L, Zheng Z, Shi Z-M, Liu L-Z, Lin MC, Chen J-Y and Jiang B-H (2019) MiR-199a Inhibits Tumor Growth and Attenuates Chemoresistance by Targeting K-RAS via AKT and ERK Signalings. Front. Oncol. 9:1071. doi: 10.3389/fonc.2019.01071 Glioma is the most malignant brain tumors in the world, the function and molecular mechanism of microRNA-199a (miR-199a) in glioma is not fully understood. Our research aims to investigate miR-199a/K-RAS axis in regulation of glioma tumor growth and chemoresistance. The function of miR-199a in glioma was investigated through in vitro and in vivo assays. We found that miR-199a in tumor tissues of glioma patients was significantly downregulated in this study. Kinase suppressor of ras 1 (K-RAS), was indicated as a direct target of miR-199a, as well as expression levels of K-RAS were inversely correlated with expression levels of miR-199a in human glioma specimens. Forced expression of miR-199a suppressed AKT and ERK activation, decreased HIF-1α and VEGF expression, inhibited cell proliferation and cell migration, forced expression of K-RAS restored the inhibitory effect of miR-199a on cell proliferation and cell migration. Moreover, miR-199a renders tumor cells more sensitive to temozolomide (TMZ) via targeting K-RAS. In vivo experiment validated that miR-199a functioned as a tumor suppressor, inhibited tumor growth by targeting K-RAS and suppressed activation of AKT, ERK and HIF-1α expression. Taken together, these findings indicated that miR-199a inhibits tumor growth and chemoresistance by regulating K-RAS, and the miR-199a/K-RAS axis is a potential therapeutic target for clinical intervention in glioma.

Keywords: miR-199a, glioma, K-RAS, chemoresistance, tumorigenesis

# INTRODUCTION

Malignant gliomas, as most common brain tumors around the world (1, 2), are classified according to their degree of malignancy as Grades I to IV (3, 4). Glioma clinical treatment includes surgery, chemotherapy and radiotherapy (5, 6). The most malignant grade IV glioma, glioblastomamultiforme (GBM), has an average life expectancy of only 15 months (7). Investigation of glioma carcinogenesis mechanisms would improve clinical diagnosis, drug therapy and prevention of glioma.

MicroRNAs (miRNAs) are a class of endogenous 18–22 nucleotides RNA molecules (8, 9), which always bind to the 3′ -untranslated region (UTR) of specific target mRNAs, and then regulate expression of several genes at the post-transcriptional level (10–14). Accumulated evidence has clearly demonstrated that aberrant miRNA expression profiles (15) and dysregulations of specific miRNAs and their target genes, are closely associated with tumor initiation and promotion in glioma (16, 17). In particular, the miR-199a has demonstrated to suppress tumor growth in a variety of cancers including esophageal, liver, and colorectal cancers. So far, the reported miR-199a downstream target genes include oncogenes PHLPP1, E2F3, FZD6/7, HK2, and MAP3K11 (18–23), which are functioned in pathogenesis of various cancers.

K-RAS, which is reported as a family member of Ras oncogene, has involved in regulation of some cellular signal transductions (24). K-RAS is implicated in the pathogenesis of various tumors, such as GBM and pilocytic astrocytoma (25, 26). Activation of K-RAS could promote the activations of several downstream molecules, such as MAPK and ERK to regulate biological processes (27, 28).

In this study for the first time in glioma, we demonstrated the loss of miR-199a, and that K-RAS is an important direct target gene of miR-199a. Results from in vitro studies in human glioma U87 and U251 cells indicated that forced expression of miR-199a downregulated K-RAS signaling and suppressed cancer development and Temozolomide (TMZ) chemoresistance. The forced miR-199a overexpression also inhibited AKT and ERK1/2 pathways, through K-RAS signaling. The in vivo studies further demonstrated that over-express of miR-199a exhibited reduced tumor growth with down-regulated K-RAS/AKT/ERK/HIF-1α signalings. These results suggested that the loss of miR-199a/K-RAS signaling in glioma plays a pivotal role in glioma progression, and it is a potential novel targets for future clinical treatment.

# MATERIALS AND METHODS

#### Specimen Collection

Human glioma specimens (n = 24) and normal brain tissues (n = 9) were collected from patients in Nanjing University Medical School, China. Samples were obtained from patients with informed consents and were histologically classified by clinical pathologist.

#### Cell Culture and Reagents

Human glioma cells (U87, U251) were cultured in DMEM medium. Antibodies against anti-GAPDH and anti-HIF-1α were purchased from Bioworld Technology (Atlanta, USA). Antibodies against antip-AKT (Ser473), anti-AKT, anti-p-ERK1/2 and anti-ERK1/2 were purchased from Cell Signaling Technology (Danvers, USA), and antibody against K-RAS was purchased from Santa Cruz (Santa Cruz, USA). TMZ (Sigma-Aldrich, USA) was used for in vitro chemosensitivity assay.

#### Real-Time PCR

RNAs were isolated from human specimens and cells using Trizol (Invitrogen, USA). To measure expression levels of miR-199a (U6 as internal control) and mRNA levels of K-RAS(GAPDH as internal control), the cDNAs were amplified by Real-time PCR with SYBR Green reagents (Vazyme, China) on a 7900HT system(Applied Biosystems), and fold changes were analyzed by relative quantification (2−11Ct).

Primers for K-RAS and GAPDH as below:

K-RAS: Forward Primer: ACAGAGAGTGGAGGATGCTTT, Reverse Primer: TTTCACACAGCCAGGAGTCTT; GAPDH: Forward Primer: ACAACTTTGGTATCGTGGAAGG, Reverse Primer: GCCATCACGCCACAGTTTC.

#### Immunoblotting

According to the manufacturer's instruction, cell lysates in this study were prepared using RIPA buffer and indicated protease inhibitors. Aliquots of protein lysates from treated cells were fractionated by SDS-PAGE, after electrophoresis, transferred to a PVDF membrane (Roche, Switzerland), and then subjected to immunoblotting analysis.

#### Cell Proliferation Assay

Indicated miR-NC and miR-199a stable cell lines were plated for 2 × 10<sup>3</sup> cells per well. To evaluate the proliferation activity

FIGURE 1 | Loss of MiR-199a in human glioma specimens. (A) Relative miR-199a expression levels were analyzed by Real-time RT-PCR in glioma specimens (n = 24) and adjacent normal tissues (n = 9). U6 RNA levels were used as an internal control; (B) Relative expression levels of miR-199a in cancer tissues at Grades I, II and III-IV (for each grade, n = 8). Data represent mean ± SD. from three replicates. \*\*Indicates significant difference at p < 0.01 when compared Grade I or Grade II group with Grade III–IV group.

of miR-199a in glioma cells, according to the manufacturer's instruction, cell proliferation rate was analyzed with CCK-8 kit (Dojindo Laboratories, Japan).

#### Migration Assay

Migration assay was conducted with migration chambers (BD Biosciences, UK). 5 × 10<sup>4</sup> cells were plated per well in the upper chamber without serum, and the lower chamber was filled with DMEM medium with 10% FBS. 18–20 h later, the bottom cells were fixed and stained, while non-invading cells were removed. Finally, cells were extracted by 33% acetic acid and quantitatively detected (OD at 570 nm).

#### Luciferase Reporter Assay

3 ′ -UTR region of K-RAS containing software predicted miR-199a-matching seed sites (WT) and corresponding mutant sites (mut) were amplified by High fidelity PCR, and inserted into pMIR-REPORTER vector (Ambion, USA). Dual-luciferase activities were analyzed in U87 cells by the Reporter Assay (Promega, USA).

#### Apoptosis Assay

Apoptosis assay (BD Pharmingen) in indicated cells were conducted according to the manufacturer's instruction with AnnexinV staining. Then samples were analyzed by flow cytometry (FACS Canto II, BD Biosciences). These data were analyzed by FlowJo software.

#### In vivo Tumor Growth Assay

Nude mice (4-weeks-old) were purchased from Animal Center (Shanghai, China), and then bred in special pathogen-free condition. Cells (5 × 10<sup>6</sup> ) were then injected subcutaneously into the posterior flanks of each nude mouse. Tumor sizes were measured by vernier caliper using the formula, that is volume

FIGURE 2 | Forced overexpression of miR-199a inhibited cell proliferation and migration in human glioma U87 and U251 cells. (A) Relative expression levels of miR-199a in U87 and U251 stable cell lines were determined by real-time RT-PCR; (B,C) Cells were plated with 2,000 cells per well in 96-well plates, and cell proliferation was determined using Cell Counting Kit-8 (CCK-8) by detecting the absorbance at 450 nm at indicated time points; (D) Migration assay of cells were performed as previously described. Data represent mean ± SD. from three replicates. \*Indicates significant difference at p < 0.05 when compared to miR-NC group; \*\*Indicates significant difference at p < 0.01 when compared to miR-NC group.

= 0.5 × (Length × Width<sup>2</sup> ). Twenty-four days later, mice were sacrificed as well as tumors were dissected. All mice used in this study were sacrificed according to the institutional guidelines and regulations.

#### Statistical Analysis

All cellular experiments were performed three times. Data were analyzed with GraphPad Prism 5 software. The correlations between miR-199a and K-RAS in human glioma tissues were analyzed by Pearson's test. The differences were considered as statistically significant at P < 0.05 by t-test.

#### RESULTS

#### Loss of MiR-199a in Human Glioma Specimens

Since mechanism of miR-199a is not fully understood in glioma, qRT-PCR analysis was then performed to determine indicated

expression levels of miR-199a in human glioma specimens. The results clarified that miR-199a expression in tumor (n = 24) tissues were significantly lower, as compared to normal (n = 9) tissues (**Figure 1A**). Furthermore, in tumor tissues of glioma patients, we showed that miR-199a expression were correlated with the clinical stages, which indicated that miR-199a in high grade tumors (n = 8, WHO Grades III-IV) were significantly lower when compared to those in low grade tumors (n = 8, WHO Grade I and n = 8, WHO Grade II) (**Figure 1B**). Thus, our results indicated that miR-199a may be a potential novel biomarker for glioma staging.

#### Forced Overexpression of miR-199a Inhibited Cell Proliferation and Migration Activity in Human Glioma U87 and U251 Cells

To overexpress miR-199a, human glioma cells U87 and U251 were infected with lentivirus expressing miR-199a, and lentivirus expressing miR-NC was used as control. Stable cell lines which termed as U87/miR-NC, U87/miR-199a, U251/miR-NC, and U251/miR-199a were established after puromycin selection, and higher miR-199a expression in U87/miR-199a and U251/miR-199a were demonstrated by qRT-PCR (**Figure 2A**). Overexpression of miR-199a markedly attenuated cell proliferation activity in U87/miR-199a (**Figure 2B**) and U251/miR-199a cells (**Figure 2C**). In addition, forced expression of miR-199a markedly reduced cell migration activity (**Figure 2D**). These results are consistent with the tumor suppressor activities of miR-199a in glioma cells.

# K-RAS Is a Direct Target of miR-199a

TargetScan software was used to predict the direct targets of miR-199a, and K-RAS was found to be a potential target (**Figure 3A**). We constructed luciferase reporter plasmids with either the putative K-RAS wild-type binding sites (WT) or seed sequence mutant sites (mut). Luciferase assay were used to investigate whether the K-RAS is a candidate target of miR-199a. Our results clarified that overexpressing miR-199a in U87 cells reduced the luciferase activity of WT K-RAS reporter by 65%, whereas it did not change the mutant luciferase activities (**Figure 3B**). Furthermore, forced overexpression of miR-199a significantly attenuated protein expression of K-RAS in U87 and U251 cells (**Figure 3C**), suggesting that in human glioma cells, miR-199a targets K-RAS directly by binding with its 3′ -UTR.

# Inverse Correlations of Lower miRNA-199a and Higher K-RAS Levels in Human Glioma Patient Tissues

To further support the notion that K-RAS oncogene is a direct target, we tested expression levels of K-RAS in human glioma specimens. Our results clarified that K-RAS expression levels were significantly higher in tumor tissues compared with normal tissues (**Figure 3D**). Furthermore, we determined the correlation between miR-199a and K-RAS oncogene levels in human glioma specimens using Pearson's correlation analysis. Inverse correlation in **Figure 3E** was found between K-RAS and miR-199a in the human glioma specimens (Pearson's correlation, r = −0.6622).

#### Forced Expression of miR-199a Reduced Activation of K-RAS Downstream Molecules AKT and ERK1/2 as Well as HIF-1α and VEGF Expression Levels in U87 Cells

The downstream molecules of RAS signaling are AKT and ERK1/2, which are linked to effectors, such as hypoxiainducible factor 1α (HIF-1α) as well as vascular endothelial growth factor (VEGF). We found that overexpression of miR-199a in U87 cells dramatically suppressed AKT and ERK1/2 activation as indicated by significantly reduced phosphorylated AKT and ERK1/2 levels and a reduction of HIF-1α protein level without change of total protein levels (**Figure 4A**).

Accumulated research have shown the importance role of HIF-1α/VEGF in the regulation of glioma angiogenesis (29, 30). Our previous study has reported that HIF-1α promotes VEGF gene expression through binding to the hypoxia response element (HRE) in the promoter region of VEGF. In addition to decreased HIF-1α, we also detected a significant reduction of VEGF mRNA levels in miR-199a-expressing U87 cells (**Figure 4B**). To test whether miR-199a inhibited VEGF expression through HIF-1α, we analyzed and compared effects of miR-199a on (1) a VEGF promoter reporter plasmid (pMAP11WT) containing the HIF-1α binding site, and (2) a mutant (pMAP11 mut) plasmid with mutation, overexpression of miR-199a inhibited the luciferase activities of the VEGF promoter reporter plasmid as shown in **Figure 4C**. Thus, our results indicated that miR-199a suppressed the gene expression of VEGF through HIF-1α.

FIGURE 5 | Overexpression of K-RAS cDNA reversed miR-199a-suppressed cell proliferation and cell migration. (A) U87 cells were co-transfected with miR-199a precursor or miR-NC, and empty vector or K-RAS cDNA without 3′ -UTR. After 72 h, immunoblotting assay was performed as described above; (B) Cells were treated as above, and cell proliferation assay was determined using CCK-8 kit; (C) U87 cells overexpressing miR-NC or miR-199a were treated as described above, cell migration was determined using 24-well BD migration chambers and quantified using a standard microplate reader (OD at 570 nm). Data represent mean ± SD. of three replicates. \*Indicates significant difference at p < 0.05 compared to miR-NC control; #Indicates significant difference at p < 0.05 compared to miR-199a and K-RAS overexpression group.

K-RAS overexpression group.

#### Restoration of K-RAS Reversed miR-199a Mediated Suppression on Cell Proliferation and Migration

To further demonstrate the role of K-RAS in miR-199a mediated effects on cell proliferation and migration, U87/miR-NC cells or U87/miR-199a cells were co-transfected with control vector (Vector) or K-RAS without 3′ -UTR. miR-199a reduced K-RAS expression significantly, and forced expression of K-RAS restored K-RAS expression as shown in **Figure 5A**. We also determined the effect on cell proliferation activity. As shown in **Figure 5B**, U87/miR-199a cells has significantly reduced cell proliferation rate, while overexpressing K-RAS in U87/miR-199a cells (U87/miR-199a/K-RAS), restored the cell proliferation rate. The cell migration was determined and quantified by a microplate reader using migration chambers (OD at 570 nm). As we expected, overexpression of K-RAS (U87/miR-199a/K-RAS) restored miR-199a-inhibited cell migration activity (**Figure 5C**). To sum up, these results suggested that miR-199a suppresses human glioma cell proliferation and migration, and forced expression of K-RAS reversed miR-199a effects.

#### Overexpression of miR-199a Rendered Cells More Sensitive to TMZ Through Targeting K-RAS

TMZ chemoresistance is the major obstacle in process of glioma chemotherapy. In this study, U87/miR-NC cells or U87/miR-199a were treated with TMZ at different concentrations for 2 days, as shown in **Figure 6A**, U87/miR-199a cells were significantly more sensitive to TMZ treatment. Furthermore, overexpressing K-RAS in U87/miR-199a cells (U87/miR-199a/K-RAS) nearly completely reversed the chemosensitivity to TMZ treatment (**Figure 6B**). We further investigated the role of miR-199a/K-RAS axis in TMZ mediated apoptosis by apoptosis analysis and caspase-3 activity assay. As shown in **Figure 6C**, TMZ treatment produced significantly higher apoptotic cell population in U87/miR-199a cells as compare to the control U87/miR-NC cells. Forced expression of K-RAS in U87/miR-199a cells (U87/miR-199a/K-RAS) nearly completely reversed the effect. Moreover, activities of caspase-3 were determined. As shown in **Figure 6D**, compared to the negative control (Bar 1, U87/miR-NC), overexpression of miR-199a (Bar 2, U87/miR-199a) significantly increased caspase 3 activity. TMZ treatmentof

control U87/miR-NC cells increased caspase3 activity (Bar 3, U87/miR-NC + TMZ), and TMZ treatment of U87/miR-199a cells strongly and significantly increased caspase3 activity (Bar 4, U87/miR-199a + TMZ). In addition, overexpression of K-RAS in U87/miR-199a cells attenuated the activation of caspase-3 (Bar 5, U87/miR-199a/K-RAS + TMZ). To sum up, our results showed that miR-199a rendered glioma cells more sensitive to TMZ through targeting K-RAS.

#### MiR-199a Suppressed Tumor Growth in vivo

To determine whether over-expression of miR-199a inhibited tumor growth in vivo, U87/miR-NC cells and U87/miR-199a cells were injected into immunodeficient nude mice. After cell injection, tumor sizes were measured. Compared to control mice injected with U87/miR-NC cells, mice injected with U87/miR-199a cells developed significantly smaller tumors from Day 14 (**Figure 7A**). Twenty-four days later, Mice were sacrificed after implantation, tumors were harvested, photographed, and weighed. A pair of representative tumors trimmed out from U87/miR-199a and U87/miR-NC groups, and the average tumor weight were shown in **Figure 7B**. Forced expression of miR-199a produced tumors with significantly lower tumor weight. We also determined the protein levels of K-RAS, p-AKT, p-ERK1/2 and HIF-1α in tumor tissues, and found that these proteins from U87/miR-199a group were significantly lower than that of U87/miR-NC group, which is consistent with in vitro data (**Figure 7C**). Our results suggested that miR-199a inhibited tumor growth by inhibiting K-RAS expression and its downstream molecules.

#### DISCUSSION

MicroRNAs (miRNAs) in diverse human cancers have been frequently indicated to be dysregulated (31, 32). The miR-199a have reported to be downregulated in multiple malignancies (33– 35). We first demonstrated that expression levels of miR-199a was downregulated in clinical glioma samples, and function as a tumor suppressor to increase sensitivity to treatment.

The reported miR-199a targets include GRP78, GSK-3β, Discoidin domain receptor 1 (DDR1), mTOR, CD44, and IκB kinase-beta. It works through targeting GRP78, a major endoplasmic reticulum chaperone, in prostate cancer cells to induce apoptosis and increase sensitivity to trichostatin A, the histone deacetylase inhibitor (36); through GSK-3β in renal cell cancer cells to decreases cell proliferation (37); through a receptor tyrosine kinase DDR1, to suppress invasiveness and migratory ability of colorectal cancer cells (38); through the mTOR and CD44 to increase sensitivity to cisplatin treatment and to reduce the number of ovarian cancer stem cells (39, 40). Finally, it works through targeting IκB kinase-beta to increase TNF-α-induced ovarian cancer cell apoptosis (41). Here, we first identified K-RAS as a novel target of miR-199a. We also confirm the inverse correlation between miR-199a and K-RAS levels in glioma specimens.

Temozolomide (TMZ) is a first-line drug for glioma treatment. A recent study has shown that miR-29c contributed to sensitize cells to temozolomide treatment by targeting O<sup>6</sup> methylguanine-DNA methyltransferases in glioma (42). On the other hand, miR-423-5p was reported to function as a oncogene and promoted chemoresistance to temozolomide in glioblastomas (43). Here we found that overexpression of miR-199a rendered cells more sensitive to TMZ through its target K-RAS. Thus, miR-199a/K-RAS signaling may be a potential new target to overcome chemoresistance to TMZ in glioma.

#### CONCLUSIONS

To sum up, we have clarified that K-RAS is a novel direct target of miR-199a. MiR-199a inhibit activity of cell proliferation, cell migration, drug chemoresistance and tumor growth by

#### REFERENCES


regulating K-RAS via AKT and ERK signalings. These results elucidated that miR-199a/K-RAS in the future may be used as a target for glioma treatment.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by Ethics Committee of Nanjing University. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Ethics Committee of Nanjing University.

#### AUTHOR CONTRIBUTIONS

WL, LW, and XG carried out the samples collection and performed the experiments. X-BJ and L-HW revised the manuscript. W-TL, LC, ZZ, Z-MS, L-ZL, and ML designed the studies. WL, LW, J-YC, and B-HJ wrote the manuscript.

#### FUNDING

This work was supported in part by National Natural Science Foundation of China (81502170, 81772951, 81803197).

inhibiting miR-143 in human prostate cancer. Oncotarget. (2017) 8:107157–66. doi: 10.18632/oncotarget.22362


production by targeting hexokinase 2 in liver cancer. Hepatology. (2015) 62:1132–44. doi: 10.1002/hep.27929


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Li, Wang, Ji, Wang, Ge, Liu, Chen, Zheng, Shi, Liu, Lin, Chen and Jiang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# LncRNA AFAP1-AS1 Supresses miR-139-5p and Promotes Cell Proliferation and Chemotherapy Resistance of Non-small Cell Lung Cancer by Competitively Upregulating RRM2

#### Edited by:

Dong-Hua Yang, St. John's University, United States

#### Reviewed by:

Shuaishuai Liu, University of Maryland, Baltimore County, United States Yunkai Zhang, Vanderbilt University Medical Center, United States

> \*Correspondence: Wei Zhao zw198626520@126.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 05 August 2019 Accepted: 07 October 2019 Published: 22 October 2019

#### Citation:

Huang N, Guo W, Ren K, Li W, Jiang Y, Sun J, Dai W and Zhao W (2019) LncRNA AFAP1-AS1 Supresses miR-139-5p and Promotes Cell Proliferation and Chemotherapy Resistance of Non-small Cell Lung Cancer by Competitively Upregulating RRM2. Front. Oncol. 9:1103. doi: 10.3389/fonc.2019.01103 Na Huang1†, Wei Guo2†, Ke Ren2†, Wancheng Li 1†, Yi Jiang<sup>1</sup> , Jian Sun<sup>1</sup> , Wenjing Dai <sup>1</sup> and Wei Zhao1,2 \*

<sup>1</sup> Department of Respiratory Medicine, The First Affiliated Hospital of Chengdu Medical College, Chengdu, China, <sup>2</sup> School of Laboratory Medicine/Sichuan Provincial Engineering Laboratory for Prevention and Control Technology of Veterinary Drug Residue in Animal-origin Food, Chengdu Medical College, Chengdu, China

Non-small cell lung cancer (NSCLC) is the leading cause of cancer-related death worldwide. This study aims to understand the underlying mechanism of lncRNA, actin filament-associated protein 1 antisense RNA 1(AFAP1-AS1) in mediating chemotherapeutic resistance in NSCLC. The levels of AFAP1-AS1 in NSCLC tissues and cells were determined using RT-PCR. The protein levels of RRM2, EGFR, and p-AKT were analyzed using Western blotting. Binding between AFAP1-AS1 and miR-139-5p was confirmed using dual luciferase reporter and RNA immunoprecipitation (RIP) assays, and binding between miR-139-5p and RRM2 was confirmed by a dual luciferase reporter assay. NSCLC cell proliferation, apoptosis, and colony formation were examined using MTT, flow cytometry, and colony formation assays, respectively. It was found that AFAP1-AS1 expression was upregulated in NSCLC tissues and cells. In addition, AFAP1-AS1 bound to and downregulated the expression of miR-139-5p, which was reduced in NSCLC tissues. Knockdown of AFAP1-AS1 and overexpression of miR-139-5p inhibited NSCLC cell proliferation, colony formation and chemotherapy resistance and increased cell apoptosis. Additionally, AFAP1-AS1 upregulates RRM2 expression via sponging miR-139-5p. Furthermore, AFAP1-AS1 enhanced NSCLC cell proliferation and chemotherapy resistance through upregulation of RRM2 by inhibiting miR-139-5p expression. Moreover, RRM2 promoted cellular chemotherapy resistance by activating EGFR/AKT. Finally, knockdown of AFAP1-AS1 significantly suppressed tumor growth and chemoresistance in nude mice. In conclusion, AFAP1-AS1 promoted chemotherapy resistance by supressing miR-139-5p expression and promoting RRM2/EGFR/AKT signaling pathway in NSCLC cells.

Keywords: AFAP1-AS1, miR-139-5p, RRM2, non-small cell lung cancer, EGFR/AKT

**83**

# INTRODUCTION

Lung cancer is the leading cause of cancer-related death worldwide (1–3). Non-small cell lung cancer (NSCLC) accounts for approximately 80% of all lung cancer cases and comprises two histological subtypes, adenocarcinoma (AD) and squamous cell cancer (SCC) (4, 5). The current overall 5-year survival rate for NSCLC is <15% due to both limited therapeutic options and recurrence (4). The prognosis of NSCLC is affected by chemotherapy resistance (6, 7). Thus, a better understanding of carcinogenesis and chemotherapy resistance is critical for developing novel therapies to treat NSCLC patients.

Long non-coding RNAs (lncRNAs) are a family of non-coding RNAs with lengths of >200 nucleotides. Accumulating evidence suggests that lncRNAs contribute to cancer initiation and progression and chemotherapy resistance (8–11). For example, the highly conserved lncRNA MALAT1 is a predictive biomarker for metastasis of lung cancer (12). Elevated LINC00473 expression often correlates with poor prognosis and is a robust biomarker for LKB1-inactivated NSCLC (13). HOTAIR is involved in the invasion and motility of lung cancer cells (14). However, MEG3 serves as a tumor suppressor in NSCLC, inhibiting cell proliferation and inducing p53-mediated cancer cell apoptosis (15).



SCC, lung squamous cell carcinoma; AC, lung adenocarcinoma; P value < 0.05 was statistically significant.

LncRNA actin filament-associated protein 1 antisense RNA 1 (AFAP1-AS1) is a 6.8-kb lncRNA located on chromosome 4p16.1. AFAP1-AS1 participates in the development of various cancers, including pancreatic ductal adenocarcinoma (16), esophageal adenocarcinoma (17), hepatocellular carcinoma (18), nasopharyngeal carcinoma (19), gallbladder cancer (20), and colorectal cancer (21). In addition, AFAP1-AS1 plays roles in NSCLC tumourigenesis by epigenetically repressing p21 expression (22, 23). However, the molecular mechanisms and global gene regulation mediated by AFAP1-AS1 and the role of AFAP1-AS1 in chemotherapy resistance in human NSCLC has not been explored.

Ribonucleoside-diphosphate reductase subunit M2 (RRM2) is the catalytic subunit of ribonucleotide reductase and modulates the enzymatic activity, which is essential for DNA replication and repair (24). RRM2 has been reported to be involved in the progression of various cancers, including gliomas (25), colorectal cancer (26), bladder cancer (27) and NSCLC (28–31). In addition, RRM2 is a prognostic biomarker for NSCLC (28– 31). Interestingly, AKT-induced tamoxifen resistance is reversed by RRM2 inhibition in breast cancer (32), suggesting that RRM2 may participate in the chemotherapy resistance of cancer cells. The abnormal overexpression or activation of AKT has been observed in cancers including lung, ovarian and pancreatic cancers (33), and AKT could be activated by epidermal growth factor receptor (EGFR) (34), implying that targeting EGFR or AKT could offer important approaches for cancer prevention and therapy. Subsequently, we investigated the effect of RRM2 on EGFR/AKT signaling.

In this study, we investigate the role of AFAP1-AS1 in NSCLC cell proliferation and chemotherapy resistance to DDP (Cisplatin) and 5-FU (fluorouracil), which are commonly used for countering progression of cancers in clinic. We also explore the function of RRM2 in the chemotherapy resistance of NSCLC cells. Our data indicate that AFAP1-AS1 expression was elevated in patients with NSCLC and that AFAP1-AS1 acts as a competing endogenous RNA for miR-139-5p, which is an important suppressor in several tumors (35–39). Knockdown of AFAP1- AS1 or overexpression of miR-139-5p inhibited the proliferation, increased the apoptosis, and attenuated the chemotherapy resistance of lung cancer cells by upregulating RRM2. In addition, knockdown of AFAP1-AS1 reduced tumor volume and weight in vivo. Taken together, AFAP1-AS1 supresses miR-139-5p and promotes cell proliferation and chemotherapy resistance of NSCLC cells by competitively upregulating RRM2 expression.

#### MATERIALS AND METHODS

#### Tissue Collection

This study was approved by the ethics committee of first affiliated hospital of Chengdu Medical College. From Feb. 2018 to Apr. 2019, a total of 44 NSCLC patients were recruited from Department of Respiratory Medicine, the First Affiliated Hospital of Chengdu Medical College Chengdu. All participants signed an informed consent form. NSCLC tissues and adjacent normal lung tissues were collected and stored at −80 ◦C until used. The drugs cisplatin (DDP), 5-fluorouracil (5-FU),

adriamycin, and paclitaxel were used for NSCLC treatment in all patients. In accordance with the Response Evaluation Criteria in Solid Tumors, we grouped the patients with a complete or partial response as responders and defined those with stable or progressive disease as non-responders. The clinicopathological characteristics of the patients with NSCLC are summarized in **Table 1**.

#### Cell Culture

The NSCLC cell lines H1975, PC-9, A549, and SPCA-1, and a human normal lung epithelial cell line BEAS-2B were purchased from the Institute of Biochemistry and Cell Biology of the Chinese Academy of Sciences (Shanghai, China). The H1975 and SPCA-1 cells were maintained in RPMI 1640 basic medium (GIBCO, Carlsbad, CA), and the PC-9 and A549 cells were cultured in DMEM (GIBCO) in a humidified incubator at 37◦C with 5% CO2. All media were supplemented with heatinactivated 10% fetal bovine serum (FBS) and antibiotics (100 U/mL penicillin and 100 mg/mL streptomycin) (GIBCO).

#### Cell Transfection

The cells were plated in dishes or plates and grown to 70% confluence for the transfection of small interfering RNA (siRNA) or plasmids using Lipofectamine 2000 (Thermo Fisher Scientific, Shanghai, China). The siAFAP1-AS1#1 and siAFAP1-AS1#2 sequences, the miR-139-5p mimic or inhibitor, the pcDNA-RRM2 plasmid, and the Lv-AFAP1-AS1 knockdown (KD) (Lv-siAFAP1-AS1#1) construct and their paired controls were synthesized by GenePharma (Shanghai, China). The pcDNA-RRM2 plasmid contained the full-length RRM2 coding mRNA sequence (NM\_001165931.1), the pcDNA-AFAP1- AS1 plasmid contained the full-length AFAP1-AS1 sequence (ENST00000608442.1), and the Lv-AFAP1-AS1 KD lentivirus expressed siRNA targeting AFAP1-AS1. The siRNA sequences targeting AFAP1-AS1 were designed as follows: siAFAP1-AS1, 5 ′ -GCA TTA TTT TGC TAA TTC AAC-3′ and the scrambled negative control siRNA was the sequence: 5′ -CCT AAC CAC AAA CTC TAC GGC-3′ (abbreviated as scramble). The inhibitor sequences targeting miR-139-5p were designed as follows: 5′ - CUG GAG ACU GCG ACU GUA GAC UGG AGA CUG CGA CUG UAG ACU GGA GAC UGC GAC UGU AGA CUG GAG ACU GCG ACU GUA GAC UGG AGA CUG CGA CUG UAG A-3′ , and the miR-139-5p mimic sequences were designed as follows: 5′ -UCU ACA GUG CAC GUG UCU CCA G-3′ , and the negative control sequences were 5′ -UCU CCG AAC GUG UCA CGU-3′ (abbreviated as pre-NC, or NC). Then, both the full length and mutant of AFAP1-AS1 (or RRM2 3′UTR) were constructed into pmirGLO plasmid for luciferase assay.

(Continued)

FIGURE 2 | luciferase reporter assay on cells transfected with AFAP1-AS1 WT, AFAP1-AS1 Mut1, and AFAP1-AS1 Mut2. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected samples. (C) RT-PCR on the miR-139-5p expression in chemoresistant tissues. Data shown as means ± S.D. #P < 0.05 compared with chemoresponsive tissues. (D) RT-PCR on the miR-139-5p expression in cancer cells. Data shown as means ± S.D. &P < 0.05 compared with BEAS-2B cells. (E–H) RT-PCR on the effect of AFAP1-AS1 knockdown on miR-139-5p mRNA expression. Data shown as means ± S.D. #P < 0.05 compared with the scramble-transfected group. (I,J) The effect of AFAP1-AS1 overexpression on miR-139-5p mRNA expression analyzed by RT- PCR. Data shown as means ± S.D. #P < 0.05 compared with the pcDNA-transfected group. (K) Cell lysate incubated with an anti-Ago2 antibody for RIP, and the AFAP1-AS1 content detected by RT-PCR. Data shown as means ± S.D. #P < 0.05 compared with the IgG control group. (L) Cell lysate incubated with Bio-AFAP1-AS1 for RIP, and the enrichment of miR-139-5p detected by RT- PCR. Data shown as means ± S.D. #P < 0.05 compared with Bio-control group. (M) The expression of AFAP1-AS1 and miR-139-5p negatively correlated in NSCLC tissues. r = −0.7686 and p ≤ 0.0001.

# RNA Extraction and Real-Time Polymerase Chain Reaction (PCR) Analyses

Total RNA was extracted from NSCLC tissues or cells with TRIzol reagent (Thermo Fisher Scientific) following the manufacturer's instructions. The expression of AFAP1-AS1, miR-139-5p, and RRM2 was analyzed with SYBR Green Master Mix (Takara, Beijing, China). Complementary DNA (cDNA) was synthesized using a PrimeScript RT Reagent Kit and gDNA Eraser (Takara). Real-time PCR was carried out on an ABI 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The primers used were as follows: AFAP1-AS1, 5′ -TCG CTC AAT GGA GTG ACG GCA-3′ (forward) and 5′ -CGG CTG AGA CCG CTG AGA ACT-3′ (reverse); miR-139-5p, 5′ -TCT ACA GTG CAC GTG TCT CCA G-3′ (forward) and 5′ -GTG CAG GGT CCG AGG T-3 ′ (reverse); U6, 5′ -TGC GGG TGC TCG CTT CGG CAG C-3′ (forward) and 5′ -GTG CAG GGT CCG AGG T-3′ (reverse); and GAPDH, 5′ -CAC CCA CTC CTC CAC CTT TG-3′ (forward) and 5′ -CCA CCA CCC TGT TGC TGT AG-3′ (reverse); RRM2, 5 ′ -GGC GCG GGA GAT TTA AAG GC-3′ (forward) and 5′ - CGG AGG GAG AGC ATA GTG GA-3′ (reverse). The relative expression levels of AFAP1-AS1, miR-139-5p, and RRM2 were calculated using the 2−11Ct method with U6 or GAPDH as the internal control.

#### Luciferase Reporter Assay

The luciferase reporter vector pGLO-basic (Promega, Beijing, China) containing the wild-type (WT) or mutant AFAP1-AS1 sequences 1/2 (Mut 1/2) were transfected into A549 cells. The pGLO plasmids containing the full-length RRM2 3′ UTR or its corresponding mutant were co-transfected with an miR-139-5p mimic or inhibitor into A549 cells. After 48 h of incubation, the cells were harvested and luciferase activity was determined using a dual luciferase assay kit (Promega).

# MTT Assay and CCK-8 Assay

Cell proliferation was measured by a 3-(4,5-dimethylthiazol-2 yl)-2,5-diphenyl-tetrazolium bromide (MTT) assay. Lung cancer cells were plated in a 96-well plate (2 × 10<sup>3</sup> cells/well). After cells were incubated with MTT (Sigma, Shanghai, China), the optical density (OD) value was determined at 450 nm. Additionally, we used a Cell Counting Kit-8 (CCK-8, Sigma) to analyse NSCLC cell viability. In the MTT and CCK-8 assays, the inhibition rate (%) = 100% × (1–OD value of the treated sample/OD value of the control sample).

### Colony Formation Assay

NSCLC cells were seeded in fresh six-well plates in a 5% CO<sup>2</sup> incubator at 37◦C and were then transfected with the indicated siRNAs. Following incubation for 2 weeks, NSCLC cells would grow into colonies, and they were fixed with methanol and stained with 0.1% crystal violet. Visible colonies were manually counted and recorded.

#### Apoptosis Analyses

Treated cells were collected, centrifuged at 2,000 rpm for 5 min and washed with PBS three times. The cells were resuspended in 100 µL of PBS, and annexin V/FITC (5 µL) and propidium iodide (PI) (1 µL) were added to each sample. After 15 min incubation at room temperature in the dark, the apoptosis of the cancer cells was analyzed on an S3e flow cytometer (Bio-Rad, Shanghai, China). Cells stained with either annexin V or PI were counted as apoptotic cells.

Caspase-3 activity was also checked in cancer cells by Caspase-3 Activity Assay Kit (Beyotime, Shanghai, China) followed the instruction.

#### RNA Immunoprecipitation (RIP) Assays

Rip experiments were performed using a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) according to the manufacturer's instructions. Antibodies against EZH2 and AGO-2 were obtained from Sigma. The AGO2 expression level was determined by immunoprecipitation and Western blotting, and the AFAP1-AS1 expression level was determined by real-time PCR.

#### Chemical Treatment in Cells

In this study, transfected NSCLC cells were incubated with DDP (solute in PBS) at concentrations of 0, 1, 2, 4, 8, and 12 µM or with 5-FU (solute in PBS) at concentrations of 0, 1, 4, 8, 16, and 32µM for 36 h. DDP and 5-FU were purchased from Sigma.

A549 and SPCA-1 cells were treated with AST1306 (Selleck, Shanghai, China) at 1µM for 24 h.

#### Generation of Drug-Resistant Cell Lines

The DDP- or 5-FU-resistant cell lines were generated by incubating NSCLC cells with increasing concentration of the indicated drugs. NSCLC cells were plated into plates and maintained in medium containing 0.2µM DDP. After 48 h incubation, the 0.2µM DDP-containing medium was discarded, and medium containing gradually increasing concentrations of DDP was added. Finally, cells resistant to 10µM DDP were obtained and named A549/DDP or SPCA-1/DDP cells.

the proliferation of A549 and SPCA-1 cells with AFAP1-AS1 knockdown. Data shown as means ± S.D. #P < 0.05 compared with the scramble-transfected cells. (C) MiR-139-5p overexpressed in A549 and SPCA-1 cells. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected group. (D,E) MTT assay on the proliferation of A549 and SPCA-1 cells with miR-139-5p overexpression. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected group. (F) A colony formation assay on the effect of miR-139-5p overexpression on cell proliferation. (G,H) Annexin V/PI staining to assess the effect of AFAP1-AS1 knockdown on the apoptosis of A549 and SPCA-1 cells. The cell apoptosis data shown as means ± S.D. #P < 0.05 compared with the scramble-transfected cells. (I,J) Annexin V/PI staining to assess the effect of miR-139-5p overexpression on the apoptosis of A549 and SPCA-1 cells. (K,L) Caspase-3 activity was identified in A549 and SPCA-1 cells, which were treated as indicated. The cell apoptosis data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected cells.

FIGURE 4 | scramble-transfected cells. (C,D) CCK-8 assay on A549 or SPCA-1 cells transfected with scramble or siAFAP1-AS1 and incubated with 5-FU for 36 h at the indicated concentrations. Data shown as means ± S.D. #P < 0.05 compared with the scramble-transfected cells. (E) AFAP1-AS1 levels determined by RT-PCR in A549 and drug-resistant A549/DDP cells as well as in SPCA-1 and SPCA-1/DDP cells. Data shown as means ± S.D. #P < 0.05 compared with parental A549 cells. (F,G) Knockdown of AFAP1-AS1 increased the DDP-induced apoptosis of drug-resistant NSCLC cells. Data presented as means ± S.D. #P < 0.05 compared with the scramble-transfected DDP-resistant cancer cells. (H) AFAP1-AS1 levels determined by RT-PCR in A549 and drug-resistant A549/5-FU cells as well as in SPCA-1 and SPCA-1/5-FU cells. Data shown as means ± S.D. #P < 0.05 compared with parental A549 cells. (I) Knockdown of AFAP1-AS1 increased the 5-FU-induced apoptosis of drug-resistant NSCLC cells. Data presented as means ± S.D. #P < 0.05 compared with the scramble-transfected 5-FU-resistant cancer cells. (J,K) Caspase-3 activity was identified in A549 and SPCA-1 cells, which were treated as indicated. #P < 0.05 compared with the scramble+PBS group.

#### Western Blot Analysis

Tissues and cancer cells were lysed using RIPA lysis buffer (Beyotime, Haimen, China) containing protease inhibitor cocktail (Roche). Approximately 20 µg of extracted protein was separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to 0.22 mm polyvinylidene difluoride (PVDF) membranes (Millipore, Shanghai, China). We blocked the PVDF membranes in 2% bovine serum albumin (BSA) and incubated them with primary antibodies against RRM2 (catalog No. #65939), EGFR (catalog No. #2085), p-AKT (catalog No. #4060), AKT (catalog No. #9272), and GAPDH (catalog No. #5174) (Cell Signaling Technology, Shanghai, China). Then, these immunoblots were incubated with horseradish peroxidase-conjugated secondary antibodies for 60 min at room temperature. GAPDH was used as the internal control.

# Pull Down Assay With Biotinylated AFAP1-AS1 DNA Probe

The biotin-labeled ABHD11-AS1 DNA probe was designed (Thermo), dissolved in binding and washing buffer and mixed with M-280 streptavidin magnetic beads (Thermo) to generate probe-coated beads according to the manufacturer's instruction. The A549 cell lysates were incubated with the probe-coated beads. Then, we used real-time PCR analysis to determine the beads-binding RNAs. The AFAP1-AS1 pull-down probe sequence was 5′ -Bio-AGT AAA CAC GCA GTT GCA CAT GGC TGG GGA GGC CTC AGA ATC ATG GCG GGA GGC GAA AGA CAC TTC TTA CGT GGC AGC AGC-3′ ; and random pulldown probe sequence used as negative control was 5′ -Bio-TGC ATC CAA GCC GAT TGC GGT AAC GTG CAT CCA AGC CGA TTG CGG TAA CG-3′ .

#### Xenograft Tumor Assay

Male athymic nude BALB/c mice were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). At 5 weeks of age, the mice were randomly divided into four groups. The animal procedure was approved by the Ethics Committee of Animal Experiments of Chengdu Medical College (Chengdu, China). A549 cells were transfected with Lvscramble or Lv-AFAP1-AS1 KD. 2 × 10<sup>6</sup> cells were inoculated subcutaneously into the mice. After 10 days, the mice were administered 3 mg/kg (body weight) DDP or PBS every 4 days for 28 days. During this period, the tumor lengths and widths were measured, and tumor volumes were calculated as follows: tumor volume = (length × width<sup>2</sup> )/2. Finally, the tumors were harvested and weighed.

#### Bioinformatic Analyses

In this study, LNCipedia version 5.2, lncBase version 2, and starBase were used to predict the potential binding sites between AFAP1-AS1 and miR-139-5p. miRBase and miRDB were applied to analyse the binding sites between miR-139-5p and the RRM2 3 ′ UTR. In addition, we predicted the RNA-binding activity by examining previous studies (40).

#### Statistical Analysis

Statistical analyses were performed using SPSS software (version 19.0). The data are expressed as means ± standard deviations (S.D.). A two-tailed Student's t-test was used to analyse difference between two groups. For multi-group comparisons, we used one-way analysis of variance (ANOVA) with a post-hoc Tukey's honestly significant difference (HSD) test. P-value of < 0.05 were considered statistically significant.

# RESULTS

#### AFAP1-AS1 Is Overexpressed in NSCLC Tissues and Cells

Firstly, NSCLC tissues and adjacent tissues were collected from hospital. The tissues were analyzed by H&E staining, and the results showed that abnormal cell over-growth appeared in tumors (**Figure 1A**). RT-PCR was performed to determine the expression of AFAP1-AS1 in NSCLC tissues and cells. It was found that the expression of AFAP1-AS1 was significantly higher in NSCLC tissues than in normal tissues (**Figure 1B**). In addition, AFAP1-AS1 was overexpressed in NSCLC tissues of patients in the chemotherapy non-response group compared to the chemotherapy response group (**Figure 1C**). Moreover, Additionally, the expression of AFAP1-AS1 was higher in NSCLC cells than in BEAS-2B cells and was highest in SPCA-1 cells and lowest in H1975 cells (**Figure 1D**).

#### AFAP1-AS1 Inhibits miR-139-5p Expression

The potential binding sites between AFAP1-AS1 and miR-139- 5p were predicted based on bioinformatic analysis (**Figure 2A**). The dual luciferase reporter assay demonstrated that the miR-139-5p mimic significantly reduced the luciferase activity of cells transfected with AFAP1-AS1 WT as well as that of cells transfected with the AFAP1-AS1 mutated type AFAP1-AS1 Mut2 (**Figure 2B**). However, the miR-139-5p mimic failed to suppress the luciferase activity of cells transfected with the other AFAP1- AS1 mutated type Mut1, suggesting that miR-139-5p may bind to more than one site on the AFAP1-AS1 Mut1 construct (**Figure 2B**). We found that the level of miR-139-5p was lower

FIGURE 5 | pre-NC-transfected cells. (C,D) CCK-8 assay on A549 or SPCA-1 cells transfected with pre-NC or the miR-139-5p mimic and incubated with 5-FU for 36 h at the indicated concentrations. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected cells. (E) RT-PCR on the expression of miR-139-5p in A549 and drug-resistant A549/DDP cells, as well as in SPCA-1 and SPCA-1/DDP cells. Data presented as means ± S.D. #P < 0.05 compared with parental A549 cells. (F,G) Overexpression of miR-139-5p increased DDP-induced apoptosis in drug-resistant NSCLC cells. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected DDP-resistant cancer cells. (H) RT-PCR on the expression of miR-139-5p in A549 and drug-resistant A549/5-FU cells, as well as in SPCA-1 and SPCA-1/5-FU cells. Data presented as means ± S.D. #P < 0.05 compared with parental A549 cells. (I) Overexpression of miR-139-5p increased 5-FU-induced apoptosis in drug-resistant NSCLC cells. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected 5-FU-resistant cancer cells. (J,K) Caspase-3 activity was identified in A549 and SPCA-1 cells, which were treated as indicated. #P < 0.05 compared with the scramble+PBS group.

in patients in the chemotherapy non-response group than in the chemotherapy response group (**Figure 2C**), and miR-139-5p was decreased in lung cancer cell lines compared with BEAS-2B cells (**Figure 2D**). Furthermore, transfection with siRNA targeting AFAP1-AS1 reduced AFAP1-AS1 expression (**Figures 2E,F**) and upregulated miR-139-5p expression (**Figures 2G,H**) in A549 and SPCA-1 cells. In contrast, pcDNA-AFAP1-AS1-mediated overexpression of AFAP1-AS1 reduced the miR-139-5p level in H1975 and PC-9 cells (**Figures 2I,J**). AFAP1-AS1 expression was significantly elevated in anti-Ago2 (Protein argonaute-2) incubated A549 cells (**Figure 2K**), and AFAP1-AS1 could directly bind to miR-139-5p (**Figure 2L**). There was a negative correlation between AFAP1-AS1 and miR-139-5p expression in NSCLC cells (**Figure 2M**). These findings indicated that AFAP-AS1 was a sponge of miR-139-5p.

#### Suppression of AFAP1-AS1 or Overexpression of miR-139-5p Inhibits the Proliferation and Increases Cell Apoptosis of NSCLC Cells

To investigate the effect of AFAP1-AS1 and miR-139-5p on the proliferation and apoptosis of NSCLC cells, A549 and SPCA-1 cells were transfected with scramble, siAFAP1-AS1, pre-NC, or the miR-139-5p mimic. Knockdown of AFAP1-AS1 suppressed cancer cell proliferation, as evidenced by the MTT assay results (**Figures 3A,B**). Similarly, overexpression of miR-139-5p decreased cell proliferation (**Figures 3C–E**) and inhibited colony formation (**Figure 3F**). As expected, suppression of AFAP1- AS1 or overexpression of miR-139-5p significantly increased the apoptosis of A549 and SPCA-1 cells (**Figures 3G–L**).

#### Knockdown of AFAP1-AS1 or Overexpression of miR-139-5p Decreases the Chemotherapy Resistance of NSCLC Cells

To analyse the effect of AFAP1-AS1 and miR-139-5p on the chemoresistance of NSCLC cells, scramble- or siAFAP1- AS1-transfected A549 and SPCA-1 cells were incubated with DDP or 5-FU. It was found that the drug-induced growth inhibition increased in a dose-dependent manner, and AFAP1- AS1 knockdown increased the inhibitory activity of DDP and 5-FU in NSCLC cells (**Figures 4A–D**), implying that suppression of AFAP1-AS1 alleviated the chemotherapy resistance of NSCLC cells. In addition, we also observed that AFAP1- AS1 was significantly overexpressed in DDP-resistant A549 and SPCA-1 cells compared with canonical A549 and SPCA-1 cells (**Figure 4E**). Furthermore, interfering with AFAP1-AS1 expression significantly increased the DDP-induced apoptosis in the drug-resistant cancer cells (**Figures 4F,G**). Similarly, AFAP1- AS1 was obviously increased in 5-FU-resistant A549 and SPCA-1 cells, and knockdown of AFAP1-AS1 also promoted 5-FUtriggered cell apoptosis (**Figures 4H–K**).

Meanwhile, miR-139-5p mimic enhanced the inhibitory activity of DDP and 5-FU on these cancer cells (**Figures 5A–D**), indicating that overexpression of miR-139-5p decreased the chemotherapy resistance of NSCLC cells to DDP and 5-FU. Although miR-139-5p expression was reduced in DDP- or 5-FUresistant NSCLC cells (**Figures 5E–H**), overexpression of miR-139-5p significantly enhanced the apoptosis of DDP- or 5-FUresistant cancer cells (**Figures 5F–K**).

#### Cooperation of miR-139-5p and AFAP1-AS1 Regulates RRM2 Expression by Targeting Its 3′ UTR

To investigate the interaction between miR-139-5p and RRM2, a luciferase reporter gene assay was performed. The binding sites between miR-139-5p and the RRM2 3′ UTR were predicted by bioinformatics (**Figure 6A**). The RRM2 3′ UTR sequences were sub-cloned into the pGLO plasmid, and A549 and SPCA-1 cells were co-transfected with the RRM2 3′ UTR plasmid and the miR-139-5p mimic or inhibitor. The results showed that the miR-139-5p mimic or inhibitor significantly decreased or increased, respectively, the luciferase activity driven by RRM2 WT; however, the miR-139-5p mimic or inhibitor did not affect the luciferase activity driven by the mutated RRM2 3′ UTR (termed RRM2 MUT) (**Figures 6B,C**). The miR-139-5p mimic or inhibitor noticeably decreased or increased RRM2 mRNA expression (**Figure 6D**) and protein expression (**Figures 6E,F**), respectively. Interestingly, the miR-139-5p mimic or inhibitor also modulated the protein levels of EGFR and p-AKT (**Figures 6E,F**). RRM2 was found to be overexpressed in NSCLC tissues (**Figure 6F**) and multidrug resistant NSCLC cells (**Figures 6G,H**). RRM2 was also found has higher mRNA level in NSCLC cells than normal. To determine the cooperation of miR-139-5p and AFAP1- AS1 on regulation of RRM2 expression, the data showed that AFAP1-AS1 significantly increased luciferase activity of RRM2 3′UTR, and elevated RRM2 protein level (**Figures 6J–M**). However, AFAP1-AS1-induced elevation of RRM2 could be reversed by overexpression of miR-139-5p in A549 and SPCA-1 cells (**Figures 6J–M**).

(Continued)

FIGURE 6 | with WT RRM2 3′ UTR or mutated 3′ UTR. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected or NC-transfected cancer cells. (D) RT-PCR on the cDNA from A549 cells to determine the effect of miR-139-5p on RRM2 mRNA expression. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected or NC-transfected A549 cells. (E,F) Western blot on the effect of miR-139-5p on protein expression of RRM2, EGFR, AKT, and p-AKT in A549 and SPCA-1 cells. (G) RT-PCR on the expression of RRM2 mRNA in NSCLC tissues (n = 44) and in adjacent normal tissues (n = 20). Data shown as means ± S.D. #P < 0.05 compared with the normal tissues. (H) The mRNA level of RRM2 in chemotherapy response (n = 4) and resistance (n = 7) NSCLC tumors by RT-PCR. The results shown as means ± S.D. #P < 0.05 compared with the response group. (I) RRM2 mRNA expression in lung cancer cells analyzed by RT-PCR. The results shown as means ± S.D. #P < 0.05 compared with BEAS-2B cells. (J,K) A dual luciferase reporter assay on A549 cells and SPCA-1 cells with WT RRM2 3′ UTR or mutated 3′ UTR. Data shown as means ± S.D. #P < 0.05 compared with the pre-NC-transfected or NC-transfected cancer cells. (L,M) Western blot on the effect of miR-139-5p and AFAP1-AS1 on protein expression of RRM2in A549 and SPCA-1 cells.

# Knockdown of AFAP1-AS1 Inhibits Cell Proliferation and Alleviates Chemotherapy Resistance Via the miR-139-5p/RRM2 Axis

To determine the role of the miR-139-5p/RRM2 axis in AFAP1-AS1-mediated cell proliferation and chemotherapy resistance, NSCLC cells were transfected with scramble, siAFAP1-AS1, siAFAP1-AS1+NC, and siAFAP1-AS1+miR-139-5p inhibitor. The miR-139-5p inhibitor reversed the suppressive effect of siAFAP1-AS1 on cell proliferation (**Figures 7A,B**) and colony formation (**Figures 7C,D**) and reversed the increased cancer cell apoptosis (**Figures 7E,F**). In addition, the miR-139-5p inhibitor reversed the suppressive effect of siAFAP1-AS1 on DDP resistance (**Figures 7G,H**) and 5-FU resistance (**Figures 7I,J**). Additionally, AFAP1- AS1 knockdown downregulated the protein expression of RRM2 and decreased the protein levels of EGFR and p-AKT. However, the miR-139-5p inhibitor reversed these effects (**Figures 7K,L**).

#### EGFR/AKT Signaling Is Involved in RRM2-Mediated Chemotherapy Resistance

To investigate the role of EGFR/AKT in RRM2-mediated chemotherapy resistance, A549 and SPCA-1 cells were transfected with the pcDNA vector, pcDNA-RRM2, pcDNA-RRM2+DMSO, and pcDNA-RRM2+AST1306 (inhibitor of EGFR). AST1306 reversed the RRM2-mediated promotion of chemotherapy resistance (**Figures 8A–D**). Additionally, AST1306 reversed the RRM2-induced upregulation of EGFR expression and p-AKT levels in A549 and SPCA-1 cells (**Figures 8E,F**). These results indicated that RRM2 enhanced the chemotherapy resistance of NSCLC cells via EGFR/AKT signaling pathway.

Moreover, the in vivo experiments show that knockdown of AFAP1-AS1 suppresses tumorigenicity and chemo-resistance of NSCLC cells in the nude mice (**Figures 9A–D**). These results indicated that AFAP1-AS1/miR-139-5p/RRM2/EGFR/AKT signaling pathway was involved in the progression of NSCLC (**Figure 9D**).

# DISCUSSION

LncRNAs are involved in many aspects of cancer development and chemotherapy resistance (8–11). To investigate whether there is an abnormal expression of lncRNA AFAP1-AS1 in NSCLC tissues and cancer cells, real-time PCR was performed. We assessed the effect of AFAP1-AS1 on the proliferation, apoptosis and chemotherapy resistance of lung cancer cells. AFAP1-AS1 could perform as a sponge of miR-139-5p in cancer progression. Suppression of AFAP1-AS1 or overexpression of miR-139-5p significantly repressed the proliferation, increased the apoptosis, and ameliorated the chemotherapy resistance of NSCLC cells by downregulating RRM2. Furthermore, downregulation of AFAP1-AS1 decreased xenograft tumor volume and weight. These findings suggested that AFAP1- AS1 could be an oncogene and induce chemotherapy resistance by modulating miR-139-5p/RRM2 signaling in NSCLC.

Accumulating evidence demonstrated that dysregulated lncRNAs are major contributors to tumourigenesis and cancer development. For example, HOTAIR plays an important role in cellular proliferation, invasion, and clinical relapse in small cell lung cancer (41). HOTAIR also mediates chemoresistance in NSCLC by regulating HOXA1 methylation and could be a potential target for new adjuvant therapy against chemoresistance (42). In addition, the p53-regulated lncRNA TUG1 affects NSCLC cell proliferation in part by epigenetically controlling HOXB7 expression (43). MEG3 acts as a tumor suppressor in NSCLC cell proliferation and induces p53-mediated cancer cell apoptosis (15). Moreover, downregulation of AFAP1-AS1 results in growth inhibition and apoptosis promotion in lung adenocarcinoma cells, indicating that this lncRNA participates in tumourigenesis (44). In NSCLC, AFAP1-AS1 increases tumourigenesis by epigenetically repressing p21 expression. AFAP1-AS1 recruits EZH2 to the p21 promoter region, resulting in downregulation of p21, which is a tumor suppressor (22, 23). Consistent with these previous findings, we found that AFAP1-AS1 was upregulated in NSCLC tissues and cells, and it was overexpressed in chemotherapy-resistant tissues, indicating that AFAP1-AS1 is a positive regulator of NSCLC development and chemoresistance. Previous studies have shown that lncRNAs can act as competing endogenous RNAs of miRNAs (45). Thus, in the present study, we predicted binding sites between AFAP1-AS1 and miR-139-5p, which is a tumor suppressor in colorectal cancer and endometrial cancer (38, 46). However, the role of miR-139-5p in NSCLC has not been explored. Luciferase reporter assays, RIP assays, and real-time PCR were performed, and the data showed that AFAP1-AS1, as a sponge, directly bound to miR-139-5p, leading to downregulation of miR-139-5p expression, and that miR-139-5p expression was

FIGURE 7 | AFAP1-AS1 promotes cell proliferation and chemotherapy resistance through the miR-139-5p/RRM2 axis. A549 or SPCA-1 cells were transfected with scramble, siAFAP1-AS1, siAFAP1-AS1+NC, and siAFAP1-AS1+miR-139-5p inhibitor. (A,B) MTT assay showed miR-139-5p inhibitor reversed the inhibitory effect of (Continued) FIGURE 7 | AFAP1-AS1 knockdown on lung cancer cell proliferation. (C) and (D) The miR-139-5p inhibitor reversed the inhibitory effect of AFAP1-AS1 knockdown on colony formation. (E,F) The miR-139-5p inhibitor reversed the inhibitory effect of AFAP1-AS1 knockdown on apoptosis of NSCLC cells. (G,H) In A549 and SPCA-1 cells incubated with DDP for 36 h, the miR-139-5p inhibitor reversed the inhibitory effect of siAFAP1-AS1 on chemotherapy resistance (DDP). (I,J) the miR-139-5p inhibitor reversed the inhibitory effect of siAFAP1-AS1 on chemotherapy resistance in cells incubated with 5-FU. (K,L)Knockdown of AFAP1-AS1 downregulated RRM2 protein expression and reduced the protein levels of EGFR and p-AKT, while miR-139-5p reversed these effects in A549, and SPCA-1 cells. All data shown as means ± S.D. #P < 0.05 indicates a significant difference between the two indicated groups.

overexpression of RRM2 enhanced chemotherapy resistance, and AST1306 reversed this effect. (C,D) In NSCLC cells incubated with 5-FU for 36 h, overexpression of RRM2 enhanced chemotherapy resistance and AST1306 reversed this effect. (E,F) In A549 and SPCA-1 cells, overexpression of RRM2 elevated the protein expression of EGFR and the level of p-AKT, and AST1306 reversed this effect. All data shown as means ± S.D. #P < 0.05 indicates a significant difference between the two indicated groups.

decreased in chemotherapy-resistant tissues. We also found that the miR-139-5p inhibitor reversed AFAP1-AS1-induced biological effects, indicating that the interaction of AFAP1- AS1 and miR-139-5p is involved in NSCLC progression and chemotherapy resistance.

In addition, we found that AFAP1-AS1 participated in positively modulating luciferase activity of RRM2 3′ UTR and RRM2 level by acting as a sponge of miR-139-5p in cancer cells, suggesting that RRM2, as AFAP1-AS1, is a oncogenic regulator. This finding is in consistent with previous reports that RRM2 is an oncogene in certain cancers (25–32). It was reported that silencing RRM2 suppresses glioblastoma cell invasion and migration by reducing the expression of metalloproteinase-2 (MMP-2) and MMP-9 (25). The abnormal overexpression or activation of AKT has been observed in many cancers, including lung, ovarian, and pancreatic cancers, and is associated with increased cancer cell proliferation and survival (33). AKT could be activated by epidermal growth factor receptor (EGFR) (34). Consequently, targeting EGFR or AKT could offer important approaches for cancer prevention and therapy. More importantly, overexpression of RRM2 in gastric cancer cells promotes their invasiveness by regulating the AKT/NF-κB signaling pathway (47), and RRM2 increases tumor angiogenesis and growth by modulating the expression of thrombospondin-1 (TSP-1) and vascular endothelial growth factor (VEGF) (48). Subsequently, we investigated the effect of AFAP1-AS1/miR-139-5p/RRM2 signaling on EGFR expression and phosphorylation of AKT. Expectedly, our data showed

groups. (D) Schema of the signaling pathways involved in AFAP1-AS1/miR-139-5p/RRM2-mediated chemotherapy resistance in NSCLC cells.

that overexpression of RRM2 promoted the proliferation, inhibited the apoptosis, and increased the chemotherapy resistance of NSCLC cells through upregulating EGFR expression and AKT phosphorylation. The EGFR inhibitor AST1306 reversed the RRM2-induced effects on cancer cells, indicating that the function of RRM2 is associated with EGFR/AKT signaling pathway. These findings suggested that miR-139- 5p inhibited the proliferation and promoted the apoptosis of NSCLC cells by upregulating RRM2/EGFR/AKT signaling pathway. The mutation of EGFR was thought main driver in NSCLC (49, 50), our data shown the mutation of EGFR in H1975 (L858R+T790M) and PC-9 (del19) has little affect on AFAP1-AS1. The underlying mechanism needs more research to lighten.

Taken together, our study demonstrates that AFAP1- AS1 expression is upregulated and miR-139-5p expression downregulated in NSCLC tissues and cells. AFAP1-AS1 promotes NSCLC development and increased chemotherapy resistance by modulating miR-139-5p/RRM2/EGFR/AKT pathway. Suppression of AFAP1-AS1 expression reduced tumor growth and attenuated chemotherapy resistance in vivo. Therefore, AFAP1-AS1 could be a promising and therapeutic target of NSCLC.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the manuscript/supplementary files.

# ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the ethics committee of first affiliated hospital of Chengdu Medical College. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by the Ethics Committee of Animal Experiments of Chengdu Medical College.

#### AUTHOR CONTRIBUTIONS

WG, KR, and NH performed the experiments, analyzed the data, and prepared the manuscript draft. YJ provided the human specimens. YJ, JS, and WD set up the experiments and repeated the key experiments. WZ and WL conceived the work, analyzed the data, and prepared the manuscript. All authors critically revised the manuscript, approved the final version, and agreed to be accountable for all aspects of the manuscript.

#### REFERENCES


#### FUNDING

This work was supported by the National Natural Science Foundation of China (81602636 and 31800154), the Fundamental Research Funds for the Central Universities (2242015K40034), Nanjing medical science and technology development project (ZKX15049), Jiangsu postdoctoral research grant (1601182B), Key project of science of Sichuan Education Department (18ZA0163 and 18ZA0164), and Natural Science Foundation of Chengdu Medical College (CYZ18-04).


lung cancer receiving chemotherapy. Tumour Biol. (2014) 35:1899–906. doi: 10.1007/s13277-013-1255-4


methylation in small cell lung cancer cells. Lab Invest. (2016) 96:60–8. doi: 10.1038/labinvest.2015.123


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Huang, Guo, Ren, Li, Jiang, Sun, Dai and Zhao. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Chronic BDE-47 Exposure Aggravates Malignant Phenotypes and Chemoresistance by Activating ERK Through ERα and GPR30 in Endometrial Carcinoma

Fan Zhang1,2†, Lin Peng3†, Yiteng Huang4†, Xueqiong Lin<sup>3</sup> , Li Zhou<sup>5</sup> \* and Jiongyu Chen1,2 \*

*<sup>1</sup> Oncology Research Laboratory, Cancer Hospital of Shantou University Medical College, Shantou, China, <sup>2</sup> Guangdong Provincial Key Laboratory for Breast Cancer Diagnosis and Treatment, Cancer Hospital of Shantou University Medical College, Shantou, China, <sup>3</sup> Department of Laboratory Medicine, Cancer Hospital of Shantou University Medical College, Shantou, China, <sup>4</sup> Health Care Center, The First Affiliated Hospital of Shantou University Medical College, Shantou, China, <sup>5</sup> Department of Gynecologic Oncology, Cancer Hospital of Shantou University Medical College, Shantou, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Nicoletta Filigheddu, University of Eastern Piedmont, Italy Ryszard Maciejewski, Medical University of Lublin, Poland*

#### \*Correspondence:

*Li Zhou zlyyzl@126.com Jiongyu Chen 513980484@qq.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *25 August 2019* Accepted: *30 September 2019* Published: *31 October 2019*

#### Citation:

*Zhang F, Peng L, Huang Y, Lin X, Zhou L and Chen J (2019) Chronic BDE-47 Exposure Aggravates Malignant Phenotypes and Chemoresistance by Activating ERK Through ER*α *and GPR30 in Endometrial Carcinoma. Front. Oncol. 9:1079. doi: 10.3389/fonc.2019.01079* Environmental exposure to certain compounds contribute to cell plasticity, tumor progression and even chemoresistance. 2,2′ ,4,4′ -tetrabromo diphenyl ether (BDE-47), one of the most frequently detected polybrominated diphenyl ethers (PBDEs) in environmental and biological samples, is a known estrogen disruptor closely associated with the development of hormone-dependent cancers. However, the effect of BDE-47 on endometrial carcinoma (EC), an estrogen-dependent cancer, remains to be elucidated. Mechanisms of estrogen receptor α (ERα) and G-protein-coupled receptor-30 (GPR30) involved in BDE-47 carcinogenesis are yet to be identified. This study aims to investigate the effect of BDE-47 on the invasive phenotype of estrogen-dependent EC cells. BDE-47-treated cells, such as Ishikawa-BDE-47 and HEC-1B-BDE-47 cells, exhibited increased cell viability and enhanced metastatic ability. *In vivo* studies showed larger tumor volumes and more metastasis in mice injected with Ishikawa-BDE-47 cells compared with parental Ishikawa cells. MTT assay showed that BDE-47 exposure could attenuate sensitivity of EC cells to cisplatin or paclitaxel treatment *in vitro*. Western blotting revealed overexpression of ERα, GPR30, pEGFR (phosphorylated epidermal growth factor receptor), and pERK (phosphorylated extracellular-regulated protein kinase) in Ishikawa-BDE-47 and HEC-1B-BDE-47 cells. Knockdown of ERα or GPR30 by small interfering RNA reversed the stimulating effect of BDE-47 on cell growth, migration and invasion of EC cells. Additionally, treatment with pEGFR or pERK inhibitor impaired cell viability, migration and invasion in Ishikawa-BDE-47 and HEC-1B-BDE-47 cells. Overall, our results indicate that chronic BDE-47 exposure triggers phenotypic plasticity, promotes progression and even chemoresistance in EC cells, at least in part, via ERα/GPR30 and EGFR/ERK signaling pathways.

Keywords: 2,2′ ,4,4′ -tetrabromo diphenyl ether, endometrial carcinoma, estrogen receptor α, G-protein-coupled receptor-30, cisplatin, paclitaxel, phosphorylated epidermal growth factor receptor, phosphorylated extracellularregulated protein kinase

# INTRODUCTION

Endometrial carcinoma (EC) is a major malignant tumor of the female reproductive system (1). The incidence of EC has been increasing in recent decades because of lifestyle change and environmental contamination (2). In China, EC was diagnosed in 63,400 patients and accounted for more than 21,800 deaths in 2016 (3). EC has been proposed to be classified into two pathogenetic groups, of which type I mostly occurs in pre- and peri-menopausal women with a history of unopposed estrogen exposure (4). However, studies found no difference in the positivity of estrogen receptor (ER) or progesterone receptor (PR) as well as the level of sex hormones between these two types of EC. Thus, type II is not completely estrogen-independent (5). Previous studies supported the role of ERα in the development of EC (4, 6). Additionally, G-protein-coupled receptor-30 (GPR30), which encodes a multi-pass membrane protein mediating the estrogen action of intracellular estrogen receptor (7, 8), has been suggested to be an indicator of EC progression (9). However, the etiology and pathogenesis of EC remain poorly defined.

Environmental exposure to certain compounds contributes to cell plasticity and tumor progression (10, 11); an example of these compounds is a group of flame retardants. These synthetic chemical additives are found in consumer products such as building materials, electronics and electrical goods, textiles, and furnishings. Polybrominated diphenyl ethers (PBDEs) are a class of flame retardants most commonly used in a variety of polymers and plastics, as well as in a broad range of consumer products. PBDEs have been produced in notable quantities (12), and consequently, their adverse health effects, environmental persistence and toxicity have raised concerns. Certain kinds of PBDEs (penta-, octa-, and deca-BDE) have been listed as prevalent organic pollutants, and have been banned or voluntarily phased out by manufacturers (13). However, consumer products containing large amounts of PBDEs are still in use and continue to release toxic chemicals into the environment (14). 2,2′ ,4,4′ -tetrabromodiphenyl ether (BDE-47), one of the most predominant PBDE congeners, is characterized by its lipophilic, bio-accumulative, degradation-resistant, and toxic properties. BDE-47 may enter the environment through volatilization and is ultimately found in dust, air, and seafood. It was reported that many products contain BDE-47 and landfill waste containing BDE-47 will persist for many years (15). BDE-47 enters the human body through oral ingestion and inhalation, and can become detectable in various types of human samples including blood, adipose tissue, and human milk (16).

Previous studies about the impact of BDE-47 mostly focused on neurological development, endocrine disruption and tumor initiation. Several studies have revealed that BDE-47 could impair neuronal differentiation (17). BDE-47 could cause DNA damage mediated by oxidative stress, and increase the in vitro migration and invasion of human neuroblastoma cells (18–21). Because of the association between PBDEs and hormone levels in humans (22), the impact of PBDEs on hormone-dependent cancers has become a topic of interest. BDE-47 was thought to be an estrogen disruptor with adverse effects on sexual behavior and reproductive function in zebra fish (23). Furthermore, BDE-47 could induce oxidative stress in MCF-7 cells by inhibiting the pentose phosphate pathway (16). An epidemiological survey reported that the serum concentration of BDE-47 in breast cancer women was significantly higher than that of controls (24). However, this pattern was not consistent across all cancers, for instance, BDE-47 could stimulate cell proliferation in human ovarian carcinoma cells OVCAR-3 but not in MCF-7 breast cancer cells (25), reflecting the complicated and inconsistent mechanisms underlying the effect of BDE-47 on different types of cancers.

Chemotherapy is commonly used to treat disseminated or recurrent EC, often after the failure of hormonal therapy. Although the management of EC has undergone a dramatic shift in recent years, and that early-stage EC has a favorable prognosis, the advanced or recurrent EC still has a poor prognosis partially because of chemoresistance. The underlying causes of drug resistance in EC are multi-factorial. Resistance to antimicrotubule agents such as paclitaxel and cisplatin (DDP) is particularly challenging given the importance of these agents in first-line treatment of EC (26). A recent study revealed that cadmium prevented the 5-fluorouracil cytotoxic effect by modifying cell cycle and apoptotic profiles in MCF-7 cells (27). Nonetheless, the potential antagonist effect of BDE-47 against chemotherapy sensitivity of EC has not been well-clarified.

Since EC is an estrogen-dependent cancer and BDE-47 could cause endocrine disruption, we hypothesized that BDE-47 might affect the progression and drug resistance of EC. In this study, the impact of BDE-47 on two human EC cell lines, Ishikawa and HEC-1B cells, was investigated. It has been found that chronic BDE-47 exposure could trigger phenotypic plasticity, promote progression, and even chemoresistance in EC cells, at least in part, via ERα/GPR30 and EGFR (epidermal growth factor receptor)/ERK (extracellular-regulated protein kinase) signaling pathways.

#### MATERIALS AND METHODS

# Cell Lines and Cell Culture

Two endometrial cancer cell lines, Ishikawa (ERαpositive/EGFR-positive), and HEC-1B (ERα-negative/EGFRpositive), were generously provided by Dr. Xiaolong Wei (Cancer Hospital of Shantou University Medical College, Shantou, China) and Dr. Bo Qiu (Southern Medical University, Guangzhou, China). Both these two cell lines have been authenticated. These cells were maintained in complete RPMI 1640 medium (Gibco, ThermoFisher Scientific Inc., California, US), supplemented with 10% fetal bovine serum (FBS, Biological Industry, Kibbutz BeitHaemek, Israel) at 37◦C in a 5% CO<sup>2</sup> incubator. To develop a chronically poisoned cell model, both Ishikawa and HEC-1B cells were exposed to 10µM BDE-47 (Lot No. 3798900, Chemservice Inc., Worms, Germany) for up to 45 days before the experiments, and were designated as Ishikawa-BDE-47 and HEC-1B-BDE-47, respectively.

#### Cell Treatment

To investigate the effect of BDE-47 on paclitaxel- and DDPinduced cytotoxicity in EC cells, Ishikawa-BDE-47 (10µM), HEC-1B-BDE-47 (10µM), and their parental cells (1 × 10<sup>4</sup> ) were treated with 0, 0.1, 1, 1.25, 5µM of paclitaxel (Bristol-Myers Squibb Company, New York, USA) and 0, 1.25, 2.5, 5, 10, 20, 50, 100µM of DDP (Hansoh pharma co. LTD, Jiangsu, China) for 48 h, respectively. After that cell viability was evaluated by MTT assays.

To further identify the cross-talk between ERα/GPR30 and EGFR/ERK signal pathway, 10µM erlotinib (No. #5083, Cell Signaling Technology Inc., Danvers, Massachusetts, US) and 20µM PD98059 (No. #9900, Cell Signaling Technology Inc., Danvers, Massachusetts, US) were used to inhibit EGFR autophosphorylation and ERK kinases for 48 h before MTT and Western blotting assay.

#### Transfection With siRNA

Twenty-four hours prior to transfection, approximately 8 × 10<sup>4</sup> cells (Ishikawa-BDE-47 or HEC-1B-BDE-47) were seeded into 6 well plates and grew to 70–80% confluence. These EC cells were then separately transfected with siERα (sc-29305, Santa Cruz Biotechnology, Inc., Dallas, Texas, US) to silence ERα expression, or with siGPR30 (sc-60743, Santa Cruz Biotechnology, Inc., Dallas, Texas, US) to silence GPR30 expression. Control siRNAs (sc-44230, Santa Cruz Biotechnology, Inc., Dallas, Texas, US) was used as the negative control for the parallel experiments. The EC cells were transfected with siRNA oligonucleotides (20µM) using lipofectamine 3000 (ThermoFisher Scientific Inc., California, US) following manufacturer's protocols, and then collected after 72 h for MTT and Western blotting assay.

#### MTT Assay

Cells were separately seeded into 96-well plates at a concentration of 4,000 cells/well, and then subjected to MTT assay. Briefly, MTT assay was performed as follows: cells were first incubated with 5 mg/ml of the sterile filtered MTT solution (Santa Cruz Biotechnology, Inc., Dallas, Texas, US) in phosphatebuffered saline (PBS) and incubated for 4 h in a moist chamber at 37◦C, followed by washing with PBS and incubation with DMSO for 10 min with shaking. Solubilized formazan product was detected at OD 490 nm using a microplate reader (Multiskan MK3, ThermoFisher Scientific Inc., California, US). The relative cell viability was calculated using the formula: (ODtreatment − ODblank)/(ODcontrol − ODblank), and the cell inhibition rate was assessed through the formula: 1 − the relative cell viability. Results were summarized by four technical replicates, and each experiment was repeated for triple times.

# Cell Migration and Invasion Assay

The transwell chamber (Lot #5011036, Falcon <sup>R</sup> Cell Culture Inserts, Krackeler Scientific lnc. Albany, New York, US) was used to detect cell migration and invasion. The bottom of the transwell chamber was made of a polycarbonate membrane with 8µm membrane pores. For invasion assays, an additional matrigel (50 µl: 50 mg/l, BD biosciences, Franklin Lakes, New Jersey, US) was used to cover the surface of the polycarbonate membrane. Each group of cells (1 × 10<sup>5</sup> cells suspended in 200 µl serum-free RPMI-1640 medium with 1% BSA) was seeded into the upper chamber, and 600 µl 10% FBS-supplemented RPMI-1640 medium was added to the lower chamber. After 24 h (migration assay) or 48 h (invasion assay), the upper chamber and cells on the upper surface of the membrane were removed. Cells transferred to the lower surface of the membrane were stained with 0.1% crystal violet, and the number of cells was counted under a Leica microscope (Model: DM3000, Wetzlar, Germany). Five fields from each sample were randomly selected for calculating stained cells (28). Results were summarized by three technical replicates, and each experiment was repeated for three times.

#### EC Xenografts

Twenty nude mice were purchased from Vital River Laboratory Animal Technology Co. Ltd (Beijing, China). The animal protocol was reviewed and approved by the Medical Animal Care and Welfare Committee at Shantou University Medical College. The body weights of the mice ranged from 20 to 25 g. The mice were randomly allocated into two groups. The right axilla of ten mice was subcutaneously injected with 4 × 10<sup>6</sup> Ishikawa or Ishikawa-BDE-47 (10µM) cells. The growth of xenograft tumors was measured every 3 days using a digital caliper, and tumor volume was calculated using the formula: Length × Width<sup>2</sup> × 0.5. After 31 days, mice were euthanized, and the xenograft tumors were dissected. Additionally, other ten mice were used to detect effect of BDE-47 on the migration ability of EC cells. In that case, 5 × 10<sup>6</sup> Ishikawa or Ishikawa-BDE-47 (10µM) cells were given intravenously through the tail. After 14 days, mice were euthanized, and potential metastasis were found and counted in the cervical, axillary, and abdominal lymph nodes, lung and liver. Lymph nodes were fixed and embedded in paraffin for histology and immunohistochemistry (IHC) staining analyses.

#### Hematoxylin-Eosin Staining and Immunohistochemistry

Lymph nodes of mice were collected, fixed in 10% phosphatebuffered formalin for 24 h, and embedded in paraffin wax. Sections of tumors (4µm in thickness) were stained with hematoxylin-eosin and pictured under a light microscope (DX45, Olympus Microsystems Ltd., Japan). IHC for Pan Cytokeratin (Kit-0009, MXB Biotechnologies, Fuzhou, China) was carried out on sections using a standard EnVision complex method. After deparaffinization and rehydration, endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 30 min. Then tissue sections were autoclaved at 121◦C in citrate buffer (pH 6.0) for 10 min, and incubated with mouse anti-Cytokeratin monoclonal antibody. IHC staining was carried out with the anti-Mouse/Rabbit (Kit-5030, MXB Biotechnologies, Fuzhou, China) and 3,3′ -diaminobenzidine as the chromogen substrate. Negative control was obtained by replacing the primary antibody with normal rabbit IgG.

# Western Blotting (WB)

For WB analysis, cells were first lysed with a cell lysis buffer containing PMSF (Phenylmethanesulfonyl fluoride, both from Beyotime, Shanghai, China) on ice for 30 min and centrifuged at 12,000 rpm for 15 min at 4◦C to remove cell debris. Proteins (50 µg) of each cell lysate were then separated by SDS-PAGE and transferred onto a PVDF membrane followed by blocking with Tris-buffered saline containing 0.05% Tween 20 (TBST) and 5% non-fat milk for 1 h at room temperature, washed 3 times for 5 min each in TBST, and incubated at 4◦C overnight with either mouse anti-GAPDH monoclonal antibody (1:3,000, ZSGB-BIO, Beijing, China), rabbit anti-phospho-EGF receptor (Tyr1148) polyclonal antibody (EGFR antibody, 1:1,000, #4404, Cell Signaling Technology Inc., Danvers, Massachusetts, US), rabbit anti-GPR30 polyclonal antibody (1:1,000, sc-48525- R, Santa Cruz Biotechnology Inc., Dallas, Texas, US), rabbit ERα monoclonal antibody (1:1,000, #2294327, Millipore Inc., Darmstadt, Germany), rabbit P-EGFR antibody (Y1148) (pEGFR antibody, 1:1,000, #4404S, Cell Signaling Technology Inc., Danvers, Massachusetts, US), rabbit P-p44/42 MAPK antibody (T202/Y204) (pERK antibody, 1:1,000, #4370S, Cell Signaling Technology Inc., Danvers, Massachusetts, US), or rabbit p44/42 MAPK antibody (ERK antibody, 1:1,000, #4695S, Cell Signaling Technology Inc., Danvers, Massachusetts, US) in blocking buffer. Following washes with TBST (3 times for 5 min each), the blots were incubated with horseradish peroxidase-labeled antirabbit (1:5,000, Novus Biologicals, Littleton, Massachusetts, US) or anti-mouse (1:3,000, Santa Cruz Biotechnology Inc., Dallas, Texas, US) IgG at room temperature for 2 h, washed with TBST, and observed through chemiluminescence (ChemiDocTM XRS+, Bio-Rad Laboratories, Inc., Hercules, California, US).

#### Statistical Analysis

Statistical analyses were performed using the SPSS 13.0 software package (SPSS Inc., Chicago, IL). The comparisons of cell viability, cell numbers representing the capacity of cell migration and invasion, and xenograft tumor size between different treatment groups were conducted using t-tests. For all tests, a P-value of <0.05 was considered significant.

#### RESULTS

#### BDE-47 Boosted Viability and Metastatic Capacity of EC Cells

Ishikawa and HEC-1B cells were incubated with different concentrations (0.1, 1, 2.5, 5, 10, 20, 40, and 80µM) of BDE-47. The positive effect of BDE-47 on cell viability was observed at concentrations of 10µM (relative cell viability at 48 and 72 h after BDE-47 treatment: 1.246, P = 0.005; 1.416, P < 0.001) and 20µM (1.218, P = 0.014; 1.319, P < 0.001) in Ishikawa cells, as well as in HEC-1B cells at concentrations of 10µM (1.282, P < 0.001; 1.416, P < 0.001) and 20µM (1.179, P < 0.001; 1.265, P = 0.001) in contrast to that in their parental cells (**Figure 1A**). Based on these data, 10µM was selected as the chronic BDE-47 treatment concentration in Ishikawa and HEC-1B cells. MTT assays showed that the cell viability of Ishikawa-BDE-47 cells was slightly increased in contrast to Ishikawa cells after 24 h (OD values: 0.42 vs. 0.35, P = 0.045) and 48 h (0.63 vs. 0.55, P = 0.031), and the increase was more obvious after 72 h (1.23 vs. 0.90, P = 0.008) and 96 h (1.55 vs. 1.23, P = 0.001). A similar trend was observed in HEC-1B-BDE-47 cells and their parental cell line, of which OD values were 0.36 and 0.27 at 24 h (P = 0.010), 0.51 and 0.36 at 48 h (P = 0.001), 1.16 and 0.84 at 72 h (P = 0.019), and 1.32 and 0.93 at 96 h (P = 0.013; **Figure 1B**).

To examine whether BDE-47 could promote metastasis in EC cells, transwell chamber assays were performed. As shown in **Figure 1C**, the average number of Ishikawa-BDE-47 cells migrating across the membrane was significantly higher than that of the parental cells (89.6 ± 8.0 vs. 44.0 ± 8.1, P < 0.001). Similarly, the average number of migrated HEC-1B-BDE-47 cells was evidently higher than that of HEC-1B cells (75.0 ± 13.7 vs. 18.4 ± 5.8, P < 0.001). Moreover, both Ishikawa-BDE-47 (number of invaded cells: 49.8 ± 8.8 vs. 11.2 ± 0.45, P < 0.001) and HEC-1B-BDE-47 (19.6 ± 2.5 vs. 6.0 ± 1.3, P < 0.001) showed more invasive ability as compared to their parental cells. These results indicated that Ishikawa-BDE-47 and HEC-1B-BDE-47 cells have increased the cell viability and metastatic capacity.

#### BDE-47 Enhanced the Growth and Metastasis of EC in vivo

In order to understand the effect of BDE-47 on EC cell growth in vivo, Ishikawa cells were subcutaneously injected into mice. As shown in **Figure 1D**, tumor volumes of Ishikawa xenografts on days 10, 22, and 31 reached 11.08 mm<sup>3</sup> (±4.38), 167.11 mm<sup>3</sup> (±37.46), and 296.78 (±65.71) mm<sup>3</sup> , respectively, while larger tumor sizes were observed in Ishikawa-BDE-47 xenografts: 39.85 mm<sup>3</sup> (±14.83, P = 0.010), 261.76 mm<sup>3</sup> (±61.91, P = 0.040), and 408.95 mm<sup>3</sup> (±54.83, P = 0.040). Additionally, potential metastasis were found and quantified. In the mice inoculated with Ishikawa-BDE-47 cells, there were 22, 10, and 4 metastatic tumors confirmed by cytokeratin IHC staining (**Figures 1E,F**, **Table 1**) in the lymph nodes of the neck, axilla and enterocoele of these mice, respectively. However, only one metastasis in the lymph nodes of the neck was found in the mice with Ishikawa xenografts (**Figures 1E,F**, **Table 1**). The cellular morphology of metastasis was assessed by HE staining (**Figure 1F**).

#### BDE-47 Lessened the Paclitaxel and DDP Cytotoxic Effects on EC Cells

We next explored whether BDE-47 exposure could attenuate sensitivity of EC cells to DDP or paclitaxel. As shown in **Figures 2A,B**, both of Ishikawa-BDE-47 and HEC-1B-BDE-47 cells consistently exhibited lower DDP sensitivity compared to their parental non-treated cells at different concentrations. The cell inhibition rates of Ishikawa-BDE-47 at the concentration of 2.5 and 5 µg/l were 0.275 and 0.436, while these numbers of Ishikawa cells were 0.555 and 0.711 (P = 0.007; P < 0.001). Similarly, the cell inhibition rates in HEC-1B at the concentrations of 2.5, 5, and 10 µg/l were 0.517, 0.609, and 0.679, but separately reduced to 0.234, 0.337, and 0.573 in HEC-1B-BDE47 cells (P = 0.002; P = 0.003; P = 0.014). Regarding paclitaxel, BDE-47 treatment also reduced its cytotoxicity to Ishikawa and HEC-1B cells. As seen in **Figure 2C**, Ishikawa-BDE-47 had significantly lower cell inhibition rates than that in Ishikawa cells, when exposed to paclitaxel (at 0.1 µg/l: 0.088 vs. 0.341, P = 0.029; 1 µg/l: 0.158 vs. 0.339, P < 0.001; 2.5 µg/l: 0.332 vs. 0.688, P < 0.001; 5 µg/l: 0.689 vs. 1.018, P < 0.001). Likewise, HEC-1B-BDE-47 displayed an

Ishikawa/Ishikawa-BDE-47 and HEC-1B/HEC-1B-BDE-47 cells. Representative images (400×) of cell migration (C1–C4) and invasion (C5–C8) are shown on the left side. (D) Primary xenograft tumors of Ishikawa cells and tumor volumes. (E) Lymph nodes containing potential metastasis from mice were found and shown. (F) Upper panel: HE staining assessed cellular morphology of lymph nodes with or without metastatic tumors. F1: 100×, F2: 400×, F3: 100×, F4: 400×). Lower panel: IHC staining of cytokeratin negative in F5 (100×) and F6 (400×), and positive in F7 (100×) and F8 (400×). CK, Cytokeratin; Con, Ishikawa cells without BDE-47 treatment. \**P* < 0.05; \*\**P* < 0.01; \*\*\**P* < 0.001.

inferior cell inhibition rate with the treatment of paclitaxel at the concentration of 5 µg/l compared to its parental cells (0.809 vs. 0.993, P = 0.002, **Figure 2D**).

#### BDE-47 Increased Viability and Metastatic Capacity of EC Cells Through the ERα/GPR30 and ERFR/ERK Signaling Pathways

Estrogens are steroid hormones that regulate a plethora of physiological processes in mammals. Activation of ERα by estrogens is a well-known mechanism underlying EC development. Expression of GPR30 has been suggested to be a novel indicator of EC progression (9). As shown in **Figure 3A**, levels of both ERα and GPR30 proteins were elevated in ERα-positive Ishikawa-BDE-47 cells when compared to that in their parental cell line. In ERα-negative HEC-1B cells, BDE-47 treatment significantly up-regulated GPR30 expression. Since activation of EGFR/ERK signaling pathway through ERα/GPR30 was present in other cancer cells (29), we detected proteins of EGFR, pEGFR, ERK, and pERK by Western blotting. Significant up-regulation of pEGFR and pERK, but not total EGFR and ERK expression, was seen in Ishikawa-BDE-47 cells compared to that in Ishikawa cells. A similar phenomenon was observed in HEC-1B-BDE-47 and its parental cells. These results indicated a cross talk between ERα/GPR30 and ERFR/ERK signaling pathways potentially underlying the increased proliferative and metastatic effect of BDE-47.



*<sup>a</sup>Total number of extracted lymph nodes.*

*<sup>b</sup>Lymph nodes without metastasis.*

*<sup>c</sup>Lymph nodes with metastasis.*

*<sup>d</sup>The proportion of lymph nodes with metastasis to all extracted lymph nodes.*

*<sup>e</sup>Total number of extracted lymph nodes from neck, axilla, or enterocoele.*

*<sup>f</sup> Lymph nodes without metastasis from neck, axilla, or enterocoele.*

*<sup>g</sup>Lymph nodes with metastasis from neck, axilla, or enterocoele.*

*<sup>h</sup>The proportion of lymph nodes with metastasis to all extracted lymph nodes from neck, axilla, or enterocoele.*

#### Down-Regulation of ERα/GPR30 Attenuated Viability and Metastatic Capacity of EC Cells Induced by BDE-47

To investigate the role of ERα/GPR30 in BDE-47 induced EC cell viability, Ishikawa-BDE-47 (ERα+/GPR30+) cells were treated with siERα and siGPR30. Because of its ER-negativity, HEC-1B-BDE-47 cells were only treated with siGPR30. As shown in **Figure 3B**, significant reductions in ERα and GPR30 expression were observed in Ishikawa-BDE-47-siER and Ishikawa-BDE-47-siGPR30 cells in contrast to that in Ishikawa-BDE-47 cells. Similarly, GPR30 expression was reduced in HEC-1B-BDE-47 siGPR30 compared to that in HEC-1B-BDE-47 cells. MTT assay also revealed significant inhibition of cell viability in Ishikawa-BDE-47-siER cells (OD values at 24, 48, 72 h: 0.42 vs. 0.40, P = 0.500; 0.63 vs. 0.48, P = 0.005; 1.17 vs. 0.66, P < 0.001) and Ishikawa-BDE-47-siGPR30 cells (OD values at 24, 48, 72 h: 0.46 vs. 0.39, P = 0.296; 0.65 vs. 0.46, P = 0.008; 1.17 vs. 0.67, P < 0.001) when compared to their parental cells. Similarly, a significant reduction in cell viability was seen in HEC-1B-BDE-47-siGPR30 in contrast to HEC-1B-BDE-47 (OD values at 24, 48, 72 h: 0.44 vs. 0.39, P = 0.367; 0.63 vs. 0.48, P = 0.012; 1.11 vs. 0.62, P < 0.001). **Figure 4A** illustrates the MTT results of siRNA at 72 h. Additionally, the transwell assay revealed an impaired capacity of metastasis in Ishikawa-BDE-47-siER (number of migrated cells: 44.0 vs. 81.4, P < 0.001; number of invaded cells: 37.0 vs. 48.2, P = 0.003, **Figures 4B–D**) and Ishikawa-BDE-47-siGPR30 (number of migrated cells: 47.8 vs. 76.4, P < 0.001; number of invaded cells: 39.8 vs. 57.6, P = 0.003, **Figures 4B–D**) compared to the corresponding parental cells. A similar trend was seen in HEC-1B-BDE-47-siGPR30 and its parental cells (number of migrated cells: 51.6 vs. 64.4, P = 0.010; number of invaded cells: 17.6 vs. 26.2, P = 0.009, **Figures 4B–D**).

#### Inhibition of pEGFR or pERK Weakened Viability and Metastatic Capacity in BDE-47-Treated EC Cells

As shown in **Figure 3B**, gene knock-down using siGPR30 or siER down-regulated the expression of pERK and pEGFR but not total ERK and EGFR. To further demonstrate the role of EGFR/ERK signaling pathway in EC cell viability induced by BDE-47, a pEGFR inhibitor or a pERK inhibitor was applied to Ishikawa-BDE-47 and HEC-1B-BDE-47 cells (**Figure 3C**). As shown in **Figure 4A**, the pEGFR inhibitor significantly down-regulated the extent of cell viability and metastasis stimulated by BDE-47 in both Ishikawa-BDE-47 and HEC-1B-BDE-47 cells. OD values at 48 h of Ishikawa-BDE-47 treated with and without the pEGFR inhibitor at 10µM were 0.36 and 0.68 (P < 0.001), indicating significant inhibition. Similarly, notable suppression was seen in HEC-1B-BDE-47 cells treated with the pEGFR inhibitor at a concentration of 10µM (OD values at 48 h: 0.34 vs. 0.64, P < 0.001). The EGFR/ERK signaling pathway is the classical MAPK pathway. The pERK inhibitor seemed to block the effect of BDE-47 on Ishikawa and HEC-1B cells. As shown in **Figure 4A**, OD values at 48 h of Ishikawa-BDE-47 was 0.68, but this was reduced to 0.38 (P < 0.001) after the pERK inhibitor treatment at a concentration of 20µM. Similarly, OD value of HEC-1B was reduced from 0.64 to 0.34 (P < 0.001) after the pERK inhibitor treatment. These results suggested that the EGFR/ERK signaling pathway might be at the downstream of ER/GPR30 and be involved in BDE-47-induced EC cell viability.

The transwell assay was used to examine the metastatic capacity of Ishikawa-BDE-47 and HEC-1B-BDE-47 cells treated with a pEGFR inhibitor at 10µM or a pERK inhibitor at 20µM for 48 h. After the addition of a pEGFR inhibitor, the number of migrated and invaded cells was reduced from 78.0 ± 8.8 to 38.8 ± 13.7 (P < 0.001), and from 63.0 ± 2.7 to 38.8 ± 5.2 (P < 0.001, **Figures 4B–D**), respectively. The same trend of cell invasion and migration was also seen in HEC-1B-BDE47 cells after the pEGFR inhibitor treatment. The number of migrated cells was reduced from 49.8 ± 6.5 to 19.2 ± 2.2 (P < 0.001), and the number of invaded cells was reduced from 25.0 ± 3.6 to 19.2 ± 2.2 (P = 0.015). Likewise, as shown in **Figures 4B–D**, treatment with a pERK inhibitor for 48 h impaired the metastatic ability of Ishikawa-BDE-47 and HEC-1B-BDE-47 cells, accompanied with reductions in the numbers of migrated and invaded cells from 77.0 ± 5.43 to 24.2 ± 3.77 (P < 0.001, **Figure 4D** left panel for Ishikawa-BDE-47), from 48.2 ± 4.92 to 20.4 ± 5.86 (P < 0.001, **Figure 4D** right panel for Ishikawa-BDE-47), from 75.0 ± 13.69

to 25.2 ± 3.49 (P < 0.001, **Figure 4D** left panel for HEC-1B-BDE-47), and from 19.6 ± 2.88 to 12.2 ± 2.05 (P = 0.001, **Figure 4D** right panel for HEC-1B-BDE-47).

# DISCUSSION

PBDEs are considered as endocrine-disrupting chemicals with thyroxine- and estrogen-like effects (30). One publication by Li et al. revealed that PBDE-209 increased the viability and proliferation of cells in several types of cancer, including breast cancer, cervical cancer, and ovarian cancer. Interestingly, PBDE-209-up-regulated phosphorylation of ERK was also observed (31). In colon cancer HCT-116 cells, BDE-99 was found to increase the cell migration and invasion as well as to trigger EMT (epithelial–mesenchymal transition), most likely via the PI3K/AKT/Snail signaling pathway (11).

Interestingly, BDE-47 could exert a differential effect on different types of cancer cells. In human neuroblastoma SH-SY5Y cells, BDE-47 had limited cytotoxicity but significantly increased the in vitro cell migration and invasion by up-regulating MMP-9 through the GPR30/PI3K/Akt signaling pathway (21). In OVCAR-3cells, BDE-47 was found to stimulate cell proliferation by activating CDK1, CDK7, E2F1, and E2F2; however, this effect was not observed in MCF-7 cells. Notably, BDE-47 had no effect on ERα protein expression in OVCAR-3 cells, while decreased ERα protein expression in MCF-7 cells. Additionally, BDE-47 had no effect on ERK phosphorylation in the two cell lines (25). In contrast, Kanaya et al. recently reported that BDE-47 stimulated proliferation of an estrogen-dependent breast cancer cell line MCF-7aroERE and induced ER-regulated genes expression, which suggested that BDE-47 acted as a weak agonist of both ERα and estrogen-related receptor α (ERRα) (32). One previous case-control study from our laboratory demonstrated that BDE-47 level was positively with breastcancer risk regardless of ER stratification (33). Similarly in this study, BDE-47 treatment enhanced the cell growth, invasion and migration in both Ishikawa and HEC-1B cells.

The differences between the above studies may attribute to cell specificity, exposure time, and moreover, a different effect for the parent compound and its metabolites. We noticed that the exposure model in Karpeta's research was short-term treatment for 72 h, while that in Kanaya's paper and the present study were longer for 5 days and up to 45 days, respectively. Furthermore, hydroxylated metabolites of PBDE were found to be more potent agonists of estrogen receptors than the parent compounds (34). For instance, BDE-47 had no effect on ERα and ERβ protein expression in OVCAR-3 cells, whereas 5-OHBDE-47 upregulated ERα only and 6-OH-BDE-47 increased both ERα and ERβ protein expression (25). Previous studies have indicated that CYP2B6, a predominant cytochrome P450s, was involved in the

formation of hydroxylated PBDEs (OH-PBDEs) (35). Although no evidence has been established for the expression of this enzyme in EC cells, detecting hydroxylated BDE-47 in the culture medium of Ishikawa-BDE-47 and HEC-1B-BDE-47 cells will help to exclude the potential effect of hydroxylated metabolites of BDE-47 on EC cell biology in the present study.

ERα, a ligand-activated transcription factor localizing in the nucleus, is expressed in approximately 80–90% of endometrioid tumors. The estrogen-ERα signaling pathway is implicated in increased uterine growth (36). The newly discovered membrane estrogen receptor GPR30, expressed in ∼80% of endometrioid tumors (5), is a specific receptor for 17β-estradiol involved in the non-genomic effect of estrogen. GPR30 was also addressed to mediate the proliferative and invasive effects of estrogen and tumorigenesis in EC cell line (37) and suggested to be an indicator of clinical outcomes of EC (9). It is well-known that the classical mechanism of estrogen action involves binding to the ERα and ERβ, which regulates transcription through direct binding at estrogen-responsive elements in the promoter region of target genes or through tethering to other well-known transcription factors such as AP1, SP1, and nuclear factor κB (38). Emerging evidence has indicated that GPR30 mediated invasion and carcinogenesis induced by 17-β estradiol in an EC cell line (37). To highlight a possible estrogen receptors involvement in the effect of BDE-47 on EC cells, we detected the status of ERα and GPR30 in these two cell lines. The upregulation of ERα and GPR30 in Ishikawa-BDE-47 cells and increased GPR30 expression in HEC-1B-BDE-47 suggest that both ERα and GPR30 are the main targets of BDE-47. Furthermore, siRNA for ERα or GPR30 in BDE-47-treated Ishikawa and HEC-1B cells attenuated the catalytic role of BDE-47 on malignant phenotypes. Since the upregulation of ERα and GPR30 in BDE-47-treated EC cells, we speculate BDE-47 carried out its cancer-promoting effect on EC cells via estrogen molecular pathways. Mitogen-activated protein kinase (MAPK) cascades are key signaling pathways involved in various cellular functions, including cell proliferation, differentiation, survival, and metastasis. EGFR/ERK is the classical MAPK signaling pathway. Recently, a link between environmental chemicals such as Bisphenol A and cancer progression has been suggested in ER-negative in flammatory breast cancer cells, in which EGFR/ERK signaling was involved (39). It is believed that the cross-talk between EGFR and ERα plays a critical role in the regulation of breast cancer development (40). Polychlorinated biphenyl 104 was found to promote migration and invasion of endometrial stromal cells by inducing the expression of MMP-3 and MMP-10, which may involve in the EGFR signaling pathway (41). The cross talk between GPR30 and EGFR/ERK signaling pathways is supported by other studies. GPR30-EGFR signaling pathway was reported to be involved in Bisphenol A promoting the proliferation of leiomyoma cells (42). One study from Zhang et al. (43) revealed that tamoxifen, known for its agonist activity on GPR30, could up-regulate the phosphorylated protein expression of both EGFR and ERK in EC cells, without altering total EGFR/ERK protein expression, indicating that the growth stimulation of EC cells in response to 4-OH tamoxifen

was via the GPR30/EGFR pathway. Another publication noted that both estradiol and tamoxifen could induce cell migration through GPR30 in EC cells with or without ERα expression, accompanied by elevated ratios of pEGFR/Total-EGFR and pERK/Total-ERK. ERα could only be partially involved in this progress (44). Unlike in either the OVCAR-3 cells or MCF-7 cells mentioned above, increased phosphorylation of EGFR and ERK were observed in the BDE-47-treated EC cells. Simultaneously, siRNA for ERα and GPR30 reduced phosphorylated EGFR and ERK expression levels. Moreover, BDE-47-induced estrogen-like effect was partially impaired in cells treated with either an EGFR inhibitor (erlotinib) or an ERK inhibitor (PD98059). The present study is the first to investigate the cross talk between EGFR/ERK and ERα/GPR30 underlined the stimulation of EC cells by BDE-47. GPR30/ERα and EGFR/ERK seem to be the general pathways mediating the impact of BDE-47 on the development of EC.

Previous studies indicated that cadmium, a metalloestrogen, influenced the 5-Fluorouracil cytotoxic effects on breast cancer cells (27, 45). Both clinical observations and experimental studies have suggested that steroid hormones and their receptors affect the therapeutic efficacy of antineoplastic drugs (46). It was demonstrated that ERα/17β-estradiol attenuates therapeutic efficacy of paclitaxel on breast xenograft tumors (47). Another in vitro study showed that the increase in ERα expression in ERα-negative Bcap37 breast cancer cells significantly increased their chemoresistance, whereas ERα activation by 17β-estradiol increased the sensitivity of natural ERα-positive T47D breast cancer cells to chemotherapeutic agents (48). Considering the upregulation of ERα and/or GPR30 in BDE-47-treated EC cell lines, we speculate that estrogen signaling pathway is involved in the chemoresistance mechanism. In the present study, the resistance to paclitaxel and DDP induced by BDE-47 was enhanced in both Ishikawa and HEC-1B cells, which suggests that not only the ER alpha status but also other molecular mechanism influence on observed results. Interestingly, upregulating GPR30 was also found in high-risk endometrial cancer patients with lower survival rates (9) and in tamoxifen-resistant breast cancer cells through the EGFR/ERK transduction pathway (49). More exhaustive and systematic studies are essential to reach deeper understandings on the cross talk between ERα/GPR30 and EGFR/ERK signaling pathway involved in BDE-47 induced chemoresistance as well as the chemoresistance differences in HEC-1-B-DBE-47 and Ishikawa-BDE-47 cells. It suggests that considering the influence of BDE-47 exposure on chemosensitivity, determining the body BDE-47 burden in EC patients might be taken into account when using chemotherapeutic drugs.

The concentration used in this BDE-47 exposure model is based on the dose-response experiment and the data shown previously in other in vitro systems (34, 50–52). Recently, Kanaya compared with in vitro breast cancer cell culture treatment at dosage of 10µM to the published human serum/tissue

#### REFERENCES


concentrations of PBDEs (BDE-47 included), and concluded that the maximum concentrations observed in human serum and tissue concentrations were ∼1,300 and 350 times lower than 10µM, respectively (32). Therefore, future study will be needed to advance understanding the effect of environmentally relevant low-dose PBDEs on the progression and chemoresistance of EC cells in vitro and in vivo, which will help to answer the critical question whether real-life BDE-47 exposure could be a risk of human EC and resistance factor for chemotherapy.

In summary, this preliminary study shows BDE-47 promotes cell growth, migration and even chemoresistance of EC cells both in vivo and in vitro. We speculate that this progress, at least in part, mimicks the effects of estrogen. The cross talk between ERα/GPR30 and the EGFR/ERK signaling pathway might be a crucial mechanism underlying the impact of BDE-47 on EC cell plasticity and chemoresistance.

#### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

#### ETHICS STATEMENT

The animal study was reviewed and approved by Medical Animal Care & Welfare Committee of Shantou University Medical College.

#### AUTHOR CONTRIBUTIONS

FZ performed the statistical analyses and drafted the manuscript. JC designed and performed the experiments. YH and LP helped to draft and revise the manuscript. XL helped to revise this manuscript. LP and LZ conceived of the study and supervised the work. All authors read and approved the final manuscript.

#### FUNDING

This work was supported by the Science and Technology Planning Project of Guangdong Province, China (No. 2014A020212287), Science and Technology Planning Project of Shantou City, China (No. [2018]121-12), and Shantou University Medical College Clinical Research Enhancement Initiative (No. [2014]29-13).

cancer: a case report and literature review. Medicine. (2017) 96:e9223. doi: 10.1097/MD.0000000000009223


function in male zebrafish (Danio rerio). Ecotoxicol Environ Saf. (2015) 111:102–8. doi: 10.1016/j.ecoenv.2014.09.037


human breast cancer MCF-7 cells. Toxicol Appl Pharmacol. (2015) 287:26–34. doi: 10.1016/j.taap.2015.05.010


cells. J Exp Clin Cancer Res. (2012) 31:42. doi: 10.1186/1756-99 66-31-42


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhang, Peng, Huang, Lin, Zhou and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Tumor Suppressor Role of Zinc Finger Protein 671 (ZNF671) in Multiple Tumors Based on Cancer Single-Cell Sequencing

Jian Zhang1,2†, Jianli Luo3†, Huali Jiang4†, Tao Xie1,2†, Jieling Zheng<sup>5</sup> , Yunhong Tian1,2 , Rong Li 1,2, Baiyao Wang1,2, Jie Lin1,2, Anan Xu1,2, Xiaoting Huang1,2 and Yawei Yuan1,2 \*

*<sup>1</sup> Department of Radiation Oncology, Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou, China, <sup>2</sup> State Key Laboratory of Respiratory Diseases, Guangzhou Institute of Respiratory Disease, Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou, China, <sup>3</sup> Department of General Disease, Health Center of Shuichun Town, Shanwei, China, <sup>4</sup> Department of Cardiovascularology, Tungwah Hospital of Sun Yat-sen University, Dongguan, China, <sup>5</sup> Department of Radiation Oncology, Nanfang Hospital, Southern Medical University, Guangzhou, China*

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, China*

#### Reviewed by:

*Fahd Al-Mulla, Genatak, Kuwait Zi-Qi Zheng, Sun Yat-sen University Cancer Center (SYSUCC), China*

> \*Correspondence: *Yawei Yuan yuanyawei@gzhmu.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *10 June 2019* Accepted: *23 October 2019* Published: *08 November 2019*

#### Citation:

*Zhang J, Luo J, Jiang H, Xie T, Zheng J, Tian Y, Li R, Wang B, Lin J, Xu A, Huang X and Yuan Y (2019) The Tumor Suppressor Role of Zinc Finger Protein 671 (ZNF671) in Multiple Tumors Based on Cancer Single-Cell Sequencing. Front. Oncol. 9:1214. doi: 10.3389/fonc.2019.01214* In humans, zinc finger protein 671 (*ZNF671*) is a type of transcription factor. However, the contribution of tumor heterogeneity to the functional role of *ZNF671* remains unknown. The present study aimed to determine the functional states of *ZNF671* in cancer single cells based on single-cell sequencing datasets (scRNA-seq). We collected cancer-related *ZNF671* scRNA-seq datasets and analyzed *ZNF671* in the datasets. We evaluated 14 functional states of *ZNF671* in cancers and performed *ZNF671* expression and function state correlation analysis. We further applied t-distributed stochastic neighbor embedding to describe the distribution of cancer cells and to explore the functional state of *ZNF671* in cancer subgroups. We found that *ZNF671* was downregulated in eight cancer-related *ZNF671* scRNA-seq datasets. Functional analysis identified that *ZNF671* might play a tumor suppressor role in cancer. The heterogeneous functional states of cell subgroups and correlation analysis showed that *ZNF671* played tumor suppressor roles in heterogeneous cancer cell populations. Western blot and transwell assays identified that *ZNF671* inhibited EMT, migration, and invasion of CNS cancers, lung cancer, melanoma, and breast carcinoma *in vitro*. These results from cancer singlecell sequencing indicated that *ZNF671* played a tumor suppressor role in multiple tumors and may provide us with new insights into the role of *ZNF671* for cancer treatment.

Keywords: ZNF671, tumor suppressor, solid tumor, single-cell sequencing, data mining

# INTRODUCTION

Cancer is a complex ecosystem composed of cells with heterogeneous functional states, leading to both therapeutic resistance, and frequent cancer recurrence or metastasis, which poses a major obstacle to cancer diagnosis and treatment (1–3). Some tumor cells have high proliferative or apoptotic capacity, some have invasion and metastasis activities, some show stem-like properties, and some exhibit a quiescent state (4, 5). These functionally heterogeneous cancer cells act cooperatively or competitively during tumor progression or metastasis, leading to distinct tumor phenotypes (6–8). Therefore, it is essential to systematically and comprehensively identify the functional states of cancer cells.

Single-cell mRNA-sequencing (scRNA-seq) provides a powerful tool for characterizing the omic-scale features of heterogeneous cell populations (9, 10). ScRNA-seq technologies permit the dissection of primary tumor cells, metastatic tumor cells, cancer stem cells (CSC), circulating tumor cells (CTC), and disseminated tumor cells in a comprehensive and unbiased manner, with no need of any prior knowledge of the cell population. ScRNA-seq has become a reference tool for analyzing the composition of cancer tissues and for establishing the characteristics of the cellular microenvironment (11). Thus, understanding single cancer cells will advance our understanding of not only therapeutic resistance but all facets of cell biology. Furthermore, the application of scRNA-seq in the clinic has the potential to change our approach to cancer management fundamentally (12).

Zinc finger protein 671 (ZNF671) is a member of the KRAB-ZF (KRAB-ZFP) family of mammalian transcriptional repressors (13–15). KRAB-ZFPs can regulate tumor cell differentiation, proliferation, apoptosis, invasion, metastasis, and transformation (16–21). Previous studies showed that ZNF671 could act as a tumor suppressor in several solid tumors (22–26). Our studies identified that ZNF671 played a tumor suppressor role in breast invasive carcinoma (BRCA), cervical squamous cell carcinoma, and endocervical adenocarcinoma (CESC), head and neck squamous cell carcinoma (HNSC), kidney renal papillary cell carcinoma (KIRP), lung adenocarcinoma (LUAD), pancreatic adenocarcinoma (PAAD), and uterine corpus endometrial carcinoma (UCEC) (26, 27). However, the roles of ZNF671 in the functional heterogeneity of cancer single cells remain unclear.

In this study, we analyzed the expression of ZNF671 in cancer scRNA-seq datasets systematically. We explored the functional role of ZNF671 in solid tumors and analyzed its expression and functional correlation in tumors. We further described the distribution of cancer single cells and explored their functional relevance in different tumor cell subgroups. Our results provide important insights into tumor heterogeneity and enhance knowledge of the tumor suppressor role of ZNF671 in solid tumors.

#### MATERIALS AND METHODS

#### Data Collection

Data were collected based on the following keywords: ("single cells" OR "single cell" OR "single-cell" OR "single-cells") AND ("transcriptome" OR "transcriptomics" OR "scRNA-seq" OR "scRNA seq" OR "RNA-sequencing" OR "RNA-seq" OR"RNA sequencing") AND ("carcinoma" OR "tumor" OR "tumor" OR "cancer" OR "neoplasm" OR "neoplastic"). According to the method used by Yuan et al. (28), three human data sets from Array Express, Sequence Read Archive (SRA), and Gene Expression Omnibus (GEO) datasets were collected and all single-cell data in these datasets were analyzed via expression quantification, quality control, and characterization of functional states.

#### Data Processing

Transcript expression quantification was performed using Salmon (version 0.9.1) with the optional parameter k (k = 31 for long reads and k = 15 for short reads). The GENCODE (Release 28, GRCh38) reference transcriptome was used to detect gcBias, seqBias, and other default parameters in the quasi-mappingbased mode. For scRNA-seq datasets with only an expression matrix, we directly converted the expression values to transcripts per million (TPM)/counts per million (CPM) values using a custom script. Expression values were log2 transformed with an offset of 1.

#### Characterizing Functional States of Cancer Single Cells

After reviewing cancer single-cell sequencing studies, 14 crucial functional states of cancer cells were selected, including angiogenesis, apoptosis, cell cycle, differentiation, DNA damage, DNA repair, epithelial–mesenchyme transition (EMT), hypoxia, inflammation, invasion, metastasis, proliferation, quiescence, and stemness using Gene Ontology, MSigDB, Cyclebase, HCMDB, and StemMapper (29–33). According to the method used by Yuan et al. (28), the activities of the 14 functional states across cancer single cells in the datasets were assessed using the Gene Set Variation Analysis (GSVA) package downloaded from http://www.bioconductor.org (34).

#### Dimensionality Reduction Using t-distributed Stochastic Neighbor Embedding (t-SNE) Analysis

According to the method used by Li et al. (35), donor files were imported into R, and expression matrices containing measured intensities at the single-cell level were extracted from the flowCore package. A subset of cells was selected for each donor at random and merged into a single expression matrix before t-SNE analysis. The beads, viability, center, offset, residual, event length, intercalator, and time channels were removed from the expression matrix. The ZNF671 protein marker was the only factor included in the t-SNE analysis, and ZNF671 intensities were transformed using the inverse hyperbolic sine (arcsinh) function.

T-SNE calculations were performed with 1,000 iterations, a perplexity parameter of 30, and a trade-off θ of 0.5, which was used to visualize similarities and the proximity of cells in a twodimensional plot. T-SNE maps were generated by plotting each event of the t-SNE dimensions in a dot-plot. ZNF671 intensities were overlaid on the dot-plot to show the expression in different cell islands and to facilitate the assignment of cell subsets to these islands. The t-SNE dimensions were characterized by t-SNE1 and t-SNE2 in the given graphs. The software is available at https://github.com/KlugerLab/FIt-SNE.

#### ZNF671 Expression and Functional State Correlation Analysis

The expression level statistics of ZNF671 in each cell were converted to normalized ranks and Next, the Kolmogorov– Smirnov liker random walk statistic, similar to the GSEA method, was used to summarize the ZNF671 expression-level rank statistics of a given signature gene set into a final enrichment score, which was used to characterize the signature activity. The enrichments of 14 signatures across cells in the scRNA-seq data were calculated, and only cells with detectable expression of ZNF671 were used. Correlations between ZNF671 expression and functional state activities were assessed using correlation analysis with false discovery rate (FDR) corrections for multiple comparisons (FDR < 0.05 and P < 0.05).

#### Cell Culture

Human GBM cell lines (U87 and U251), the A375 melanoma cell line, and triple-negative breast cancer cell lines (MDA-MB-231 and BT-549) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Cells were maintained at 37◦ , 5% CO<sup>2</sup> in 10% DMEM (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum.

#### Western Blot Analysis

After cells were transfected with the pEnter-ZNF671 or pEntervector plasmids (Vigene Biosciences, Shandong, China) for 48 h, RIPA lysis buffer (Beyotime, Shanghai, China) was used to isolate proteins. Proteins were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (SDS-PAGE, Beyotime), transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA, USA), and incubated with primary anti-ZNF671 (1:500; Proteintech, Chicago, IL, USA), E-cadherin (1:500, BD Biosciences), Vimtenin (1:500, BD Biosciences), and anti-GAPDH (1:1,000, Proteintech, Chicago, IL, USA).

#### Migration and Invasion Assays

Transwell plates (8-µm pores) (Costar/Corning, Lowell, MA) were used for Transwell migration or invasion assays. 5 × 10<sup>4</sup> (migration assay) or 1 × 10<sup>5</sup> (invasion assay) cells resuspended in serum-free medium were placed in the upper chamber of each insert, either uncoated or coated with Matrigel (BD Biosciences). The lower chamber contained culture medium with 10% FBS to act as a chemoattractant. The cells were incubated for 12 or 24 h and were then fixed and stained. Cells on the undersides of the filters were observed and counted under 200× magnification.

#### Statistical Analysis

Statistical analysis was performed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). Differences between two groups were analyzed using the two-tailed unpaired Student's t-test; P < 0.05 was considered statistically significant.

# RESULTS

#### scRNA-seq Dataset Features

As shown in **Table 1**, a total of eight cancer-related ZNF671 scRNA-seq datasets were included in the study. They contained 14 functional states of 13941 cancer single cells from glioblastoma (GBM; n = 623), glioma (brain; n = 2259), glioma (PDX; n = 167), astrocytoma (AST; n = 5097), oligodendroglioma (ODG; TABLE 1 | Features of the scRNA-seq datasets searched with *ZNF671*.


n = 4043), lung adenocarcinoma (LUAD; n = 126), melanoma (MEL; n = 1257), and breast cancer (BRCA; n = 369).

#### ZNF671 Functional States in the scRNA-seq Datasets

Expression analysis showed that ZNF671 was obviously downregulated in GBM, glioma, AST, ODG, LUAD, MEL, and BRCA (**Figure 1**), which indicated that ZNF671 might play an important role in tumor progression. To further explore the functional role of ZNF671 in different cancers, 14 crucial functional states of cancer cells, including angiogenesis, apoptosis, cell cycle, differentiation, DNA damage, DNA repair, EMT, hypoxia, inflammation, invasion, metastasis, proliferation, quiescence, and stemness were summarized and analyzed. As shown in **Figure 2**, the expression of ZNF671 and the activity of each functional state across single-cell datasets in different cancers were explored using an interactive bubble chart. The upper bar plot shows a summary of the association between the functional state and the number of single-cell datasets. We found that the expression of ZNF671 had a significant negative regulation for angiogenesis, apoptosis, EMT, hypoxia, invasion, and quiescence, which was consistent with our previous research (26, 27). These results indicate that ZNF671 might play a suppressor role in tumor development.

#### The Different Roles of ZNF671 in Cancers

We next explored the functional roles of ZNF671 in cancers, and analyzed the correlation between ZNF671 expression and functional state. We found that ZNF671 was positively associated with DNA damage (R = 0.18; ∗∗∗P < 0.001), apoptosis (R = 0.13; <sup>∗</sup>P < 0.05), DNA repair (R = 0.10; <sup>∗</sup>P < 0.05) in GBM; with stemness (R = 0.11; <sup>∗</sup>P < 0.05) and inflammation (R = 0.06; <sup>∗</sup>P < 0.05) in AST; with proliferation (R = 0.29; ∗∗P < 0.01), quiescence (R = 0.23; <sup>∗</sup>P < 0.05), and differentiation (R = 0.21; <sup>∗</sup>P < 0.05) in MEL; with inflammation (R = 0.17; ∗∗∗P < 0.001), metastasis (R = 0.16; ∗∗∗P < 0.001), stemness (R = 0.15;

Glioblastoma (GBM), Glioma (Brain), Glioma (PDX), Astrocytoma (AST), Oligodendroglioma (ODG), Lung adenocarcinoma (LUAD), Melanoma (MEL), and Breast cancer (BRCA). \*\**p* ≤ 0.01 compared with the control using Student's *t*-test.

∗∗P < 0.01), hypoxia (R = 0.13; ∗∗P < 0.01), EMT (R = 0.13; <sup>∗</sup>P < 0.05), and differentiation (R = 0.08; <sup>∗</sup>P < 0.05) in ODG; with EMT (R = 0.12; <sup>∗</sup>P < 0.05) and hypoxia (R = 0.12; <sup>∗</sup>P < 0.05) in glioma (brain); and with stemness (R = 0.18; <sup>∗</sup>P < 0.05), and hypoxia (R = 0.18; <sup>∗</sup>P < 0.05) in BRCA (**Figures 3**, **4**).

However, ZNF671 was negatively associated with DNA damage in ODG (R = −0.11; ∗∗P < 0.01); with hypoxia (R = −0.55; <sup>∗</sup>P < 0.05), EMT (R = −0.50; <sup>∗</sup>P < 0.05), apoptosis (R = −0.49; <sup>∗</sup>P < 0.05), angiogenesis (R = −0.48; <sup>∗</sup>P < 0.05), and quiescence (R = −0.43; <sup>∗</sup>P < 0.05) in glioma (PDX), and with inflammation (R = −0.20; <sup>∗</sup>P < 0.05) and differentiation (R = −0.18; <sup>∗</sup>P < 0.05) in BRCA (**Figures 3**, **4**). These results indicated that ZNF671 plays a different functional role in cancers and that the functional difference could be associated with the functional populations of cancer cells.

#### The Different Roles of ZNF671 in Different Cell Groups

To determine the functionally heterogeneous roles of ZNF671 in cancer cells, we inferred that single cells exhibited widespread heterogeneity in terms of their functional states in cancer. We applied t-SNE to reduce the non-linear dimensionality of the cancer cell data and placed different cell clusters on a t-SNE map (**Figure 5**), which indicated that the cell groups might be associated with the functional heterogeneity of cancer.

To reveal the roles of ZNF671 in different cell groups, we further the explored functional roles and correlations of ZNF671 in different cancer subgroups. As shown in **Figure 6**, ZNF671 expression was positively associated with DNA repair, DNA damage, and apoptosis but negatively associated with angiogenesis, differentiation, and proliferation in MGH30 cell groups of GBM, while ZNF671 expression was positively associated with proliferation in MGH31 cell groups of GBM. In glioma (brain), ZNF671 expression was negatively correlated with angiogenesis in MUV1, with DNA repair, DNA damage, and cell cycle in MUV5, with DNA repair in BCH836, and with apoptosis in BCH869. ZNF671 expression was positively correlated with hypoxia in MUV10, BCH836, and BCH869 in glioma (brain). In glioma (PDX), ZNF671 expression in BCH869 correlated negatively not only with hypoxia but also with EMT, apoptosis, angiogenesis, and quiescence. In AST, ZNF671 expression was positively correlated with stemness in MGH45 and MGH56, with invasion in MGH61, and with inflammation in MGH64, and it was negatively correlated with cell cycle and invasion in MGH45, with angiogenesis in MGH57, and with invasion in MGH64. In ODG, ZNF671 expression was positively

FIGURE 2 | Relevance of *ZNF671* across 14 functional states in distinct cancers. The upper bar chart shows the number of datasets in which *ZNF671* is significantly related to the corresponding state. In the bubble chart in the second section, a results table is used to display the basic information of all single-cell datasets in the selected cancer type and the corresponding correlations with the 14 functional states. Glioblastoma (GBM), Glioma (Brain), Glioma (PDX), Astrocytoma (AST), Oligodendroglioma (ODG), Lung adenocarcinoma (LUAD), Melanoma (MEL), and Breast cancer (BRCA).


FIGURE 3 | Functional relevance of *ZNF671* in primary solid tumors. *ZNF671* plays different functional states in different single-cell datasets. Glioblastoma (GBM), Glioma (Brain), Glioma (PDX), Astrocytoma (AST), Oligodendroglioma (ODG), Lung adenocarcinoma (LUAD), Melanoma (MEL), and Breast cancer (BRCA). \*\*\**p* ≤ 0.001; \*\**p* ≤ 0.01; \**p* ≤ 0.05 compared with the control using Student's *t*-test.

adenocarcinoma (LUAD), Melanoma (MEL), and Breast cancer (BRCA).

correlated with metastasis, hypoxia, inflammation, and apoptosis in MGH36 and with inflammation in MGH60 but negatively correlated with apoptosis in MGH54 and with quiescence in MGH93. Similarly, in MEL, ZNF671 expression was positively correlated with stemness in tumor78, with proliferation and stemness in tumor79, with proliferation and differentiation in tumor88, and with inflammation in tumor89. However, ZNF671 expression was negatively correlated with DNA repair in tumor78, DNA damage and angiogenesis in tumor80, and cell cycle in tumor89. In LUAD, ZNF671 expression was positively correlated with DNA repair in MBT15 but negatively correlated with metastasis and invasion in PT45. In BRCA, ZNF671 expression was only positively correlated with DNA damage in CSL KO xenograft tumor (**Figure 6**, all <sup>∗</sup>P < 0.05; ∗∗P < 0.01).

# ZNF671 Inhibits Cell EMT, Migration, and Invasion in vitro

To determine the functional roles of ZNF671 in cancer cells, we performed Western blot assay and migration and invasion assays using U87, U251, A375, MDA-MB-231, and BT-549 cell lines transfected with ZNF671 or vector plasmids. As shown in **Figure 7A**, Western blot analysis validated that ZNF671 protein was obviously upregulated after transfection of ZNF671 plasmid. Furthermore, the overexpression of ZNF671 was associated with increased expression of the epithelial marker Ecadherin and decreased expression of the mesenchymal marker Vimentin. Transwell assays showed that overexpression of ZNF671 inhibited cancer cell migration and invasion in vitro (**Figures 7B–D**). These findings indicate that ZNF671 inhibits the EMT, migration, and invasion of U87, U251, A375, MDA-MB-231, and BT-549 cells in vitro.

#### DISCUSSION

ZNF671, which contains C2H2-type zinc fingers (ZFs) and a Krüppel-associated box (KRAB) domain, is a member of the KRAB-ZF (KRAB-ZFP) transcriptional family. KRAB-ZFPs are involved in regulating angiogenesis (36), apoptosis (37–39), the cell cycle (40, 41), inflammation (42), invasion and metastasis (43, 44), and stemness (45). Our previous studies demonstrated that ZNF671 is a tumor suppressor that is epigenetically silenced by DNA methylation in nasopharyngeal carcinoma, BRCA, CESC, HNSC, KIRP, LUAD, PAAD, and UCEC (26, 27). However, there is limited information regarding the role of ZNF671 in cancer progression and development, and there have been no systematic studies of the role of ZNF671 in cancer's heterogeneous functional states.

In this study, we found a total of eight solid tumorrelated ZNF671 scRNA-seq datasets, including GBM, glioma, AST, ODG, LUAD, MEL, and BRCA. ScRNA-seq functional state analysis showed that ZNF671 played a tumor suppressor role and/or an oncogenic role in angiogenesis, apoptosis, cell cycle, differentiation, DNA damage, DNA repair, EMT, hypoxia, inflammation, invasion, metastasis, proliferation, quiescence, and stemness. The different functional states in tumors may be associated with the inherent heterogeneity of the tumor. However, the synthetic analysis of eight solid tumors showed that ZNF671 was negatively associated with angiogenesis, apoptosis, EMT, hypoxia, invasion, and quiescence. Western blot and transwell assays showed that ZNF671 inhibited EMT, migration, and invasion of CNS cancers, lung cancer, melanoma, and breast carcinoma in vitro. These results suggested a crucial tumor suppressor role for ZNF671 in the progression of these cancers, which was consistent with our previous studies (26, 27).


and Breast cancer (BRCA). \*\**p* ≤ 0.01; \**p* ≤ 0.05 compared with the control using Student's *t*-test.

To further explore the heterogeneous functional state of ZNF671 in cancers, we applied t-SNE to describe the distribution of cells. We found different cell clusters on a t-SNE map and proposed that these cell subgroups might lead to cancer functional heterogeneity. Functional analysis of the cancer cell subgroups validated that the heterogeneous cell populations had different roles in cancer progression and development, which provided us with a fine level of resolution for cancer treatment. However, there were still several limitations. First, this study was based on current scRNA datasets, and several scRNA datasets only contain data for hundreds of single cells, so more cells should be considered for analysis. Second, we found the ZNF671 inhibits angiogenesis, apoptosis, EMT, hypoxia, invasion, and quiescence in CNS cancers, lung cancer, melanoma, and breast carcinoma. Moreover, we only identified that ZNF671 suppresses cell EMT, migration, and invasion in vitro. The angiogenesis, apoptosis, hypoxia, and quiescence functional states need be identified further, and the suppressor role of ZNF671 in vivo needs to be explored further.

In conclusion, this study systematically evaluated the tumor suppressor role of ZNF671 based on scRNAseq datasets. Our findings revealed that ZNF671 is a tumor suppressor in LUAD, BRCA, GBM, glioma, AST, ODG, and MEL. However, the mechanism of ZNF671's tumor suppressor role remains unknown, and further studies are needed to clarify this issue. Our results provide new insights into the role of ZNF671 in multiple tumors and identifies ZNF671 as a novel target for cancer treatment.

#### DATA AVAILABILITY STATEMENT

Publicly available datasets were analyzed in this study. This data can be found here: http://www.bioconductor.org.

# AUTHOR CONTRIBUTIONS

JZha, JLu, and HJ designed the research. TX, JZhe, YT, RL, BW, JLi, AX, and XH acquired and analyzed the data. JZha, HJ, and YY wrote the manuscript.

#### FUNDING

This work was supported by grants from the Social Science and Technology Development Key Project of Dongguan

# REFERENCES


(201750715046462); Guangzhou Key Medical Discipline Construction Project Fund (B195002004042); Open Funds of State Key Laboratory of Oncology in South China (KY013711).

#### ACKNOWLEDGMENTS

We thank professors Ying Sun, Jun Ma, and Na Liu (State Key Laboratory of Oncology in South China, Collaborative Innovation Center of Cancer Medicine; Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy; Sun Yat-sen University Cancer Center, Guangzhou 510060, P.R. China) for supporting this work. We would like to thank the native English-speaking scientists of Elixigen Company (Huntington Beach, California) for editing our manuscript.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhang, Luo, Jiang, Xie, Zheng, Tian, Li, Wang, Lin, Xu, Huang and Yuan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Methyl-Cantharidimide Inhibits Growth of Human Hepatocellular Carcinoma Cells by Inducing Cell Cycle Arrest and Promoting Apoptosis

Xiangzhong Huang1†, Wen Xie2†, Xiaofan Yu<sup>2</sup> , Caiyun Fan<sup>2</sup> , Jin Wang<sup>2</sup> , Yi Cao<sup>2</sup> and Jianxiang Li <sup>2</sup> \*

*<sup>1</sup> Department of Interventional Therapy, Affiliated Jiangyin Hospital, Medical College of Southeast University, Jiangyin, China, <sup>2</sup> School of Public Health, Medical College of Soochow University, Suzhou, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Haichang Li, The Ohio State University, United States Xiaozhuo Liu, University at Buffalo, United States*

> \*Correspondence: *Jianxiang Li aljxcr@suda.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *22 August 2019* Accepted: *28 October 2019* Published: *15 November 2019*

#### Citation:

*Huang X, Xie W, Yu X, Fan C, Wang J, Cao Y and Li J (2019) Methyl-Cantharidimide Inhibits Growth of Human Hepatocellular Carcinoma Cells by Inducing Cell Cycle Arrest and Promoting Apoptosis. Front. Oncol. 9:1234. doi: 10.3389/fonc.2019.01234* Methyl-Cantharidimide (MCA) is a derivative of cantharidin which has potential anticancer activity. This study investigates the effect of MCA on the growth and metastasis of human hepatocellular carcinoma (HCC) cells. Human HCC HepG2 and Hep3B2.1-7 cells, and normal hepatocytes (L02) were treated with a series of concentrations of MCA. The inhibition ability of these cells was examined by CCK-8 assay. Cell cycle and cell apoptosis were determined using Flow Cytometry. The effect of MCA on cell migration and invasion was evaluated through scratch wound healing and transwell migration assays. Furthermore, Western blot was used to evaluate biomarkers associated with cell cycle and apoptosis. It was found that: (i) MCA inhibited cell proliferation in HCC cells in a dose- and time-dependent manner, especially in HepG2 cells; (ii) MCA arrested HCC cells in G-1 phase cell cycle; (iii) MCA induced HCC cells apoptosis; (iv) MCA inhibited the migration ability of HCC cells; and (v) MCA treatment significantly increased cleaved-caspase3 and decreased NF-κB protein in HCC cells. These results suggest that MCA has cytotoxic effect on HCC cells by inducing cell cycle arrest and promoting apoptosis. MCA could be developed as an previous anticancer drug for the treatment of human hepatocellular carcinoma.

Keywords: methyl-cantharidimide (MCA), cantharidin (CTD), hepatocellular carcinoma, cell growth, invasion

#### INTRODUCTION

Hepatocellular carcinoma (HCC) is a primary malignant tumor in the liver. It is the third leading cause of cancer deaths worldwide. The annual incidence of HCC in North America and Western Europe ranges from 10 to 15 cases per 100,000 people, while in parts of Africa and Asia, the incidence is much higher and range from 50 to 150 cases per 100,000 people (1). The median survival following diagnosis ranges from ∼6 to 20 months. Currently, tumor resection is an option of choice for the treatment of HCC. However, since most of the diagnosis is made in late stages of the disease, HCC patients have few opportunity to undergo surgery. Thus, there is an urgent need for early diagnosis and effective therapeutic medication (2, 3). In recent years, there were several reports suggest that some traditional Chinese medicines showed remarkable anti-cancer activity (4–7). Cantharidin (CTD) is a biologically active ingredient isolated from Chinese blister Beetle, Mylabris phalerata. It is used in traditional Chinese medicine for more than 2000 years to treat various diseases including cancers (8– 14). Several recent reports have identified potential mechanisms of action of cantharidin in different types of human cancer cells, including hepatocellular carcinoma cells (15–21). Although CTD is a natural toxin possessing potent anti-tumor properties, its use in cancer treatment has been limited because of severe side effects due to its highly toxic nature. Analogs of chemically modified CTD have been synthesized to achieve a comparable anti-tumor property with less toxic effect (22).

Methyl-cantharidimide (MCA, 2,3a,7a-Trimethyl-hexahydro -4,7-epoxido-isoindol-1,3-dion, C11H15NO3, molecular weight 209.242) is one such derivative of CTD. MCA is chemically synthesized and developed as a powerful anticancer drug with less toxicity (23). However, the anticancer effect of MCA and its mechanism of action has not been evaluated. This study investigates the anticancer effect of MCA using hepatocellular carcinoma cell models. Human hepatocellular carcinoma cell lines, HepG2 and Hep3B2.1-7 and normal hepatocyte LO<sup>2</sup> cells were used. Various concentrations of MCA were tested in these cells to determine its tumor growth inhibitory effect. The mechanism of anticancer activity of MCA was explored by determining the change of cell cycle, cell apoptosis and cell migration ability. The activity of key-players that regulate cell apoptosis, caspase-3 and Nuclear Factor kappa-light-chainenhancer of activated B cells (NF-κB), were evaluated.

#### MATERIALS AND METHODS

#### Cell Culture

Human hepatocellular carcinoma cell lines, HepG2 and Hep3B2.1-7, were purchased from Chinese Type Culture Collection, Chinese Academy of Sciences, Shanghai, China. Normal human hepatic cell L02, was a gift from the Department of Toxicology, School of Public Health, Soochow University, Suzhou, China. Cells were cultured in complete Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, 100µg/ml streptomycin and penicillin (all obtained from Hyclone, Suzhou, China) in an incubator maintaining at 37 ± 0.5◦C in a humidified atmosphere with 5% CO2. Cells were trypsinized after reaching confluence and aliquots were kept frozen at −80◦C. Cells were sub-cultured and those in 3–4 passages were used in experiments described below. All experiments were repeated three times.

# Methyl-Cantharidimide (MCA)

Methyl-Cantharidimide (MCA) was a gift of Sihuan Bioengineering Co., Ltd., Jiangsu, China. A stock solution (20 mM) was freshly prepared by dissolving MCA in dimethyl sulfoxide (DMSO, PanReac Applichem, Germany) and diluted in desired concentrations in complete DMEM. HepG2, Hep3B2.1- 7, and LO<sup>2</sup> cells were treated in vitro with a series of final concentrations of MCA or with the solvent DMEM as control.

# Cytotoxicity Essay (IC50)

Two-hundred µl aliquots of HepG2, Hep3B2.1-7 and L02 cells in DMEM complete medium (∼3000 cells each) were distributed into 96-well plate and cultured for 24 h at 37 ± 0.5◦C. Then, 200 µl MCA solution was added to give a final concentration of 50, 100, 200, 400, and 800µM. The cells were cultured for 24, 48, and 72 h. The proliferation ability of the cells in each well was assessed using a CCK-8 assay kit (Dojindo, China) according to manufacturer's instructions. Briefly, 20 µl of CCK-8 solution was added to each well and the cells were incubated for 4 h at 37 ± 0.5◦C. The plates were then read in the standard plate reader (FilterMax F5, Molecular Devices, USA) at a reference wavelength of 450 nm. The percent inhibition of growth in cells treated with MCA was calculated as follows: % Inhibition = [A450(drug) – A450(blank)]/[A450(control) – A450(blank)] × 100%. The IC30 that was obtained for HepG2 cells was 137.56µM MCA. This dose was used in subsequent experiments.

# Cell Cycle Evaluation

Two-hundred µl aliquots of HepG2 and Hep3B2.1-7 cells in complete DMEM medium (∼1 × 10<sup>5</sup> cells each) were distributed in 6-well plates and cultured for 24 h at 37 ±−0.5◦C. Then, the cells were treated with 137.56µM MCA (IC30 concentration obtained for HepG2 cells) for 48 h, collected by trypsinization, washed twice with cold phosphate buffered saline (PBS), suspended in cold 70% methanol and left at −20◦C overnight. The cells were then washed twice with cold PBS and stained with PBS solution containing 20µg/ml PI and 50µg/ml of RNaseA for 30 min. The cell cycle analysis was carried out using a flow cytometer (Beckman coulter, Shanghai, China) (24).

# Cell Apoptosis Detection

Annexin V-FITC apoptosis detection kit (KeyGEN Biotech, Shanghai, China) was used to evaluate cell apoptosis. Twohundred µl aliquots of HepG2 and Hep3B2.1-7 in complete DMEM medium (∼1 × 10<sup>5</sup> cells each) were distributed in 6 well plates and cultured for 24 h. Then, the cells were treated with 137.56µM MCA (IC30 concentration obtained for HepG2 cells) for 48 h. The cells were collected by trypsinization, incubated with Annexin V in a buffer containing propidium iodide for 15 min. The percent cells in apoptosis were then determined using a flow cytometer (Beckman coulter, Shanghai, China) (25).

# Scratch Wound Healing Assay

Two hundred microliters aliquots of HepG2 and Hep3B2.1-7 cells in complete DMEM medium (∼2 × 10<sup>5</sup> cells each) were distributed in 6-well plates and cultured for 24 h at 37◦C. Then, the cells were treated with 137.56µM MCA (IC30 concentration obtained for HepG2 cells) for 48 h. Cells were allowed to grow up to 100% confluence and a scratch was made in the plate using with a P10 pipette tip. The cells were cultured in fresh serumfree DMEM medium. images were collected at 0 and 24 h under an inverted microscope (Olympus, Germany) and quantitatively analyzed using the NIH Image J software.

#### Transwell Migration Assay

HepG2 and Hep3B2.1-7 cancer cells and MCA treated cells (2 × 10<sup>5</sup> ) were seeded in the upper chambers (pore size, 8µm) of the 6-well plate (Corning, USA) in 1 ml serum-free medium. The lower chambers were filled with 2 ml complete medium with 10% FBS, and the plate was incubated under standard conditions for 24 h. After removing the cells in the upper surface of the membrane with a cotton swab, cells in the lower chamber were fixed with methanol and stained with 0.5% crystal violet solution. The images were taken using an inverted microscope (Olympus, Germany and analyzed using NIH Image J software.

#### Western Blot Analysis

Approximated 2 × 10<sup>5</sup> HepG2 cells were treated with 137.56µM MCA (IC30 concentration obtained for HepG2 cells) for 48 h. Protein extracts were prepared by lysing the cells in lysis buffer containing 50 mM Tris (pH 7.4), 150 mM sodium chloride,



*All experiments were repeated three times. Data are mean* ± *Standard Deviation.* \**p* < *0.05: Compared to L02 cells;* #*p* < *0.05: Compared to 24 h.*

TABLE 2 | Alteration of cell cycle in HepG2 and Hep3B2.1-7 HCC cells treated with MCA.


*All experiments were repeated three times. MCA treatment with IC30 concentration (137.56*µ*M) for 48 h. Data are mean* ± *Standard Deviation.* \**p* < *0.05.*

obtained from Beyotime, Shanghai, China). The cell lysates were centrifuged at 14,000 × g for 5 min at 4◦C and the supernatant containing solubilized proteins was collected. The protein concentration in all samples was determined by using the BCA protein assay kit (Beyotime, Shanghai, China). From each sample, equal amount of protein (40 µg per lane) was loaded, separated by 10% sodium dodecyl sulfate polyacrylamide gel (SDS–PAGE) and then transferred to polyvinylidene difluoride (PVDF) membranes (Millipore Corporation, Billerica, MA, USA). The membranes were blocked for 2 h in 5% fat-free dry milk (Yili Industrial, Inner Mongolia, China) containing Tween 20-Tris-buffered saline (TTBS). The membranes were then incubated with primary antibodies (rabbit monoclonal anti-NF-κB antibody, rabbit monoclonal anti-cleaved caspase 3 antibody and rabbit monoclonal anti-GADPH, Abcam, Cambridge, USA) overnight at 4◦C. They were washed three times in TTBS and incubated further with horseradish peroxidase-conjugated antibodies (Beyotime, Shanghai, China) for 1.5 h at room temperature. This was followed by washing the membranes three times with TTBS. The immunoreactive proteins on the membranes were detected using enhanced chemiluminescence reagents (Millipore Corporation) and G-BOX Chemi XRQ (Syngene, UK). The blots were quantified and normalized with the level of GADPH to

1% Triton X-100, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate and 1 mM phenyl-methyl-sulfonyl fluoride (all


correct the differences in loading of the proteins in MCA-


*All experiments were repeated three times. MCA treatment with IC30 concentration (137.56*µ*M) for 48 h. Data are mean* ± *Standard Deviation.* \**p* < *0.05.*

FIGURE 1 | Flow cytometry on cell cycle analysis in HepG2 and Hep3B2.1-7 cells following treatment with MCA for 48 h. (A) Effect of MCA on cell cycle of HepG2 cells. (B) Effect of MCA on cell cycle of Hep3B2.1-7 cells. \**p* <0.05 via control.

treated cells.

#### Statistical Analysis

The results from three independent experiments were pooled and analyzed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA). The data were presented as mean ± SD. Statistical significances of differences were analyzed by one way analysis of variance (ANOVA) test followed by Sidak's multiple comparison post-test. A p-value of <0.05 was considered as significant difference between MCA treated and untreated cells.

#### RESULTS

# MCA Inhibits the Proliferation of HCC Cells With Little Effect on the Proliferation of Normal Hepatic Cells

The effect of MCA on the proliferation of HCC and normal hepatic cells was evaluated using CCK 8 assay. The inhibition rate of growth in L02, HepG2 and Hep3B2.1-7 cells treated with 50, 100, 200, 400 and, 800µM MCA were presented. In all three types of cells, there was a decrease in growth which was positively correlated with not only the increased concentration of MCA but also the duration of treatment from 24 to 72 h. The IC50 observed in HepG2 and Hep3B2.17 HCC cells and normal LO<sup>2</sup> hepatic cells treated with MCA for 24, 48, and 72 h were presented in **Table 1**. The IC50 observed in LO2, HepG2 and Hep3B2.1-7 cells, after 48 h treatment with MCA were at concentrations of 363.56 ± 44.46, 227.00 ± 6.11, and 273.44 ± 18.52µM, respectively. This result indicated that MCA inhibits the proliferation of HepG2 and Hep3B2.1-7 cells with little effect on the proliferation of normal hepatocyte LO<sup>2</sup> cells.

#### MCA Induces Cell Cycle Arrest at G1/S of HCC Cells

Since MCA inhibits HCC cell proliferation, we next examine whether MCA affects cell cycle progression in HCC cells by flow cytometry. HepG2 and Hep3B2.1-7 cells were treated with and without MCA for 48 hrs. The IC30 concentration of MCA obtained from HepG2 cells in the above experiment 137.56µM was used. It was found that MCA treatment blocks the progression of cell cycle from G1- to S-phase in both HepG2 and Hep3B2.1-7 HCC cells, resulting in increased cell

FIGURE 2 | Flow cytometry on cell apoptosis analysis in HepG2 and Hep3B2.1-7 cells following treatment with MCA for 48 h. (A) Apoptosis of untreated HepG2 cells for 48 h. (B) Apoptosis of HepG2 cell treated with MCA for 48 h. (C) Apoptosis of untreated Hep3B2.1-7 cells for 48 h. (D) Apoptosis of Hep3B2.1-7 cell treated with MCA for 48 h.

number in G1-phase. The percent of HepG2 cells at G1 phase was 60.23 ± 0.70 which was significantly higher than in solvent-treated cells, which was 54.58 ± 1.67 (p < 0.05). Similarly, the percent of Hep3B2.1-7 cells at G1-phase was 50.81 ± 0.76, which was significantly higher than that in solvent-treated cells, 47.48 ± 0.62 (p < 0.05). There was no significant difference in percent cells in S- and G2-phase in both cell types (p > 0.05). This result indicates that

FIGURE 3 | Scratch wound healing assays performed in HepG2 and Hep3B2.1-7 cells treated with conditioned medium containing MCA or vehicle control for 48 h (A). Quantitative analysis of wound closure (B). HepG2 and Hep3B2.1-7 cells were treated with the indicated conditioned medium containing MCA or vehicle control for 48 h in a transwell assay (C). Cells that migrated through transwells were stained with crystal violet and photomicrographed. Quantitative analysis of migrated cells perfield (D). \**p* < 0.05, \*\**p* < 0.01 via Control.

MCA induces cell cycle arrest at G1/S phase (**Table 2** and **Figures 1A,B**).

#### MCA Promotes Apoptosis of HCC Cells

The effect of MCA on cell apoptosis of HCC cells was evaluated by flow cytometry. The early, late and total cell apoptosis in HepG2 and Hep3B2.1-7 cells, treated with or without 137.56µM MCA for 48 h were presented in **Table 3** and **Figure 2**. The data showed a significant increase in total percent of cells of apoptosis in both HepG2 and Hep3B2.1- 7 cells: 12.34 ± 2.81 and 7.49 ± 0.42% in HepG2 and Hep3B2.1-7 cells, respectively compared with solvent-treated cells, 2.83 ± 1.04 and 3.70 ± 1.48%, respectively. This result indicates that MCA treatment promotes apoptosis of HCC cells.

#### MCA Inhibits Cell Migration

To investigate the effect of MCA in the migratory ability of HCC cells, we performed scratch wound healing and transwell cell migration assays. Scratch wound healing assay indicated that MCA significantly inhibited the migration ability of HepG2 and Hep3B2.1-7 cells (**Figures 3A,B**). As shown in **Figures 3C,D**, further transwell migration assay demonstrated that the migrated cells were significantly decreased in MCA treated HepG2 and Hep3B2.1-7 cells compared to the untreated cells.

#### MCA Up-Regulates the Expression of Cleaved Caspase-3 and Down-Regulates the Expression of NF-κB Proteins

Because MCA induces apoptosis of HCC cells, we explored whether the molecules that regulate cell apoptosis were involved. Western blot analysis on the protein expression of cleaved caspase-3 and NF-κB was performed and the results were presented in **Figure 4**. The data showed significantly increased level of caspase-3 protein (p < 0.05) and significantly decreased level of NF-κB protein (p < 0.05) in HepG2 cells treated with 137.56µM MCA for 48 hrs. This result suggested that MCA up-regulates the expression of Cleaved Caspase-3 and downregulates the expression of NF-κB proteins.

#### DISCUSSION

Hepatocellular carcinoma is the primary malignant tumor in liver. It is usually associated with liver cirrhosis, viral hepatitis, aflatoxin exposure, and alcohol abuse. Typically, HCC is diagnosed in late stages and thus, patients have few opportunity for surgical resection or transplantation (2, 3). Currently, there is no ideal treatment option for the cure of the disease. In recent years, there have been increasing efforts to develop drugs for the treatment and prevention of HCC. Cantharidin (CTD) is a biologically active ingredient isolated from the Chinese blister beetle (Mylabris phalerata) and has been used in traditional Chinese medicine for over 2000 years to treat various diseases including cancers (8–14). However, the dose of CTD used for cancer treatment is very close to its toxic dose and its use had been

extracted from HepG2 cell. \**p* < 0.05 via Control.

hampered in clinics. Nonetheless, researchers have identified the pharmacological and toxicological properties of CTD (26–30), and started to synthesize and develop bioactive derivatives that have anti-cancer potential with low toxicological profile (31).

MCA is one of such synthetic derivatives of CTD (23). Its anti-cancer effect and the mechanism of action particularly in hepatocellular carcinoma cells has not been delineated. In this study, MCA was tested in a series of concentrations to determine its growth inhibitory effect in different human HCC cells, HepG2, and Hep3B2.1-7 cells, and compared with the normal LO<sup>2</sup> hepatic cells. The 50% growth inhibitory effect of MCA in HepG2 and Hep3B2.1-7 cells was observed at a concentration of 226.82 and 273.18µM, respectively, while that in normal hepatic LO<sup>2</sup> cells was at 379.41µM. This result suggests that normal hepatic cells required a higher concentration of MCA to elicit 50% growth inhibitory effect. In other words, MCA was selectively toxic to HCC cells compared with that in normal hepatic cells. We also found that treatment of HepG2 and Hep3B2.1-7 HCC cells with IC30 dose of MCA induces cell cycle G1/S arrest, resulting in accumulation of cells in G1 phase of the cell cycle. Flow cytometric analysis showed that the IC30 dose of MCA induced significantly higher percent of cell apoptosis in HepG2 and Hep3B2.1-7 cells compared with normal hepatic cells. These observations suggested that MCA exert growth inhibitory effect and induce cell apoptosis in HCC cells. Transwell migration assay demonstrated that the migrated cells were significantly decreased in MCA treated HepG2 and Hep3B2.1-7 cells compared to the untreated cells. Caspases are crucial mediators of cell apoptosis, among them, caspase-3 is a frequently activated protease, catalyzing the specific cleavage of key cellular proteins. Caspase-3 is indispensable for apoptotic chromatin condensation and DNA fragmentation. Thus, caspase-3 is essential for certain processes associated with the formation of apoptotic bodies, but it may also function before or at the stage when commitment to loss of cell viability is made (32, 33). NF-κB is a primary transcription factor which plays a key role in several cellular processes including cell proliferation and survival (34–36). In this study, 48 h after treatment of HepG2 cells with MCA, the protein level of cleaved-caspase3 was significantly increased while that of NF-κB was significantly decreased. These results suggested

#### REFERENCES


that both cleaved caspase-3 and NF-κB played important roles in MCA-induced cell apoptosis in HCC cells. Our results are in consistent with other study which showed that MCA can inhibit the proliferation of HCC cells and prolong the survival of mice with hepatocellular carcinoma (37). Studies on sodium cantharidate, another derivative of cantharidin, have also shown that sodium cantharidate can inhibit the growth of liver cancer cells and induce cell apoptosis (38). Therefore, the inhibitory effect of CTD derivatives on HCC cells is most likely associated with inhibiting cancer cell growth and inducing cancer cell apoptosis by activating caspace-3 and inhibiting NF-κB. The mechanism of action of MCA in vivo warrants further investigation.

#### CONCLUSIONS

Our results suggest that MCA can inhibit cell proliferation and migration of cancer cells. This study provides a rational for developing MCA as a therapeutic agent for the treatment of hepatocellular carcinoma.

#### DATA AVAILABILITY STATEMENT

Data will be available upon request by writing to the corresponding author.

# AUTHOR CONTRIBUTIONS

XH performed MTT assay and write the first draft. XH, WX, CF, and XY performed flow cytometry, cell migration, and invasion assays. CF and JW performed Western blotting, YC performed statistics. JL conceived the study, supervise experiments and results analysis. All authors edited the manuscript and approved the final version.

#### ACKNOWLEDGMENTS

We are grateful to Dr. Vijayalaxmi, Department of Radiology, University of Texas Health Science Center, San Antonio, TX, USA for her helpful comments on the manuscript.

progression and inducing apoptosis. Biol Pharm Bull. (2006) 29:2388– 94. doi: 10.1248/bpb.29.2388


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Huang, Xie, Yu, Fan, Wang, Cao and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

**130**

# hsa\_circ\_0007841: A Novel Potential Biomarker and Drug Resistance for Multiple Myeloma

Meng Gao† , Chengyuan Li † , Han Xiao, Hang Dong, Siyi Jiang, Yunfeng Fu\* and Liying Gong\*

The Third Xiangya Hospital of Central South University, Changsha, China

Purpose: Circular RNA (circRNA) is a key regulatory factor in the development and progression of human tumors. However, the working mechanism and clinical significance of most circRNAs remain unknown in human cancers, including multiple myeloma (MM).

#### Edited by:

Pascale Cohen, Université Claude Bernard Lyon 1, France

#### Reviewed by:

Mingli Liu, Morehouse School of Medicine, United States Qing Xu, Tongji University, China Fengxia Wu, Capital Medical University, China

#### \*Correspondence:

Yunfeng Fu fuyfeng427@163.com Liying Gong gongliying1987@163.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 02 June 2019 Accepted: 31 October 2019 Published: 19 November 2019

#### Citation:

Gao M, Li C, Xiao H, Dong H, Jiang S, Fu Y and Gong L (2019) hsa\_circ\_0007841: A Novel Potential Biomarker and Drug Resistance for Multiple Myeloma. Front. Oncol. 9:1261. doi: 10.3389/fonc.2019.01261 Patients and Methods: This study employs high-throughput circRNA microarray with bioinformatics to identify differentially expressed circRNAs in patients with MM. The hsa\_circ\_0007841 expressions were observed in the MM tissues of 86 patients. Drug-resistant cell lines and pathological features were also detected. In addition, the relationship between hsa\_circ\_0007841 expressions in the MM tissues and the pathological features of patients with MM were evaluated and role of hsa\_circ\_0007841 as a potential biomarker and therapeutic target was assessed.

Results: The results show that in the MM cell lines and drug-resistant cell lines, hsa\_circ\_0007841 expression was significantly upregulated, which was closely associated with disease prognosis. Specifically, hsa\_circ\_0007841 upregulation was correlated with chromosomal aberrations such as gain 1q21, t (4:14) and mutations in ATR and IRF4 genes. This finding was corroborated in large samples. Finally, bioinformatics analysis showed that eight differentially expressed miRNAs and 10 candidate mRNAs interacted with hsa\_circ\_0007841, shedding some new light on the basic functional research.

Conclusion: This study may be the first to report that hsa\_circ\_0007841 is significantly upregulated in MM. It also suggests that hsa\_circ\_0007841 may be a novel biomarker for MM and its involvement in the progression of MM.

#### Keywords: circular RNAs, biomarker, diagnosis, prognosis, multiple myeloma

#### INTRODUCTION

Non-coding RNA (ncRNA) consists of long non-coding RNA (lncRNA), short microRNA (miRNA/miRs), and circular RNA (circRNA), the latter being the most common in the eukaryotic transcriptome. Unlike the conventional linear RNA (including 5′ and 3′ terminals), circRNA has a closed loop structure generated by back-splicing of RNA introns and/or exons. After the discovery of circRNA in viral RNAs in 1970 (1), it was later found in eukaryocytes (2). In recent years, high-throughput sequencing combined with transcriptome analysis has indicated the large abundance of circRNA in eukaryocytes (3).

Circular RNAs, repleted with miRNA binding sites, serve as sponge molecules for miRNAs and inhibit the binding of miRNAs to their target genes (4, 5). They bind competitively to miRNAs through the sponging action, thereby abolishing the inhibitory effect on the transcription of downstream target genes (6). Besides sponging action, circRNAs can also regulate gene expressions in parents and influence gene transcription or even protein translation by interacting with the proteins. Moreover, circRNAs are large RNA molecules with closed loop structures. Most ncRNAs and circRNAs are involved in the regulation of transcription and posttranscriptional gene expression. They play important roles in cancer progression, metastasis, and therapeutic response (7). Given the specificity of circRNAs in the disease state and their stability in the body fluid, circRNAs may be used in the diagnosis of cancer (8, 9).

Multiple myeloma (MM) is a hematologic malignancy caused by an abnormal proliferation of bone marrow plasma cells. In the clinical setting, MM is the second most common hematologic malignancy and accounts for 10% of all hematologic malignancies. Its incidence rate reaches 2–3 per 100,000. The ratio of affected male to female patients is 1.6:1. The features of MM are diverse but include anemia, bone pain, renal insufficiency, infection, hemorrhage, neurological symptoms, hypercalcemia, and amyloidosis (10). The development of MM is usually accompanied by a series of genetic variations, such as cytogenetic abnormalities, primary or secondary chromosomal translocation, and some oncogenes activations (11). Identifying these alterations is highly valuable for understanding the pathogenesis of tumors and predicting prognosis and therapeutic response. Some genetic mutations are correlated with the poor outcome of MM, including chromosomal variations t (11:14), t (14:16), and t (14:20) (12). The identified mutations include CCND1 and DNA repair pathway-related genes (TP53, ATM, ATR, and ZFHX4) (13). In contrast, some mutations predict positive outcomes such as mutations in the IRF4 and EGR1 genes (14). Individualized therapy for MM based on biomarkers can increase therapeutic efficacy while reducing toxicity (15). Therefore, some biomarkers for MM can be used as predictive and prognostic indicators to guide diagnosis and treatment.

In the present study, the bioinformatics method was combined with high-throughput sequencing in small samples. By using the circRNA database, circRNAs that might influence the treatment and prognosis of patients with MM were preliminarily screened. Next, real-time quantitative polymerase chain reaction (qRT-PCR) was applied for sample amplification. It was found that hsa\_circ\_0007841 was significantly upregulated in patients with MM and MM cell lines. The correlation between hsa\_circ\_0007841 expressions and clinicopathological features of such patients with MM was determined. It was determined whether the hsa\_circ\_0007841 expression could be used as a diagnostic and prognostic indicator for MM. The results provide the basis for identifying novel prognostic markers and therapeutic targets for MM.

#### MATERIALS AND METHODS

#### Clinical Data

From January 2012 to January 2018, 86 MM patients and 30 IDA patients treated at the Third Xiangya Hospital of Central South University were included as the case group. Their bone marrow samples and clinical data were collected. There were 53 males and 33 females in the case group, and the median age of onset was 55 years (range, 44–78 years). The diagnosis, staging, and risk stratification of MM were performed according to the National Comprehensive Cancer Network (NCCN). All of the cases had complete clinical and pathological data (**Table S1**). Due to a lack of normal bone marrow donors, 30 patients with iron deficiency anemia (IDA) were chosen as controls, and their bone marrow samples were collected to avoid sample variation. The bone marrow samples were cryopreserved at −80◦C. The collection of all samples was approved by the ethics committee of Xiangya third hospital (Approval number: 2016121), and the consent acquired was both written and informed consent.

The microarray data has been deposited in public, community-supported repositories (GEO, GSE133058).

# Cell Culture

Normal human mononuclear cell line THP-1 and MM cell lines KM3, U266, and RPMI-8226 were provided by the basic laboratory of Central South University Xiangya School of Medicine. Drug-resistant cell lines KM3/BTZ, U266/BTZ, and RPMI-8226/BTZ were acquired by inducing tolerance through stepwise increase of drug concentrations. The cells were cultured in the 1,640 culture medium (HyClone, Logan, UT, USA) supplemented with 10% fetal bovine serum (ExCell Biology, Shanghai, China), 50 U/ml penicillin, and 50 g/ml streptomycin (HyClone). The cells were incubated in a 37◦C, 5% CO<sup>2</sup> incubator and harvested in the log phase of growth.

#### RNA Extraction

Firstly, the bone marrow samples from 3 MM patients were sorted by magnetic activated cell sorting (MACS) using anti-CD138 MicroBeads (Miltenyi, Germany), and the plasma cells were enriched. According to the instruction manual of TRIzol reagent (Invitrogen, USA), total RNA extraction was performed from the enriched plasma cells and plasma cells in the normal human bone marrows. The extracted products were preserved at −80◦C. The RNA concentration and activity were first detected by NanoDrop ND-1000 (NanoDrop, USA). Then RNA purity and integrity were detected by denaturing formaldehyde agarose gel electrophoresis.

#### High-Throughput Sequencing of circRNAs and Identification of Differentially Expressed circRNAs

Sample preparation and microarray hybridization were performed according to Arraystar's standard protocols. First, linear RNAs in the extracted total RNA were removed by using RNaseR (Epicentre, USA), and the circRNAs were enriched. Then, using random primers, the enriched circRNAs were amplified and transcribed into fluorescent cRNAs. Alignment and hybridization were performed between the labeled circRNAs and Arraystar Human circRNA Array V2 (8x15K, Arraystar), followed by washes with the cleaning solution kit (Agilent, USA). The microarrays were scanned with the Agilent Scanner G2505C. The microarray images were analyzed with Agilent Feature Extraction software (version 11.0.1.1). The quantiles

#### TABLE 1 | Primers designed for qRT-PCR validation of candidate circRNAs.


were normalized and subsequent data processing was performed using limma package in R. Differentially expressed circRNAs were identified based on the fold change and P value. Significantly upregulated and downregulated genes were those with fold change > 2.0 and P < 0.05.

#### Verification of Quantitative RT-PCR

Sample RNA was reversely transcribed into cDNA using SuperScript III Reverse Transcriptase (Invitrogen, Grand Island, NY, USA). Real-time qRT-PCR (Arraystar) was implemented by using ViiA 7 Real-time PCR System (Applied Biosystems) and 2 × PCR Master Mix. Each mix (10 µL) contains 2 × Master Mix 5 µL, specific primer F 0.5 µL, specific primer R 0.5 µL, and 2 µL cDNA. The mix was added into each well of the 384-Well PCR Plate, which was then placed in the realtime PCR instrument for PCR reaction. The reaction conditions were as follows: incubation at 95◦C for 10 min, then 40 cycles of 10 s at 95◦C and 1 min at 60◦C. GAPDH was used as internal reference and its expression was normalized. 1Ct value reflected the circRNA expression level. The expression levels of each target gene were calculated and normalized by that of the internal reference. The primers used for each target gene are shown in **Table 1**.

#### Functional Analysis of circRNA

The circRNA/microRNA interaction was predicted using Arraystar's home-made miRNA target prediction software based on miRanda and TargetScan. To establish circRNA-miRNA network, we searched MREs on circRNAs using the software, then selected the miRNAs according to seed match sequences. The graph of the circRNA/miRNA network was drawn with the help of Cytoscape 3.01 (16, 17). GO analysis was conducted using Kyoto Encyclopedia of Genes and Genomes GO analysis was also used to determine the biological functions of the adjacent protein-coding genes of the target circRNAs (18).

#### Statistical Analysis

SPSS 20.0 software was used for statistical analysis. The data were described as mean ± standard deviation. Mann-Whitney test was used for pair wise comparison, and Kruskal-Wallis H-test for multiple comparison. For the analysis of clinical data, Kaplan-Meier survival curve was plotted, and log-rank test was used for statistical hypothesis test. The correlation between circRNA expressions and chromosomal variation and genetic mutations was analyzed by the chi-squared test. P < 0.05 indicated significant difference.

# RESULTS

#### Analysis of circRNA Profiles

To identify the differentially expressed circRNAs in MM patients, high-throughput circRNA microarray was applied to 3 MM patients and 3 IDA patients. Thousands of circRNAs were detected, including 4,727 upregulated circRNAs and 5,283 downregulated circRNAs (**Figure 1A**). There were 147 differentially expressed circRNAs (fold change > 2), including 131 upregulated and 16 downregulated circRNAs. The block diagram represents the normalized intensity of the two groups (**Figure 1B**), and the volcano plot indicates the differential expressions of circRNAs (**Figure 1C**).

#### hsa\_circ\_0007841 Was Upregulated in MM Cells

Four significantly upregulated circRNAs were then verified by qRT-PCR in 20 MM patients and 10 IDA patients. As shown in **Figure 2B**, the expression of hsa\_circ\_0007841 was upregulated to the highest degree. hsa\_circ\_0007841 is localized to chr3:127778944-127779504 and generated by back-splicing of exons 6 and 7 in the sec61a1 gene (**Figure 2A**). We further detected expressions of hsa\_circ\_0007841 in the bone marrow of 66 MM patients and 20 IDA patients (**Figure 2C**). The results showed that hsa\_circ\_0007841 may be a potential biomarker for MM. Moreover, hsa\_circ\_0007841 expression was also detected in THP-1 mononuclear cells, MM cell lines (KM3, U266, 8226), drug-resistant cell lines (KM3/BTZ, U266/BTZ, 8226/BTZ) (**Figure 2E**) and patients with MM with BTZ resistance (**Figure 2D**). hsa\_circ\_0007841 was selectively expressed in the MM cell lines but lowly expressed in the THP-1 mononuclear cells, and the difference was of statistical significance (P < 0.05). In bortezomib-resistant cell lines (KM3/BTZ, U266/BTZ), the expression of hsa\_circ\_0007841 was significantly higher than that in KM3 and U266 cells. And also significantly increased in patients with MM with bortezomib (BTZ) resistant (N = 21) than bortezomib (BTZ) sensitive (N = 36). The above results indicated that the upregulation of hsa\_circ\_0007841 may be involved in the bortezomib tolerance of MM patients.

#### Correlation Between hsa\_circ\_0007841 Expressions and Clinicopathological Parameters of MM Patients

According to the 2018 NCCN Guidelines, the correlation between hsa\_circ\_0007841 expressions and age, gender, type, staging, and risk stratification (IMWG) of MM patients was analyzed. The clinical correlation between hsa\_circ\_0007841 and MM was established. The results showed (**Table 2**) that hsa\_circ\_0007841 was correlated with typing (P = 0.002), cytogenetic mutation (P = 0.025), bone destruction (P = 0.014), R-ISS staging (P < 0.001), but not correlated with gender, age, percentage of plasma cells in the bone marrow and kidney injury.

FIGURE 1 | Overview of the microarray signatures. (A) Cluster analysis diagram of differentially expressed circRNAs. (B) The block diagram represents the normalized intensity of the two groups. (C) Volcano plot showing the significantly deregulated genes in MM samples. MM, multiple myeloma.

Expression of hsa\_circ\_0007841 in the bone marrow of 86 MM patients and 30 controls. (D) Expression of hsa\_circ\_0007841 in bortezomib (BTZ) sensitive (N = 36) and bortezomib (BTZ) resistant (N =21) patients. (E) Expression of hsa\_circ\_0007841 in MM patients, cell lines, and controls. MM, multiple myeloma; qRT-PCR, Real-time quantitative polymerase chain reaction. \*P < 0.05, \*\*\*\*P < 0.01.

# The Role of hsa\_circ\_0007841 in the Diagnosis of MM

ROC curves can reflect the specificity and sensitivity of continuous variables comprehensively. ROC curves were plotted for the MM group and control group to assess the diagnostic potential of hsa\_circ\_0007841 in MM patients (**Figure 3A**). The sensitivity and specificity of hsa\_circ\_0007841 were verified by this method, and the AUC value was calculated as 0.907 (95% CI 0.8476–0.9663). Therefore, hsa\_circ\_0007841 may be used as a diagnostic marker for MM.

# Prognostic Value of hsa\_circ\_0007841 in MM

The survival curve was constructed to analyze the correlation between hsa\_circ\_0007841 and the prognosis of MM patients (**Figure 3B**). All 86 patients adhered to the follow-ups. Using


TABLE 2 | Correlations between the relative expression of hsa\_circ\_0007841 and

MM, multiple myeloma; SD, standard deviation; \*P < 0.05.

Stage 3 39 12.290 ± 4.211

the median hsa\_circ\_0007841 expression in the bone marrow of MM patients as threshold, all 86 patients were divided into low and high hsa\_circ\_0007841 expression groups. The prognostic value of hsa\_circ\_0007841 in MM patients was assessed based on progression-free survival (PFS). The survival analysis showed that the upregulation of hsa\_circ\_0007841 was significantly correlated with poor prognosis of MM patients (log-rank P = 0.0206).

Prognostic biomarkers can be used to predict the possibility of recurrence and patient survival. As the prognosis of MM patients is closely related to genetic mutations, the correlation between hsa\_circ\_0007841 expressions in bone marrow and cytogenetic mutation was analyzed. It was found that the upregulation of hsa\_circ\_0007841 was correlated with chromosomal variations gain 1q21 (P = 0.039) and t (4;14) (P = 0.025) and mutations in ATR and IRF4 genes, but not correlated with del (17p) or mutations in TP53 (**Table 3**).

#### Prediction and Annotation of Target miRNA and mRNA Networks of hsa\_circ\_0007841

miRNA target prediction software TargetScan and miRanda from Arraystar was used to predict the circRNA—miRNA mRNA network. The miRNAs and candidate mRNAs binding to hsa\_circ\_220 were identified. The circRNA-miRNA interaction network was predicted on the CircInteractome database (**Figure 4A**). A total of 8 differentially expressed miRNAs and 10 candidate mRNAs were predicted to interact with hsa\_circ\_0007841. As shown in **Figure 4B**, the predicted mRNAs might be involved in multiple pathways. Therefore, it was necessary to further investigate the role of hsa\_circ\_0007841 in the progression of MM.

#### DISCUSSION

The incidence of MM has been rising yearly due, in part, to the aging population. Although new therapies have greatly improved the prognosis of MM, it remains incurable. One of the major contributing factors to this is the high heterogeneity of MM cells that leads to disease recurrence and drug resistance in patients (19). Individualized therapy based on biomarkers can maximize the efficacy and reduce disease recurrence as well as drug resistance. Some of the biomarkers already identified for MM may be useful in the selection of the most suitable therapy to alter the clinical outcome (20).

circRNA is the sponge molecule for miRNA that indirectly regulates the expression of miRNA target genes; thus, playing an important role in the onset and development of human diseases. As biomarkers, circRNAs are now widely used for disease diagnosis. Recently, circRNAs have received an increasing amount of attention in the biomedical literature. Vo et al. analyzed over 2,000 human tumor specimens derived from different tissues by exome capture and studied the circRNA profiles (21). Boutros and He systematically analyzed the expression characteristics of circRNAs in localized prostate cancer and designed the shRNA library for high-throughput sequencing. They reported that circCSNK1G3 competitively bind to miR-181b/d and exhibit fundamental function on four prostate cancer cells (22). Shen et al. reported the upregulation of circFAT1, formed by exon 2 in the FAT1 gene, in human osteosarcoma (OS). Their study demonstrated the oncogenicity of circFAT1 in OS and its potential value as a therapeutic target in OS (23). In addition, specifically expressed circRNAs have been identified in hepatocellular carcinoma (24), colorectal cancer (25), and lung cancer (26). These observations suggest the idea that circRNAs may be used as potential diagnostic and prognostic biomarkers. However, the clinical value of circRNAs in MM remains unclear.

We screen carefully the key circRNA based on the following considerations: (1) the significant difference in FC (fold change) value will be considered priority; (2) each sample has the original signal value, the tendency to select the original signal value is relatively large, and the average within the group is stable; (3) small P value and combined with the original signal value average within the group; (4) the appropriate length is

TABLE 3 | Cytogenetic aberration status distribution between low/high hsa\_circ\_0007841 expression groups of MM patients.


MM, multiple myeloma; \*P < 0.05.

recommended within 200–2,500 bp, in order to facilitate the later functional test; (5) try to choose the exonic type. In the present study, high-throughput sequencing was used to identify differentially expressed circRNAs in the bone marrow tissues of patients with MM. Several circRNAs were found to be upregulated, four significantly upregulated circRNAs were then verified by qRT-PCR in small clinical sample and MM cell lines, include hsa\_circ\_0010402, hsa\_circ\_0004777, hsa\_circ\_0080212, hsa\_circ\_0007841, and the hsa\_circ\_0007841 expression showed the most significant increase. This result was later verified by qRT-PCR in large samples (n = 86). To the best of our knowledge, this finding is the first confirmation of the expression, diagnostic, and prognostic value of hsa\_circ\_0007841 in MM. By combining these parameters with the clinicopathological indicators of patients with MM, we found a significant difference in hsa\_circ\_0007841 expressions among patients with different typing and staging. We further analyzed disease progression during the survival period between 1 and 4 years. We found that hsa\_circ\_0007841 overexpression was correlated with poor prognosis in patients with MM. Moreover, since the circ-miRNA axis has a considerable impact on the onset and development of human diseases, it has been speculated that hsa\_circ\_0007841 might influence the occurrence of MM by mediating the expression of oncogenes through its miRNA target. Annotation and functional prediction have shown that hsa\_circ\_0007841 interacted with eight miRNAs and 10 target mRNAs. Among these, hsa\_circ\_0007841 is the sponge molecule of hsa-miR-29b-2-5p. Rossi et al. found that miR29b overexpression inhibited osteoclast differentiation and reversed the MM cellstriggered osteoclast activation, which delayed the progression of MM. Furthermore, miR-29b caused the apoptosis of the BTZ-induced MM cells through activation of the feedback loop of transcription factor Sp1 (27). In the present study, hsa\_circ\_0007841 overexpression was found to be correlated with osteolytic bone destruction in MM and it was overexpressed in BTZ-resistance cell lines in MM. These results agree with previous studies, further showing the close connection between hsa\_circ\_0007841 and its diagnostic and prognostic value in patients with MM.

Drug resistance is the one that causes the most trouble during the therapy of MM in clinic settings and need urgent solution (28, 29). In our study, hsa\_circ\_0007841 expression was also detected in drug-resistant cell lines (KM3/BTZ, U266/BTZ, 8226/BTZ) and MM patients. hsa\_circ\_0007841 also is the sponge molecule of hsa-miR-199a-3p. Fornari et al. found that miR-199a-3p regulated mTOR and c-Met to influence the doxorubicin sensitivity of human hepatocarcinoma cells (30). Lei et al. found that miR-199a-3p affected the multi-chemoresistance of osteosarcoma through targeting AK4 (31). Although, a large biological signal pathways of circRNAs involved in drug resistance are still unknown. More mechanisms and functions of chemoresistance-related circRNAs need to be further mined for advance of MM therapy, which may offer new approaches to reverse drug resistance.

Prognostic biomarkers can be used to assess the probability of disease recurrence and the prediction of clinical outcomes. Since prognostic indicators are not formulated by considering

the factor of treatment, they can be used to guide individualized therapy. It has been shown that some genetic mutations are closely related to the poor outcome of MM, including chromosomal alterations t (4; 14), t (14; 16), and t (14; 20). Due to chromosomal translocation, oncogenes such as MMSET/FGFR3, CCND3, CCND1, MAF, and MAFB are still regulated by the IgH gene enhancer and their expressions increase (11, 32). As a result, the cyclin D family members are upregulated, leading to the dysregulation of the G1/S checkpoint. In the present study, hsa\_circ\_0007841 was closely related to chromosomal variations and genetic mutations in patients with MM. However, the present study is a single-center small-sample-size experiment, from which the results had limitations in guiding clinical treatment. In the future, the sample size should be enlarged. Further detection of its expression changes at different stages of disease progression is needed. Cell and animal experiments are needed to further explore its biological functions. Based on this fact, hsa\_circ\_0007841 may be used to predict the prognosis, recurrence, and drug resistance in patients with MM.

This study has evaluated the diagnostic and prognostic value of hsa\_circ\_0007841 as a circRNA in MM and the results provide the basis for preclinical research. Currently, the factors of age and complications are mainly considered for individualized therapy of MM, However, to further improve treatment outcome, molecular information will be needed to guide treatment. The endogenous competitive mechanism of circRNAs provides the theoretical basis for research on the replacement of antinucleotide chemotherapy. However, studies on circRNAs are still at the preliminary stage and very little has been hitherto understood about their functions and working mechanisms. But there is a high potential for clinical translational research with the advancement of research and development of RNA-targeted drugs based on circRNAs.

#### DATA AVAILABILITY STATEMENT

This manuscript contains previously unpublished data. The name of the repository and accession number(s) are not available.

#### REFERENCES


#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by 2016121. The patients/participants provided their written informed consent to participate in this study.

#### AUTHOR CONTRIBUTIONS

MG and CL designed and performed the study. HD and LG wrote the manuscript with inputs from all authors. YF and SJ performed the analytic calculations and statistical analysis. All authors provided critical feedback and helped to shape the research, analysis, and manuscript. We thank HX for her contribution to the collection of data.

#### FUNDING

This study was supported by Hunan Provincial Natural Science Foundation of China (Grant No. 2017JJ3463).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2019.01261/full#supplementary-material


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Gao, Li, Xiao, Dong, Jiang, Fu and Gong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Canmei Formula Reduces Colitis-Associated Colorectal Carcinogenesis in Mice by Modulating the Composition of Gut Microbiota

#### Huayue Zhang1†, Dengcheng Hui 2†, Yuan Li <sup>1</sup> , Guangsu Xiong<sup>3</sup> \* and Xiaoling Fu<sup>1</sup> \*

*<sup>1</sup> Department of Medical Oncology, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China, <sup>2</sup> Department of Cirrhosis, Shuguang Hospital, Shanghai University of Traditional Chinese Medicine, Shanghai, China, <sup>3</sup> Endoscopic Center, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Jianfeng Cai, University of South Florida, United States Guoxiang Cai, Fudan University Shanghai Cancer Center, China*

#### \*Correspondence:

*Xiaoling Fu fuxiaoling111@163.com Guangsu Xiong xiongguangsu@ shyueyanghospital.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *09 August 2019* Accepted: *16 October 2019* Published: *19 November 2019*

#### Citation:

*Zhang H, Hui D, Li Y, Xiong G and Fu X (2019) Canmei Formula Reduces Colitis-Associated Colorectal Carcinogenesis in Mice by Modulating the Composition of Gut Microbiota. Front. Oncol. 9:1149. doi: 10.3389/fonc.2019.01149* The gut microbiota, including pathogenic microorganisms and probiotics, has been involved in tumor initiation and progression by regulating the components of intestinal flora. Canmei formula (CMF), a traditional Chinese medicine, chronicled in the Chuang Yang Jing Yan Quan Shu, has been clinically used as an adjuvant therapy to treat patients with colorectal carcinoma (CRC) in China. In this study, we investigate the treatment effect of CMF in the azoxymethane (AOM) and dextran sodium sulfate (DSS) induced and high-fat diet augmented colitis-associated colorectal cancer *in vivo*, and explore its mechanism of action. We found that CMF treatment relieved the inflammation and alteration of the gut microbiota and significantly inhibited the development of intestinal adenoma. Linear discriminant analysis showed that the flora diversity in the normal mice, model mice and CMF treatment mice was different. At the family level, the relative abundance of Desulfovibrionaceae decreased in CMF groups. The relative abundance of Desulfovibrionaceae were lower in the CMF groups than in model group, whereas Rikenellaceae and Alistipes were increased. Altogether our results indicate that CMF treatment ameliorate colitis-associated colorectal carcinogenesis by modulating the composition of the gut microbiota *in vivo*.

Keywords: Canmei formula (CMF), traditional Chinese medicine, gut microbiota, AOM/DSS, colorectal carcinogenesis

# INTRODUCTION

Colorectal carcinoma (CRC) is the second leading cause of cancer-related death worldwide (1). There are about 1.4 million new CRC patients and 700,000 deaths each year (2). Colorectal adenoma (CRA) refers to any lesion that originates from the surface of colorectal mucosa and protrudes into the intestinal cavity. It is a precancerous lesion of CRC and has the characteristics of high recurrence and high incidence of cancerous change (3). Microscopic removal is currently the main treatment for CRA, which can effectively reduce the incidence of CRC and reduce mortality

**140**

by 67% (4). However, microscopic resection did not reduce the recurrence rate of CRA. Aspirin, celecoxib, and cyclooxygenase-2 inhibitors are used clinically to treat inflammatory bowel disease, but the efficacy is limited according to the large sample clinical data (5). To date, there is no therapeutic agent specific for this disease. It is critical to explore drugs that can safely and effectively prevent and treat the recurrence of CRA.

In China, traditional Chinese medicine (TCM) has a long history dating back several thousands of years ago. Results from both pre-clinical laboratory and human studies have indicated that TCM has been widely used in anticancer therapy. Previous studies have indicated that PHY906, a modified formulation derived from Huang-Qin-Tang, could ameliorate chemotherapyinduced GI tract toxicity (6, 7). Qingjie Fuzheng granules (QFGs) has been proven to be able to inhibit proliferation and induce apoptosis of CRC cells (8). These studies indicate that TCM play important roles in cancer treatment and could be expected to become a promising therapeutic candidate for tumor.

Over the past decade, CRC have emerged as one of the most studied human conditions link to gut microbiota. Studies indicate that changes in the composition of gut microbiota, one of the most important internal environmental factors, are closely associated with the progression of CRC (9, 10). At present, the study on the secretion of toxin-activated inflammation by the intestinal flora is the most in-depth study of CRA carcinogenesis (3). The changes of intestinal flora are positively correlated with colorectal cancer. The secretion of toxins by the bacteria is a key factor for activating inflammation and oxidative stress pathways (11, 12). A number of studies have confirmed that Enterococcus faecalis, Escherichia coli, Enterotoxin-producing Bacteroides fragilis (ETBF), Streptococcus bovis, Fusobacterium nucleatum, and Helicobacter pylori are the main species that can induce cancer (3, 12, 13). Among them, Fusarium nucleatum induces uncontrolled intestinal epithelial cell growth by activating βcatenin; E. coli produces colibactin with genotoxicity (3).

Canmei formula (CMF), a classical traditional Chinese herbal formulation with Titanium and Marci Hieronymi, is widely used to treat colorectal diseases. The active constituents in CMF, including Citric acid, DL-TYROSINE, L(-)-Carnitine, L-Tyrosine, Ambrosic acid and others, have been reported (14, 15). However, there is no study on the effect of CMF in modulating gut microbiota. In our previous studies, 16sDNA sequencing technique was used to compare the change of gut microbiota induced by CMF treatment in AOM/DSS and high-fat dietinduced CRC mice (16, 17). This study provides evidence of the modulations of gut microbiota through CMF treatment, which will help understanding the host-microbe interactions during the treatment of CRC.

#### MATERIALS AND METHODS

#### Preparation of the Extracts for CMF

Mume Sieb and Marci Hieronymi were purchased from Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine. Mume Sieb was the fruit of Prunus mume. The Marci Hieronymi is a dry body of the larvae of the camphor family, Bombyx mori linnaeus, 4 ∼ 5 years old, infected with Beauveria bassiana. They were formally identified by the department of pharmacy, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China.

CMF was formulated with Titanium and Marci Hieronymi in a ratio of 1:1. All the herbs were purchased from Yueyang hospital herbal pharmacy. Formulation was performed as described below. Briefly, 980 g mixture was extracted twice for 1.5 and 1 h for each time. The filtrates were concentrated in a vacuum at 60◦C and dried to obtain the CMF extract powder of 335.2 g, the rate of the extract was 32.1%.The extract obtained in this part is CMF-L/H. The same process was repeated twice for the CMF extract. Then the liquid was poured through the column of the macroporous resin, and then the resin column was rinsed with pure water, 20% ethanol and 90% ethanol. After alcohol precipitation, the filtrateswere concentrated in a vacuum at 60◦C and dried to obtain the CMF extract powder of 60.3 g, the rate of the purified product was 18%. The extract obtained in this part is used as CMF-A. The quality of preparations was controlled according to the guidelines from Chinese food and drug administration (CFDA).

#### Liquid Chromatograph and Mass Spectrometry

The solution obtained by the above preparation method was stored at 4◦C. For HPLC analysis, the solution was centrifuged at 12,000 rpm for 10 min, then the supernatant was filtered through a 0.22µm membrane before injection into the HPLC system. Analysis were performed by using Dionex UltiMate 3000 HPLC system (ThermoFisherScientific, USA) with a diode array detector. The C18 column (150 × 2.1 mm, 2.5µm) was used with a flow rate of 0.3 ml·min−<sup>1</sup> . The injection volume was 5 µl, and the column temperature was maintained at 40◦C. The mobile phase was composed of (A) aqueous formic acid (0.1%, v/v) and (B) acetonitrile (0.1%, v/v) under following gradient elution: 95% A from 0 to 0.5 min, 95-2% A from 0.5 to 15 min, 2% A from 15 to 17 min, 2-95% A from 17 to 17.5 min, 95% A from 17.5 to 25 min. Mass spectrometry was performed on a Q Exactive high-resolution benchtop quadrupole Orbitrap mass spectrometer (Thermo Fisher Scientific, USA) using a heated electrospray ionization (HESI) source for ionization of the target compounds in positive and negative ion modes. The key parameters were as follows: ionization voltage, +3.5 kV/−3.2 Kv; sheath gas pressure, 40 arbitrary units; auxiliary gas, 10 arbitrary units; heat temperature, 350◦C. For the compounds of interest, a scan range of m/z 120–1,500 was chosen. Resolution for higher energy collisional dissociation cell (HCD) spectra was set to 17,500 at m/z 120 on the QExactive.

#### Animal Experiments and Grouping

The animal procedures and care in this study was approved by the IRB of Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine. Seven-week-old female C57/BL6 mice were purchased from Shanghai Jihui Experimental Animal Breeding Co., Ltd. After arrival, the animals were quarantined for the first 7 days. Under controlled conditions: humidity 50 + 10%, 12/12 h light/dark cycle, and temperature ∼23◦C, all animals were housed in plastic cages (6 mice/cage)and provided with free drinking water and basic diet. AOM, the colonic carcinogen, was purchased from Sigma-Aldrich, USA. DSS with a molecular weight of 36,000–50,000 (No.216011080) was purchased from MPBIO Company, USA. For the induction of tumorigenesis, mice were injected with AOM (12 mg/kg body weight) intraperitoneally. Starting 5 day after the injection, animals received 2.5% DSS in the drinking water for 5 days. At the same time, the mice were given a high-fat diet. Then mice were given normal drinking water and normal diet for 2 weeks and subjected to two more DSS treatment cycles. Subsequently, mice were randomized into experimental and control groups. Group 1, NC, n = 8. Experimental groups included 5 groups: group 2: treatment with AOM and DSS (MC, n = 8); group 3: treatment with AOM, DSS, and Aspirin (Aspirin, n = 8); group 4: treatment with AOM, DSS, and CMF ethanol extract (CMF-A, n = 8); group 5: treatment with AOM, DSS, and CMF aqueous extract (CMF-L, n = 8); group 6: treatment with AOM, DSS, and CMF water extract high dose group(CMF-H, n = 8). Aspirin was given every day at the dose of 1.4 mg/kg/d. CMF-L and CMF-H were given at the doses of 3.65 and 7.3 g/kg/d, respectively. CMF-A was given at the doses of 0.657 g/kg/d. In the clinical practice of Chinese herbal medicine, the prescription dose of CMF is usually 60 g of herbal materials per day. When this human dose was converted into an animal dose (a person of 60 kg, and a conversion factor of 9.1 between human and mouse), it was equivalent to the low dose (3.65 g/kg/d) used in this study. All animals were sacrificed at the 18th week after the corresponding drug was administrated.

#### Colon Tissue Collection and Processing

After mice were sacrificed, the colorectal adenoma tissues were rapidly isolated, freed of adherent tissues and rinsed with phosphate buffer saline. Subsequently, tumor biopsy specimens of 1 cm-length were fixed with 4% paraformaldehyde for 24 h, embedded in paraffin and sectioned into 4-µm-thick slices. Slices were hydrated and stained with hematoxylin and eosin (H&E) followed by microscopic examination. Then portions of adenoma tissues were homogenized (10% W/V) in (20 mM Tris-HCI containing 1 mM EDTA, PH 7.4) by a glass homogenizer. The homogenates were centrifuged at 3,000 × g for 20 min at 4◦C and supernatants were stored at −80◦C.

#### Real Time Quantitative Reverse Transcription PCR

According to the modified methods as described, the RNA extracts of the colorectal tissues were prepared (8). By normalizing target mRNA Ct values to those for GAPDH (Ct), the relative amount of target mRNA was determined by the comparative threshold (Ct) method. The primer sequences used were as follows:

MUS-NF-κB-F (bp1297): (ATGCACCGTAACAGCAGGAC); MUS-NF-κB-R (bp1405C): (TGTCATCCGTGCTTCCAGTG); MUS-IL-17C-F (bp269) (AGGAGGTGCTGGAAGCTGAC); MUS-IL-17C-R (bp391C) (TGCATCCACGACACAAGCAT);

GAPDH-F (5′ -GTGAGGCCGGTGCTGAGTAT-3′ ); GAPDH-R (5′ -GTGCAGGATGCATTGCTGAC-3′ );

# ELISA Assay

Concentration of IL-17C in serum and NF-κB in supernatants of colonic tissues were quantified by IL-17C assay kit and NFκB assay kit (Shanghai Pusheng Biological Technology Co., Ltd. China) according to manufacturer's instruction.

#### Fecal DNA Extraction and Illumina Miseq Sequencing

Genomic DNA was extracted from every stool sample using the FastDNA Spin Kit for Soil (MP Biomedical, LLC, catalog 116560- 200) following the manufacturer's instructions.V3 and V4 region of the 16s rDNA sequencing was amplified by Primer F (5′ -AAC GGGAAGACAACGTACGG-3′ ) and Primer R (5′ -CAGATG CAGGAGGACATGTC-3′ ) with barcode sequence. Library was constructed following the manufacturer's instructions of the Ion Plus Fragment Library Kit and sequenced by Ion S5 Sequencer.

#### Bioinformatics Analysis

Reads was filtered and chimera sequence was removed.

The read pairs were de-multiplexed based on their unique molecular barcodes and overlapping reads were merged using USEARCH v7.0.1001 software. Operational taxonomic unit (OTU) picking was conducted using the QIIME (Quantitative Insights Into Microbial Ecology) software package. By using UCLUST, 16S rRNA gene sequences were clustered at a similarity cutoff value of 97%. By the SILVA reference database (https://www.arb-silva.de/download/), matching of OTUs to bacteria was conducted. By mapping the demultiplexed reads to the identified OTUs, abundances were recovered. Alpha andbeta diversity plots were also generated using QIIME. The non-weighted UniFrac approach was used to measure the beta diversity between five bacterial cave communities. The relative abundance of transporter genes was predicted and performed by picrust analysis. Independent of the taxonomic analysis, aclosed-reference OTU picking protocol (QIIME) and the Greengenes database (http://greengenes.lbl.gov) were used to pre-cluster at 97% identity, and 97% of the OTUs were picked. The obtained OTU table was normalized by 16S rRNA copy number, and the functional genes were predicted according to the Kyoto Encyclopedia ofGenes and Genomes (KEGG) catalog. By using SPSS version 15.0 statistical software (IBM, Armonk, NY, USA), Fisher's test and the Mann-Whitney U testwere used to analyze, respectively, the gender distribution of subjects and intergroup differences at the genus level in each subcluster. The linear discriminant analysis (LDA) effectsize (LEfSe) method with default settings was used to analyze the differences among groups in phyla, class, order, family, and genera levels.

#### Statistical Analysis

SPSS software version 15.0 was used for statistical analysis. The data were presented as mean ± SD. Unpaired students' t-test was used to compare the means of the two groups. One-way analysis of variance and Adonis were used to compare the means of more than two groups. A level of P < 0.05 was considered as statistically significant.

#### RESULTS

#### Identification of the Main Components in the Extract of CMF

To identify the components of Mume Sieb and Marci Hieronymi extract, Q Exactive high-performance benchtop quadrupole-Orbitrap LC-MS/MS was performed. The main components in the extract of CMF were identified according to the elemental composition data determined from accurate mass measurements and comparison with the literature data. Fortyone compounds were identified in the CMF extract and demonstrated in **Table 1**. In addition, to exclude the possibility of manufacturing problems relating to procession, extraction, handling, and storage, two batches of CMF were evaluated by HPLC. We found that the magnitude, number, and retention time of the peaks were similar between the two batches (**Figure 1**). These results confirmed the stability of the CMF extraction process.

TABLE 1 | Major chemical constituents identified in CMF.


# CMF Attenuated AOM/DSS-Induced Colitis-Associated Tumorigenesis

To evaluate the protective effect of CMF on the development of colitis-associated cancer, the well-established AOM/DSS and high-fat diet-induced colitis-associated colorectal tumor model in C57BL/6 mice were used (**Figure 2A**). Initially, the toxicity of CMF was tested in vivo. We did not find weight loss or behavioral changes in all mice. These results indicate that each of those extracts of CMF were well-tolerated (**Figure 2B**). AOM/DSS colorectal cancer mice were induced as described. All mice were treated with CMF once a day. After CMF treatment, tumor development and growth was inhibited in AOM/DSS colorectal cancer mice (**Figures 2C–F**). As a control, aspirin treatment effectively inhibited the occurrence of intestinal adenomas of 3 mm and above, but had no inhibitory effect on the occurrence of adenomas smaller than 3 mm (**Figure 2C**). Different extracts of CMF can significantly inhibit the growth of intestinal glands. **Figures 2D–F** shows that the ethanol extract of CMF significantly inhibited the occurrence of intestinal adenoma, and there is no difference in the occurrence of adenoma in the colon and small intestine. Histological examination revealed that mucosal inflammation, aberrant crypt foci adenoma, ulcer, or dysplasia in the colon tissues of the model group. Histological studies indicated that there was a large adenocarcinoma inside mucosa with several abnormal cells exhibiting cylindrical shape, large nuclei, increasing nuclear/cytoplasmic (N/C) ratio and cellular cleavagein AOM/DSS group (**Figure 2G**). CMF remarkably relieved the condition. These results suggest that CMF treatment inhibits inflammationassociated carcinogenesis in the AOM/DSS and high-fat diet mice.

#### CMF Treatment Repressed NF-κB and IL-17C Signaling in AOM/DSS Induced Mice

As NF-κB and IL-17C play important roles in the development of CRC, the mRNA expression of IL-17C and NF-κB collected from CRC mice were detected by Quantitative real-time PCR (**Figures 3A,B**). IL-17C was upregulated in the colorectal tissues of AOM/DSS mice. However, treatment with different extracts of CMF, including water extract and alcohol extract, effectively suppressed the mRNA expression of IL-17C. Drugs treatment, including aspirin, induced the mRNA expression of NF-κB. ELISA assay was performed to detect the protein expressions of inflammatory factors. As it is illustrated in **Figure 3C**, the protein levels of IL-17C in the serum of AOM/DSS-induced mice were significantly highest among mice with other drug intervention. The level of NF-κB significantly reduced in the tumor tissues of CMF-A, CMF-H, CMF-L, and Aspirin groups when compared to MC group (**Figure 3D**). All these results indicate that CMF treatment represses the expression of NF-κB and IL-17C.

# CMF Treatment Regulated Gut Microbiota

To examine the effect of CMF on the regulation of components of gut microbiota, high-throughput pyrosequencing was performed with an Illumine MiSeq platform to generate 39584 high quality and valid sequences from 16 samples from different groups. We conducted a bar-coded pyrosequencing run to analyze the changes of gut microbiota in the five studied groups. In total, 39,584 available reads and 1461 OUTs were obtained. Shannon diversity and rarefaction curves are shown in **Figures 4A,B**. The rarefaction curves were consistent with the current sequencing, demonstrating that most of the flora diversity has been captured in all samples. What's more, the overlap of OUTs between groups revealed that 482 OUTs coexisted in both the MC group and CMF-A group, 496 OUTs coexisted in both the NC group and CMF-A group, 514 OUTs coexisted in both the NC group and MC group and 467 OUTs coexisted in both the MC group and Aspirin group (**Figures 4C,D**). In addition, we examined the flora community richness and diversity in mice. No statistical difference was observed for the ACE, Chao, Shannon, and Simpson indices among these groups (**Figures 4E–H**), although Simpson index were lower, but other indices are higher in NC group and the different CMF extract groups compared with MC group. Nonmetric multidimensional scaling (NMDS) analysis indicated the changes of overall structure of gut microbiota among different groups (**Figure 4I**). These results showed that the

mean ± SD of mice in each group.

diversity of bacterial population of the NC group was higher than that of the MC group. There was a statistical difference between the CMF-H group and the MC group. These results suggested that CMF treatment increases the diversity of the intestinal flora.

# Analysis of the Relative Abundance of Microbiota at the Family Level and Genus Level

The family muribaculaceae, bacteroidales, and prevotellaceae were the most prevalent taxa in these groups (**Figure 5A**).

It was found that these bacteria are dominant bacteria at the genus level (**Figure 5B**). Intraindividual changes were detected in these groups. Compared with the model group, the abundance of a small number of bacteria increased or decreased to varying degrees, and most of the bacteria remain unchanged. At the family level, abundance of Desulfovibrionaceae decreased in groups that were treated with CMF. Conversely, a transient CMF-treatment increase several family microbiota, including Rikenellaceae, Erysipelotrichaceae, Lactobacillaceae, Streptococcaceae, and Tannerellaceae. At the genus level, the abundance of genus Alistipes, Faecalibaculum, Lactobacillus, and Parabacterioides, were up-regulated after the treatment of CMF. The genus Parasutterella was found down-regulated in CMF treatment groups. It is worth mentioning that genus Desulfovibrionaceae\_uncultured acted to AOM/DSS-mediated gut microbiota dysbiosis had statistical difference (**Figure 5F**). There arechanges in the

gut microbiota but there is no statistical difference in these changes (**Figures 5C–E**).

To systematically examine the alteration of taxa abundance according to CMF treatment, we conducted LEfSe (LDA Effect Size). The microbial cladogram indicates that the gut microbiome significantly altered in Bacteroidaceae and Staphylococcaceae, ranging from family to genus level (**Figures 6A,C**). At the genus level, the relative abundance of Turicibacter, Bacteroides, Bacteroidaceae, Faecalibaculum, Erysipelatoclostridium, and Staphylococcus were enriched in mice treated with the CMF-A diet (**Figures 6B,D**).

#### DISCUSSION

Traditional Chinese medicine (TCM) have been used for the treatment of diseases and health conditions for several thousands of years. In the treatment of CRC, the application of TCM has

Frontiers in Oncology | www.frontiersin.org

received worldwide attention. Many studies have demonstrated the anti-tumor effect of TCM in CRC therapy. For example, Wei et al. showed that JianPi JieDu decoction (JPJD) is able to inhibit tumorigenesis, metastasis, as well as angiogenesis of tumors through the mTOR/HIF-1α/VEGF pathway (18). Wu et al. demonstrated that Gegen Qinlian decoction (GGT) can ameliorate gut-toxicity through anti-inflammatory pathways, inhibition of neutrophil migration, anti-oxidative stress via the Nrf2/Keap-1 pathway in the colon (19). Wing lam et al. reported that PHY906 reduced the CPT-11-induced gastrointestinal toxicity by inhibiting multiple steps of inflammation and the promotion of intestinal progenitor cell repopulation (20). Extensive studies have revealed that TCM antitumor drugs work by inhibiting tumor proliferation, promoting tumor apoptosis, and affecting diverse molecular targets (21–23).

The alterations of cancer genetic and molecular targets in CRC treatment have been extensively reported. Recent studies have reported that the regulation of gut microbiota may affect carcinogenesis (24–28). Intestinal flora is a direct factor of intestinal cancer. In this study, we found that CMF can ameliorate colitis-associated colorectal carcinogenesis in mice by modulating the fecal microbiota. The gut microbiota, including Fusobacterium (F.) nucleatum (29) and Firmicutes, Bacteroidetes Ley et al. (30), plays an important role in human health. In addition, many factors will affect and destroy the balance of intestinal microbiota, including the invasion of antigens, activation of immune cells, and production of cytokines. Therefore, investigating the relationship between CMF and the gut microbiota in AOM/DSS-induced colitis in miceis critical.

In our study, the anti-inflammatory role of CMF in the in vivo model of colitis-associated colorectal carcinogenesis induced by AOM/DSS and high-fat diet in C57BL/6 micewere investigated. Two different solvent extraction components of CMF were used to treat these mice to evaluate its therapeutic effect. The results showed that each of these extracts of CMF were welltolerated. Histopathology findings were consistent with previous experimental results (31). The CRC mice treated with AOM/DSS and high-fat diet presented more severe condition than CMFtreated mice. In contrast, administration of CMF decreased tumor number and tumor size. Especially the ethanol extract of CMF can effectively inhibit the occurrence of colitis-associated colorectal carcinogenesis in the colon and small intestine.

Previous studies showed that tumor-prone mice colonized with E. coli and B. fragilis and increased IL-17 expression that promotes colon tumorigenesis in AOM mice (3, 32). It was reported that spontaneous activation of NF-κB was detected in tissues of human colorectal cancer (31, 33).We found that CMF could decrease the expression level of IL-17C and inhibit the activity of NF-κB.

High-throughput sequencing determined microbiological composition in mice. Most of the diversity of microorganisms was captured in all samples. The ACE, Chao, Simpson, and Shannon indices reveal that the microbial diversity in control mice and CMF treated mice were greater than in the model mice (16). The microflora analysis show that there are significant differences between the control mice and the model mice. There was a statistical difference between the ethanol extract of CMF treated mice and the model mice. This result indicated that the ethanol extract of CMF could increase the diversity of the intestinal flora. We evaluated phylum, class, order, family, and genus, the gut microbiota community in all samples. From the experiment, Muribaculaceae, Bacteroidales, and Prevotellaceae were the dominant family found in the gut. In accordance with family criteria, these flora occupied a main tier in the genus level. Focus on certain bacteria, statistical results reveal that CMF reversed the gut microbiota dysbiosis in CRC mice, including inhibiting the growth of Desulfovibrionaceae and promoting the Rikenellaceae. Desulfovibrionaceae, a kind of sulfate-reducing and endotoxin-producing bacteria, were mostly enriched in mice with long term high-fat feeding. Previous studies found that the family Desulfovibrionacea is positively associated with obesity (9, 34–36). This phenomenon was also found in our study. Wu et al. found that relative abundances of Rikenellaceae increased in the AOM/DSS mice (32). This result was in consistent with our study. We speculate that Rikenellaceae might be a bacterium that could cause cancer.

Because the results of our study showed that the alcohol extract of CMF is more effective than that of the water extract, we analyzed the difference of flora of the mice treated with the alcohol extract of CMF and the model mice. The results showed that the gut microbiome significantly altered in Bacteroidaceae and Staphylococcaceae, ranging from family to genus level. As shown in our results, at genus level, the relative abundance of Turicibacter, Bacteroides, Faecalibaculum, Erysipelatoclostridium,

# REFERENCES


and Staphylococcus were enriched in mice treated with the CMF-A diet. In a previous study, the abundance of Bacteroidaceae had an important influence in the active progression of colitis in DSS-treated mice (16), which was in contrast to our study. Liu et al. found that the abundance of Turicibacter decreased in obesity mice (37, 38). Similarly, the Turicibacter was also reduced in AOM/DSS and high-fat diet mice in this study. Therefore, we suspect that Turicibacter, Bacteroides, Faecalibaculum, Erysipelatoclostridium, and Staphylococcus were the beneficial bacteria in the gut. But the specific role of these gut microbiota needs to be further investigated.

In summary, we report that CMF, a two-herb formulation, has reduced the incidence of CRC by inhibiting inflammatory factors and regulating gut microbiota. In addition, the mechanism of action of CMF is through the direct action on inflammatory factors or the indirect effect on regulating gut flora. Understanding the role of gut microbiota would be helpful for the prevention of CRC development.

# DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

#### ETHICS STATEMENT

The animal study was reviewed and approved by IRB of Yueyang Hospital Affiliated with Shanghai University of Traditional Chinese Medicine. Written informed consent was obtained from the owners for the participation of their animals in this study.

#### AUTHOR CONTRIBUTIONS

XF, HZ, and GX contributed to study design, data interpretation, and manuscript preparation. YL contributed to animal rearing. DH contributed to data acquisition and analysis.

#### FUNDING

This work was supported by the National Natural Science Foundation of China (81403360). The funders play no role in data collection and analysis, design, decision to publish, or preparation of the manuscript.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhang, Hui, Li, Xiong and Fu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Survivin Promotes Piperlongumine Resistance in Ovarian Cancer

Xing-Wei Nan1†, Li-Hua Gong1†, Xu Chen1†, Hai-Hong Zhou<sup>1</sup> , Piao-Piao Ye<sup>1</sup> , Yang Yang<sup>2</sup> , Zi-Hao Xing<sup>2</sup> , Meng-Ning Wei <sup>2</sup> , Yao Li <sup>2</sup> , Sheng-Te Wang<sup>2</sup> , Kun Liu<sup>2</sup> , Zhi Shi <sup>2</sup> \* and Xiao-Jian Yan1,3 \*

*<sup>1</sup> Department of Gynecology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China, <sup>2</sup> Guangdong Provincial Key Laboratory of Bioengineering Medicine, Department of Cell Biology and Institute of Biomedicine, National Engineering Research Center of Genetic Medicine, College of Life Science and Technology, Jinan University, Guangzhou, China, <sup>3</sup> Center for Uterine Cancer Diagnosis & Therapy Research of Zhejiang Province, Women's Hospital and Institute of Translational Medicine, Zhejiang University School of Medicine, Hangzhou, China*

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, China*

#### Reviewed by:

*Qing Wang, Sun Yat-sen University, China Shuangping Liu, Dalian University, China*

#### \*Correspondence:

*Zhi Shi tshizhi@jnu.edu.cn Xiao-Jian Yan yxjbetter@126.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *27 September 2019* Accepted: *15 November 2019* Published: *29 November 2019*

#### Citation:

*Nan X-W, Gong L-H, Chen X, Zhou H-H, Ye P-P, Yang Y, Xing Z-H, Wei M-N, Li Y, Wang S-T, Liu K, Shi Z and Yan X-J (2019) Survivin Promotes Piperlongumine Resistance in Ovarian Cancer. Front. Oncol. 9:1345. doi: 10.3389/fonc.2019.01345* Ovarian cancer is one of the most fatal female malignancies while targeting apoptosis is critical for improving ovarian cancer patients' lives. Survivin is regarded as the most robust anti-apoptosis protein, and its overexpression in ovarian cancer is related to poor survival and apoptosis resistance. Piperlongumine (PL) extracted from peppers is defined as an active alkaloid/amide and exhibits a broad spectrum of antitumor effects. Here, we demonstrate that PL induces the rapid depletion of survivin protein levels via reactive oxygen species (ROS)-mediated proteasome-dependent pathway *in vitro*, while exerting a remarkable inhibitory influence on the proliferation of ovarian cancer cells. Overexpression of survivin raises the survival rate of ovarian cancer cells to PL. Moreover, PL inhibits ovarian cancer cells xenograft tumor growth and downregulates survivin *in vivo*. Our findings reveal a previously unrecognized mechanism of PL in suppressing survivin expression as well as survivin promotes piperlongumine resistance in ovarian cancer and suggest that ROS-mediated proteasome-dependent pathway can be exploited to overcome apoptosis resistance triggered by aberrant expression of survivin.

#### Keywords: piperlongumine, survivin, ovarian cancer, ROS, proteasome

#### INTRODUCTION

Ovarian cancer is the most fatal female reproductive tract cancer and the seventh most common cancer among women all over the world (1). It was evaluated that 22,240 ovarian cancer cases would be diagnosed while concurrently 14,070 women would succumb to the disease in 2018 (2). Between 1976 and 2015, owing to effective approaches to the clinical treatment, including aggressive cytoreductive surgery followed by system chemotherapy, the mortality rate of ovarian cancer declined by 33% (2). However, despite the markable achievement of ovarian cancer during the last decade, the prognosis for women with advanced disease remains poor and chemoresistance appears to be a significant obstacle constraining the clinical application (3). Progress in curbing the incidence and mortality of human ovarian cancer can be propelled by the development of novel agents.

Survivin, one of the inhibitor of apoptosis proteins family (IAP) firstly found in 1997, may accelerate the progress of cancer via promoting the insurgence of mutations as well as inducing resistance to chemotherapy (4). While undetectable in terminally differentiated adult tissues, the survivin gene has been proved to be overexpressed in lots of cancers including ovarian cancer (5) in which survivin overexpression is detected in 74% of case and is related to the advanced clinical stage (6). Furthermore, our previous studies showed that the cytoplasmic protein expression level of survivin was an independent prognosis marker for ovarian cancer, downregulating survivin expression could enhance apoptosis (7, 8). Thus, survivin could be a promising target for apoptosis-based treatment in ovarian cancer therapy (9).

Piperlongumine (PL) extracted from the long pepper (Piperlongum L.) belongs to a biological alkaloid, embodying potential selective cytotoxic and antitumor properties over cancer cells of several histo-types including ovarian cancer (10). Our previous studies firstly demonstrated that PL facilitated G2/M cell cycle arrest, impeded the proliferation of human ovarian cancer cells, and induced reactive oxygen species (ROS) mediated apoptosis (11). Our previous cell-based experiments (11) and data published (12) indicated that the effect of PL to suppress the growth of tumors without general toxicity and to perturb redox and ROS homeostasis verify the PL treatment as a potential therapeutic opportunity against ovarian cancer.

However, the detailed molecular mechanism of the PL-mediated antitumor effect is not well-understood and still requires further investigation. Here, we firstly demonstrate that PL depleted the expression of survivin protein via a ROSmediated proteasome-dependent pathway in ovarian cancer cells. We further evaluate the efficacy of PL to inhibit tumor growth in vivo as a potentially effective therapy for ovarian cancer treatment.

#### MATERIALS AND METHODS

#### Cell Culture

The cancer cell lines A2780, OVCAR-3, and HEK293T were cultured in DMEM (Gibco, NY, USA). Ten percent fetal bovine serum (Gibco, NY, USA) and antibiotics (10 mg/mL streptomycin and 10,000 U/mL penicillin) were additionally supplemented. The aforementioned cell lines were cultivated in a 37◦C incubator at 5% CO2.

#### Chemicals and Antibodies

PL, N-Acetyl-cysteine (NAC) and Chloroquine (CQ) were obtained from Sigma Chemical Co (St. Louis, MO, USA). MG132 was from ApexBio. The HRP secondary antibodies, anti-survivin (#2808), and anti-PARP (#9542) were obtained from Cell Signaling Technologies (Danvers, MA, USA). Anti-Vinculin (BM1611) antibody was obtained from Boster, China. Anti-β-actin (KM9001), and anti-GAPDH (KM9002) antibodies were obtained from Tianjin Sungene Biotech Co., China. MG132, CQ, and NAC were added 1 h before PL treatment in all cotreated experimented.

#### Western Blot

Cells were lysed in RIRA buffer at 4◦C for around 30 min followed by centrifugation at 13,200 × rpm for 10 min to remove nuclei and other cell debris. Total protein concentration was detected by the Micro BCA Protein Assay Kit (Sangon Biotech, C503061) and the lysates were either used immediately or stored at −80◦C. 10–12% SDS-PAGE gels were used to separate the proteins extracted before and thereby separated proteins were completely transferred to polyvinylidene difluoride (PVDF) membranes. Five percent BSA was used to block the membranes for 1 h and the indicated primary antibodies were added and incubated overnight. Membranes were probed with the chemiluminescent detection reagents, and reactive bands were visualized using UVP ChemStudio PLUS (Analytikjena) (13, 14).

#### RT-PCR

Total RNA was extracted from 5 × 10<sup>6</sup> to 5 × 10<sup>7</sup> cells using the RaPure Total RNA Mini Kit (Magen, #R4011-02), treated with DNAseI to eliminate genomic DNA, and quantitated using the Epoch spectrophotometer. Reverse transcription (RT) followed instructions provided by StarScript II First-strand cDNA Synthesis Kit-II (GenStar, #A214-05). The resulting cDNA was used as a template for the amplification of target gene transcripts by RT-PCR, using HieffTM qPCR SYBR <sup>R</sup> Green Master Mix (YEASEN, #11201ES08) on the Hema9600 PCR machine. After 35 amplification cycles, reaction products were analyzed and β-actin RNA was used as a loading control. The primer sequences were as follows: survivin forward: CCGAC GTTGCCCCCTGC; survivin reverse: TCGATGGCACGGC GCAC; β-actin forward: AAATCGTGCGTGACATTAAGC; β-actin reverse: CCGATCCACACGGAGTACTT (15, 16).

#### Lentivirus Production and Infection

pDONR201-survivin was inserted into pCDH-Neo-Venus/DEST via LR clonase (Invitrogen, #11791) to form pCDH-Neo-Venussurvivin plasmid. At a 4:3:1 ratio, the lentiviral transfer vector, packaging plasmids psPAX2 and pMD.2G were transferred to the HEK293T cells to produced lentivirus. The PEI reagent was performed for transfection. Then, the viral supernatant was, respectively, collected 24, 48, 72, 96 h following transfection, filtered through a 0.20µm filter, and concentrated. Using the polybrene (Solarbio, H8761), A2780, OVCAR-3 cells were transfected with pCDH-Neo-Venus-survivin or pCDH-Neo-Venus/DEST, followed by incubation with 48 h for the subsequent trials (17, 18).

### Cell Viability Assay

It was assessed using a 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT) assay. Briefly, ovarian cancer cells were seeded in 96-well plates, 4,500 cells per well, supplementing with 100 µL of medium and exposed to PL at different concentrations for 72 h, followed by the addition of 10 µL of MTT solution (5 mg/mL) per well to the medium and incubation 4 h at 37◦C. After we discarded the medium, 50 µL of DMSO per well was used to elute the blue MTTformazan product, and absorbance of the solution was read at 570 nm (19, 20).

#### Nude Mice Xenograft Assay

Balb/c nude female mice with 4–5 weeks of age and 20–22 g weight were purchased from the Guangdong Medical Laboratory Animal Center. For in vivo experiments, 4 × 10<sup>6</sup> A2780 cells in 100 µL of medium were injected subcutaneously into the left and right shoulders of each mouse. After the subcutaneous tumors reached a size of 0.3 × 0.3 cm<sup>2</sup> , mice were randomized to treatment with PL (20 mg/kg) intraperitoneally or 0.5% methylcellulose alone every day. In the end, the tumors were got rid of mice and weighed after they were sacrificed. The bodyweight and the tumor volume (V) of mice were recorded every day. Formula: V = π/6 (1/2 (A + B))<sup>3</sup> , the rate of inhibition (IR) = 1 – Mean tumor weight of experiment group/Mean tumor weight of control group × 100% (21, 22).

#### Statistical Analysis

Statistical significance of differences was determined by student's t-test. Differences were defined as significant at P < 0.05.

# RESULTS

#### PL Downregulates the Expression of Survivin in Ovarian Cancer Cells

Our previous study firstly demonstrated that PL facilitated G2/M cell cycle arrest, impeded the proliferation of human ovarian cancer cells, and induced ROS dependent apoptosis in ovarian cancer (11). Due to the ability to inhibit the cell apoptosis, survivin may take part in various processes associated with the tumor development and progression, facilitated metastasis and angiogenesis, as well as favor cell cycle progression (9). Because

of the apoptotic effect of PL induced in ovarian cancer cells, we explored the negative regulation of PL on survivin expression. As shown in **Figures 2A,C**, A2780 and OVCAR-3 cells were exposed to increasing concentrations of PL (up to 20µM) for 6 h, and the survivin protein was detected by western blot assay. PL effectively depleted the survivin expression in both A2780 and OVCAR-3 cells at the low micromolar concentration (10 or 20µM) in contrast to the control cells. Furthermore, as demonstrated in **Figures 1B,D**, downregulation of survivin was significant as soon as 6 h following PL treatment. Together, these evidences manifested that PL induces the reduction of survivin ovarian cancer cells in vitro, which may contribute to its apoptotic effect.

# PL Does Not Alter mRNA Expression of Survivin

To deliberate the discipline of PL-mediated survivin reduction, our group detected whether PL regulates the survivin mRNA expression in ovarian cancer cells. A2780 and OVCAR-3 cells were exposed to increasing concentrations of PL for 6 h or treated for different time scales at the same concentration of PL (10 or 20µM), and the expression levels of survivin mRNA were examined by RT-PCR. In **Figures 2A–D**, the survivin mRNA expression levels were not altered both on concentration and time scales, which were not similar to the depletion of survivin protein levels. These findings strongly supported the notion

that the PL regulates survivin gene expression at the nontranscriptional level.

# PL Induces Proteasome-Dependently Degradation of Survivin

The aforementioned findings indicated that PL might induce depletion of survivin at the post-transcription level. The possible discipline expounding PL-induced decrease of survivin protein expression may associate with its rapidly increased ubiquitinproteasome degradation. In order to verify this supposition, our group preincubated A2780 and OVCAR-3 cells with proteasome inhibitors MG132 (1µM). The results presented in **Figures 3A,C** demonstrated that MG132 reversed PL-mediated survivin depletion compared with control cells, hinting that PL-mediated degradation of survivin occurs via ubiquitinproteasome mechanism.

Otherwise, the autophagy-lysosome system and the ubiquitinproteasome system are the essential intracellular proteolytic pathways in eukaryotes (23). To clarify whether the degradation of survivin protein was related to the induction of autophagy, CQ, a potent inhibitor of autophagy, was used to verify this hypothesis. However, in contrast to the result of the MG132, the expression levels of survivin stayed unchanged in response to autophagy inhibition **(Figures 3B,D)**, indicating that the cellular autophagic flux is not involved in the PL-mediated degradation of survivin expression. Collectively, these results showed that PL induces the depletion of survivin via the proteasomedependent pathway.

# PL Induces ROS-Dependently Apoptosis and Survivin Depletion

Lots of anticancer agents perform the antitumor property by the activation of intracellular ROS. Indeed, it has been elucidated that PL-induced apoptotic cell death in mammary tumors and sarcoma are closely related to the increased elevation of the intracellular ROS level (24). Of note, our previous study found that PL caused a prominent rise in ROS levels and induced significant apoptosis in ovarian cancer cells (11). Meanwhile, co-administration of PL and NAC fully restored the PL-mediated increase in ROS and inhibited the depletion of survivin **(Figures 4A,B)**. Taken together, these data offered additional support to our hypothesis that ROS generation is significant for PL-mediated ovarian cancer cells apoptosis and survivin degradation.

taking advantage of pCDH-Neo-Venus-survivin lentivirus, while pCDH-Neo-Venus/DEST was adopted as the control. The Western blotting analysis was used to confirm the result of infection. As shown in **Figures 5A,C**, A2780 and OVCAR-3 cells infected with pCDH-Neo-Venus-survivin exhibited a much higher survivin expression. To detect the role of survivin in the apoptosis effect of PL, stable cells overexpressing survivin were treated with the increasing concentrations of PL and assayed to examine the survival rate **(Figures 5B,D)**. In the presence of PL, the IC<sup>50</sup> value of OVCAR-3 cells was 4.59µM. However, in survivin overexpression cells, the IC<sup>50</sup> value was 9.65µM, which was 2.10-fold higher than the control group. Consistently, similar results were also demonstrated in A2780 cells. Collectively, these results suggested that survivin overexpression raises the survival rate of ovarian cancer cells.

# Overexpression of Survivin Raises the Survival Rate of Ovarian Cancer Cells to PL

To explore the role of survivin involved in PL-induced ovarian cancer apoptosis, we performed survivin infection

# PL Inhibits A2780 Xenograft Tumor Growth and Downregulates Survivin in vivo

To evaluate the antitumor effects of PL in vivo, a xenografts model of A2780 cells in BALB/c mice was conducted.

Intraperitoneal injection of PL at doses of 20 mg/kg for 15 days decreased A2780 tumors weight as well as the tumor volume vs. vehicle control **(Figures 6A,B,E)**. Notably, PL administration was endurable without severe weight loss **(Figures 6C,D)**. Additionally, immunoblotting analyses of the tumor tissues showed that survivin was significantly downregulated after PL exposure for 15 days **(Figure 6F)**, which was in accordance with the results in vitro. In short, these data indicated that PL suppresses the xenograft tumor growth in vivo accompanied by a decreased survivin level.

# DISCUSSION

Because of the advanced stage at the time of diagnosis (80%) and a high recurrence rate (70–80%), human ovarian cancer surpassing other gynecological cancers lies in the fifth-leading cause of cancer-related deaths. The first-line treatment for it is the tumor cytoreductive surgery combined with platinum-taxane chemotherapy (25, 26) or neoadjuvant chemotherapy followed by debulking surgery (27). Although ∼80% of patients will have a response to the frontline therapies, 25% of women within 6 months will experience resistant cancer recurrence, and the overall 5-year survival rate ranges between 30 and 40% on a global scale (1). Thus, the explore of applicable treatment drugs remains a crucial goal for carrying out a better outcome.

Owing to the abnormal expression of anti-apoptotic proteins resulting in antitumor treatments less effective or ineffective, the reactivation of cell apoptosis becomes a potential therapy to conquer antitumor drug resistance. Survivin, the smallest member of IAP proteins, participates in the inhibition of apoptosis, affects the proper process of mitosis and promotes angiogenesis or even DNA repair (28). Generally, abnormal expression of survivin is associated with decreased apoptosis, increased tumor recurrence, poor prognosis, and high chemoresistance in human ovarian cancer (29, 30). Downregulating the expression of survivin elevated the sensitivity of ovarian cancer to chemotherapy and promoted apoptosis (7, 31). Thus, elucidating the regulatory mechanisms of survivin protein expression in cancers may propose mechanistic clues to future therapeutic strategies (32, 33). In general, survivin

FIGURE 6 | PL inhibits A2780 xenograft tumor growth and downregulates survivin *in vivo*. Each mouse was injected subcutaneously with A2780 cells (4 × 10<sup>6</sup> in 100 µL of medium) under the left and right shoulders. When the subcutaneous tumors were ∼0.3 cm × 0.3 cm<sup>2</sup> in size, mice were randomized into two groups and received an intraperitoneal injection of the vehicle alone (0.5% methylcellulose) or PL (20 mg/kg) every day. The body weight and tumor volume were recorded. After the experiment, the mice were anesthetized, and tumor tissue was excised from the mice and weighed. The tumor volume (A), original tumors (B), body weight (C), tumor weight (D), summary data (E), and Western blot analysis of survivin (F) were shown. β-actin served as a loading control. The band quantification was performed using ImageJ software. \**P* < 0.05 vs. corresponding control.

combines with XIAP to indirectly suppress the function of caspases. XAF1 binds to XIAP resulting in the activation of ubiquitin-protein isopeptide ligase (E3) activity of XIAP, then promotes survivin ubiquitination and degradation (34). Conversely, the activation of ERK leads to the combination obstruction between survivin and the XIAP-XAF1 E3 ligase complex, which exactly stabilizes survivin proteins (35). The proapoptotic F-box protein FBXL7 acting as a ubiquitin E3 ligase interacts with Glu-126 to modulate the polyubiquitylation of survivin and thereby enhance its proteasomal degradation (36). Nonetheless, aurora kinase A (AURKA) restricts survivin to combine with FBXL7 E3 ubiquitin ligase through the negative regulation on FOXP1-FBXL7 axis, which upregulates the survivin expression (37). lncRNA LINC00473 (LNC473) remarkably suppresses the ubiquitination of survivin through recruiting deubiquitinase USP9X and then enhanced the stability of survivin (38). The Culling 9 (CUL9) is a putative E3 ligase and deplete the protein abundance of survivin via enhancing its ubiquitylation (39). Additionally, it has been reported that the ubiquitin-like-modifier proteins FAT10 (40), the heat shock protein (HSP90) (41) and CSN5/JAB1 (42) directly bind to survivin and therefore are resistant to ubiquitin-proteasome degradation. Therefore, the therapeutic regiment targeting the survivin axis is pushing the door of ovarian cancer. Such treatments include regiments that target transcription or posttranslational level, vaccines based on cytotoxic activities of immune-cells and gene therapy methods that suppress survivin's function (43).

Piperlongumine, a biologically active alkaloid/amide isolated from peppers, has selectively excellent ability to inhibit tumor cells via the induction of oxidative stress and genotoxicity, as well as the significant oral bioavailability in mice burden tumor without unbearable systemic toxicity (44). We have previously reported that exposure of ovarian cancer cells to PL selectively suppressed cell viability in a concentration- and time-dependent manner (11). We deeply take an insight into the underlying mechanism of PL-induced apoptosis in ovarian cancer and find that survivin plays a vital role in the regulation of PL-mediated apoptosis in this study. Here, we firstly elucidated that PL induced apoptosis as well as the rapid depletion of survivin in ovarian cancer in vitro and vivo. Furthermore, overexpression of survivin raised the survival rate of ovarian cancer cells, implying that PL induced apoptosis of ovarian cancer via inhibiting survivin expression. In addition, our data suggested that the levels of survivin mRNA stayed unchanged under PL administration, indicating that PL induced depletion of survivin at the nontranscriptional level. This notion may lead to a rational concern about the strategy modulating either ubiquitin-proteasome or autophagy because both of them are the two primary proteolytic

#### REFERENCES


mechanisms. Then, a blockage of proteasome activity reversed the reduction of survivin by PL, as compared with keeping stable under autophagy inhibition. Otherwise, the previous studies also observed that the elevated levels of intracellular ROS induced by PL involved in the process of apoptosis in ovarian cancer. Considering that ROS-mediated signaling in cancer cells has long been thought to participate in malignant transformation, carcinogenesis initiation, and the survival of cancer cells, it is reasonable that the ROS-dependent pathway may take part in the regulation of survivin (45). In agreement with this notion, we observed that the apoptosis and the depletion of survivin under PL were reversed by NAC.

As all the data we assembled suggest, it is reasonable to regard the ROS-mediated proteasome-dependent mechanism as the major system responsible for the depletion of survivin under PL treatment. PL is on the way to transform ovarian cancer patients' outcomes and looking for finding its place in ovarian cancer therapeutic strategies.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

#### ETHICS STATEMENT

The animal study was reviewed and approved by the ethics committee of Jinan University.

#### AUTHOR CONTRIBUTIONS

X-WN, L-HG, XC, ZS, and X-JY designed the experiments, performed the experiments, analyzed the data, and wrote the paper. H-HZ, P-PY, YY, Z-HX, M-NW, YL, S-TW, and KL conducted the experiments. All authors read and approved the final manuscript.

#### FUNDING

This study was supported by funds from the National Key Research and Development Program of China No. 2017YFA0505104 (ZS), the National Natural Science Foundation of China Nos. 81772540 (ZS), 81503293 (X-JY), the Science and Technology Foundation of Wenzhou City, Zhejiang Province, China No. 2015Y0304 (L-HG), the Technology Development Funds of Wenzhou City No. Y20190014 (X-JY), and the Traditional Chinese Medicine Science and Technology Foundation of Zhejiang Province, China No. 2020ZB144 (X-JY).

CA Cancer J Clin. (2018) 68:284–96. doi: 10.3322/caac. 21456

3. Herzog TJ, Monk BJ. Bringing new medicines to women with epithelial ovarian cancer: what is the unmet medical need? Gynecol Oncol Res Pract. (2017) 4:13. doi: 10.1186/s40661-017- 0050-0


of specificity protein transcription factors. Cancer Prev Res. (2017) 10:467–77. doi: 10.1158/1940-6207.CAPR-17-0053

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Nan, Gong, Chen, Zhou, Ye, Yang, Xing, Wei, Li, Wang, Liu, Shi and Yan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# MiR-204-3p Inhibited the Proliferation of Bladder Cancer Cells via Modulating Lactate Dehydrogenase-Mediated Glycolysis

Jinan Guo1,2, Pan Zhao2,3, Zengqin Liu2,3, Zaishang Li 1,2, Yeqing Yuan1,2, Xueqi Zhang1,2 , Zhou Yu1,2, Jiequn Fang1,2 \* and Kefeng Xiao1,2 \*

*<sup>1</sup> Department of Urology, Shenzhen Urology Minimally Invasive Engineering Center, The Second Clinical Medical College of Jinan University, Shenzhen People's Hospital, The First Affiliated Hospital of South University of Science and Technology of China, Shenzhen, China, <sup>2</sup> Shenzhen Public Service Platform on Tumor Precision Medicine and Molecular Diagnosis, Shenzhen Urology Minimally Invasive Engineering Center, The Second Clinical Medical College of Jinan University, Shenzhen People's Hospital, The First Affiliated Hospital of South University of Science and Technology of China, Shenzhen, China, <sup>3</sup> Clinical Medical Research Center, Shenzhen Urology Minimally Invasive Engineering Center, The Second Clinical Medical School of Jinan University, Shenzhen People's Hospital, The First Affiliated Hospital of South University of Science and Technology of China, Shenzhen, China*

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, China*

#### Reviewed by:

*Jianqing Ruan, Soochow University Medical College, China Yan Deng, The Chinese University of Hong Kong, China Jigang Wang, China Academy of Chinese Medical Sciences, China*

#### \*Correspondence:

*Jiequn Fang sz2011@sohu.com Kefeng Xiao xiao.kefeng@szhospital.com*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *16 August 2019* Accepted: *29 October 2019* Published: *29 November 2019*

#### Citation:

*Guo J, Zhao P, Liu Z, Li Z, Yuan Y, Zhang X, Yu Z, Fang J and Xiao K (2019) MiR-204-3p Inhibited the Proliferation of Bladder Cancer Cells via Modulating Lactate Dehydrogenase-Mediated Glycolysis. Front. Oncol. 9:1242. doi: 10.3389/fonc.2019.01242* MicroRNAs (miRNAs) are endogenous non-coding RNAs that negatively regulate the expression of downstream targeted mRNAs. Increasing evidence has suggested that miRNAs act as tumor suppressors or oncogenes to interfere the progression of cancers. Here, we showed that miR-204-3p was decreased in bladder cancer tissues and cell lines. Down-regulation of miR-204-3p was significantly associated with a poor prognosis in bladder cancer patients. Overexpression of miR-204-3p inhibited proliferation and induced apoptosis in bladder cancer cells. Furthermore, miR-204-3p was found to bind to the 3′ -untranslated region (UTR) of the lactate dehydrogenase (LDHA), which consequently reduced the expression of both mRNA and protein of LDHA. Interestingly, overexpression of miR-204-3p decreased glucose consumption and lactate production of bladder cancer cells. Overexpression of LDHA relieved the growth inhibition and cell apoptosis enhancement by miR-204-3p in bladder cancer cells. These results demonstrated that miR-204-3p negatively modulated the proliferation of bladder cancer cells via targeting LDHA-mediated glycolysis. MiR-204-3p might be a promising candidate for designing anticancer medication.

Keywords: bladder cancer, miR-204-3p, lactate dehydrogenase, glycolysis, proliferation

# INTRODUCTION

Bladder cancer (BC) is one of the most common malignancies in the urological system (1). There are approximately 356,000 new cases and 145,000 deaths annually around the world (1–3). The high occurrence and mortality of BC make it a big threat to the health of patients. Although remarkable progress has been made in the treatment of BC including surgical resection and chemotherapy, the 5-year overall survival of BC remains poor (4, 5). Therefore, it is urgent to identify novel biomarkers and targets of BC.

**163**

MicroRNAs (miRNAs) are a class of single-stranded noncoding RNAs with the length of 22–24 nucleotides (6, 7). MiRNAs negatively regulate the gene expression via binding with the 3 ′ -untranslated region (UTR) of targets, which leads to the degradation or translation inhibition of mRNAs (8). MiRNAs are reported to be involved in both physiological and pathological conditions including cell proliferation, differentiation, and apoptosis (9, 10). Recent research found that aberrant expression of miRNAs was associated with the tumorigenesis of BCs (9– 12). Therefore, detecting the abundance of miRNA might provide novel evidence for the diagnosis of BC and benefit the outcome of cancer patients. To get a whole picture about the expression of miRNAs in BC, we reviewed the previous publications relevant to the altered miRNAs in BC and performed a meta-analysis. The data showed that miR-204-3p was significantly down-regulated in the urine supernatant of BC patients (13). Interestingly, miR-204-3p was frequently found to be down-regulated in cancers, which acted as a tumor suppressor in tumorigenesis (14). Abnormal expression of miR-204-3p was observed in gastrointestinal stromal tumors, breast cancer, and hepatocellular carcinoma (HCC) (15, 16). However, the expression and function of miR-204-3p in BC remain largely unknown.

Aerobic glycolysis has been considered as a distinct hallmark of cancer cells, which represents that cancer cells addictively depend on glycolysis to metabolize glucose even in oxygen-rich condition (17–19). Therefore, understanding the mechanisms that contribute to the glycolysis process will provide novel cues for therapeutic strategies for cancer treatment. The lactate dehydrogenase (LDHA) is a critical enzyme of the glycolysis, which catalyzes the production of lactate (20). Up-regulation of LDHA has been found in a variety of human cancers, which is associated with cell growth, metastasis, and poor prognosis of cancer patients (21–24). Notably, miRNAs are reported to target LDHA and negatively regulate the expression of LDHA. For example, repression of LDHA by miR-34a inhibited the glycolysis and suppressed the growth of breast cancer cells (20). Recent study showed that miR-142-3p targeted LDHA, which reduced the glycolysis and growth of HCC cells (25). These reports suggest that the negative modulation of LDHA by miRNA is an important strategy to regulate the progression of cancers.

In this study, we investigated the function of miR-204-3p in BC. The data showed that miR-204-3p was down-regulated in BC tissues and cell lines. Overexpression of miR-204-3p decreased the proliferation of BC cells via modulating LDHAmediated glycolysis.

#### MATERIALS AND METHODS

#### Meta-Analysis

The studies that reported the alternation of miRNAs in BCs was searched using the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) and EBI ArrayExpress databases with the keywords "human" and "bladder" and "miRNA." For a meta-analysis, priority was given to the peer-reviewed publications that investigated the expression of miRNAs in both the BC and corresponding normal controls. For the studies without data of control group, only cell-culture-based results were excluded. Data extraction from the selected publications contained the first author's name, publication time, sample size, patient's age, type of case, origin of the studied population, tumor stage, detection method, and cutoff values for down-regulation or up-regulation. The quality of the publications was evaluated by the Newcastle-Ottawa scale, which was generally used for assessing the quality of nonrandomized studies in meta-analyses. Each article was subjected to assessment with eight methodology items with the score ranging from 0 to 9. The higher score indicated better quality of the publication. The articles with score of 7 or more were recruited for the meta-analysis.

#### Clinical Samples and Cell Lines

A total of 60 paired BC and adjacent normal tissues were obtained from the BC patients who underwent surgery in the Shenzhen People's Hospital. The tissues were confirmed by three pathologists independently. None of the patients received radiotherapy or chemotherapy before the surgical resection. All the tissues were maintained in liquid nitrogen until use. All the experimental procedures were approved by the Research Ethics Committee of Shenzhen People's Hospital, and each patient signed the written informed consent.

The normal human urothelial SV-HUC-1 cells were purchased from the Institute of Cell Research, Chinese Academy of Sciences (Shanghai, China). BC cell lines including SW780, J82, UMUC3, 5637, and T24 were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA). Cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 medium (Gibco, Thermo Fisher Scientific, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO, USA), 100 U/ml of penicillin and 100 mg/ml of streptomycin (Invitrogen, Carlsbad, CA, USA) at 37◦C with 5% CO2.

#### Reverse Transcriptase Quantitative PCR Analysis

MiRNAs were extracted from the cell lines or bladder tissues using the miRNeasy Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instruction; 0.5 µg of RNA was reversely transcripted into cDNA with the PrimeScript RT Master Mix (Takara, Dalian, China). Real-time PCR analysis was performed with the SYBR Select Master Mix (Takara, Dalian, China) on the ABI7500 platform (Applied Biosystems, Foster City, CA, USA). The expression of U6 RNA and GAPDH was detected as the respective endogenous control. The expression level of miR-204-3p and LDHA was normalized to U6 and GAPDH and calculated using the 2−11CT method. The primers were as follows: miR-204-3p, F, 5′ -ACACTCCAGCTG GGGCTGGGAAGGCAAAGGG-3′ and R, 5′ -CTCAACTGG TGTCGTGGA-3′ ; LDHA, F, 5′ -AGCCCGATTCCGTTACCT-3′ and R, 5′ -CACCAGCAACATTCATTCCA-3′ ; U6, F, 5′ -CTC GCTTCGGCAGCACA-3′ and R, 5′ -AACGCTTCACGAATT TGCGT-3′ ; and GAPDH, F, 5′ -TGACGCTGGGGCTGGCAT TG-3′ and R, 5′ -GCTCTTGCTGGGGCTGGTGG-3′ .

#### Cell Viability

The proliferation of BC cells with the transfection of control miRNA or miR-204-3p mimics was evaluated by the Cell Counting Kit-8 assay (CCK-8, Beyotime, Shanghai, China). Briefly, 10 µl of CCK-8 reagent was added into the medium at the indicated time points of 1, 2, 3, 4, and 5 days and incubated for 4 h at 37◦C. The absorbance value (optical density [OD]) of each well at 450 nm was detected with a microplate reader (Bio-Rad, CA, USA). The results were obtained from three independent experiments.

# Cell Colony Formation

The BC cells expressing the indicated miRNA were seeded in a 6-well plate with the density of 1,000 cells per well. Cells were cultured with the RPMI-1640 medium for 2 weeks. And then the medium was discarded, and the cells were washed twice with phosphate-buffered saline (PBS). Cells were fixed with methanol for 15 min at room temperature (RT). After being washed with PBS, the colonies were stained with 1% crystal violet, and the number of colonies was counted with light microscopy.

#### Western Blot

Protein was extracted from the BC cells with the NP-40 lysis buffer (150 mM of NaCl, 1% NP-40, 50 mM of Tris–HCl (pH 8.0), and 1 mM of EDTA) in the presence of proteinase inhibitor (Millipore, Braunschweig, Germany). The protein concentration was evaluated using the bicinchoninic acid (BCA) assay kit (Beyotime, Shanghai, China); 15 µg of proteins was loaded and separated by the 15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto the polyvinylidene difluoride (PVDF) membrane (Millipore, Braunschweig, Germany). The membrane was firstly blocked with 5% non-fat milk at RT for 1 h, followed by incubation with primary antibodies at RT for 2 h and by incubation of the membrane with goat anti-mouse IgG secondary antibody (1:3,000, Bio-Rad, CA, USA), and then analyzed using the chemiluminescent horseradish peroxidase (HRP) conjugated substrate (Millipore, Braunschweig, Germany). GAPDH was used as the loading control. Antibodies including anti-LDHA (#2012, 1:3,000, Cell Signaling Technology, Inc., Danvers, MA, USA) and anti-GAPDH (#5174, 1:3,000, Cell Signaling Technology, Inc., Danvers, MA, USA) were commercially obtained.

#### Luciferase Reporter Assay

The wild-type (WT) or mutant (MUT) 3′ -UTR of LDHA containing the putative binding sites of miR-204-3p was inserted into the pmiR-GLO vector. BC cells were transfected with the control miRNA or miR-204-3p mimics in the presence of luciferase reporter vector. After transfection for 48 h, cells were harvested, and the luciferase activity was determined with the dual-luciferase reporter system (Promega, Madison, WI, USA). The experiment was performed in triplicate.

### Measurement of the Glucose Consumption and Lactate Production

BC cells transfected with the corresponding miRNAs were cultured in RPMI-1640 medium without phenol red (Gibco, New York, NY, USA) for 24 h, and then the medium was collected. The glucose consumption and lactate production were determined using the Glucose Assay Kit (GAHK20-1KT; Sigma-Aldrich, USA) and Lactate Assay Kit (BioVision, CA, USA) according to manufacturers' instructions, respectively. The protein level in each group was measured using the BCA assay kit (Beyotime, Shanghai, China) for the normalization.

#### Statistical Analysis

Results were presented as mean ± standard deviation from three independent experiments. The data were analyzed with the SPSS 19.0 software (SPSS Inc., Chicago, IL, USA). Significant differences between/among treatment groups were analyzed using unpaired Student's t-test or one-way ANOVA followed by Dunnett's post hoc test. P < 0.05 was considered to be statistically significant.

# RESULTS

# MiR-204-3p Was Down-Regulated in Bladder Cancer Tissues and Cell Lines

To obtain a whole picture of miRNA expression in BC, a miRNA meta-analysis was performed using previous publications. The data found that miR-204-3p was significantly aberrantly expressed in the urine supernatant of the BC patients (**Figure 1A**). To confirm this result, the expression of miR-204- 3p was detected by reverse transcriptase quantitative PCR (RTqPCR) with paired BC tissues and corresponding normal tissues. The result showed that the level of miR-204-3p was significantly decreased in BC tissues in comparison with that of the normal counterparts (**Figure 1B**). To further confirm the aberrant expression of miR-204-3p in BC, the abundance of miR-204-3p in several BC cell lines was examined. As indicated in **Figure 1C**, miR-204-3p was remarkably decreased in BC cells, compared with the normal cells, including SW780, J82, UMUC3, 5637, and T24. These results suggested the down-regulation of miR-204-3p in BC. To further characterize the association between miR-204-3p expression and the clinicopathological features, the abundance of miR-204-3p in BC patients with or without lymph node metastasis was compared. The data revealed that lower expression of miR-204-3p was correlated with positive lymph node metastasis (**Figure 1D**). Consistently, decreased expression of miR-204-3p was also observed in BC patients with larger tumor size (**Figure 1E**). These results demonstrated that miR-204-3p was down-regulated in BC and associated with a poor prognosis of cancer patients.

#### Overexpression of MiR-204-3p Inhibited the Proliferation and Induced Apoptosis of Bladder Cancer Cells

As miR-204-3p was decreased in BC, to investigate the effect of aberrant expression of miR-204-3p on the growth

miR-204-3p in 60 paired bladder cancer tissues and the corresponding adjacent normal tissues was determined by RT-qPCR. (C) The expression of miR-204-3p in bladder cancer cells and normal bladder epithelial cell SV-HUC-1 was determined by the RT-qPCR analysis. (D) The expression of miR-204-3p in the bladder cancer patients with or without lymph node metastasis was determined by RT-qPCR. (E) The expression of miR-204-3p in bladder cancer tissues stratified by tumor size was determined by RT-qPCR. \*\**P* < 0.01 and \*\*\**P* < 0.001. RT-qPCR, reverse transcriptase quantitative PCR.

of BC cells, SW780 and 5637, which harbored the lowest expression of miR-204-3p among all the cell lines we tested (**Figure 1C**), were transfected with miR-204-3p mimics to upregulate the expression of miR-204-3p. The overexpression of miR-204-3p in both SW780 and 5637 cells was detected by RT-qPCR (**Figure 2A**). The influence of miR-204-3p on the proliferation of BC cells was determined by CCK-8 assay. As showed in **Figures 2B,C**, the growth of both SW780 and 5637 cells was significantly inhibited with overexpressed miR-204-3p. Consistently, the colony formation assay indicated that overexpression of miR-204-3p dramatically decreased the number of colonies compared with those of the control cells (**Figure 2D**). In addition, to further confirm the negative regulation of overexpressed miR-204-3p on the growth of BC cells, the apoptosis rate of both SW780 and 5637 cells was examined with the fluorescence-activated cell sorting (FACS) analysis. The result suggested that ectopic expression of miR-204-3p significantly enhanced the apoptosis of BC cells (**Figure 2E**). These data demonstrated that overexpression of miR-204-3p inhibited the growth of BC cells, which

suggested the potential tumor suppressive function of miR-204- 3p in BC.

#### Lactate Dehydrogenase Was a Target of MiR-204-3p in Bladder Cancer Cells

To further explore the underlying molecular mechanism by which miR-204-3p modulated the growth of BC cells, the targets of miR-204-3p were predicted using the miRDB database (http://mirdb.org). Among the candidates, LDHA was found as a possible target of miR-204-3p (**Figure 3A**). To confirm the potential binding between miR-204-3p with the LDHA 3′ -UTR (position 515-521), the WT or MUT 3′ -UTR of LDHA containing the putative binding sites of miR-204-3p was inserted into the pmiR-GLO vector. Luciferase reporter assay was performed by co-transfecting negative control miRNA or miR-204-3p mimics with WT or MUT 3′ -UTR of LDHA. The results showed that overexpression of miR-204-3p significantly reduced the luciferase activity of the WT 3′ -UTR of LDHA; however, no remarkable decrease was observed when cells were transfected with MUT 3 ′ -UTR of LDHA (**Figures 3B,C**). This result indicated that miR-204-3p bound the 3′ -UTR of LDHA.

To detect whether the binding of miR-204-3p at the 3′ -UTR affected the mRNA stability of LDHA, both SW780 and 5637 cells were transfected with control miRNA or miR-204-3p mimics, and the mRNA level of LDHA was examined with the RT-qPCR assay. As shown in **Figure 3D**, overexpression of miR-204-3p significantly reduced the mRNA level of LDHA in BC cells. Additionally, the protein level of LDHA in SW780 and 5637 cells expressing miR-204-3p was also investigated by Western blot using the anti-LDHA antibody. The data showed that the level of LDHA was down-regulated with the transfection of miR-204-3p in SW780 and 5637 cells (**Figure 3E**). These results demonstrated that LDHA was a target of miR-204-3p and negatively regulated by miR-204-3p in BC cells.

#### MiR-204-3p Suppressed the Glycolysis of Bladder Cancer Cells by Modulating Lactate Dehydrogenase

LDHA was critical for the glucose metabolism of cancer cells. Considering the negative regulation of miR-204-3p on the expression of LDHA, the influence of miR-204-3p on the glycolysis of BC cells was determined by measuring the glucose consumption and lactate production. The results showed that overexpression of miR-204-3p significantly decreased the glucose uptake of SW780 and 5637 cells (**Figure 4A**). Consistently, the generation of lactate of BC cells was also dramatically reduced

\*\**P* < 0.01 and \*\*\**P* < 0.001. LDHA, lactate dehydrogenase; UTR, untranslated region; RT-qPCR, reverse transcriptase quantitative PCR.

in the presence of high level of miR-204-3p (**Figure 4B**). To further confirm these results, miR-204-3p was knocked down in SW780 and 5637 cells by transfecting miR-204-3p antagomir. The reduction of miR-204-3p was verified by RTqPCR assay (**Figure 4C**). The glycolysis of BC cells harboring depleted miR-204-3p was explored. As shown in **Figure 4D**, down-regulation of miR-204-3p remarkably elevated the glucose consumption of both SW780 and 5637 cells. Increased lactate production was also obtained with the decreased expression of miR-204-3p (**Figure 4E**). These results indicated that miR-204-3p was a negative regulator of the glycolysis of BC cells.

To further investigate whether the inhibitory effect of miR-204-3p on the proliferation of BC cells was achieved via LDHA, CCK-8 assay was performed by transfecting LDHA in miR-204-3p overexpressed cells. The results showed that highly expressed miR-204-3p suppressed the proliferation and growth and induced cell apoptosis of both SW780 and 5637 cells, while rescue the expression of LDHA attenuated the inhibitory effect of miR-204-3p on the proliferation growth of BC cells and the enhanced effect of miR-204-3p on the cell apoptosis of BC cells (**Figures 4F–I**). These data demonstrated that miR-204-3p decreased the expression of LDHA, inhibited the glycolysis, and suppressed the growth of BC cells.

*(Continued)*

FIGURE 4 | miR-204-3p antagomir transfection was determined by the RT-qPCR. (D,E) The effects of miR-204-3p knockdown on the glucose uptake and lactate generation ofthe SW780 and 5637 cells were determined by Glucose Assay Kit and Lactate Assay Kit, respectively. (F,G) The cell viability of SW780 and 5637 with co-transfection with miRNAs and plasmid vectors was determined by CCK-8 assay. (H) The cell growth of SW780 and 5637 with co-transfection with miRNAs and plasmid vectors was determined by colony formation assay. (I) The cell apoptosis of SW780 and 5637 cells with co-transfection with miRNA and plasmid vectors was evaluated by flow cytometry. *N* = 3; \*\**P* < 0.01 and \*\*\**P* < 0.001. RT-qPCR, reverse transcriptase quantitative PCR; CCK-8, Cell Counting Kit-8.

#### DISCUSSION

Although remarkable progression has been made in the treatment of BC with the application of surgery and chemotherapy, the prognosis of BC is still poor. Moreover, most patients are diagnosed at the advanced stage and showed unsatisfactory response to treatments. Notably, increasing evidence has suggested the critical roles of miRNAs in the initiation and progression of cancers (9–11). Here, we showed that miR-204-3p was decreased in both BC tissues and cell lines, which was associated with the poor prognosis of BC patients. Further gain-of-function analysis demonstrated that overexpression of miR-204-3p suppressed the growth of BC cells, which suggested the potential suppressive function of miR-204-3p in modulating the progression of BC.

The involvement of miR-204-3p in cancer has been highlighted by the studies that aberrant expression of miR-204-3p can be used to distinguish the malignant progression in gastric cancer (16) and breast cancer (26). Significant decrease of miR-204-3p was observed in HCC, which suppressed the growth of HCC via targeting fibronectin 1 (15). Recent study by Chen et al. showed that xanthohumol, a prenylated chalcone potential for anticancer therapy, up-regulated the expression of miR-204-3p in glioma cells and induced the cell apoptosis through modulating the insulin-like growth factor (IGF)-binding protein (27). In clear cell renal cell carcinoma (ccRCC), miR-204-3p was found to be down-regulated by the ERβ-suppressed circular RNA ATP2B1, which increased the expression of fibronectin 1 and promoted the invasion of ccRCC cells (28). These findings suggested the potential tumor suppressive function of miR-204-3p in cancers. In the present study, our data revealed that miR-204-3p was decreased in BC tissues compared with the normal tissue. Down-regulation of miR-204-3p was correlated with lymph node metastasis and larger tumor size. Further investigation might be of interest to explore the relationship between the expression of miR-204-3p with the 5-year overall survival of BC patients. Consistent with the decreased level of miR-204-3p in BC, overexpression of miR-204-3p significantly suppressed the proliferation and induced apoptosis of BC cells. In vivo tumorigenesis assay might be necessary to further confirm the modulating of miR-204-3p on the growth of BC cells in the future study.

The function of miRNAs was achieved via regulating the expression of downstream targets. In this study, the possible targets of miR-204-3p were predicted with the bioinformatics, and LDHA was identified as one of the binding partners of miR-204-3p. Overexpression of miR-204-3p significantly decreased both the mRNA and protein levels of LDHA in BC cells. Consistently, overexpression of miR-204-3p reduced the glucose consumption and lactate production of BC cells. LDHA deficiency results in a decrease in lactate production accompanied by inhibition of cell migration and invasion. Therefore, LDHA has been the target of different miRNAs in cancer cells to regulate the tumorigenesis. Among them, miR-142-3p was reported to target LDHA and inhibited the aerobic glycolysis of HCC (25). In breast cancer, miR-30-5p inhibited the cell growth and metastasis through suppression of LDHAmediated glucose metabolism (29). These results collectively demonstrated that reprograming the glycolysis via targeting LDHA has been a promising way to modulate the progression of cancers.

Our study still presented several limitations. Firstly, the relationship between miR-204-3p and the overall survival of the BC patients has not been determined in the study, which needs further examination. Secondly, the present study lacks the in vivo animal studies, and the effects of miR-203-3p on the in vivo BC growth should be considered in future studies. Thirdly, the targets of miR-204-3p were not limited to LDHA, and other potential targets may be explored in future studies.

In conclusions, our results demonstrated that miR-204-3p was down-regulated in BC and correlated with the progression of the BC patients. Overexpression of miR-204-3p decreased the growth of BC cells via down-regulating LDHA, which consequently suppressed the glucose metabolism of BC cells. These data indicated that targeting the miR-204-3p-LDHA pathway might interfere with the tumorigenesis of cancer cells.

#### DATA AVAILABILITY STATEMENT

All datasets for this study are included in the article/supplementary material.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the Institutional Review Board of Shenzhen people's Hospital. The patients/participants provided their written informed consent to participate in this study.

#### AUTHOR CONTRIBUTIONS

JG and KX designed the whole study. JG, PZ, ZLiu, and ZLi performed the experiments. YY, XZ, and ZY performed the statistical analysis. JF and KX drafted the manuscript.

#### FUNDING

This study was supported by the grants from Sanming Project of Medicine in Shenzhen (SZSM201412014), the Science and Technology Foundation of Shenzhen (JCYJ20170307095620828 and JCYJ20160422145718224), and the Shenzhen Urology Minimally Invasive Engineering Center (GCZX2015043016165448). All sources of funding received for the research being submitted.

#### REFERENCES


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Guo, Zhao, Liu, Li, Yuan, Zhang, Yu, Fang and Xiao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Binding of PD-L1 and Akt Facilitates Glioma Cell Invasion Upon Starvation via Akt/Autophagy/F-Actin Signaling

Ruo Qiao Chen1†, Xiao Hong Xu2†, Feng Liu<sup>2</sup> , Chun Yang Li <sup>2</sup> , Yuan Jun Li <sup>2</sup> , Xiang Rui Li <sup>1</sup> , Guo Yong Jiang<sup>1</sup> , Feng Hu<sup>3</sup> , Di Liu<sup>4</sup> , Feng Pan<sup>4</sup> \*, Xin Yao Qiu<sup>2</sup> \* and Xiao Qian Chen<sup>2</sup> \*

<sup>1</sup> School of Basic Medicine, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China, <sup>2</sup> Department of Pathophysiology, School of Basic Medicine, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China, <sup>3</sup> Department of Neurosurgery, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China, <sup>4</sup> Department of Urology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China

#### Edited by:

Pascale Cohen, Université Claude Bernard Lyon 1, France

#### Reviewed by:

Anupam Mitra, University of California, Davis, United States Jiajun Fan, Fudan University, China

#### \*Correspondence:

Feng Pan panfeng@hust.edu.cn Xin Yao Qiu qiu.xinyao1991@outlook.com Xiao Qian Chen chenxq@mails.tjmu.edu.cn

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 06 August 2019 Accepted: 15 November 2019 Published: 03 December 2019

#### Citation:

Chen RQ, Xu XH, Liu F, Li CY, Li YJ, Li XR, Jiang GY, Hu F, Liu D, Pan F, Qiu XY and Chen XQ (2019) The Binding of PD-L1 and Akt Facilitates Glioma Cell Invasion Upon Starvation via Akt/Autophagy/F-Actin Signaling. Front. Oncol. 9:1347. doi: 10.3389/fonc.2019.01347 Glioma, especially glioblastoma, is pathologically characterized by high aggressiveness, which largely contributed to the ineffectiveness of current therapies. It has been recently reported that intrinsic PD-L1 can regulate tumor malignancy, whereas underlying mechanisms remain mostly unclear. Here, we report a novel mechanism by which PD-L1 promotes glioma cell infiltration. In orthotopic glioma models, PD-L1 expression was up-regulated predominantly in glioma cells in the infiltrating front. For PD-L1-overexpressed glioma cells, PI3K/Akt and actin regulations were among the top six most altered signaling pathways as detected by RNA-sequencing. PD-L1 significantly activated Akt/F-actin signaling while suppressed autophagic signaling upon cell starvation. Mechanistically, PD-L1 preferentially bound to Akt among various PI3K/Akt signaling proteins. Serial truncation identified the interaction between the 128-237aa fragment of PD-L1 and the 112-480aa fragment of Akt, which facilitates the membrane translocation/activation of Akt, and was unaffected by Perifosin (specific p-Akt inhibitor targeting Akt PH-domain). Taken together, our data indicate that in glioma cells, PD-L1 is induced to prevent autophagic cytoskeleton collapse via Akt binding/activation, facilitating glioma cell invasion upon starvation stress.

#### Keywords: glioblastoma multiforme, CD274, p62, autophagic influx, ischemia

# INTRODUCTION

Glioma, the most common primary intracranial tumor, originates from various types of cells. The most malignant glioma is glioblastoma, or glioblastoma multiforme (GBM), which accounts for around 21% of glioma and can be primary or secondary from grade II-III glioma. Recently, glioma is also classified into three types based on its molecular markers: 1p/19q deletion, IDH mutation and TERT promoter mutation. Among them, grade IV glioma mainly has the TERT promoter mutation accompanied by EGFR/PTEN mutation. Current treatments for glioma remain conventional surgery, radiotherapy and chemotherapy. However, glioma is extremely easy to relapse after treatment, with an average survival time of only 14 months for GBM patients after treatment (1). It is urgent to find new and effective therapies in order to prolong the survival time of GBM patients.

**172**

The pathological features of GBM are necrosis, high invasiveness and microvascular hyperplasia (2), which are closely related to abnormal energy metabolism. It is generally considered that solid malignant tumors including glioma suffer ischemia/hypoxia throughout their growth. For example, it is estimated that glioma tissues have wide but heterogeneous hypoxia with oxygen concentrations ranging from 0.1 to 2.5% (3). Clinical data reveal that severer GBM necrosis closely correlates to faster GBM progress and shorter patient survival time (1, 2). Upon energy stress such as ischemia, autophagy is a major mechanism to maintain cellular energy homeostasis, which is highly regulated by conserved autophagic influx signaling involving PI3K/Akt/mTOR, p62, Beclin-1 and LC3 (4–6). Higher activity levels of autophagy are widely detected in glioma tissues, particularly around necrotic tissues (7). It is reported that inducing autophagy can either improve the efficacy of chemotherapy (8) or induce chemotherapy-resistance in glioma (9). Large-scale gene knockout screening shows that mitochondrial energy-metabolism genes are necessary for anoxic adaptive growth of a variety of tumors, including GBM (10). Specifically, Kim et al. (11) reported in recent research an increase of mitochondrial SHMT2 (serine hydroxymethyltransferase) in ischemic necrosis tissues of GBM patients, this enhances the anaerobic energy metabolism of GBM cells and promotes survival of GBM cells under ischemic environment. The mechanisms by which ischemia stimulates GBM development are extremely complicated and far from clear.

Programmed cell death ligand-1 (PD-L1, i.e., CD274 or B7H1) is a key negative regulator for immune inhibitory axis signaling controlling T-lymphocyte infiltration inside solid tumors. Previous studies reported that PD-L1 is widely expressed in glioma cell lines (12, 13) and most human glioma specimens (14). However, data of PD-L1 levels and subcellular distributions in human glioma tissues vary greatly (15). Most studies reported that higher PD-L1 expression was correlated with higher glioma grades (16–18) and worse prognosis (17, 19–22), while opposite results were also reported (14, 17, 23). PTEN mutation/deletion (in 36% of glioma) is closely associated with higher PD-L1 expression in glioma (24). Present evidence suggests that PI3K/Akt is a major pathway controlling PD-L1 expression in cancer cells. For mTOR, a key downstream signaling molecule in PI3K/Akt pathway, mTOR complex 1 (mTORC1) mainly mediates PI3K/Akt-induced cell autophagy (25) and mTORC2 mediates Akt-induced cell survival (26).

PD-L1 is recently considered to be oncogenic. PD-L1 knockdown significantly decreases tumor volume in murine ovarian cancer, melanoma (27), murine medulloblastoma (28), and U87 glioma in nude mice (29), while PD-L1 overexpression promotes glioma development (29). Mechanistically, PD-L1 regulates cell growth, proliferation, apoptosis, autophagy, migration and invasion in various cancers via modulating PI3K/Akt/mTOR and Ras/Erk/EMT signalings (27, 29–31). Further exploration of biological roles and mechanisms of PD-L1 in glioma will provide therapeutic cues including targeted immunotherapies for glioma.

In the present study, we first reported that PD-L1 was mostly prominent in those highly aggressive infiltrating glioma cells in vivo. Our data revealed that PD-L1 promoted glioma cell infiltration via starvation-induced Akt/autophagy/Factin signaling. Particularly, we dissected PD-L1-Akt binding fragments and elucidated how PD-L1/Akt interactions activated its downstream cascades.

# MATERIALS AND METHODS

#### Drugs and Antibodies

Cloroquine (CQ, Selleck Chemicals, Shanghai, China) was used at a final concentration of 25µM. LY294002 (PI3K inhibitor, Cell Signaling Technology, MA, USA) was used at a final concentration of 0–50µM. Perifosine (inhibitor of Akt PHdomain, Selleck Chemicals, Shanghai, China) was used at a final concentration of 0–25µM. Primary antibodies against phospho-Akt (Ser473, #D9E), total Akt (#9272), phospho-mTOR, total mTOR,pP62, phospho-P70S6K, LC3B were purchased from Cell Signaling Technology (MA, USA). Antibodies against PD-L1/CD274 (PA5-28115, Thermo Fisher, IL, USA), PD-L1 (ABM4E54, Abcam, Cambridge, UK), GST (Z-5, Santa Cruz Biotechnology, TX, USA), Beclin1 (Santa Cruz Biotechnology, TX, USA), N-terminal GFP (Sigma), β-actin (20536–1-AP, Proteintech, Wuhan, China), Na+/K<sup>+</sup> ATPase α1 (Proteintech, Wuhan, China), β-tubulin (Proteintech, Wuhan, China) were commercially purchased.

# Plasmids and Transfection

GFP-LC3 plasmid was a gift from Dr. He Li (Huazhong University of Science and Technology). GSTkRas/PTEN/Akt1/Akt2/Akt3/EGFR VIII/PKA plasmids were gifts from Dr. Haian Fu (Emory University). PD-L1-EGFP expressing-plasmid was purchased from GeneChem (Shanghai, China). Full-length and truncated human PD-L1 coding cDNA was PCR amplified from PD-L1-EGFP plasmids and cloned into NV (N-terminal Venus 1-157aa) vector including NV-PD-L1 FL (full length 1-290 aa), NV-PD-L1-T1-259 (truncate 1-259aa), T128-259, T128-239, T1-178. PD-L1 cloned into pDEST-26 or p-FU-Venus vector was used to express GST-PD-L1 and Venus-PD-L1 FL1-290 or T19-290. Full-length and truncated human Akt1 cDNA was PCR amplified from GST-Akt1 plasmid and cloned into p-FU-Venus vector for Venus-Akt1, Akt1-T1-111, and Akt1-T122-480. Akt1-Y176A and Akt1-K14R mutants were constructed by using overlapping-PCR from p-FU-Venus-Akt1 plasmid. Transfection of plasmids was performed by using Lipofectamine 2000 (11668, Life Technologies, CA, USA) according to the manufacturer's instructions.

# RNA-Sequencing and Enrichment Analysis

Three 100-mm dishes of U251/PD-L1 and U251/Vec stable cell lines were subjected to total mRNA isolation, cDNA libraries construction and sequencing at Illumina HiSeq sequence platform (PE150) with 6G clean data by Novogene Bioinformatics Institute (Shanghai, China). Differential gene expression analysis between U251/PD-L1 and U251/Vec groups was performed using the DESeq2 R package (1.10.1). The resulting P-values were adjusted using the Benjamini and Hochberg's approach for controlling the false discovery rate. Differently expressed genes (DEGs) with P < 0.005, adjusted P < 0.05 (Padj) and absolute changing fold ≥1.2 were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) enrichment analysis. KEGG pathways and GO terms with Padj < 0.05 were considered significantly enriched by DEGs. We have submitted our data to NCBI (https:// www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE107581).

#### Orthotopic Mouse Glioma Model

All animal handling and experiments were performed in accordance with NIH guidelines and approved by the Ethics Committees of Huazhong University of Science and Technology. The mice were group housed in the Animal Core Facility of Tongji Medical College under a 12 h light/dark cycle with ad libitum access to food and water. Briefly, adult Kunming male mice (18–20 g) were anesthetized with chloral hydrate (350 mg/kg) and a burr hole was drilled in the skull 0.5 mm posterior to the bregma and 2.0 mm lateral to the midline. A 10-µl Hamilton syringe (26 gauge, Reno, NV) containing 20,000 G422 cells (mouse GBM cell line) in 1 µl of PBS was advanced to a depth of 3.5 mm from the skull surface and then withdrawal 0.3 mm. Cell suspension was delivered at the rate of 1 µl/min. After cell implantation, the needle was left in place for 6 min before withdrawal. After 6–14 d of cell inoculation, the mice were perfused with 4% paraformaldehyde (PFA) and the brains were paraffin-embedded.

#### Hematoxylin-Eosin (HE) Staining and Immunohistochemistry (IHC)

IHC was performed as previously reported (29). Paraffinembedded mouse brain tissues (bearing tumor) were cut into 4 µm-thick slices for H&E staining and IHC analysis. Briefly, the slices were deparaffinized in xylene and antigenretrieved by microwave processing. After 1 h of blocking with 5% bovine serum albumin in PBS, the slices were incubated with primary antibodies (PD-L1, Abcam, UK) overnight at 4 ◦C, followed by corresponding secondary antibody incubation (Polink-1 HRP DAB Detection System, ZSGB-BIO, China). The immunoreaction was visualized with diaminobenzidine tetrachloride. The brain images were scanned with an automatic slice scanning system-SV120 (Olympus, Tokyo, Japan). The tumor parenchyma rim was delineated with black dashed ellipse circle, while the infiltration frontiers was delineated with blue or white dashed irregular circle.

#### Cell Culture and Starvation

Human glioblastoma cell lines U251, LN229, and human embryonic kidney 293T cell line were purchased from American Tissue Culture Collection (MA, USA) or China Center for Type Culture Collection (Wuhan, China). U251, LN228, U87MG with stable PD-L1 overexpression (U251/vec or PD-L1, LN229/vec or PD-L1) or knockout (U251/sgGFP or sgPD-L1) were generated as previously described (29). All cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco, CA, USA) supplemented with 10% FBS (Gemini, CA, USA) and 1% Penicillin-Streptomycin Solution (Hyclone, Thermo, Beijing, China). Fresh Earle's balanced salt solution (EBSS, GIBCO BRL, USA) media was used to induce cell starvation at 24 h after transient transfection or initial seeding. Cells were washed with EBSS media for three times and then incubated with EBSS media for various time points.

#### Western Blotting Analysis

Western blotting analysis was performed as previously reported (29). Briefly, the cell lysates were collected and dispersed in radio-immunoprecipitation assay lysis buffer containing phenylmethane-sulfonyl fluoride. Equal amounts of total proteins were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and electro-transferred onto nitrocellulose filter membranes (Merck Mil- lipore, Cork, Ireland). The blots were incubated with corresponding primary and secondary IRDye 800 or IRDye 680 CW-conjugated goat anti-rabbit or anti-mouse IgG antibodies (LI-COR Biosciences, Lincoln, USA). The labeled bands were visualized and quantified by Odyssey Infrared Imaging System (LI-COR Biosciences, MA, USA).

# Palloidin Staining and Immunofluorescence

Immunofluorescence staining was performed as previously reported (29). Cells in 35-mm culture dishes were fixed, permeabilized, blocked, and then incubated with primary and corresponding Dylight 488-labeled secondary antibodies (Abbkine, CA, USA). F-actin was stained with rhodaminephalloidin (Yisheng Bioengineering Institute, China) according to the manufacturer's instructions. Hoechst 33342 was used to stain the nucleus. For paraffin-embedded tissues, rat brain slices were deparaffinized, rehydrated, antigen unmasked, blocked with 5% bovine serum albumin (BSA) and then incubated with primary antibodies and corresponding Dylight 488/594 labeled secondary antibodies. Micrographs were taken under the same conditions with a conventional fluorescent microscope (Olympus, Tokyo, Japan).

#### GFP-LC3 Punctate Quantification

U251/sgGFP or U251/sgPD-L1 cells were transiently transfected with pGFP-LC3 plasmids for 24 h. Then, the cultures were subjected to EBSS treatment for 12 h and fixed. U251 cells expressing GFP-LC3 were randomly photographed under 400× magnifications under the same conditions. Cells with five or more GFP-LC3 vacuole dots (puncta) were considered autophagypositive. An average percentage of puncta-positive cells from nine fields/culture (total cells > 500) was calculated and used for statistical analysis.

#### Glutathione S-Transferase (GST)-Pull Down Assay

GST-pull down assay was performed as previously described (29). 293 cells were transfected with indicated plasmid at 1:1 for 2 d. Cell lysates were extracted with GST-lysis buffer and 400 µg of total soluble proteins from each sample were incubated with 30 µl of glutathione sepharose bead slurry (GE Healthcare Life Sciences, Piscat-away, USA) overnight at 4◦C. After extensive washing, the immunoprecipitates were subjected to Western blotting analysis. Anti-GST, anti-N-GFP or anti-PD-L1 antibodies were used to probe corresponding proteins.

#### Membrane and Cytosol Protein Detection

U251 cells were transfected with p-FU-Venus or p-FU-Venus-PD-L1 plasmid for 24–36 h. Membrane and cytosol proteins were extracted using Membrane and Cytosol Protein Extraction Kit (Beyotime Biotechnology, Shanghai, China) according to manufature's instructions. Membrane and cytosol proteins were then subjected to Western blotting with Na+/K<sup>+</sup> ATPase α1 and β-tubulin as internal control of membrane and cytosol proteins respectively.

#### Statistical Analysis

All experiments were repeated independently for at least three times. The values were expressed as means ± SEM. Unpaired Student's test was used to compare between two groups of in vitro experiments. Paired Student's test was used to compare between two groups of animal experiments. Comparisons among multiple groups were performed through one-way ANOVA with Student–Newman–Keuls post-test. P < 0.05 was considered statistically significant.

# RESULTS

#### PD-L1 Is Prominently Elevated in Invasive Frontier GBM Cells in vivo

To ascertain the relationships between intrinsic PD-L1 and glioma cell invasive behavior during GBM development, we examined PD-L1 expression in the whole brain at various stages of GBM development in a highly aggressive orthotopic

FIGURE 1 | PD-L1 is preferentially elevated in invasive frontier GBM cells in vivo. Mouse G422 glioma cells (20,000 cells) were inoculated into right striatum of Kunming mice. 6, 10, and 12 days after initial cell implantation, the mice were perfused with PFA and paraffin-embedded (3 mice/group). Serial mouse brain slices were cut and the slices with maximal tumor-load were used for PD-L1 (3 slice/brain) or H&E staining (3 slice/brain). (A) Representative PD-L1 staining of glioma on day 6. Inner black-dot ellipse delineated the boundary of tumor parenchyma while the outer irregular dot circle delineated the infiltration boundary. Left panels showed the whole brain and right panels showed enlarged square images. Arrows indicated the invasive frontier glioma cells. (B) Representative H&E staining of glioma at day 6. (C,E) Representative PD-L1 staining of glioma at day 10 and 12. (D,F) Representative H&E staining of glioma at day 10 and 12.

FIGURE 2 | the following creteria: P < 0.005, Padj < 0.05 and absolute fold change ≥1.2 (n = 3). (A) DO (Disease Ontology) enrichment analysis results showed that 42 of significantly altered-genes upon PD-L1 overexpression were associated with malignant glioma (indicated by red line). (B) BP (biological process) from GO enrichment analysis results showed that PD-L1 overexpression was closely associated with various cell migration pathways (indicated by red lines). (C) CC (cellular component) from GO enrichment analysis results showed that PD-L1 overexpression was closely associated with several actin signaling pathways (indicated by red lines). (D) KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis results showed that PI3K-Akt and actin regulations are among the top altered pathways by PD-L1 overexpression (indicated by red lines). (E) Overall survival of glioma patients with PD-L1 expression from database http://gepia.cancerpku.cn. Survival significant longer for patients with low PD-L1(p = 0). (F–H) Correlation analysis of PD-L1 with PIK3CG, SQSTM1 (p62), and ACTB (β-actin) showed that mRNA levels of PD-L1 were positively associated with those of PIK3CG, p62, and β-actin (p = 0) in glioma patient tissues (http://gepia.cancer-pku.cn).

mouse G422 model (**Figure 1**). PD-L1 as well as H&E staining clearly showed that the glioma tissue grew rapidly from day 6 to 12 (D6, D10, D12) after initial G422 cell-implantation. The tumor parenchyma in the left hemisphere was delineated with black-dot ellipse, and the tumor frontiers was delineated with blue (PD-L1, **Figures 1A,C,E**) or white-dot irregular circle (H&E, **Figures 1B,D,F**). Clearly, invasive GMB cells could migrate far away from its parenchyma mass at the early stage (D6) during GBM development. Along with the growth of glioma parenchyma, infiltration frontiers were also enlarged correspondingly. The frontier (blue) and inner (black) square micrographs in brain coronal sections were enlarged in right panels. It is clear that PD-L1 was prominently elevated in invasive G422 cells (indicated by arrows) in GBM infiltration frontier (**Figures 1A,C,E**). Such evidence establishes a strong positive correlation between PD-L1 and the aggressiveness of GBM cells.

### PD-L1-Altered Gene Expression Are Highly Enriched in Migration and PI3K/Akt-Actin Signaling

To explore the underlying mechanisms of PD-L1+-GBM cells in the infiltration frontier, mRNA sequencing was conducted in PD-L1-overexpressed U251 glioma cells. Differential expressed genes (DEGs) between U251/PD-L1 and U251/Vec cells (i.e., CD274 vs. Vec) were selected by three criteria (i.e., P < 0.005, Padj < 0.05 and absolute fold change ≥1.2) and the selected DEGs were subjected to various systemic analysis. DO (Disease Ontology) enrichment showed that PD-L1-altered gene expression was significantly associated with malignant glioma (**Figure 2A**, indicated by red line). Gene ontology (GO) enrichment analysis showed that PD-L1-altered gene expression was largely associated with cell migration (BP: biological process, **Figure 2B**, indicated by red lines), actin-structure functions (CC: cellular component, **Figure 2C**, indicated by red lines), PI3K-Akt signaling pathway and Regulation of actin cytoskeleton (KEGG: Kyoto Encyclopedia of Genes and Genomes, **Figure 2D**, indicated by red lines). Further analysis of human glioma database (http://gepia.cancer-pku.cn) revealed that PD-L1 (CD274) was negatively associated with the prognosis of glioma patients (p = 0, **Figure 2E**). In addition, PD-L1 expression was significantly positively correlated with PI3K (p = 0, **Figure 2F**), SQSTM1 (p62, a negative marker of autophagic influx signaling) (32) (p = 0, **Figure 2G**) and ACTB (β-actin) (p = 0, **Figure 2H**) in human glioma tissues. The evidence together indicates that PD-L1 greatly affects glioma cell invasion and PI3K/Akt-actin signaling.

# PD-L1 Regulates Akt-p62-Autophagic Influx Signaling Upon Starvation

It is well known that GBM tissues suffer severe ischemia that causes necrosis and promotes GBM cell aggressiveness (1–3), suggesting that energy-deprivation stress is a major driving force for glioma cell invasion. Since PI3K/Akt signaling is pivotal in promoting cell survival and suppressing autophagy (6, 33), we further investigated the role of PD-L1 in regulating Akt-autophagic influx signaling in glioma cells with a EBSSinduced starvation model. Western blot results showed that PD-L1 overexpression did not evidently alter p-Akt/mTOR/p70S6K signaling under normal culture conditions (**Figure 3A**, EBSS-0 h and **Figure S1**). Upon EBSS incubation, p-Akt and p62 (a negative maker for autophagy influx) was prominently reduced in U251/Vec cells (**Figures 3A,B**), while LC3II was evidently increased (–CQ2, **Figure 3C**) as detected by Western blot. Cells with PD-L1 overexpression significantly reversed EBSS-induced p-Akt reduction, p62 reduction as well as LC3II elevation compared to their corresponding Vec controls (indicated by red boxes, **Figures 3A–C**). Consistently, fluorescent immunostaining showed that PD-L1 evidently reduced LC3B and Beclin-1 while increased p62 in starved LN229 glioma cells at EBSS-6 h (**Figure 3D**).

We further examined the effects of endogenous PD-L1 on autophagy influx signaling in glioma cells by knocking-out PD-L1 using CRISPR/Cas9 technique. Western blot results demonstrated that endogenous PD-L1 was prominently decreased in stable U251/sg-PD-L1 cell lines (single colnes #2 and #3, **Figure 4A**). Under normal culture conditions, PD-L1 knockout did not evidently affected p-Akt (**Figure 4B**) or p-mTOR (**Figure S2**) in U251 cells. Upon EBSS incubation (6 and 12 h), LC3II was significantly elevated in U251/sgPD-L1 cells as compared to its corresponding U251/sgGFP controls (**Figure 4C**). Consistently, fluorescent images of GPF-LC3-puncta clearly showed that the average number of GFP-LC3-puncta (representing autophagosome) per cell was significantly increased in U251/sgPD-L1 cells compared to its U251/sgGFP control at EBSS-12h (**Figure 4D**).

#### PD-L1 128-237aa Fragment Interacts With Akt 112-480aa Fragment

In order to dissect the mechanism of PD-L1 action, we screened the binding partners of PD-L1 with various key upstream signaling proteins in PI3K/Akt pathway by GSTpull down experiment. The results clearly showed that PD-L1 preferentially bound to AKT1 and AKT2 compared to KiRAS, PTEN-1, PTEN-2, AKT3, EGFRVIII and PKA under normal

culture conditions (**Figure 5A**) as well as EBSS-4h incubation (**Figure 5B**). Then, we further dissected the exact binding domain of PD-L1 for its AKT binding. We made a series of PD-L1 (full length 290 aa, FL) truncates by deleting 260–290 aa (PD-L1 T1-259), 179-290 aa (PD-L1 T1-178) from its C-terminal and 1-127 aa (PD-L1 T128-259), 1-127/238-290 (PD-L1 T128- 259) (**Figure 5C**), 1-18 aa (PD-L1 T19-259) (**Figure 5D**) based on PD-L1 structure (upper panel, **Figure 5E**). GST-pull down results showed that PD-L1 FL1-290, T19-290, and T128-237 were the major truncates binding to AKT1 (indicated by red rectangles, **Figures 5B,C**). These results identified that PD-L1 128-237 fragment was required for its Akt binding.

We then seeked to identify the binding domain or key amino acid (aa) of Akt based on its structure (lower panel, **Figure 5E**). GST-pull down clearly showed that AKT1-FL (1-480 aa), T112- 480 but not T1-111 bound to PD-L1 (**Figure 5F**). Single amino acid mutant of AKT Y176A (loss of TNK2 binding and membrane localization) (34) and K14R (substantial reduction of

FIGURE 5 | Identification of PD-L1-Akt binding domains. Venus (or N-terminal Venus)-tagged plasmids were co-transfected with GST-tagged plasmids in 293T cells. Equal amount of total soluble proteins were subjected to GST-pull down analysis. (A,B) Western blot analysis of the binding between Venus-PD-L1 and GST-KiRas, PTEN-1, PTEN-2, AKT1, AKT2, AKT3, EGFRVIII, and PKA under normal culture conditions (A) and EBSS-4 h (B). WCL, whole cell lysate. (C,D) Western blot analysis of the binding between GST-AKT1 and NV-PD-L1 truncates. FL, full length; T, truncate. (E) Schematic diagram of PD-L1 and AKT truncates. (F) Western blot analysis of the binding between GST-PD-L1 and Venus-AKT truncates. All experiments were repeated at least three times independently.

ubiquitination and loss of PIP3 binding) (35) did not evidently affect its interaction with PD-L1 (**Figure 5F**), suggesting that PD-L1/Akt interaction occurred in the cytosol via independent AKT K14 and Y176 sites.

#### PD-L1 Facilitates Akt Membrane-Translocation and F-Actin Formation in Starved Glioma Cells

Since PD-L1 selectively interacted with Akt 112–480 fragment but not its PH domain (1–111 aa) (**Figure 5F**), it is interesting to further figure out how PD-L1 promoted Akt activation upon starvation (**Figure 3A**). Full activation of Akt depends on Akt translocation from cytoplasm to cell membrane (via its PH domain) and PIP3/PDK action on cell membrane (36, 37), which can be pharmacologically inhibited by specific inhibitor Perifosin (37) and LY294002, respectively. Administration of LY294002 (GFP, **Figure 6A**) and Perifosine (GFP, **Figure 6B**; Vec, **Figure 6C**) evidently reduced p-Akt in U251/GFP cells. In PD-L1-overexpressed U251 cells, however, LY294002 but not Perifosin could reduce p-Akt level (PD-L1, **Figures 6A–C**). Such evidence suggests that PD-L1 protects the membranetranslocation of Akt. Western blot results demonstrated that membrane-bound Akt was evidently increased upon PD-L1 overexpression compared to Vec control (**Figure 6D**). Consistently, fluorescent immunostaining of Akt showed that membranous Akt (indicated by arrows) was more evident in PD-L1-overexpressed U251 cells, which was well co-localized with F-actin (Palloidin staining) (**Figure 6E**).

Finally, we examined the effects of PD-L1 on F-actin structure in starved glioma cells. Fluorescent images of Palloidin staining clearly showed that F-actin formation was evidently increased in PD-L1-overexpressed U251 cells under normal culture conditions as well as EBBS-12 h incubation (**Figure 7A**). Taken together, the evidence suggested that PD-L1 facilitated Akt translocation to cell membrane and Akt activation to modify cellular morphology via F-actin cytoskeleton during energydeprivation stress.

# DISCUSSION

In our present study, we discovered that PD-L1 elevation occurred predominantly in highly aggressive glioma cells. RNA sequencing revealed that high levels of PD-L1 in glioma cells mainly mediated cell migration and PI3K/Akt/actin signaling. PD-L1 reduced starvation-induced Akt inhibition, autophagic influx and F-actin collapse in glioma cells. Mechanistically, PD-L1 directly interacted with Akt, and this process involved PD-L1 128-237aa and Akt 112-480aa. The binding of PD-L1 to Akt facilitated membrane-translocation of Akt and thus elicited downstream biological effects in frontier glioma cells. These findings reveal a novel invasive mechanism in GBM.

PD-L1 is a pivotal negative regulatory molecule at the immune checkpoint axis and has complicated biological functions besides immune regulation (38). Detection of PD-L1 level in cancer cells is of primary importance for predicting its biological functions as well as prognosis of immune therapy in various cancers such as melanoma. Previous studies have reported that PD-L1 is widely expressed in most glioma cell lines including U251 (12, 29) and human glioma tissues (14), indicating important biological functions of PD-L1 in glioma. Here, we clearly demonstrated that PD-L1+-glioma cells were highly aggressive (**Figure 1**). Compared to rat/C6 and nude mouse/U87 glioma models, in which 1,000,000 or 250,000 glioma cells were implanted in the striatum and only a few glioma cells migrated outside from the tumor rim at day 14 or 24 later (29), in our orthotopic glioma model, only 20,000 G422 cells were microinjected into the striatum of mouse brain. In very short time (6 d) after initial glioma cell injection, PD-L1+-glioma cells had already migrated far from the glioma center (**Figures 1A,B**). Afterwards, invasive PD-L1+-glioma cells quickly reached the outer surface of cerebral cortex at day 10 (**Figures 1C,D**) and middle line of two hemispheres at day 12 (**Figures 1E,F**) after the initial glioma cell implantation. Obviously, the G422 glioma model mimics well the invasive pathological process of GBM. Since PD-L1 in cell membrane (39), exosome (40), and cytoplasm (29) are all pivotal for cancer development and progress, we can infer that the elevation of PD-L1 levels in those infiltrating glioma cells heavily contribute to their invasiveness.

To systematically reveal the role of PD-L1 in glioma cells, we conducted RNA sequencing in PD-L1-overexpressed human glioma U251 cells, which might mimic the status of higher PD-L1 levels in those infiltrating glioma cells. Bioinformatics analysis of DEGs clearly pointed out that increased PD-L1 in glioma was closely associated with glioma malignancy (**Figure 2A**), as well as distinct signaling pathways in invasive malignant glioma, such as cell migration, PI3K/Akt and actin organization (**Figures 2B–D** and **Figures S1, S3**). In a database of 674 glioma human samples, mRNA levels of PD-L1(i.e., CD274) were significantly positively correlated to those of PI3K (i.e., PI3KCG) and β-actin (i.e., ACTB) (**Figures 2F,H**), supporting our RNAsequencing data. Also, bioinformatics analysis of glioma patient samples in opensource database showed that higher PD-L1 expression indicates poorer patient survival (**Figure 2E**), supporting that PD-L1's increase in infiltrating glioma cells is of clinical importance.

It is well known that GBM tissues suffer severe ischemic starvation, which is an important pathological factor driving aggressiveness (1, 2). PI3K/Akt is a pivotal signaling pathway promoting cellsurvival and suppressing autophagy under normal and ischemic conditions, while ischemic starvation is a major pathological stimulus for autophagy and actin cytoskeleton collapse (41, 42). It is conceivable that boundary GBM cells encounter energy-deprivation and higher PD-L1 expression may prevent GBM cell death and promote invasion via Aktautophagy-actin signaling. Indeed, bioinformatics analysis of our RNA-sequencing data showed that PD-L1 affected autophagic signaling (**Figure S2**). By applying the glioma patient database, we found that mRNA levels of PD-L1 were significantly positively correlated to those of p62 (i.e., SQSTM1) (**Figure 2G**). Consistently, our experimental results demonstrated that PD-L1 significantly resumed EBSS starvation-induced Akt-autophagy signaling in glioma cells in vitro (**Figure 3**) but had minor effects under normal culture conditions (**Figure 4B** and **Figure S4**).

The effects of PD-L1 overexpression on Akt increase and LC3 decrease were verified in vivo in a orthotopic rat/C6 glioma model (**Figure S5**). Further, via double-fluorescent immunostaining, we verified that higher PD-L1 levels in individual GBM cells were associated with lower Beclin1 and LC3 levels in GBM patient tissues (**Figure S6**). Such evidence consistently supported that PD-L1 suppressed autophagy via Akt induction/activation. In addition, PD-L1 increased F-actin formation and Akt/Factin co-localization beneath cell membrane in glioma cells upon starvation (**Figures 6E**, **7A**). In other cancer cells such as sarcoma cells, melanoma and ovarian cancer cells, PD-L1 also regulates Akt-mediated autophagic signaling (27, 31). This data largely supported our proposed PD-L1-Akt-p62 autophagy-actin-invasion mechanism in frontier GBM cells. Further collection of evidence is helpful to elucidate the exact invasive mechanism of GBM in vivo.

Importantly, we identified a novel mechanism by which PD-L1 regulate PI3K/Akt signaling. Among various upstream signaling proteins in PI3K/Akt pathway, we found that PD-L1 preferentially bound to Akt isoforms, and this became more evident upon EBSS incubation (**Figure 5** and **Figure S7**). Moreover, we dissected the binding element of PD-L1 and Akt. Our data revealed that the 128-237aa fragment of PD-L1 and the 112-480aa fragment of Akt were responsible for PD-L1/Akt interaction (**Figure 5**). Mutation of K14 and Y176, two active sites for the membrane-translocation of Akt, did not alter PD-L1/Akt interactions (**Figure 5F**), suggesting that PD-L1/Akt interaction occurred mostly in the cytoplasm. Upon PD-L1 overexpression, membrane-bound Akt was prominently increased (**Figure 6D**), supporting that cytoplasmic PD-L1/Akt interactions facilitated Akt translocation from cytoplasm to cell membrane, which is required for full activation of Akt (36). Interestingly, LY294002 (inhibiting p-Akt via PI3K) but not Perifosin (inhibitor of membrane-translocation of Akt and p-Akt via PH domain) (37) could abolished Akt phosphorylation in the presence of PD-L1 overexpression (**Figures 6A–C**). Such evidence suggested that the binding of PD-L1 to non-PH domain of Akt may prevented Perifosin action via some unknown mechanism, thus strengthened the attachment of Akt PHdomain to cell membrane and facilitated its full phosphorylation (**Figure 7B**). Further study on the mechanisms of PD-L1/Akt interactions requires more experimental data in the future.

In summary, our findings indicate the following model for PD-L1 actions in GBM (**Figure 7B**): Along with the rapid GBM growth, inner GBM cells suffer severe starvation, which causes cytoskeleton collase; while frontier GBM cells suffer less severe starvation, which induces PD-L1 expression and its binding to Akt non-PH domain. This reinforces Akt's membrane-binding and leads to increased full activation of Akt as well as its downstream p62/LC3/F-actin signaling, preventing starvationinduced Akt inhibition and facilitating GBM cell invasion. By combining our previous findings that PD-L1 promotes EMT of glioma cells, we speculate that PD-L1/Akt/autophagy/F-actin is a key driving force for GBM aggressiveness, which serves as a potential therapeutic target for GBM.

#### REFERENCES


#### DATA AVAILABILITY STATEMENT

The datasets generated for this study can be found in the NCBI (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE107581).

#### ETHICS STATEMENT

The animal study was reviewed and approved by Experimental Animal Ethics Committee, Tongji Medical College, Huazhong University of Science and Technology.

#### AUTHOR CONTRIBUTIONS

RC, XX, XQ, FL, CL, YL, XL, and GJ performed the experiment. RC, FP, XQ, and XC data analysis, wrote the manuscript, and contributed to discussion. FH and DL contributed to discussion. RC, FP, XQ, and XC designed and wrote the manuscript.

#### FUNDING

This work was supported by grants from the National Nature Science Foundation of China (Grant Nos. 81471386, 81672504 to XC, Grant No. 81873854 to FP, Grant No. 81602202 to FH). The Fundamental Research Funds for the Central Universities, HUST (Grant No. 2017KFYXJJ048 to XC). Natural Science Foundation of Hubei Province (No. 2017 CFB639), Hubei Province Health and Family Planning Scientific Research Project (No. WJ2015Q002) and Science and Technology Planning Project of Wuhan (No. 2017060201010202) to FP. We declare all sources of funding received for the research being submitted.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2019.01347/full#supplementary-material


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Chen, Xu, Liu, Li, Li, Li, Jiang, Hu, Liu, Pan, Qiu and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# FAP-a and GOLPH3 Are Hallmarks of DCIS Progression to Invasive Breast Cancer

Li-Na Yu1,2,3, Zhen Liu<sup>4</sup> , Yan Tian1,2,3, Pei-Pei Zhao1,2,3 and Xing Hua<sup>5</sup> \*

<sup>1</sup> Department of Pathology, Nanfang Hospital, Guangzhou, China, <sup>2</sup> Department of Pathology, School of Basic Medical Sciences, Southern Medical University, Guangzhou, China, <sup>3</sup> Guangdong Provincial Key Laboratory of Molecular Tumor Pathology, Guangzhou, China, <sup>4</sup> Department of Pathology, School of Clinical Medicine, Guizhou Medical University, Guiyang, China, <sup>5</sup> Department of Pathology, Guangzhou Red Cross Hospital, Medical College, Jinan University, Guangzhou, China

Biological markers that could predict the progression of ductal carcinoma in-situ (DCIS) to invasive breast cancer (IDC) are required urgently for personalized therapy for patients diagnosed with DCIS. As stroma was invaded by malignant cells, perturbed stromal-epithelial interactions would bring about tissue remodeling. With the specific expression of the fibroblast activation protein-alpha (FAP-a), Carcinoma-associated fibroblasts (CAFs) are the main cell populations in the remodeled tumor stroma. Golgi phosphoprotein 3 (GOLPH3), a documented oncogene possessing potent transforming capacity, is not only up-regulated in many tumors but also an efficient indicator of poor prognosis and more malignant tumors. The present study aimed to retrospectively evaluate the pathological value of FAP-a and GOLPH3 in predicting the recurrence or progression of DCIS to invasive breast cancer. Immunohistochemical techniques were applied to investigate the expression of FAP-a GOLPH3 in 449 cases of DCIS patients received extensive resection and with close follow-up, but not being treated with any form of chemo- or radio-therapy. The combination of FAP-a and GOLPH3 in predicating the recurrence or progression of DCIS into invasive breast cancer was specifically examined. The study demonstrated that the overexpression of FAP-a in stromal fibroblasts and GOLPH3 in carcinoma cells are highly predictive of DCIS recurrence and progression into invasive breast cancer. Both FAP-a and GOLPH3 have high specificity and sensitivity to predict the recurrence of DCIS. Moreover, the combination of FAP-a and GOLPH3 tends to further improve the specificity and sensitivity of DCIS recurrence by 9.72–10.31 and 2.72–3.63%, respectively. FAP-a and GOLPH3 serve as novel markers in predicting the recurrence or progression of DCIS into invasive breast cancer.

Keywords: FAP-a, GOLPH3, DCIS, recurrence, breast cancer

# INTRODUCTION

With the development and aid of high quality screening mammography, the probability of detected ductal carcinoma in situ (DCIS) has raised rapidly (1). DCIS declares roughly 20% of all screening-detected breast cancer recently (2). Breast conserving therapy combined with or without radiotherapy has been generally accepted as the standard and efficient method to treat DCIS (3). Although DCIS is a type of non-invasive breast cancer, in which carcinoma cells are

#### Edited by:

Dong-Hua Yang, St. John's University, United States

#### Reviewed by:

Hamid Morjani, Université de Reims Champagne-Ardenne, France Shanchun Guo, Xavier University of Louisiana, United States

> \*Correspondence: Xing Hua nana1800@smu.edu.cn

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 25 July 2019 Accepted: 29 November 2019 Published: 17 December 2019

#### Citation:

Yu L-N, Liu Z, Tian Y, Zhao P-P and Hua X (2019) FAP-a and GOLPH3 Are Hallmarks of DCIS Progression to Invasive Breast Cancer. Front. Oncol. 9:1424. doi: 10.3389/fonc.2019.01424

**186**

controlled by the basement membrane of breast duct, in reality, DCIS remains a heterogeneous cluster of pre-invasive breast cancers with various types of morphology, expansion, and malignant capacities (4).

DCIS remains the status of the carcinoma precursor lesion of invasive ductal carcinoma (IDC), but mechanisms under the conversion are largely unknown (5–7). In order to predict the fate of DCIS more accurately, to develop invasive cancer or remain at DCIS status, efforts must be made continuously to investigate the molecular aspects of the disease. In the last few years, several prognostic factors have emerged as possible contenders to help stratify risk (8). For example, a study by Martins et al. (9) demonstrated that monocarboxylate transporter 4 (MCT4) and stromal caveolin 1 (Cav-1) were differentially expressed in IDC and DCIS. Suppression of Cav-1 and activation of MCT4 in the stroma were indicators of overproduction of oxidative stress and glycolysis, indicating the progression from DCIS status into IDC.

The prolyl-specific serine proteinase fibroblast activation protein alpha (FAP-a), a type II integral membrane protein (10), is a M(r) 95,000 cell surface antigen brought about instantaneously by induced stromal fibroblasts during embryogenesis (11), which is the latter stages of wound healing (12), in some pathologic status where fibrous tissue growth is a conspicuous feature (13), and is occasionally present in normal fibroblast or pancreatic a-cells (14, 15). As shown in our previous studies, immunohistochemical analyses were applied to evaluate the expression of FAP-a in normal mammary tissues and usual ductal hyperplasia (UDH), DCIS, DCIS with microinvasive (DCIS-MI) and IDC. We further hypothesized the participation of FAP-a in designating the microemboli's formation, facilitating the pathological process of breast cancer (16). Cremasco et al. (17) showed that breast tumors contain a variety of FAP expressing stromal cells with dichotomy function, phenotype, and location. Busek et al. (18) also found that FAP is a promising therapeutic target. In this study, we found that FAP-a is specifically expressed in the stromal fibroblasts of IDC and the tumor-host interface of the invasive front of DCIS-MI. However, FAP-a is negatively expressed in stromal fibroblasts of normal mammary tissues and UDH. We are the first study that proposed and demonstrated FAP-a's potential as a novel biomarker for pathological diagnosis of DCIS with microinvasion. Although high FAP-a expression was detected in stromal fibroblasts of IDC and tumor-host interface at the invasive front of DCIS-MI, its correlation with the chemo- or radio-therapy treatment has not been studied because of the limited sample size.

Previous investigations have demonstrated that overexpression of human chromosome 5p13 contributed to tumor formation, progression, and metastasis. The expression of Golgi phosphoprotein 3(GOLPH3) was found to be correlated with copy number of 5p13 in human lung cancer, and gain or loss of function studies have proved that GOLPH3 was a genuine oncogene possessing strong transforming activity, which was induced in cancers with 5p amplification (19). Further researches on the clinical importance of GOLPH3 revealed that GOLPH3 expression was not only increased in oral tongue cancer and glioma, but also was indicating poor prediction and more malignant tumors (20–22). These results further indicated GOLPH3's significant role in the progression of various types of tumors. In our previous works, it was revealed that overexpression of GOLPH3 promoted the progression of prostate cancer from hormone sensitive stage to hormone stage, and GOLPH3's potential as a novel biomarker and potential target for antagonizing castration resistant prostate cancer (23). Zeng et al. showed that high expression of GOLPH3 was associated with low overall survival rate of breast cancer patients, and overexpression of GOLPH3 increased the proliferation and tumorigenicity of human breast cancer cells (24).

In the present study, immunohistochemical techniques were applied to evaluate the role of GOLPH3 in carcinoma cells and FAP-a in the stroma in recruited DCIS patients received combined treatment of wide-excision and close follow-up, but haven't received any type of chemo- or radio-therapy; which further explored the potentials of FAP-a and GOLPH3 as indicative marker during the natural progression of DCIS. To this end, we specifically examined the correlation between FAPa and GOLPH3 expression, their clinico-pathological values in predicting the recurrence or progression of DCIS into invasive breast cancer.

### MATERIALS AND METHODS

#### Case Selection

We retrospectively recruited 449 patients who had received treatment of mastectomy between May 1994 and May 2013 using the surgical oncology breast cancer database. Archives were acquired from the documentation of the Department of Pathology and Medical Records Room, the Fourth Affiliated Hospital of Jinan University and Nan Fang Hospital of Southern Medical University in Guangzhou, China. This study was approved by the Ethics Committee of Guangzhou Red Cross Hospital.

Formalin-fixed and further paraffin-embedded (FFPE) tissue sections were prepared from 449 patients with DCIS. All patients undergoing surgical excision were treated by the same surgeon, and all had negative margins (= or > 10 mm) at the end of surgical excision or re-excision. None of the patients received radiation treatment or tamoxifen therapy. The time of diagnosis and recurrence was recorded as the time of surgery that initiated the pathologic diagnosis. Data analysis indicated the endpoint of treatment based on the appearance of local recurrence of DCIS, invasive breast carcinoma and without pathological symptoms at the last time of follow-up. As to each subject, histologic patterns (comedo, cribriform, micropapillary, papillary, or solid), presence of necrosis, inflammation and nuclear grade were kept on record. Routine morphological analysis of hematoxylin and eosin stained tissue sections were blindly reviewed by five pathologists together and a diagnostic agreement was finalized.

#### Immunohistochemistry

The expression of GOLPH3 in tumor cells and FAP-a in tumor stroma were evaluated by immunohistochemical staining. Firstly, sections were deparaffinized by reduced xylene and then conducted antigen retrieval in Tris-EDTA retrieval buffer Yu et al. Hallmarks of Invasive Breast Cancer

(PH 9.0) (Dako, Carpinteria, CA) before immunohistochemical staining with a Dako autostainer. Immunohistochemical staining for GOLPH3 and FAP-a was completed as described previously. The GOLPH3 primary antibodies (clone 19112-1-AP, 1:100 dilution) were obtained from the ProteinTech Group, and the FAP-a primary antibodies (clone 427819, 1:50 dilution) were obtained from R & D systems. Positive and negative staining were carried out for each staining. The degrees of staining were graded based on the following criteria. Grade 1 was taken as semi-quantitative and negative staining; Grade 2 was taken when either diffuse weak staining or <30% strong staining appeared; Graded 3 was awarded when 30% or more of the cells have been stained.

#### Statistics

The time from surgery to recurrence (TTR) was recorded and evaluated by using Kaplan-Meier survival curves. The comparison of stratified survival curves used log-rank tests. Cox proportional hazards regression was applied to analyze the correlation among FAP-a, GOLPH3 expressions, and TTR with other potential diagnostic markers of TTR. All hypotheses testing was conducted using the Fisher exact test or the Kruskal-Wallis test, based on the discrete or continuous essence of any other factor. Correlation analysis among FAP-a, GOLPH3 staining intensity, and DCIS recurrence rate was determined using the Spearman test.

P-values were all two-sided, and p < 0.05 was considered as statistical significance. Statistical analysis was accomplished using the SPSS statistical analysis software version 19.0, and tables and graphs were generated.

# RESULTS

#### Informative Analysis of the DCIS Patient Cohort

As shown in **Table 1**, 449 women were with a median age of 52 years (range 22–87 years) when diagnosed with DCIS. The median of follow-up time was 169.84 months (14.15 years). Pathological variables including grade, histologic pattern and necrosis were recorded accordingly. Meanwhile, follow-up on DCIS recurrence pattern and DCIS deterioration into invasive breast cancer were also recorded. The immunophenotype of GOLPH3 and FAP-a is correlated with each clinicopathological parameter listed in **Table 1**. As shown in **Figure 1**, the FAPa protein expression was mainly detected in the cytoplasmic fractions of stromal fibroblasts, while GOLPH3 in was detected the cytoplasm of breast carcinoma cells. In these 449 DCIS patients, 110 of which appeared certain degree of recurrence (49 progression into DCIS, while the other 61 progression into invasive breast cancer), resulting in an final recurrence rate of 24.50%. Therefore, the progressive rate of DCIS into invasive breast cancer at the present cohort study was 13.59 percent (61/449), which is relevant to the expected rate of 12–15%. The median time of recurrence turned out to be 83.25 months (DCIS with 33.45 months, invasive breast cancer with 115.36 months).

TABLE 1 | Levels of FAP-a and GOLPH3 expression in relation to clinicopathologic variables.


# Stromal FAP-a Is Associated With DCIS Recurrence

As shown in **Table 2**, immunostaining of stromal FAP-a was scored semiquantitative as 3 (high), 2 (low), and 1 (absent of staining) based on the staining intensity. In 49 DCIS recurrence cases, the percentages of grade 3, 2, and 1 of stromal FAP-a immunostaining were 65.31% (32/49), 18.37% (9/49), and 16.33% (8/49), respectively. In 61 DCIS patients who recurred to IDC, percentages of grade 3, 2, and 1 of FAP-a immunostaining were 65.57% (40/61), 19.67% (12/61), and 14.75% (9/61), respectively. In 339 DCIS patients without recurrence, the incidence of high, low, and absent stromal FAP-a immunostaining was 7.96% (27/339), 3.54% (12/339), and 88.50% (300/339), respectively. Spearman test showed no significant correlation between FAP-a staining intensity and DCIS recurrence in DCIS recurrence patients. Among 132 FAPa positive DCIS patients, 41 recurred to DCIS, 52 recurred, and progressed into invasive breast cancer, while the remaining 39 cases experienced no recurrence. FAP-a (high/low) is expressed in 84.55% (93/110) of the DCIS recurrence patients, while only 11.50% (39/339) in DCIS patients without recurrence. This difference was statistically significant (P < 0.001). By contrast, no statistical difference was detected of FAP-a expression between DCIS recurrence and invasive progression. Kaplan-Meier curves for overall recurrent rate, rate of recurrent to DCIS, and progression into invasive breast cancer data were revealed in **Figure 2**. The left panel suggested that for the patients with higher expression of FAP-a, the overall recurrent rate will be higher. The middle panel suggested that the higher expression of FAP-a may be correlate with the higher rate of recurrent to DCIS. The right panel suggested that the patients with

higher expression of FAP-a may have the higher possibility of progression to invasive breast cancer. It was worth noting that the presence of stromal FAP-a was positively correlated with DCIS recurrence.

#### GOLPH3 Expression Is Associated With DCIS Recurrence

As shown in **Table 3**, immunostaining of stromal GOLPH3 was scored semiquantitative as 3 (high), 2 (low), and 1 (absent of staining) based on the staining intensity. In 49 DCIS recurrence cases, the incidence of high, low, and absent GOLPH3 immunostaining was 69.39% (34/49), 20.41% (10/49), and 10.20% (5/49), respectively. In 61 DCIS patients who recurred to IDC, the incidence of high, low, and absent GOLPH3 immunostaining was 62.30% (38/61), 16.39% (10/61), and 21.31% (13/61), respectively. In the 339 DCIS cases without recurrence, the percentages of grades 3 (high), 2 (low), and 1 of GOLPH3 immunostaining were 6.78% (23/339), 4.13% (14/339), and 89.09% (302/339), respectively. Spearman test showed no significant correlation between the GOLPH3 staining intensity and DCIS recurrence in DCIS recurrence patients. Among the 132 GOLPH3 positive DCIS patients, 44 of them underwent a recurrence to DCIS, 48 recurred and progressed into invasive breast cancer, while the remaining 37 cases did not experienced no recurrence. GOLPH3 (high/low) is expressed in 83.64% (92/110) of the DCIS recurrence patients while in 10.91% (37/339) of the DCIS without recurrence. This difference was statistically significant (P < 0.0001). By contrast, no statistical difference was detected regard of GOLPH3 expression between DCIS recurrence and the invasive progression of DCIS. Kaplan-Meier curves for overall recurrence, DCIS recurrence, and progression into invasive breast cancer were shown in **Figure 3**. Note that the presence of GOLPH3 is specifically associated with DCIS recurrence.

### The Combination of FAP-a and GOLPH3 Expression Is Associated With DCIS Recurrence

The immunostaining of FAP-a in stromal fibroblasts and GOLPH3 in carcinoma cells was scored semiquantitatively as 3 (high), 2 (low), and 1 (absent of staining) based on the staining intensity, respectively. The staining intensity 2 and 3 were considered as positive and the staining intensity 1 was considered as negative. As shown in **Table 4**, 41 of the 49 DCIS recurrence patients showed both FAP-a positive in stromal fibroblasts and GOLPH3 positive in carcinoma cells (FAP-a+GOLPH3+), three of them showed FAP-a−GOLPH3+, five showed FAP-a−GOLPH3−, while none of them showed FAP-a+GOLPH3−. Among the 61 DCIS patients who recurred to IDC, 48 showed FAP-a+GOLPH3+, four showed FAPa <sup>+</sup>GOLPH3−, nine showed FAP-a−GOLPH3−, while none of them showed FAP-a−GOLPH3+. Moreover, FAP-a+GOLPH3<sup>−</sup> existed in 80.91% (89/110) of DCIS recurrence patients, while only 1.18% (4/339) of DCIS patients without recurrence (4/339) showed FAP-a+GOLPH3<sup>+</sup> with statistically significant difference (P < 0.0001). Pearson test demonstrated a positively correlative relationship among FAP-a, GOLPH3 co-expression and DCIS recurrence. As shown **Figure 4**, Kaplan-Meier curves were revealed for overall recurrence, DCIS recurrence, and invasive recurrence progressing into invasive breast cancer. Note that the co-expression of FAP-a and GOLPH3 are specifically associated with DCIS recurrence.

# Specificity and Sensitivity of FAP-a and GOLPH3 in Diagnosis of DCIS Recurrence

The immunostaining of FAP-a in stromal fibroblasts and GOLPH3 in carcinoma cells was scored semiquantitatively as 3 (high), 2 (low), and 1 (absent of staining) based on the staining intensity, respectively. The staining intensity 2 and 3 were considered as positive and the staining intensity 1 was considered as negative. As shown in **Table 5**, the specificity and sensitivity of FAP-a in the diagnosis of DCIS recurrence is 88.50 and 84.55%, respectively. The specificity and sensitivity of GOLPH3 in the diagnosis of DCIS recurrence is 89.09 and 83.64%, respectively. When combined FAP-a and GOLPH3, the specificity and sensitivity in diagnosis of DCIS recurrence increased to 98.82 and 87.27%, respectively.




FIGURE 3 | Kaplan-Meier curves for GOLPH3 state and duration of recurrence among DCIS patients. (Left panel) The expression of GOLPH3 is positively associated with the induction of overall recurrence. (Middle panel) The expression of GOLPH3 is correlated with the increase of DCIS recurrence. (Right panel) The expression of GOLPH3 is associated with the induction in DCIS progression to IDC. High (score = 3), low (score = 2), and absent (score=1). p-values (log rank test) are as shown.


TABLE 4 | Specificity and sensitivity of FAP-a and GOLPH3 in diagnosis of DCIS recurrence.

# DISCUSSION

DCIS is considered as a precursor to invasive breast cancer. In this study, we demonstrated that 24.50% of DCIS patients undergo an ipsilateral local recurrence, and 13.59% of these recurrences are invasive in a sample of 449 DCIS patients. The risk of DCIS patients of local recurrence appearance or progression into invasive breast cancer after their treatment was still unpredicted at the present stage (25). The ability to develop a predictive model would present to be an enormous clinical advance to identify a molecular marker that would have predictive potential of DCIS's prognosis, either DCIS develops invasive cancer or stays DCIS.

Although several molecular factors have emerged as predictive marker of DCIS in the last few years, it remains challenging to clarify the prognostic significance of those bio-markers. In most studies, sample sizes were limited, and patients had undergone either endocrine treatment or radiotherapy (8, 9). In this study, all of the patients undergone only surgical excision and were treated by the same surgeon, and all of them are with negative margins to the end of surgical excision. None of the patients received radiation therapy or tamoxifen treatments, which potentiates us to evaluate the predictive potential of FAP-a and GOLPH3 during the natural history of DCIS disease progression.

Tissue remodeling stimulated by the interactions of perturbed stromal-epithelial cells is brought about during the carcinogenesis of DCIS recurrence (16). Although these interactions turns to of great importance in tumorigenesis, their underlying molecular mechanisms in tumorigenic processes are not fully understood. Most studies demonstrated that carcinoma-associated fibroblasts (CAFs) remain the major stromal cell population participating in the epithelial-stromal interaction in the remodeled micro environment, which were mainly responsible for production of FAP-a (5, 26, 27). In this study, 84.55% of the DCIS recurrence showed high/low expression of FAP-a expression in stromal fibroblasts, while only 11.50% of the DCIS without recurrence showed high/low FAP-a expression. Significant difference in the protein expression of FAP-a was detected between the DCIS recurrence group and without recurrence group. We have previously demonstrated that 82.72% (67/81) stromal fibroblasts in the invasive front of DCIS with microinvasion at the tumor-host interface showed significant FAP-a positivity, and all of the 67 cases of IDC exhibited strongly positive FAP-a staining in stromal fibroblasts. Thus, FAP-a was potentiated in the promotion of the formation of microemboli, which promotes the invasion of breast cancer. Hence, the micro environmental reprogramming of cancer revealed as a key step in DCIS recurrence, and it may of

TABLE 5 | Association of the FAP-a and GOLPH3 with DCIS recurrence.


greater importance to elucidate stromal metabolism other than characteristic of carcinoma cells to predict the prognosis of DCIS patients by characterizing.

Interestingly, our study elucidated that the presence of GOLPH3 in carcinoma cells is also specifically associated with DCIS recurrence. In this study, 83.64% of the DCIS recurrences demonstrated high/low GOLPH3 expression in carcinoma cells, while only 10.91% of the DCIS without recurrence showed high/low GOLPH3 expression. In addition, a significant difference was found in the expression of GOLPH3 protein between the DCIS recurrence group and without recurrence group. Several studies have shown that GOLPH3 is related to the prognosis of cancer. It has been reported that high GOLPH3 expression is associated with poor overall survival in patients with breast cancer and that GOLPH3 overexpression increases the proliferation and tumorigenicity of human breast cancer cells (24). In prostate cancer, patients with high levels of GOLPH3 will have shorter survival time (28). Zhang et al. (29) has found that GOLPH3, which is up-regulated in prostate cancer tissues, can be useful for predicting biochemical recurrencefree survival and overall survival in patients with prostate cancer. Moreover, GOLPH3 may be of significant potential as predictive marker of DFS and OS in subjects with diagnosed prostate cancer (23). Not only prostate cancer, but also GOLPH3 has been proved to be highly expressed in NSCLC tissues, indicating that GOLPH3 may be a useful diagnostic factor for NSCLC (30). Tang et al. found that high expression of GOLPH3 usually indicates poor survival of breast cancer and weak resistance to chemotherapy (31). GOLPH3L is an accessory homolog of GOLPH3. It had been found that the prognosis of ovarian cancer patients with high expression of GOLPH3L was lower than that of patients with low or no expression of GOLPH3L (32).

Meanwhile, the detection of GOLPH3 expression might serve as a reliable predictive marker. Same is reasonable that to determine the changes of GOLPH3 expression may also facilitate to comprehensively understand the incidence of progression in patients with prostate cancer.

To our understanding, a group of proteins constitutes in the trans-Golgi matrix were tightly regulated in the trans-Golgi and demonstrated to be responsible for anterograde and retrograde Golgi trafficking, as well as collaborating with the cytoskeleton and supporting the Golgi structure (33). GOLPH3 would transiently be transported to the surface of cytoplasm in the trans-Golgi and represented a first-in-class Golgi oncoprotein. Holly C. Dippold indicated that trafficking out of the Golgi relied on the interaction of GOLPH3, MYO18A, F-actin, and PtdIns (4) P. This process is controlled by MYO18A and Factin and their transmission to the Golgi regulated by GOLPH3 and PtdIns (4) P. This evidence suggests that GOLPH3 traffics the tensile force responsible for successful tubule and vesicle formation (20). Kenneth L. Scott raised the hypothesis that GOLPH3 interacted with VPS35 and the tetramer to recycle receptor for key molecules, thus regulating downstream mTOR signaling (19). Tenorio et al. found that the cytosolic and membrane components of the three breast cell lines had biochemical differences in GOLPH3, that was that, in cancer cells, part of the overexpressed GOLPH3 was modified by differentiation (34).

We have shown here that the presence of FAP-a in stromal fibroblasts and GOLPH3 in carcinoma cells are reliable biomarker of DCIS recurrence and progression into invasive breast cancer. Both FAP-a and GOLPH3 have high specificity and sensitivity in predicting the recurrence of DCIS. The specificity and sensitivity of FAP-a in diagnosis of DCIS recurrence was 88.50 and 84.55%, respectively. The specificity and sensitivity of GOLPH3 in the diagnosis of DCIS recurrence was 89.09 and 83.64%, respectively. However, the combination of FAP-a and GOLPH3 had higher specificity and sensitivity that the specificity and sensitivity was 98.82 and 87.27%, respectively. Compared with FAP-a and GOLPH3, the combination of FAP-a and GOLPH3 refined the specificity and sensitivity of potent prediction of DCIS recurrence by 9.72– 10.31 and 2.72–3.63%, respectively. Thus, we identified that FAP-a and GOLPH3 might potentiate as a reliable biomarker for predicting the DCIS recurrence and progression of DCIS into invasive breast cancer. Though FAP-a and GOLPH3 showed significant correlation both in DCIS recurrence and in DCIS without recurrence patients, the complex interplay

#### REFERENCES


between FAP-a and GOLPH3 in DCIS recurrence remains poorly understood.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

#### AUTHOR CONTRIBUTIONS

XH and L-NY: study concepts. L-NY, ZL, YT, and P-PZ: study design and statistical analysis. YT, P-PZ, and ZL: data acquisition. L-NY: quality control of data and algorithms. L-NY and ZL: data analysis and interpretation. XH: manuscript preparation. ZL and L-NY: manuscript editing. XH, L-NY, ZL, and P-PZ: manuscript review.

collagen I derived gelatin cleavage sites. Biochemistry. (2008) 47:1076–86. doi: 10.1021/bi701921b


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Yu, Liu, Tian, Zhao and Hua. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# TRPM7 Induces Mechanistic Target of Rap1b Through the Downregulation of miR-28-5p in Glioma Proliferation and Invasion

Jingwei Wan1,2, Alyssa Aihui Guo3,4, Indrajit Chowdhury <sup>3</sup> , Shanchun Guo<sup>5</sup> , Jacqueline Hibbert <sup>1</sup> , Guangdi Wang<sup>5</sup> and Mingli Liu<sup>1</sup> \*

*<sup>1</sup> Department of Microbiology, Biochemistry and Immunology, Morehouse School of Medicine, Atlanta, GA, United States, <sup>2</sup> Department of Neurosurgery, The Second Xiangya Hospital, Central South University, Changsha, China, <sup>3</sup> Department of Obstetrics and Gynecology, Morehouse School of Medicine, Atlanta, GA, United States, <sup>4</sup> University of South Carolina SOM Greenville, Greenville, SC, United States, <sup>5</sup> Department of Chemistry, Xavier University, New Orleans, LA, United States*

Objectives: Our previous findings demonstrate that channel-kinase transient receptor potential (TRP) ion channel subfamily M, member 7 (TRPM7) is critical in regulating human glioma cell migration and invasion. Since microRNAs (miRNAs) participate in complex regulatory networks that may affect almost every cellular and molecular process during glioma formation and progression, we explored the role of miRNAs in human glioma progression by comparing miRNA expression profiles due to differentially expressed TRPM7.

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, China*

#### Reviewed by:

*Priyanka Gupta, University of Alabama at Birmingham, United States Prasanna Ekambaram, University of Pittsburgh, United States*

> \*Correspondence: *Mingli Liu mliu@msm.edu*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *20 August 2019* Accepted: *28 November 2019* Published: *17 December 2019*

#### Citation:

*Wan J, Guo AA, Chowdhury I, Guo S, Hibbert J, Wang G and Liu M (2019) TRPM7 Induces Mechanistic Target of Rap1b Through the Downregulation of miR-28-5p in Glioma Proliferation and Invasion. Front. Oncol. 9:1413. doi: 10.3389/fonc.2019.01413* Methods: First, we performed miRNA microarray analysis to determine TRPM7's miRNA targets upon TRPM7 silencing in A172 cells and validated the miRNA microarray data using A172, U87MG, U373MG, and SNB19 cell lines by stem-loop RT-qPCRs. We next determined whether TRPM7 regulates glioma cell proliferation and migration/invasion through different functional domains by overexpressing wild-type human TRPM7 (wtTRPM7), two mutants with TRPM7's α-kinase domain deleted (1kinase-DK), or a point mutation in the ATP binding site of the α-kinase domain (K1648R-KR). In addition, we determined the roles of miR-28-5p in glioma cell proliferation and invasion by overexpressing or under expressing miR-28-5p *in vitro*. Lastly, we determined whether a Ras-related small GTP-binding protein (Rap1b) is a target of miR-28-5p in glioma tumorigenesis.

Results: The miRNA microarray data revealed a list of 16 downregulated and 10 upregulated miRNAs whose transcripts are significantly changed by TRPM7 knock-down. Cell invasion was significantly reduced in two TRPM7 mutants with inactive kinase domain, 1kinase, and K1648R transfected glioma cells. miR-28-5p overexpression suppressed glioma cells' proliferation and invasion, and miR-28-5p under expression led to a significant increase in glioma cell proliferation and migration/invasion compared to that of the controls. miR-28-5p suppressed glioma cell proliferation and migration by targeting Rap1b. Co-transfection of siRap1b with miR28-5p inhibitor reduced the glioma cell proliferation and invasion, caused by the latter. Conclusions: These results indicate that TRPM7's channel activity is required for glioma cell growth while the kinase domain is required for cell migration/invasion. TRPM7 regulates miR-28-5p expression, which suppresses cell proliferation and invasion in glioma cells by targeting Rap1b signaling.

Keywords: TRPM7, Rap1b, miR-28-5p, glioma, proliferation, invasion, prognosis

#### INTRODUCTION

High grade malignant gliomas, also called glioblastoma multiforme (GBM), the most common and aggressive primary brain tumor, are devastating, uniformly fatal tumors for which no effective therapies currently exist. We have reported that TRPM7 channels, a subfamily member of the transient receptor potential (TRP), regulate glioma stem cell (GSC) growth/proliferation through STAT3 and Notch signaling pathways (1). Because of traditional treatments' limited success in prolonging the overall survival of GBM patients, treatments that target genomic abnormalities, epigenetics, and epigenetic modulators have been attracting an increasing amount of attention for their influence in many tumors including glioma. miRNAs, one of the epigenetic effectors, reversibly regulate transcription through binding to complementary sequences of mRNA and silencing its translation into proteins. miRNAs are small noncoding RNAs, generally 19– 22 nucleotides in length, that modulate protein-coding genes by binding to the 3'-untraslated region (UTR) of the target mRNA, therefore disrupting transcription (2–4). Many studies have shown that miRNA are expressed in a variety of human tumors and exert dramatic functions in human tumorigenesis and metastasis (5). miRNAs can function as either tumor suppressor by inhibiting the expression of oncogene or tumor promoter by reducing the expression of tumor suppressor gene (6, 7). While a specific miRNA may simultaneously regulate different targets, namely a given miRNA's upstream regulation can involve different regulators at different steps of mRNA biogenesis, a single protein target can be regulated by different miRNAs. Therefore, miRNAs participate in complex regulatory networks that affect almost every cellular and molecular processes in tumor initiation and progression (6). Since miRNAs have significant roles in human tumorigenesis and metastasis (5), identification of aberrantly expressed miRNAs is a crucial initial step in illustrating miRNA-mediated tumorigenic pathways.

Our in-depth data analysis from miRNA microarray data revealed a list of 16 downregulated and 10 upregulated miRNAs whose transcripts are statistically significant with fold changes >2 by TRPM7 knock-down in A172 glioma cells. Among these, the miRNA of hsa-miR-28-5p (miR-28-5p) has been shown to exert crucial influence on tumor growth and migration by modulating AKT (8), ERK (9), and IGF-1 (10) signaling pathways. Although oncogenic and invasive roles of TRPM7 (11, 12) and tumorinhibitory role of miR-28-5p (13) have been reported in multiple studies, how TRPM7 regulates miRNA and downstream target genes are unclear. To seal the gap between TRPM7 and miRNA in the existing literature, we determined the changes in miRNA in response to reduced TRPM7 expression in glioma cells. The aim of the present study was to investigate the functional roles of miR-28-5p to elucidate the molecular mechanism of TRPM7's regulation of miR-28-5p in multiple glioma cell lines with different genomic mutational status. Our data revealed that miR-28-5p was inversely regulated by TRPM7 expression in glioma cells; overexpression of miR-28-5p significantly inhibited glioma cell growth, migration, and invasion; the gene encoding Ras-related protein Rap1b, positively correlated with TRPM7, is a promising candidate target gene of miR-28-5p and may serve as a predictive marker for the poor survival of glioma patients (14, 15). Our study elucidated the crucial roles of TRPM7/miR-28-5p/Rap1b axis in gliomagenesis.

#### MATERIALS AND METHODS

Antibody and reagents: Primary antibodies used included anti-TRPM7 (Abcam, Cat no. ab232455, Cambridge, MA) and antiβ-actin (Sigma-Aldrich, St. Louis, MO, Cat no. A3854). Rabbit polyclonal Rap1b (36E1, Cat no. 2326) and mouse HA (Cat no. 2367) antibodies were purchased from Cell Signaling Technology (Danvers, MA). All secondary antibodies used for Western blot were purchased from Calbiochem (La Jolla, CA).

#### Plasmid, siRNA, and miRNAs

The wild-type human TRPM7 (wtTRPM7) or constructs in which the α-kinase domain was deleted (1kinase) or rendered inactive with a point mutation in the ATP binding site of the αkinase domain (K1648R) were provided by Dr. Carsten Schmitz, University of Colorado, Denver, CO. All constructs (wtTRPM7, 1kinase, K1648R) were tagged with a hemagglutinin (HA) at the N-terminal. Control scrambled siRNA (On-TARGETplus Non-targeting siRNA, Cat no. D-001810-01-05) and ON-TARGETplus SMARTpool siRNA (Cat no. L-005393-000005) targeting TRPM7 were purchased from Dharmacon (Lafayette, CO). Control scramble siRNA (Cat no. sc-37007) and siRNA targeting Rap1b (Cat no. sc-41854) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The scrambled siRNAs, with no homology to any known sequence were used as controls. miR-28-5p mimic (Cat no. 4464066), miR-28-5p inhibitor (Cat no. 4464084), and control were purchased from Life Technology (Carlsbad, CA).

#### Cell Culture

Human glioblastoma cell line, A172, was obtained from ATCC (Manassas, VA, USA). Other glioma cell lines, U87MG, U373MG, and SNB19, were kindly provided by Dr. Yancey G. Gillespie at the University of Alabama at Birmingham (UAB), Birmingham, AL. Human glioblastoma cell line SF767 was kindly provided by Dr. Hui-Kuo Shu at Emory University, Atlanta, Georgia. All cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM, Sigma-Aldrich, St. Louis, MO) plus 10% fetal bovine serum (FBS), 50 units/ml penicillin, and 50µg/ml streptomycin at 37◦C.

# Transfection of siRNA and Expression of Wild-Type Human TRPM7 and Mutant Constructs in Glioma Cells

When the glioblastoma cells reached about 50–75% confluency, the appropriate amount of siRNAs specific to TRPM7, Rap1b, and control scramble siRNA with the final concentration of 100 nM, and miR-28-5p mimics, inhibitor, and controls with final concentration of 30 nM were transfected using Lipofectamine RNAiMAX reagent in serum free OptiMEM-1 medium (Invitrogen, Carlsbad, CA) according to the manufacture's instruction. After six hours of transfection, cells were grown in Dulbecco's Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS) (16, 17) further for 24–72 h as indicated in each experiment. 48 or 72 h post transfection, targets knockdown were assessed by RT-PCR or Western blot, respectively. Various glioma cells at 50–75% confluency were transfected with a pcDNA4/TO plasmid that allowed protein expression of wt hTRPM7 and hTRPM7 mutants for the α-kinase deletion or lacking of phosphotransferase activity by lipofectamine 3000 transfection reagent (Invitrogen, Carlsbad, CA) according to the manufacture's instruction. The transiently transfected glioma cells expressing wt hTRPM7, 1kinase, and K1648R hTRPM7 constructs were maintained in DMEM containing 10% FBS (16, 17) for further growth for 24–72 h. The overexpression of TRPM7 and its mutants were assessed by HA expression. All studies were done in triplicates.

#### MTT Assay

All glioma cells were seeded at 1 × 10<sup>4</sup> cells in 100 µl of medium per well into 96-well plates and were transfected with 100 nM specific siRNA or control using Lipofectamine reagent for indicated times. Ten micorliter of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reagent (Sigma-Aldrich, St. Louis, MO, the ratio of MTT reagent to medium is 1:10) was added into each well and incubated in the dark at 37◦C for 2–4 h. Absorbance at 570 nm was measured using 690 nm as the reference using a CytoFluorTM 2300 plate reader.

#### Cell Migration and Invasion Assay

The migration and invasion potential were assessed as previously described (1, 18). Briefly, cell culture chambers with 8µm pore size of polycarbonate membrane filters (Corning, USA) were used for cell invasion assays with the filters pre-coated with matrigel (50 µl, 1.25 mg/ml). Each of the glioma cell lines were transfected with or without 100 nM siRNA for 48 h, harvested and seeded with 1% FBS medium in the upper chambers that were soaked in bottom chambers filled with 500 µl whole medium (DMEM and 10% FBS). After another 24 h of incubation at 37◦C, matrigel and cells on the upper surface of the filter were wiped off thoroughly with a Q-tip. Cells attached on the lower surface of the membrane filters were fixed with 4% paraformaldehyde/PBS for 10 min and stained with 0.5% crystal violet/methanol for 10 min. The cells were then counted under light microscopy with 10x magnification in 3–4 random fields. Cell numbers under different treatments were normalized to the appropriate controls. Assays were done in triplicate of samples and were performed in two independent experiments.

#### miRNA Microarray

TRPM7-siRNA was employed to selectively suppress the expression of TRPM7 channels in A172 cells. Then, total RNA was extracted using the miRNeasy Mini Kit (Qiagen, Valencia, CA) and was subjected to the human miRNA array assay by LC Sciences (Houston, TX, www.LCsciences.com) using MRA-1001B2 version miRHuman 22, containing 2,632 standard mature miRNA unique probes with 3 repeats, 50 controls (8– 32 repeats) based on Sanger miRBase Release 22 (http://www. mirbase.org/). PUC2PM-20B and PUC2MM-20B are the control probes for quality controls of chip production, sample labeling, and assay conditions. Fold changes and P-values were calculated using Student's t-test. A P < 0.05 with a fold change >2.0 was considered to be a significant dysregulation. In-depth data analysis from miRNA microarray data showed a list of 16 downregulated and 10 upregulated miRNAs whose transcripts are statistically significant with fold changes >2 by TRPM7 knock-down.

# Real-Time RT-PCR Analysis

Total RNA isolation, cDNA synthesis, and PCR amplification were performed as previously described (19). Cell pellets were stored in Trizol reagent and homogenized in fresh Trizol. Total RNA was isolated from cells using a miRNeasy Kit (Qiagen, Valencia, CA) and quantified using the Nanodrop N-1000 by Agilent Biosystems (Santa Clara, CA). Purified total RNA (0.75 µg) was reverse transcribed using iScript cDNA Synthesis Kit according to the manufacture's protocol (Bio-Rad Laboratories, Inc., Hercules, CA). Reverse transcription was performed by using random hexamers at 25◦C for 5 min, 42◦C for 30 min, and 85◦C for 5 min. After diluting 10 times, the cDNA was then amplified using iQ SYBR Green Supermix (Bio-Rad Laboratories, Inc.) according to the manufacture's protocol under the following conditions: activation of the Taq DNA polymerase at 95◦C for 3 min, 40 cycles at 95◦C for 10 s (denaturation), and 61◦C for 45 s (combined annealing and extension). The quantitative gene analysis utilized the CFX Connect Real Time PCR Detection System. Each condition was conducted in biological triplicates, and each individual biological replicate was amplified in technical triplicates. Relative expression for each gene was evaluated using the 2−11Ct Livak method, and GAPDH was used as the reference gene (20). We used the melting curve analysis to assess whether or not the intercalating dye qPCR assays have produced single, specific product. The single peak was observed for each specific gene, which represented as a pure single amplicon, indicating the specificity of each primer for each specific gene.

### Stem-Loop Pulsed Reverse Transcription: A Highly Sensitive RT-PCR Method for the Detection and Quantification of miRNAs

The miRNA validation was performed using stem-loop pulsed RT-PCR with some modifications as described before (21). The RT primer for miR-28-5p reverse transcription, forward and reverse primers for RT product amplification were designed based on miR-28-5p's sequence: AAGGAGCUCACAGUCUAUUGAG (http://www.mirbase. org/). For each reaction, "no RNA" master mix comprised of 10 mM dNTP, 5µM RT primer (see **Table 1**), and appropriate water, was heated at 65◦C for 5 min and incubated on ice for 2 min. Then, the "no RNA" master mix was combined with RT master mix containing first-strand buffer, 0.1M DTT, 4 units RNaseOUT, and 50 units of SuperScript III reverse transcriptase. Then the pulsed RT was performed under the following conditions: load thermal cycler and incubate for 30 min at 16◦C, pulsed RT of 60 cycles at 30◦C for 30 s, 42◦C for 30 s and 50◦C for 1 s, and incubate at 85◦C for 5 min to inactivate the reverse transcriptase. Finally, the RT product was amplified using iQ SYBR Green Supermix (Bio-Rad) as described above.

#### Western Blotting

Cells were lysed with lysis buffer (50 mM HEPES, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 10% glycerol, 1% Nonidet P-40, 100 mM NaF, 10 mM sodium pyrophosphate, 0.2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 10µg/ml aprotinin, and 10µg/ml leupeptin). Samples were separated by SDS/PAGE, and separated proteins were transferred to nitrocellulose membranes and identified by immunoblotting. Primary antibodies were obtained from commercial sources and were diluted at the ratio of 1:1,000 according to manufacturer's instruction. Blots were developed with Supersignal Pico or Femto substrate (Pierce). A densitometric analysis of the bands was performed with the ImageQuant program (Bio-Rad).

#### Bioinformatics Analysis

Kaplan-Meier analysis of overall survival according to the TRPM7, Rap1b mRNA expression and Pearson correlation between TRPM7 and Rap1b were obtained from microarray analysis on 454 glioma patients in the TCGA data set (http:// www.betastasis.com/glioma/tcga\_gbm/). P-value is based on log rank test.

#### Statistical Analysis

The results obtained in this work were expressed as mean ± SD of at least 2 independent experiments done in triplicate. Paired Student t-test or one-way ANOVA tests were performed for data analysis, and significant difference was defined as p < 0.05.

# RESULTS

#### TRPM7 Regulates Glioma Cell Proliferation and Migration/Invasion Through Different Functional Domains

We have reported that the activation of TRPM7 channels plays an important role in the growth and proliferation of human glioma cells (1). In the current study, we further investigated whether or not changes in glioma cell proliferation and migration might be caused by channel domain-mediated and/or kinase domain-mediated TRPM7 activation. To this end, A172 cells were transfected with (a) 5 µg of wild-type human TRPM7 (wtTRPM7 or M7-WT); (b) constructs in which the α-kinase domain was deleted (1kinase or M7-DK) or rendered inactive with a point mutation in the ATP binding site of the αkinase domain (K1648R, or M7-KR); all the cells were allowed to grow from 24 to 72 h as indicated. Note that in vitro autophosphorylation assays have been done to confirm that the phosphotransferase activity of the M7-WT channels are easily detectable and much reduced in the absence of mutant channel activities under standard conditions by Dr. Carsten Schmitz's group who provided us the constructs (17). The effects of TRPM7 on glioma cell proliferation and invasion were determined using MTT assays and transwell invasion assays, respectively. As shown in **Figure 1A**, A172 cells with 1kinase (M7-DK) and K1648R (M7-KR) mutants expressed the TRPM7 protein, as demonstrated by Western blot, in which TRPM7 expression at the lysate level was similar to that of wtTRPM7 expression (**Figure 1A**). The cells overexpressing wt TRPM7 (M7-WT), M7- KR, or M7-DK grew significantly more (P < 0.05) than the cells transfected with control (**Figure 1B**) in a time-dependent manner from 24 to 72 h. Cell proliferation was not significantly changed when cells were transfected with 1kinase or K1648R compared to those transfected with wt TRPM7, which indicates that TRPM7 channel, rather than kinase activity, is required for A172 cell growth (**Figure 1B**). Cell invasion was increased up to 3 times in A172 cells transfected with wt TRPM7 compared to


that of the control; by contrast, the cell numbers that penetrated the Matrigel-coated transwell were significantly reduced in cells transfected with 1kinase (M7-DK) to 30% or with K1648R (M7- KR) to 50% at 72 h compared to control, indicating that TRPM7 kinase activity is required for changes in cell migration and invasion (**Figures 1C,D**). These results suggest that the channel activity of TRPM7 is required for cell growth while the kinase domain is required for cell migration/invasion.

#### TRPM7 Negatively Regulates miR-28-5p in Glioma Cells

#### The Expression Levels of hsa-miR-26b-5p, hsa-miR4530, and hsa-miR-28-5p Were Significantly Changed by TRPM7 Knock-Down

To test whether or not miRNAs are regulated by TRPM7 expression, 100 nM of siRNAs specific to TRPM7 (siTRPM7) or control scramble siRNA (siCtrl) were transfected into A172 cells for 72 h for specifically selective suppression of TRPM7 channel expression. Total RNA from triplicate biological samples, either with scramble control (siCtrl, A423, A424, and A425) or siTRPM7 (A427, A428, and A434) (**Figure 2A**), were extracted and parallelly subjected to human miRNA array assay by LC Sciences (Houston, TX, www.lcsciences. com). In-depth data analysis from miRNA microarray data revealed 16 downregulated and 10 upregulated miRNAs whose transcripts are statistically significant with fold changes >2 by TRPM7 knock-down (**Figure 2A**). Among them, microRNAs hsa-miR-26b-5p (miR-26b-5p), hsa-miR-4530 (miR-4530), and hsa-miR-28-5p (miR-28-5p) have been shown to exert crucial influence on tumor growth and migration by modulating AKT (8), ERK (9), and IGF-1 (10) signaling pathways. In our case, hsa-miR-26b-5p, hsa-miR-4530, and hsa-miR-28-5p were significantly changed by TRPM7 knock-down.

#### Validation of miRNA Array Results by siTRPM7

Next, miR-26b-5p, miR-4530, and miR-28-5p were further validated by real-time miRNA quantification using stem-loop

RT-PCR (21–27) in glioma cell lines A172 and SNB19 (**Figures 2B,C**). A set of endogenous RNAs SNOD48, SNOD44, SNOD47, and U6 served as endogenous controls in real-time miRNA quantification to normalize miRNA gene expression. miR-26b-5p and miR-28-5p were significantly upregulated while miR-4530 was significantly downregulated upon TRPM7 silencing by siTRPM7 in both A172 and SNB19 cells as compared to each of their miRNA endogenous controls (∗P < 0.05, ∗∗P < 0.01, Student t-test). The data indicated that the expressions of hsa-miR-26b-5p, hsa-miR-4530, and hsa-miR-28- 5p in the two cell lines are consistent with those of miRNA microarray data. Since miRNA's discovery decades ago, miR-26- 5p, together with miR-21-5p and miR-30-5p was only detected in breast cancer and may not have significant roles in glioma; while miR-28-5p was found to be important in multiple malignancies including glioma. Therefore, we chose miR-28-5p to further elucidate the TRPM7-mediated pathways in glioma.

#### Validation of miRNA Array Results by Overexpressing TRPM7

The constructs of wt hTRPM7, 1 kinase, and K1648R were introduced into and expressed in glioma cell lines A172, U373MG, SNB19, and U87MG for 48 h. Total RNA and protein lysate were isolated, and TRPM7 levels were detected by qRT-PCR. GAPDH was used as an internal control to monitor the presence of amplified cDNA in samples. Optimum mRNA expression of wt hTRPM7 and the mutants, 1kinase and K1648R, was observed after 48 h. Overexpression of TRPM7 by M7-WT, M7-KR, and M7-DK dramatically increased TRPM7 mRNA as compared to that of the pcDNA4.2 control (**Figure 2D**, upper left, <sup>∗</sup>P < 0.05, ∗∗P < 0.01, Student t-test) in each of four cell lines, indicating the system's high transfection efficiency. Western blots by overexpression of individual TRPM7 constructs for 72 h confirmed the data from qRT-PCR, which showed significantly increased TRPM7 in M7-WT, M7-KR, and M7-DK compared to the control (**Figure 2D**, right). Unlike proteins that showed consistent expression of M7-KR and M7-DK as M7-WT across the cell lines, there are varying expression at mRNA levels of M7- WT and M7-KR and M7-DK. This indicates that transcription levels are frequently not reflected at the protein level due to protein instability and lower rate of mRNA transcription compared to protein translation in mammalian cells (28). The levels of miR-28-5p were determined by stem-loop RT-PCR. The results showed that the overexpression of wt hTRPM7 decreased miR-28-5p expression by 53, 50, 57, and 51% in A172, U373MG, SNB19, and U87MG, respectively; on the other hand, K1648R mutants increased miR-28-5p expression by 1.5, 2.67, 2.74, and 2.45 times. 1 kinase mutants exhibited similarly increased miR-28-5p expression pattern as K1648R mutants did by 1.7, 2.64, 2.78 and 2.1 times, respectively (**Figure 2D**, lower left). These results suggest that wild type TRPM7 may have different functions from its mutants.

#### miR-28-5p Inhibits Glioma Cell Proliferation and Invasion

To understand the biological effect of miR-28-5p on the proliferation of glioma cells, miR-28-5p expression was manipulated by transfecting A172, U87MG, U373MG, and SNB19 cells with either miR-28-5p mimics, inhibitors, or controls (ctrl or miR-NC) for 72 h to increase or decrease miR-28-5p expression levels. The stem-loop RT-PCR assay confirmed that miR-28-5p was significantly increased in A172, U87MG, U373MG, and SNB19 cells that were transfected with miR-28-5p mimics as compared to the expression levels in cells transfected with miR-NC (**Figure 3A**, left). Similarly, miR-28-5p was significantly decreased in the above four glioma cell lines that were transfected with miR-28-5p inhibitors as compared to the expression levels in cells transfected with miR-NC (**Figure 3A**, left). We assessed TRPM7 protein expression levels simultaneously when overexpressing miR-28-5p mimics and inhibitors by Western blot, and we confirmed the negative correlation between TRPM7 and miR-28-5p (**Figure 3A**, right).

We then examined the effects of miR-28-5p on glioma cell growth using MTT assay. As shown in **Figures 3B–E**, the overexpression of miR-28-5p by transient transfection of miR-28-5p mimics dramatically inhibited the growth rate of A172, U87MG, U373MG, and SNB19 glioma cell lines, where cell proliferation decreased in A172 (**Figure 3B**), U87MG (**Figure 3C**), U373MG (**Figure 3D**), and SNB19 (**Figure 3E**) by a maximum of 51.7, 52.5, 48.9, and 53.3% on day 4, respectively. In contrast, reduced miR-28-5p expression by miR-28-5p inhibitor augmented the growth of the four glioma cell lines above, where cell proliferation increased in A172 (**Figure 3B**), U87MG (**Figure 3C**), U373MG (**Figure 3D**), and SNB19 (**Figure 3E**) by a maximum of 32.0, 83.8, 31.1, and 45% on day 4, respectively. Next, we evaluated the possible roles of miR-28-5p in glioma cells invasion. Matrigel transwell invasion assays were performed on U87MG, U373MG, and SNB19 cell lines that were transiently transfected with miR-NC, miR-28-5mimics, and miR-28-5p inhibitors, and the results showed that the ectopic expression of miR-28-5p inhibited the number of invaded cells by 50, 60, and 40%, respectively as compared with that of the controls; on the other hand, miR-28-5p inhibitors enhanced the number of invaded cells by 50, 200, and 80%, respectively (**Figure 3F**).

# TRPM7 Positively Regulated Rap1b Expression in Glioma Cells

The biological roles of miRNAs in human tumors depend on their specific targets; therefore, we searched the literature for potential targets of miR-28-5p. We found 52 publications that related miR-28-5p with human tumors, with most of them serving as a tumor suppressor. Several important tumor-related genes were defined as the direct target for miR-28-5p, such as AKT in gastric cancer (8), N4BP1 in ovarian cancer (29), IL-34 (30) and IGF-1 (10) in hepatocellular carcinoma, and Rap1b in renal cell carcinoma (31). Among these, Rap1b has drawn our attention for the following reasons. First, Rap1b has been found to play a role in the glioma cell proliferation and migration (15, 32). Second, the bioinformatics analysis from TCGA (http://www.betastasis.com/ glioma/tcga\_gbm/) indicated both TRPM7 (**Figure 4A**, upper panel) and Rap1b (**Figure 4A**, lower panel) predict glioma patients' poor prognosis. Third, TRPM7 and Rap1b mRNA expression levels are correlated by Pearson's correlation analysis (r = 0.515) (**Figure 4B**). To confirm this correlation, we transfected A172, U87MG, U373MG, and SNB19 cells either with synthesized specific small interfering RNAs (siTRPM7) targeting TRPM7 mRNA to decreasing TRPM7 expression or with TRPM7 expression vector along with its mutants K1648R and 1 kinase to increase the TRPM7 expression. The transfection efficiency of TRPM7 knock-down and overexpression were examined by qPCR (**Figure 2D**, upper left and **Figure 4D**, upper right) and Western blot (**Figure 4D**, left and lower right). The results indicated that the inhibition of TRPM7 expression by siTRPM7 markedly reduced Rap1b expression at both mRNA (**Figures 4C,D**, upper right) and protein levels (**Figure 4D**, left and lower right); while the overexpression of TRPM7 dramatically enhanced the expression of Rap1b at both mRNA (**Figure 4E**) and protein levels (**Figure 4F**) in all four glioma cell lines. Again, the mutants K1648R and 1 kinase exhibited a noticeable difference in Rap1b expression pattern at the protein levels compared to that of wt TRPM7, which reflect differences in their functional roles (**Figure 4F**).

# Rap1b Elevates Glioma Cell Proliferation and Invasion

We first determined whether or not Rap1b protein, a Rasrelated small GTP-binding protein that acts as GTPase in several signaling cascades, is expressed in glioma cells. As shown by Western blot, all of the glioma cell lines A172, SF767, SNB19, U373MG, and U87MG expressed the Rap1b protein (**Figure 5A**, upper panel). To confirm whether or not Rap1b functions as an oncogene in glioma, we assessed the effects of Rap1b

Student *t*-test). The data were presented as the fold changes relative to the corresponding controls and are mean ± SD of triplicate samples performed in three independent experiments. (F) A Matrigel transwell invasion assay, using U373MG, SNB19, and U87MG cell lines that were transiently transfected with miR-NC, miR-28-5p mimics, and miR-28-5p inhibitors were conducted exactly same as described in Figures 1C,D. The cell numbers were counted on the bottom of Matrigel-coated transwell chamber at 72 h post-transfection. A representative experiment was shown. The ectopic expression of miR-28-5p inhibited the number of invaded cells while inhibition of miR-28-5p expression increased the invasion compared to the controls. The data were represented as fold changes relative to controls are the mean ± SD of triplicate samples performed in two independent samples (\**P* < 0.05, \*\**P* < 0.01, Student *t*-test). Photomicrographs were taken at 10x magnification.

on glioma cell proliferation and invasion in vitro using MTT assay or Matrigel transwell invasion assay after transfecting A172, U87MG, U373MG, and SNB19 with synthesized specific small interfering RNAs (siRap1b) targeting Rap1b mRNA. The efficiency of the transfection was detected by the expression of total Rap1b protein, as examined by Western blot. As shown in **Figure 5A**, Rap1b siRNA markedly exhibited the inhibition of Rap1b expression (**Figure 5A**, lower panel). Next, MTT and cell invasion assay were conducted to examine the effects of Rap1b downregulation on the proliferation and invasion in A172, U87MG, U373MG, and SNB19, respectively. The results showed that transfection with siRap1b resulted in a significant decrease in proliferation in an approximate time-dependent manner when compared with the respective siRNA control groups in the four glioma cell lines; the cell growth viability was reduced by a maximum of 44.4% at 96 h in A172 cells, 37.4% at 96 h in U87MG cells, 46.5% at 72 h in U373MG cells, and 54.3% at 72 h in SNB19 cells (**Figure 5B**). The number of invasive cells crossed the Matrigel transwell was dramatically reduced by 40, 50, 80, and 55% in A172, U87MG, U373MG, and SNB19 cells at 72 h post-transfection, respectively (**Figure 5C**). Taken together, these results suggest that Rap1b increases glioma cell line proliferation and invasion, and promotes glioma tumorigenesis and metastasis.

(*p* = 3.52e−5 for the former, *p* = 0 for the latter, logrank tests). The data in upper panel of (A) showed that glioma patients with high TRPM7 expression (*n* = 165) have short overall survival compared to those with low TRPM7 expression (*n* = 164). The glioma patients with high Rap1b (lower panel) predict the poor prognosis of glioma patients as well (A, lower panel). (B) By Pearson correlation analysis via TCGA (http://www.betastasis.com/glioma/tcga\_gbm/), a clearly direct correlation was observed between TRPM7 and Rap1b expression levels in 356 glioma patients with both genes are detectable (*r* = 0.515). (C–F) To confirm this correlation, we transfected A172, U87MG, U373MG, and SNB19 cells either with synthesized specific small interfering RNAs (siTRPM7) targeting TRPM7 mRNA to decreasing TRPM7 expression or with TRPM7 expression vector along with its mutants 1 kinase and K1648R to increase the TRPM7 expression. The transfection efficiency of TRPM7 knock-down and overexpression were examined by qPCR [see Figure 2D and (D), upper right] and Western blot (D left and lower right). The lower right panel of (D) is the densitometry analysis of the bands performed using the ImageQuant program. The results showed that the inhibition of TRPM7 expression by siTRPM7 markedly reduced Rap1b expression at both mRNA in all four cell lines (C, \**P* < 0.05, Student *t*-test) and protein levels in U87MG and SNB19 cells (D). While overexpression of TRPM7 dramatically enhanced the expression of Rap1b at both mRNA (E, \*\**P* < 0.05 compared to control-Ctrl, #*P* < 0.05 compared to wild type-WT, one-way ANOVA) and protein levels (F) in all four glioma cell lines. The right panel of (F) is the densitometry analysis of the bands performed using the ImageQuant program. Again, the mutants 1 kinase and K1648R exhibited a noticeable difference in Rap1b expression pattern at both mRNA and protein levels compared to that of wt TRPM7, which reflect differences in their functional roles. The data are expressed as the fold changes relative to the corresponding controls and are the mean ± SD of triplicated samples preformed in three independent experiments (C–D). All the data represent one of the three independent experiments with similar results (D–F).

#### miR-28-5p Directly Targets and Downregulates Rap1b Expression in Glioma Cells

#### miR-28-5p Expression Is Inversely Correlated With Rap1b mRNA and Protein Expression in Glioma Cells

To further explore the association between miR-28-5p and Rap1b in glioma cells, we analyzed the expression of the endogenous Rap1b after transiently transfecting cells with either (a) miR-28- 5p mimics, (b) miR-28-5p inhibitor, or (c) control miR-28-5p-NC in A172, U87MG, U373MG, and SNB19 cells. The transient transfection efficiency was determined by qRT-PCR and Western blot. As shown in **Figure 6A**, Rap1b mRNA expression levels were significantly reduced by miR-28-5p mimics and significantly enhanced by miR-28-5p inhibitor in each of the four glioma cell lines. Similarly, the Rap1b protein expressions displayed the same patterns as those of mRNA expressions in U87MG, U373MG, and SNB19 cells (**Figure 6B**). These results confirm that miR-28-5p regulates both the expression levels of Rap1b mRNA and protein.

#### miR-28-5p Inhibits Glioma Cell Proliferation and Invasion by Directly Targeting Rap1b

To further clarify whether or not tumor suppressive roles of miR-28-5p were dependent on Rap1b expression, 100 nM of siRap1b was co-transfected with 30 nM of miR-28-5p inhibitor in A172, U87MG, U373MG, and SNB19 glioma cell lines to assess glioma cells' proliferation by MTT assay and glioma cell invasion by Matrigel transwell invasion assay. As expected, the reduction of Rap1b expression by siRap1b effectively reduced glioma cell proliferation induced by miR-28-5p inhibitor during 24 to 96 h post-transfection by a maximum of 52, 41.8, and 69.2% for A172, U87MG, and U373MG at 96 h, 60% for SNB19 at 48 h (**Figure 6C**). Similarly, siRap1b markedly reduced miR-28-5p inhibitor-induced invasion by 35.6, 42.1, and 39.0% in U87MG, U373MG, and SNB19 cells at 72 h, respectively (**Figure 6D**, <sup>∗</sup>p < 0.05, ∗∗p < 0.01, one-way ANOVA). These results suggest that Rap1b is able to at least partially restore the function of miR-28-5p, and these functional restoration assays support that miR-28-5p functions upstream of Rap1b and is dependent on Rap1b.

densitometry analysis for each Western blot performed using the ImageQuant program are on the right sides correspondingly. (B,C) Rap1b elevates glioma cell proliferation and invasion. MTT and cell invasion assay were conducted to examine the effects of Rap1b on the proliferation and invasion after transfecting A172, U87MG, U373MG, and SNB19 with synthesized specific small interfering RNAs (siRap1b) targeting Rap1b mRNA. The results showed that transfection with siRap1b resulted in a significant decrease in proliferation (B) and invasion (C, Photomicrographs were taken at 10x magnification) compared with the respective siRNA control groups in the four glioma cell lines. These results suggested that Rap1b increases glioma cell line proliferation and invasion, and Rap1b may perform tumor promoting roles in glioma growth and metastasis [\**P* < 0.05, one-way ANOVA for proliferation assay in (B), Student *t*-test for invasion assay in C]. All data are from triplicate samples performed in two different independent experiments.

#### DISCUSSION

In the current study, we utilized four glioma cell lines, A172, U87MG, U373MG, and SNB19, based on their different molecular characteristics. U373MG has a p53 gene mutation (33), while A172, U87MG (33), and SNB19 are PTEN-mutant cell lines (34, 35). In addition to harboring PTEN and CDKN2A (p16INK4a) mutation as in A172, U87MG has another CDKN2C (p18INK4c) mutation (36). Although U373MG cells may share common origins with SNB19, these two cell lines appear to have evolved to exhibit distinct karyotypes and drug sensitivities (37); thus, they serve as culture models with different molecular background for comparison in this study.

TRPM7 channel is nonselectively permeable to many divalent cations such as Ca2+, Mg2+, and Zn2+, which comprises the TRPM7-mediated inward current. In addition, it has the unique characteristic of α-kinase domain at its carboxylterminal (17). The kinase domain of TRPM7 is not essential for the activation of its channel, but structural changes in the kinase domain could alter the sensitivity of channel activation to divalent cation (17). TRPM7-dependent divalent cations, growth factors (GFs), hypoxia and/or other signals can open TRPM7 channel and increase cell proliferation. In this regard, the major activated signaling pathways activated include PI3K/AKT/mTOR (38–40), Src/MAPKs/JNK, p38, and ERK (41–45). Depleted TRPM7 resulted in the inactivation of STAT3 and the disruption of proliferation of thymocytes (46). On the other hand, the opening of TRPM7 channels causes a local increases in Ca2+/Mg2<sup>+</sup> concentration, which affects the recruitment/targeting of TRPM7 kinase substrates (16). TRPM7-kinase phosphorylates downstream eEF2-k on Ser78 in human (Ser77 in mouse eEF2-k), causing increased stability of eEF2-k, strengthening inhibitory eEF2 at Thr56 phosphorylation and resulting in an increased translational efficiency when Mg2<sup>+</sup> is available (16). The channel kinase of TRPM7 participates in the regulation of cell migration by modulating phosphorylation of annexin 1 (47–50), myosin IIA (51–53) and calpain (54) in response to various stimuli in tumor cells. In the present study, we evaluated TRPM7 channels' properties using a deletion mutant of human TRPM7 in which the entire kinase domain was removed following amino acid 1569 (TRPM7 1kinase or M7-DK) and a mutant with a point mutation in the ATP binding site of the αkinase domain (K1648R, or M7-KR). We found that the channel activity of TRPM7 is required for cell growth while the kinase domain is required for cell migration/invasion. This finding is important for designing and developing drugs targeting TRPM7 in human malignancies.

Among 2,588 mature and 1,881 precursor human miRNA sequences in 12 miRNA databases, 365 miRNAs were found to be deregulated in glioblastoma (55). The genes targeted by the deregulated miRNAs in glioblastoma include many pathways such as cell growth/proliferation, apoptosis, invasion and metastasis, angiogenesis, autophagy, and drug resistance (56). miR-28-5p was found to interfere with genes involved in cell replication and cell cycle checkpoints (57). In glioblastoma, the role of miR-28-5p has not yet been fully explored, but its structurally related miR-708 was demonstrated to inhibit glioblastoma cell proliferation by targeting EZH2, AKT1, MMP2, CCND1, Parp-1, and Bcl-2 (58). Our results suggest that miRNA changes demonstrate different glioma tumorigenicity, and miR-26b-5p, miR-4530, and miR-28-5p are regulated by TRPM7 and participate in glioma progression. In the present study, we further elucidated the TRPM7/miR-28-5p/Rap1b axis in gliomagenesis using multiple cell lines with different mutational status. Our results showed that miR-28-5p inhibits glioma cell proliferation and invasion by targeting its downstream Rap1b gene, and miR-28-5p is regulated by TRPM7.

Functional analyses of miR-28-5p revealed tumor suppressive properties caused by the inhibition of Rap1b, E2F6, IGF-1, IL-34, and AKT. A study reported that miR-28-5p inhibited cell proliferation and migration by directly suppressing Rap1b gene in renal cell carcinoma (31). In prostate cancer, the miR-28-5p targeted E2F6 and induced apoptosis in DU-145 cells (7). The ectopic miR-28-5p expression downregulates insulinlike growth factor 1 (IGF1) protein and the expression of miR-28-5p correlates negatively with IGF1 protein level in hepatocellular carcinoma (HCC) cells, which indicate that miR-28-5p-IGF1-PI3K/AKT pathway may play an important role in the development of HCC (10). IL-34 is a direct downstream target of miR-28-5p whose expression is inversely correlated with IL-34 expression (30). Lastly, miR-28-5p may act as a tumor suppressor gene, which inhibited the invasion and metastasis of gastric cancer by inhibiting the activation of the AKT signaling pathway (8). By contrast, miR-28 also served as a tumor promoting factor in some specific tumors. For instance, it is a thrombopoietin receptor (TpoR, MPL) targeting miRNA, which is overexpressed in the platelets of patients with myeloproliferative neoplasms (MPNs) and its negative role in megakaryocyte differentiation leads to MPNs through the downregulation of MPL (59). In the process of ovarian cancer development and progression, miR-28-5p downregulates N4BP1, forces cancer cells to enter S phase, and promotes ovarian cancer cell proliferation and invasion (29). Taken together, miR-28-5p can function as either tumor suppressor genes or oncogenes, reflecting cell type-specific and tissue-specific roles of this miRNA.

miRNAs are involved in diverse biological processes by binding to different regions of the target mRNA sequences, such as the 3′ -untranslated regions, coding sequences, or 5′ untranslated regions. The same hairpin RNA structure can produce mature miRNAs from each strand, named 5p and 3p, that can bind to different mRNAs (60). Almeida et al. (60) found that strand-specific 5p and 3p of miR-28-5 have distinct functions in colorectal cancer cells, where miR-28-5p suppressed cell proliferation, resulting in apoptosis and G1 arrest in the cell cycle, while miR-28-3p had no influence on proliferation in vitro. In addition, miR-28-5p and miR-28-3p had opposite effects on migration and invasion in vitro, with miR-28-5p decreasing cell migration and miR-28- 3p increasing cell migration, which appears to be independent of cell growth. Overall, miR-28-3p overrides miR-28-5p to promote colorectal cancer cell metastases in vivo. The contrasting biological effects caused by miR-28-5p and miR-28-3p might be partly due to their binding to different targets where miR-28-5p targets CCND1 and HoxB3 while miR-28-3p targets Nm23-H1. In HCC, Kaplan-Meier's analysis showed that HCC patients with miR-28-5p overexpression, but not with miR-28- 3p overexpression, had better overall survival rate. Girardot et al. (59) identified two miRNAs that are closely related to miR-28. miR-151 and miR-28 share 80% sequence identity, while miR-708 share 68% sequence identity with miR-28 and 71% with miR-151. All three miRNAs targeted the MPL 3′ UTR for translational inhibition. However, the combination of miR-28 with miR-151 and miR-708 did not produce synergic inhibition due to all targeting the same sequence in the 3′UTR of MPL.

The normal physiological role of the small GTP-binding protein Rap1 is to antagonize Ras mitogenic signals. However, in 1998, Altschuler et al. first revealed that Rap1 is a conditional oncoprotein (61). Later, Rap1 was demonstrated to be required for pancreatic and prostate cancer cell metastasis and angiogenesis but not for the proliferation properties of these cancer cells (62). In 2014, Sayyah's group published the first study showing that this small G-protein is required for glioblastoma cell growth in vitro by activating downstream signaling of G-protein-coupled receptor and RhoA utilizing Rap1 knockdown techniques. However, they stated that this critical role was caused by the isoform Rap1a, not Rap1b (32). Rap1a and Rap1b are 95% homologous, but differences in subcellular localization of two isoforms and their mechanisms of activation resulted in their distinctive functions and roles in glioblastoma cell proliferation (63, 64). In this study, we found Rap1b demonstrated the mitogenic roles in both glioma cell proliferation and invasion.

A positive correlation has been shown between TRPM7 and Rap1b in **Figure 4A**. There are several possible mechanistic links between the two proteins. First, they may connect to each other by miR-28-5p. In this study, we showed that TRPM7 downregulated miR-28-5p, while miR-28-5p downregulated Rap1b; this is likely one mechanism of a positive correlation between TRPM7 and miR-28-5p. Second, TRPM7 and Rap1b may be connected by Notch signaling. In hematopoietic stem cell development, Integrin-mediated cell adhesion was promoted by Notch-regulated Rap1b signaling (65); while our published data also demonstrated that TRPM7 regulated the Notch pathway in gliomagenesis (1).

Taken together, the discovery of cellular and molecular targets modulated by the tumor suppressor miR-28-5p and its upstream regulator TRPM7 molecule provides key insights into the potential mechanisms of glioma tumorigenesis and suggests novel therapeutic targets for glioma treatments.

# DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

# AUTHOR CONTRIBUTIONS

SG, GW, and ML designed the study protocol. JW and AG performed experiments based on glioma cell cultures and evaluated the data with the help of ML. ML performed the biostatistical evaluation of the data. AG and ML wrote the manuscript with contributions and final approval by all authors. IC, SG, JH, and GW contributed to the critical reading and revision of the manuscript.

# FUNDING

This study was supported by NIH NIGMS GM121230 to ML. GW was supported by NIMHD grant U54MD007595. This study was partly supported by NIH/NINHD MD007589. The funding body had no role in the design, collection, analysis and interpretation of the study's data, and in writing the manuscript.

# ACKNOWLEDGMENTS

We thank Dr. Carsten Schmitz (University of Colorado, Denver, CO) for providing us with all constructs (wtTRPM7, 1kinase, K1648R) that were tagged with a HA at N-terminal. We are grateful to Dr. Yancey G. Gillespie at the University of Alabama at Birmingham (UAB) for providing us U87MG, U373MG, and SNB19 cell lines, to Dr. Hui-Kuo Shu at Emory University for providing SF767 cell line.

# REFERENCES


atrophy and modulates extracellular signal-regulated kinase activity. AJP Cell Physiology. (2019) 316:C567–81. doi: 10.1152/ajpcell.00234.2018


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wan, Guo, Chowdhury, Guo, Hibbert, Wang and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# BJ-B11, an Hsp90 Inhibitor, Constrains the Proliferation and Invasion of Breast Cancer Cells

Kaisheng Liu<sup>1</sup> , Juan Chen<sup>2</sup> , Fang Yang<sup>1</sup> , Zhifan Zhou<sup>1</sup> , Ying Liu<sup>1</sup> , Yaomin Guo<sup>1</sup> , Hong Hu<sup>1</sup> , Hengyuan Gao<sup>1</sup> , Haili Li <sup>1</sup> , Wenbin Zhou<sup>1</sup> \*, Bo Qin<sup>2</sup> \* and Yifei Wang<sup>3</sup> \*

*<sup>1</sup> Shenzhen People's Hospital, The First Affiliated Hospital of Southern University of Science and Technology, The Second Clinical Medical College of Jinan University, Shenzhen, China, <sup>2</sup> Shenzhen Nanshan District Shekou People's Hospital, Shenzhen, China, <sup>3</sup> Institute of Biomedicine, College of Life Science and Technology, Jinan University, Guangzhou, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Haitao Mao, UT Southwestern Medical Center, United States Tao Hu, University of Maryland, Baltimore, United States*

#### \*Correspondence:

*Wenbin Zhou zhouwb1016@163.com Bo Qin qinbozf@163.com Yifei Wang twang-yf@163.com*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *03 September 2019* Accepted: *03 December 2019* Published: *18 December 2019*

#### Citation:

*Liu K, Chen J, Yang F, Zhou Z, Liu Y, Guo Y, Hu H, Gao H, Li H, Zhou W, Qin B and Wang Y (2019) BJ-B11, an Hsp90 Inhibitor, Constrains the Proliferation and Invasion of Breast Cancer Cells. Front. Oncol. 9:1447. doi: 10.3389/fonc.2019.01447* Breast cancer is the leading cause of cancer-related deaths in women; however, its underlying etiology remains largely unknown. In this study, we systematically analyzed breast cancer tissues using comprehensive iTRAQ labeled quantitative proteomics, identifying 841 differentially expressed proteins (474 and 367 significantly over- and under-expressed, respectively), which were annotated by protein domain analysis. All the heat shock proteins identified were upregulated in breast cancer tissues; Hsp90 upregulation was also validated by RT-qPCR and immunohistochemistry, and high Hsp90 protein levels correlated with poorer survival. Hsp90AA1 overexpression promoted MDA-MB-231 cell proliferation, whilst BJ-B11, an Hsp90 inhibitor, hampered their invasion, migration, and proliferation in a time and dose-dependent manner and induced cell cycle arrest and apoptosis. BJ-B11 inhibited the expression of epithelial-mesenchymal transition (EMT) marker in MDA-MB-231 cells, whereas Hsp90AA1 promoted its expression. Moreover, BJ-B11 inhibited tumor growth in xenograft model. Altogether, Hsp90 activation is a risk factor in breast cancer patients, and BJ-B11 could be used to treat breast cancer.

Keywords: breast cancer, Hsp90, BJ-B11, proliferation, invasion, migration, EMT

# INTRODUCTION

Breast cancer is the most frequently diagnosed cancer in women (1), and its incidence has increased in most developing countries over the past few decades (1, 2). Although patients with breast cancer have a high 5-year survival rate following treatment, the survival rate decreases rapidly for patients with more advanced disease (3–5). Triple-negative breast cancers (TNBCs) account for 15% of all breast cancers and lack estrogen, progesterone, and ERBB2 receptor expression (6). TNBCs are poorly differentiated, and there are no specific treatment guidelines for this breast cancer subgroup (7, 8); therefore, biomarkers and more effective medical therapies are urgently required.

Breast cancer is coordinately controlled by regulatory networks; thus, understanding these networks could help identify candidates for the diagnosis, prediction, and therapy of breast cancer. Proteomics approaches are often used to acquire a comprehensive and quantitative profile of protein expression. Isobaric tags for relative and absolute quantitation (iTRAQ) is a novel and unbiased approach to simultaneously quantify relative protein abundance (9); in particular, the method enables protein quantification during various developmental stages (10).

Hsp90 is highly expressed in various cancers (11). It is responsible for the stability and function of client proteins, including Akt, IKKα, B-Raf, and GSK3β, which are critical for cell survival and proliferation (12). Therefore, Hsp90 is a potential therapeutic target and diagnostic marker for cancer (13, 14). BJ-B11 is a novel Hsp90 inhibitor that reportedly exhibits antitumor activity in myeloid leukemia and esophageal carcinoma (15, 16); however, its antitumor activity in breast cancer has not yet been investigated.

In this study, we investigated whether Hsp90 was associated with breast cancer and whether BJ-B11 affected the functions of breast cancer cells. Our findings suggested that Hsp90 could be a candidate for the early diagnosis, prognosis, and therapy of breast cancer and that BJ-B11 could be used to treat breast cancer.

#### MATERIALS AND METHODS

#### Primary Breast Cancer Samples

Tumor tissue and adjacent normal tissue samples were collected at the Department of Thyroid and Breast Surgery, Shenzhen People's Hospital, with the informed consent of the patients. The study (LL-KT-2015002) was approved by the Ethics Committee of the hospital. The clinicopathological information regarding the samples is detailed in the **Table S1**. Tumor tissues and normal tissues (10 mg) were homogenized for proteomics and iTRAQ labeling followed by LC-MS/MS analysis. The protein with iTRAQ ratio (tumor tissue/normal tissue) < 0.83 or > 1.2 (P < 0.05) was considered to be significantly differentially expressed.

### Cell Culture and Reagents

The human breast cancer cell line MDA-MB-231 was obtained from the American Type Culture Collection (Manassas, VA, USA) and cultured in DMEM/F12 supplemented with 10% FBS, 100µg/mL streptomycin, and 100 unit/mL penicillin in a humidified incubator in a 5% CO<sup>2</sup> atmosphere at 37◦C.

BJ-B11 was prepared in our lab, as previously described (17), and the 10 mmol/L BJ-B11 stock solution in DMSO was stored at 4◦C. Plasmids expressing wild-type Hsp90AA1 were provided by SAGENE (Guangzhou, Guangdong, China). Mouse anti-E-cadherin (cat: 14472), rabbit anti-vimentin (cat: 3932), and mouse anti-β-actin (cat: 3700) antibodies were purchased from CST (MA, USA).

#### Cell Viability and Apoptosis Assay

CCK-8 (Dojindo, Japan) was used to detect cell viability. Cell apoptosis induced by BJ-B11 was determined using AnnexinV/PI (KeyGEN, Nanjing, China) staining, followed by flow cytometry (Beckman Coulter, CA, USA) according to the manufacturer's instructions.

# Cell Cycle Analysis

Cells were treated with BJ-B11 for 48 h, harvested in cold PBS, fixed in 70% ethanol, and stored overnight at 4◦C. The cells were then washed twice with cold PBS, resuspended in 50µg/mL PI staining reagent containing 100µg/mL RNase and 0.1% Triton X-100 for 30 min in the dark, and analyzed by flow cytometry (Becton-Dickinson, CA, USA).

#### Real-Time Quantitative Polymerase Chain Reaction (RT-qPCR)

Total RNA was extracted using TRIZOL (Thermo Fisher Scientific) and subjected to qRT-PCR using the primers shown in **Table S2**). Gene expression was normalized against GAPDH using the relative <sup>11</sup>CT method and is reported as relative expression compared to the control.

# Cell Invasion Assay

A total of 2 × 10<sup>4</sup> MDA-MB-231 cells treated with or without BJ-B11 were added to Transwell inserts and cultured in an incubator for 16 h. Cells inside the insert were cleaned thoroughly with a cotton swab, while those on the underside were fixed in 4% paraformaldehyde for 5 min and stained with 0.5% crystal violet solution. At least five random fields were counted per insert, and each group consisted of three replicates.

#### Tissue Microarray

Human breast tissue (HBreD077Su01, Shanhai Xinchao, China) and breast cancer tissue (HBreD140Su05, Shanhai Xinchao, China) microarrays consisting of 77 adjacent non-malignant tissue samples and 140 breast cancer tissue samples, were stained with rabbit anti Hsp90 (4874, CST, USA). Immunohistochemical staining was carried out according to the manufacturer's instructions. Slides were evaluated for their positive staining rate (0, negative; 1, 1–25%; 2, 26–50%; 3, 51–75%; and 4, 76– 100%) and the staining intensity of the positively stained cells (0, none; 1, weak; 2, moderate; and 3, strong). Samples were grouped according to the H score, which was the product of the "staining intensity" and "staining positive rate" scores: low expression group, < 8; and high expression group, ≥ 8. Two investigators evaluated each tissue section independently.

#### The Cancer Genome Atlas (TCGA) Data Analysis

The expression level and survival of Hsp90AA1 and Hsp90AB1 in breast cancer were analyzed using the UALCAN platform.

# Xenograft Model

Female BALB/c Nude Mice (6-week-old) were obtained from the Guangdong Medical Laboratory Animal Center. They were maintained in an air-conditioned room with controlled temperature of 21 ± 2 ◦C, and humidity of 30–70% in a 12 h light/darkness cycle regulation and were fed laboratory chow and water ad libitum. All animal experiments were approved by the Animal Ethics Committee of Shenzhen People's Hospital (No. LL-KY-2019512). The athymic female nude mice were injected with about 5 × 10<sup>6</sup> MDA-MB-231 cells subcutaneously. When the tumors were measurable, the mice were randomly assigned

**Abbreviations:** Hsp90, heat shock protein 90; TNBC, triple-negative breast cancer; iTRAQ, isobaric tags for relative and absolute quantitation; GO, Gene Ontology; RT-qPCR, real-time quantitative polymerase chain reaction; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; EMT, epithelial-mesenchymal transition.

into treatment group receiving oral BJ-B11 (20 mg/kg, daily for a total of 27 days) or the control group receiving oral vehicle alone. All mice were euthanized at day 28, and the tumors were excised and weighed to evaluate tumor growth inhibition.

#### Statistical Analysis

Data are presented as the mean ± SD, and Student's unpaired ttests were used for the statistical analysis of differences between two groups. Differences between groups were analyzed using Prism 6 (GraphPad, Inc., San Diego, CA). P < 0.05 were considered statistically significant.

# RESULTS

# Identification of Hsp90 as a Diagnostic Marker

We screened proteins that were differentially expressed between tumor tissues and adjacent normal tissues using iTRAQ with ratio (tumor:adjacent normal tissue) thresholds of >1.2 or <0.83, which indicated higher or lower protein expression in tumor tissue than in adjacent normal tissue, respectively. A total of 841 differentially expressed proteins were identified, among which 474 and 367 were up- and down-regulated in breast cancer tissues (**Figure 1A**).

To understand the functions of these differentially expressed proteins, they were classified according to their protein domains (P < 0.001; **Figure 1B**). Heat shock proteins showed the most significant change of all the upregulated domains, with Hsp90AA1, Hsp90AB1, TRAP1, HspA5, HspB1, HspE1, HspD1, HspA1B, HspA8, HspA9, and HspA4 identified in breast cancer tissue (**Table S3**). RT-qPCR revealed that Hsp90AA1 and Hsp90AB1 mRNA levels increased in breast cancer (**Figure 1C**), consistent with TCGA data (**Figure 1D**). These results suggest that the highly expressed Hsp90 could act as a diagnostic marker for breast cancer.

# Identification of Hsp90 as a Prognostic Marker

We then performed tissue microarray analysis, finding that Hsp90 protein levels increased in breast cancer tissue (P = 0.014; **Figure 2A** and **Table S4**). Moreover, Hsp90 expression was significantly correlated with clinical tumor grade (**Table S5**); therefore, we analyzed the survival rate of patients with breast cancer. High Hsp90 levels predicted poor survival (**Figure 2B**), consistent with TCGA data (**Figure S1**). Thus, the univariate and multivariate analyses suggested that Hsp90 could serve as a prognostic indicator for breast cancer (**Table S6**).

(D) Hsp90AA1 and Hsp90AB1 mRNA levels in healthy and breast cancer tissues from TCGA database (\*\**P* < 0.01).

### BJ-B11 Induces Apoptosis and Cell Cycle Arrest in Breast Cancer Cells

Since our findings suggested that Hsp90 was involved in breast cancer, we examined whether manipulating the Hsp90 gene affected the proliferation of breast cancer cells. Hsp90AA1 overexpression increased MDA-MB-231 cell proliferation (**Figure 3A**), whereas treating the cells with BJ-B11, an Hsp90 inhibitor, for 24, 48, or 72 h inhibited their growth in a time- and dose-dependent manner (**Figure 3B**) and induced dose-dependent apoptosis (**Figure 3C**) and G2/M cell cycle arrest (**Figure 3D**). Taken together, Hsp90AA1 promoted the proliferation of breast cancer cells, whilst BJ-B11 could be used to treat breast cancer.

#### BJ-B11 Inhibited Invasion and Migration of Breast Cancer Cells

MDA-MB-231 cells were cultured with or without BJ-B11 in Transwell inserts for 16 h. BJ-B11 inhibited the invasion of the MDA-MB-231 cells (**Figure 4A**) and inhibited Hsp90, thus significantly suppressing MDA-MB-231 cell migration (**Figure 4B**). We also analyzed epithelial-mesenchymal transition (EMT) markers related to invasion and migration, finding that BJ-B11 upregulated E-cadherin and downregulated vimentin (**Figure 4C**). Conversely, Hsp90AA1 overexpression upregulated vimentin and downregulated E-cadherin and occludin (**Figure 4D**), suggesting that Hsp90 plays a vital role in EMT in breast cancer. Taken together, these data strongly suggest that BJ-B11 inhibits cell invasion and migration by affecting EMT in breast cancer.

# BJ-B11 Inhibited Tumor Growth in vivo

We further tested whether BJ-B11 could suppress cell growth of breast cancer in vivo. First, we established xenograft models by subcutaneous injection of MDA-MB-231 cells into the right flanks of mice. We then tested the anti-tumor effects of BJ-B11 on MDA-MB-231 cancer cells. Nude mice bearing MDA-MB-231 tumor xenografts were treated with 20 mg/kg BJ-B11 (n = 7) or physiological saline (n = 7) for 27 days. Bodyweight was measured each day before the administration of BJ-B11. Results showed that BJ-B11 inhibited tumor growth significantly in vivo (**Figures 5A,B**).

# DISCUSSION

This study investigated breast cancer proteome using iTRAQ to obtain a global view and identify therapeutic targets for breast cancer. We identified 841 differentially expressed proteins and showed that heat shock proteins could be candidate biomarkers for the early diagnosis and therapy of breast cancer. Moreover, this is the first study to investigate the antitumor activity of BJ-B11 in TNBCs, showing that BJ-B11 could be used to treat breast cancer.

Multiple heat shock proteins, including Hsp90AA1, Hsp90AB1, TRAP1, HspA5, HspB1, HspE1, HspD1, HspA1B, HspA8, HspA9, and HspA4, were significantly upregulated in breast cancer tissue. Hsp70, HspA8, HspA9, HspA5, and Hsp110s may constitute up to 3% of the total protein in unstressed human cells (18). Moreover, Hsp70 is upregulated in various human cancers and associated with tumorigenesis (19, 20). HspA8 is overexpressed in cancer cells and it belongs to the Hsp70 family (21). Its depletion in RL-95-2 and HEC-1B cells was shown to suppress cell growth and promote apoptosis, suggesting that HspA8 could be a candidate biomarker for endometrial carcinoma (22).

Hsp90 is an ATP-dependent molecular chaperone (23). Reportedly, Hsp90 expression levels are associated with disease progression and survival in melanoma, gastrointestinal stromal tumors and non-small cell lung cancer (24, 25). Moreover, high Hsp90 expression has been associated with decreased survival in breast cancer (26) whilst its inhibition can suppress growth and promote apoptosis in breast cancer cells, suggesting that Hsp90 could act as both a biomarker and a therapeutic target for breast cancer (27). A phase II study of 17-AAG in breast cancer showed that Hsp90 inhibitors exhibit significant anticancer activity (28). Previously, we demonstrated that the Hsp90 inhibitor SNX-2112 suppressed MCF-7 cell proliferation and induced apoptosis (27), whilst the Hsp90 inhibitor PU-H71 has been shown to induce a complete response in TNBC models (29). BJ-B11 is a novel Hsp90 inhibitor that can inhibit cancer cell proliferation and exhibits anti-HSV activity (15, 16). To the best of our knowledge, this is the first study to investigate the antitumor activity of this molecule in TNBCs; BJ-B11 inhibited the proliferation, invasion, and migration of breast cancer cells, which may be associated with EMT. In addition, BJ-B11 showed significant antitumor activity in vivo. However, the underlying mechanism requires further clarification.

In summary, our study provides new insights into the molecular changes that occur in breast cancer. Hsp90AA1, Hsp90AB1, TRAP1, HspA5, HspB1, HspE1, HspD1, HspA1B, HspA8, HspA9, and HspA4 were validated as potential biomarkers for breast cancer tissue. In particular, Hsp90 plays important roles in breast cancer development and could be a

candidate biomarker for the early diagnosis and therapy of breast cancer. Moreover, BJ-B11 could open new, potential therapeutic alternatives for breast cancer.

# DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

# ETHICS STATEMENT

The studies involving human participants were reviewed and approved by Ethics Committee of Shenzhen People's Hospital. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Ethics Committee of Shenzhen People's Hospital.

#### AUTHOR CONTRIBUTIONS

KL conceived, designed the experiments, and wrote the manuscript. KL, YL, and YG performed the experiments and analyzed the data. KL, JC, FY, ZZ, HH, HG, HL, WZ, BQ, and YW contributed reagents, materials, and analysis tools.

#### FUNDING

This research was supported by the National Natural Science Foundation of China (81802749), the Science and Technology Foundation of Shenzhen (JCYJ20180301170047864, JCYJ20180228164300106), and the Cultivating Fund Project of Shenzhen People's Hospital (No. SYLY201704).

#### REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2019.01447/full#supplementary-material

Figure S1 | Cumulative survival analysis of breast cancer patients with low or high Hsp90AA1 (A) or Hsp90AB1 (B) expression.

Table S1 | Sample information.

Table S2 | Primers used in this study.

Table S3 | Hsp family members upregulated in breast cancer tissue.

Table S4 | Differential Hsp90 expression in healthy and breast cancer tissues.

Table S5 | Correlation between Hsp90 expression and clinicopathological characteristics.

Table S6 | Univariate and multivariate analyses of factors correlated with the overall survival of breast cancer patients.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Liu, Chen, Yang, Zhou, Liu, Guo, Hu, Gao, Li, Zhou, Qin and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Erastin Reverses ABCB1-Mediated Docetaxel Resistance in Ovarian Cancer

Hai-Hong Zhou1†, Xu Chen1†, Lu-Ya Cai 1†, Xing-Wei Nan<sup>1</sup> , Jia-Hua Chen<sup>1</sup> , Xiu-Xiu Chen<sup>1</sup> , Yang Yang<sup>2</sup> , Zi-Hao Xing<sup>2</sup> , Meng-Ning Wei <sup>2</sup> , Yao Li <sup>2</sup> , Sheng-Te Wang<sup>2</sup> , Kun Liu<sup>2</sup> , Zhi Shi <sup>2</sup> \* and Xiao-Jian Yan1,3 \*

*<sup>1</sup> Department of Gynecology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China, <sup>2</sup> Guangdong Provincial Key Laboratory of Bioengineering Medicine, Department of Cell Biology & Institute of Biomedicine, National Engineering Research Center of Genetic Medicine, College of Life Science and Technology, Jinan University, Guangzhou, China, <sup>3</sup> Center for Uterine Cancer Diagnosis & Therapy Research of Zhejiang Province, Women's Hospital and Institute of Translation Medicine, Zhejiang University School of Medicine, Hangzhou, China*

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, China*

#### Reviewed by:

*Junjian Wang, Sun Yat-sen University, China Chaochu Cui, Xinxiang Medical University, China*

#### \*Correspondence:

*Zhi Shi tshizhi@jnu.edu.cn Xiao-Jian Yan yxjbetter@126.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *01 November 2019* Accepted: *26 November 2019* Published: *19 December 2019*

#### Citation:

*Zhou H-H, Chen X, Cai L-Y, Nan X-W, Chen J-H, Chen X-X, Yang Y, Xing Z-H, Wei M-N, Li Y, Wang S-T, Liu K, Shi Z and Yan X-J (2019) Erastin Reverses ABCB1-Mediated Docetaxel Resistance in Ovarian Cancer. Front. Oncol. 9:1398. doi: 10.3389/fonc.2019.01398* Overexpression of drug efflux transport ABCB1 is correlated with multidrug resistance (MDR) among cancer cells. Upregulation of ABCB1 accounts for the recurrence of resistance to docetaxel therapy in ovarian cancer with poor survival. Erastin is a novel and specific small molecule that targets SLC7A11 to induce ferroptosis. In the present research, we explored the synergistic effect of erastin and docetaxel in ovarian cancer. We confirmed that the co-delivery of erastin with docetaxel significantly decreased cell viability, promoted cell apoptosis, and induced cell cycle arrest at G2/M in ovarian cancer cells with ABCB1 overexpression. Mechanistically, erastin dominantly elevated the intracellular ABCB1 substrate levels by restricting the drug-efflux activity of ABCB1 without alteration of the expression of ABCB1. Consequently, erastin can reverse ABCB1-mediated docetaxel resistance in ovarian cancer, revealing that the combination of erastin and docetaxel may potentially offer an effective administration for chemo-resistant patients suffering from ovarian cancers.

Keywords: erastin, docetaxel, ABCB1, ovarian cancer, ferroptosis

# INTRODUCTION

Ovarian cancer threatens women's health with high morbidity and mortality and is the leading cause of death in gynecological malignancies (1). The combination of aggressive cytoreductive surgery followed by system chemotherapy is an effective way of treatment (2, 3). Despite how ∼80% of patients respond well to such therapies, ∼25% of women faced resistant cancer recurrence within 6 months (4). Multidrug resistance (MDR) in ovarian cancer is recognized as the primary cause of failing chemotherapeutic treatments and low survival rates in humans (5). ABCB1 (P-glycoprotein/MDR1), a glycosylated 170-kDa transmembrane protein encoded by the MDR1 gene (6), has emerged as a central drug transport and the best-studied drug transporter in MDR (7). ABCB1 overexpression in cancers contributes to reduced intracellular chemotherapeutics accumulation and brings about resistance against a wide variety of the recently available antineoplastic agents like taxanes (docetaxel), vinca alkaloids (vinblastine), and anthracyclines (doxorubicin) (8–11). Upregulated ABCB1 has been confirmed as the primary protein resulting in MDR in ovarian cancer treated with paclitaxel and related taxane drugs (12–15). Inhibition of ABCB1, therefore, will restore the sensitivity of ABCB1-substrate chemotherapeutic agents, such as docetaxel and doxorubicin. However, drugs under tests that are aimed at ABCB1 to reserve MDR are far from satisfactory for clinical use.

Docetaxel, a new member of the taxane family, has been widely applied in ovarian cancer treatment, especially during first-line chemotherapy in replacement of paclitaxel. It can be delivered alone or together with other chemotherapy drugs, such as carboplatin, to inhibit the microtubule, thus inducing cell cycle arrest (16, 17). Despite the promising anti-cancer effect of docetaxel, nowadays, ovarian cancer has emerged a growing risk of resistance to it. Upregulation of ABCB1 explains part of the resistance mechanism.

Erastin is a small molecule that induces ferroptosis which is a non-apoptotic iron-dependent mode of cell death with smaller mitochondria and increasing membrane density (18). It is the most efficient inhibitor of SLC7A11 at low micromolar concentrations (19). It specifically targets SLC7A11 to prevent cystine import and cause GSH depletion (18, 20). The anticancer property of erastin has been proved to be incredibly effective in a variety of cancers, such as liver cancer, lung cancer, gastric cancer, breast cancer, osteosarcoma, etc. (21–24). Combination therapy of erastin with other chemotherapeutic agents like cisplatin shows a significant synergistic effect in a number of cancers. Nevertheless, whether erastin can enhance the antiovarian cancer effect of docetaxel remains unknown. In this study, we revealed that erastin can overcome docetaxel resistance and present as a magical molecule to augment docetaxel efficacy in ovarian cancer by inhibition of ABCB1. This finding may serve as a potential strategy to transform a traditional ferroptosis inducer to elevate therapeutic efficiency for human cancers in the future.

# MATERIALS AND METHODS

#### Cell Culture and Reagents

Human ovarian cancer cells—A2780/Taxol cells—were created by continuous incremental taxol selection in A2780 cells. Cells were grown at 37◦C/5% CO<sup>2</sup> in DMEM. Then, 10% fetal bovine serum (FBS) and antibiotics (100 U/ml penicillin and 100 ng/ml streptomycin) were added. Erastin and ferrostatin-1 were obtained from ApexBio. Deferiprone was purchased from MCE. Docetaxel was bought from Hengrui Medicine. Rhodamine 123 and verapamil were from Sigma-Aldrich. The anti-β-tublin (KM9003T) antibody was purchased from Tianjin Sungene; the Anti-vinculin (bm1611) antibody was obtained from Boster; the anti-PARP (9542) antibody was purchased from Cell Signaling Technologies; the anti-Mcl-1(RLT2679) antibody was from Ruiying; and the Anti-ABCB1 (SC-13131) antibody was purchased from Santa Cruz Biotechnology.

#### Cell Viability Assay

Cells were seeded in a 96-well plate, including 6,000 cells, where each well-presented in 100 µl of medium with indicated drugs for 72 h. At the end of the reaction period, 10 µl

of 5 mg/ml 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide MTT solution per well was added to the medium for another 4 h. After discarding the whole medium carefully, 100 µl of dimethyl sulfoxide (DMSO) was added to dissolve formazan grains. The absorbance value was read at 570 nm. A Bliss method was used to calculate the IC<sup>50</sup> (25, 26).

#### Apoptosis Assay

Cell apoptosis was detected by a flow cytometry (FCM) assay. In short, cells were washed in cold PBS twice, stained with the binding buffer mixed with Annexin V-FITC and propidium iodide (PI) for more than 15 min in the dark, and then detected by FCM. Fluorescence was valued at an excitation wavelength of 480 nm by 530 and 585 nm filters. The early and late apoptosis rates were measured by FlowJo software (27, 28).

#### Cell Cycle Assay

Cells were collected after 48 h of drug incubation and set on cold 70% ethanol for more than 30 min. The ethanol was then discarded and cleaned with cold PBS before staining with PI (50µg/ml) for 15 min before being measured by FCM with an excitation wavelength of 480 nm through an FL-2 filter (585 nm). ModFit LT 3.0 software (Becton Dickinson) was used to quantify the date (29, 30).

#### Western Blot Analysis

Cells were collected and lysed in a RIPA buffer for about 30 min at 4◦C. The lysates were then centrifuged at 13,200 × rpm for 10 min to obtain supernatants without cell debris and nuclei. A total of 10% SDS-PAGE gels were used to separate the total proteins and thereby transferred proteins to polyvinylidene difluoride membranes, and 5% BSA was applied to block the membranes for 1 h and then incubated with the indicated primary antibodies overnight. Finally, the proteins were detected by the chemiluminescent detection reagents and a chemstudio plus imaging system (31, 32).

#### Rhodamine 123 Accumulation Assay

Cells were treated with or without inhibitors for 1 h before incubation with rhodamine 123 at a dose of 10µM for an extra 2 h. Verapamil is used as a positive inhibitor of ABCB1. Fluorescent images were taken under fluorescence microscopy. Following that, cells were harvested and cleaned with PBS three times and analyzed with FCM as previously described to measure the fluorescence intensity (33, 34).

#### Docking Protocol

Firstly, we obtained the 3D chemical structure of erastin through an online search on PubChem (a national library of medicine).

Secondly, we searched the RCSB Protein Data Bank to acquire the human ABCB1 structure, which has been reported to have an active binding site (PDB ID: 6QEX). Then, a docking experiment was executed with Discovery Studio 2.5. The most stable pose with a top-scoring of the ABCB1 complex was selected (35, 36).

#### Statistical Analysis

A student's t-test was done for date comparison among every group. A P-value of <0.05 was considered to indicate statistical significance.

# RESULTS

#### ABCB1-Overexpressing Ovarian Cancer Cells Are Resistant to Docetaxel and Erastin

The structure of erastin was listed in **Figure 1A**. As shown in **Figures 1B,C**, the growth rate of ovarian cancer cells is suppressed by docetaxel in a dose-dependent manner. A2780/Taxol, overexpressing the ABCB1 gene, was resistant to docetaxel with the IC<sup>50</sup> value of 377.54 nM, which was nearly 100-fold higher than A2780 cells. To examine whether erastin induced cytotoxicity in A2780 and A2780/Taxol cells, we exposed cells to increasing concentrations of erastin for 72 h. Consistent with the tolerance toward docetaxel, A2780/Taxol cells possessed more resistance to erastin with the IC<sup>50</sup> value of 24.98µM, while A2780 cells were more sensitive to erastin with a lower IC<sup>50</sup> value of 2.60 nM. Such an interesting co-resistance phenomenon indicated that erastin and docetaxel may share a similar drugresistance mechanism.

#### Erastin Induces Ferroptosis in Ovarian Cancer Cells

To investigate whether erastin induces ferroptosis in ovarian cancer cells, two ferroptosis inhibitors, ferrostatin-1 and deferiprone, were applied to reverse its cytotoxicity. As presented in **Figures 2A,B**, erastin-induced cell death could be partially reversed by both ferrostatin-1 and deferiprone in both A2780 and A2780/Taxol cells, indicating that erastin can induce ferroptosis in ovarian cancer cells and that there might also be other forms of cell death that can be induced by erastin.

#### Erastin Enhances the Sensitivity of Docetaxel in the ABCB1-Overexpressing Ovarian Cancer Cells

Cancer chemotherapy usually combines drugs for treatment. To explore the combinational effect of erastin and docetaxel in ovarian cancer cells, we co-administrated erastin and docetaxel in both A2780 cells and A2780/Taxol cells. As presented in **Figures 3A,B**, erastin dose-dependently decreased the IC<sup>50</sup> values of docetaxel in A2780/Taxol cells, while there was nearly no change in A2780 cells, indicating that erastin can enhance

FIGURE 4 | Erastin inhibits the drug efflux activity of ABCB1. A2780 and A2780/Taxol cells were treated with the increasing concentration of erastin for 48 h (0, 1, 3, and 10 as well as 0, 3, 10, and 30µM, respectively), and the protein expression of ABCB1 was detected by Western blot (A). Cells were incubated with 10µM rhodamine 123 for another 2 h at 37◦C after being pre-treated with the indicated concentrations of erastin and verapamil for 0.5 h at 37◦C, as measured by FCM and photographed by fluorescent microscope. The representative graphs (B), charts (C), and quantified data (D) are shown. Data are mean ± SD of three independent experiments. \**p* < 0.05 vs. corresponding control.

the sensitivity of docetaxel only in the ABCB1-overexpressing ovarian cancer cells.

### Erastin Inhibits the Drug Efflux Activity of ABCB1

To find out whether erastin enhances the sensitivity of docetaxel in the ABCB1-overexpressing ovarian cancer cells is due to downregulation of the expression of ABCB1 or the inhibition of ABCB1 activity, we valued the protein expression of ABCB1 as well as the intracellular aggregation level of rhodamine 123 (ABCB1 substrate) in the pre-incubation of erastin or absence of erastin. The protein expression level of ABCB1 was obviously higher in A2780/Taxol cells than that in A2780 cells (**Figure 4A**), and erastin did not alter the protein expression of ABCB1 in A2780/Taxol cells. Furthermore, the intracellular rhodamine 123 accumulated in A2780/Taxol cells was at a dramatically lower level compared with A2780 cells, and erastin dosedependently increased the intracellular rhodamine 123 levels only in A2780/Taxol cells but not in A2780 cells (**Figures 4B–D**), indicating that erastin can inhibit the drug efflux activity of ABCB1.

# Erastin Enhances Docetaxel-Induced Apoptosis in the ABCB1-Overexpressing Ovarian Cancer Cells

To further examine the sufficiency of erastin in combination with docetaxel in ovarian cancer cells, cells were incubated under different conditions for 48 h, and the apoptosis rate was detected by FCM. Besides, the related proteins were measured by Western blot. Co-administration of erastin and docetaxel significantly increased the apoptosis rate (both early and late apoptosis) in A2780/Taxol cells but not in A2780 cells (**Figures 5A,B**). Additionally, as shown in **Figure 5C**, the co-treatment group showed more increase of cleaved PARP (C-PARP) protein and a higher decrease of Mcl-1 protein than those in either docetaxel or erastin alone group only in A2780/Taxol cells. There was nearly no alteration in A2780 cells, suggesting that erastin can enhance docetaxel-induced apoptosis in the ABCB1 overexpressing ovarian cancer cells.

#### Erastin Enhances Docetaxel-Induced Cell Cycle Arrest in the ABCB1-Overexpressing Ovarian Cancer Cells

To examine the merging effect of erastin and docetaxel in ovarian cancer cells, the distribution change of cell cycle was measured by FCM with PI staining. As demonstrated in **Figures 6A,B**, the co-treatment group of erastin and docetaxel dramatically induced more accumulation in the sub-G1 and G2/M phase in comparison with erastin or docetaxel alone treatment only in A2780/Taxol cell but not in A2780 cells, suggesting that erastin can enhance the docetaxel-induced cell cycle arrest in the ABCB1-overexpressing ovarian cancer cells.

# Model for Binding of Erastin to ABCB1

Docking studies were carried out to demonstrate the binding mechanism of erastin to ABCB1. Firstly, the human ABCB1 crystal structure downloaded online was initially bound with

FIGURE 5 | Erastin enhances docetaxel-induced apoptosis in the ABCB1-overexpressing ovarian cancer cells. A2780 (A) cells were treated with 1µM erastin, 10 nM docetaxel alone or in combination for 48 h. A2780/Taxol (B) cells were treated with 10µM erastin and 300 nM docetaxel alone or in combination for 48 h. The apoptosis was detected by FCM with Annexin V/PI staining. The proportions of Annexin V+/PI– and Annexin V+/PI+ cells indicated the early and late stages of apoptosis. The protein expression was examined by Western blot after lysing cells, and β-tublin was used as loading control. The representative charts, quantified results, and Western blot results (C) of three independent experiments are shown. DTX, Docetaxel. \*\**p* < 0.01 vs. corresponding control.

and 10 nM docetaxel alone or in combination for 48 h. A2780/Taxol (B) cells were treated with 10µM erastin and 300 nM docetaxel alone or in combination for 48 h. The distribution of cell cycle was detected by FCM with PI staining. The representative charts and quantified results of three independent experiments are shown. DTX, Docetaxel. \**p* < 0.05 and \*\**p* < 0.01 vs. corresponding control.

taxol (PDB ID: 6qex). As demonstrated in **Figures 7A,B**, there existed special interactions between erastin and human ABCB1 (for example, hydrogen bonding), accounting for the stable affinity. The pyridine ring and pyrimidine ring of erastin connected with Phe983 via π-π stacking. The ethyl of erastin interacted with Phe336 through π-σ stacking. The hydron bonds exist between Phe343, Leu339, Met69, Ile340, Gln946, Gly62, Gln195, Leu65, Met949, Met986, Ala987, Phe728, Gln725, Tyr953, and erastin. Meanwhile, the hydrophobic pocket was formed by His61, Gly64, Thr199, Ser344, and TYR310 (**Figure 7C**), which stabilized the other part of erastin. In short, erastin directly binds to ABCB1 by a special chemical structure connection to suppress the pump activity of ABCB1.

#### DISCUSSION

Overexpression of ABCB1 is identified as one of the chief components of cancer chemotherapy failure (13, 37). A recent study declaimed that ABCB1 overexpression is guilty of olaparib resistance (38). Olaparib is a PARP inhibitor that is currently emerging as a promising treatment for ovarian cancer patients with BRCA mutation. Patients with a high expression of

nitrogen, blue; and oxygen, red), and erastin is shown as a ball and stick model with the atoms colored (carbon, gray; hydrogen, white; nitrogen, blue; oxygen, red; and chlorine, green). Yellow lines indicate π-π and π-σ stacking. The dotted green line indicates a hydrogen bonding interaction. The dotted blue line represents the interaction site of erastin and ABCB1 (C).

ABCB1 may not prospectively benefit from olaparib according to the single BRCA expression context. Therefore, strategies to inhibit ABC transporter proteins are identified as a potentially promising approach to suppress drug efflux in order to overcome drug tolerance. However, we have failed to figure out approved drugs suitable for clinical use for the inhibition of ABCB1, inducing delighted outcomes. A series of trials carried out with third-generation drugs have not been confirmed to have favorable outcomes yet (39). Therefore, developing more effective inhibitors of ABCB1 is urgently required.

Ferroptosis, a newly discovered cell death mode, is characterized by the aberrant accumulation of lipid peroxides in an iron-dependent way (40). Inhibitors of either SLC7A11 or glutathione peroxidase 4 (GPX4) can trigger ferroptosis. Some of the corresponding drugs are erastin and RSL3. Cancer cells often exhibit an increased iron demand to facilitate cell growth (41, 42), indicating that cancer cells may be more vulnerable to ferroptosis. Our work offers data to support the anti-ovarian cancer effect of erastin. Ferroptosis also plays a significant role in the drug resistance of cancer therapy. Firstly, ferroptosis is a non-apoptosis cell death pattern. Faced with the resistance caused by apoptosis-inducing chemotherapy drugs, ferroptotic reagents can serve as a promising strategy in reversing such therapeutic inefficiency. Secondly, persistent drug-tolerant cancer cells are sensitive to ferroptosis (9). In our study, we verified a decrease in the IC<sup>50</sup> value of docetaxel in the presence of erastin in the ABCB1-overexpressing ovarian cancer cells. The accumulation of rhodamine 123 confirmed that erastin significantly antagonized the drug-efflux function of ABCB1. Moreover, it has been reported that the safety of erastin in xenograft modes is acceptable (43, 44). Inhibition of SLC7A11 by erastin also increased the sensitivity of cisplatin in cancer cells (19, 45), suggesting the combination of erastin and chemotherapeutical reagents can be sufficient for cancer therapy.

In summary, our results demonstrated that erastin could reverse ABCB1-mediated docetaxel resistance in ovarian cancer, suggesting that the combination of erastin and docetaxel may expand the limited options for chemo-resistant ovarian cancers.

#### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

H-HZ, XC, L-YC, ZS, and X-JY designed the experiments, performed the experiments, analyzed the data, and wrote the paper. X-WN, J-HC, X-XC, YY, Z-HX, M-NW, YL, S-TW, and KL performed the experiments. All authors read and approved the final manuscript.

#### FUNDING

This work was supported by funds from the National Key Research and Development Program of China No. 2017YFA0505104 (to ZS), the National Natural Science Foundation of China Nos. 81772540 (to ZS), and 81503293 (to X-JY), the Science and Technology Program of Guangdong No. 2019A050510023 (to ZS), the Technology Development Funds of Wenzhou City No. Y20190014 (to X-JY), and the Traditional Chinese Medicine Science and Technology Foundation of Zhejiang Province, China No. 2020ZB144 (to X-JY).

in ovarian cancer revealed by bioinformatics analyses. Cancer Med. (2019) 8:606–16. doi: 10.1002/cam4.1964


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhou, Chen, Cai, Nan, Chen, Chen, Yang, Xing, Wei, Li, Wang, Liu, Shi and Yan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Pigment Epithelium-Derived Factor Promotes the Growth and Migration of Human Esophageal Squamous Cell Carcinoma

De-Rong Tang, Cheng-Lin Li, Ke-Ping Xu, Qing-Quan Wu, Qi-You Chen, Jun-Jie Lv, Jian Ji, Bao Zang, Chen Chen, Biao Gu and Jian-Qiang Zhao\*

*Department of Thoracic Surgery, The Affiliated Huaian No.1 People's Hospital of Nanjing Medical University, Huai'an, China*

Pigment epithelium-derived factor (PEDF) is an oncogene found in various types of cancers. However, how PEDF affects the development of human esophageal squamous cell carcinoma (ESCC) is unknown. This study investigates the role of PEDF in ESCC cell proliferation, migration, and cell cycle both *in vitro* and *in vivo.* The PEDF expression was examined in patient tumor samples and ESCC cell lines. Short hairpin RNA technology was used to inhibit the PEDF expression in ESCC EC9706 and KYSE150 cells. *In vitro* cell proliferation and migration assays were performed. The effects of PEDF on tumor growth and progression were examined *in vivo* in murine subcutaneous xenograft tumor models. It was found that PEDF was overexpressed in esophageal cancer cells and patient tumor tissues compared to normal control samples. PEDF enhanced cell cycle progression and inhibited cell apoptosis. Knock down of PEDF inhibited esophageal cell proliferation and migration *in vitro*. Moreover, Inhibition of PEDF significantly reduced tumor growth and tumor size *in vivo*. These results indicate that PEDF induce tumorigenesis in ESCC and can be a potential therapeutic target for cancer treatment.

Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Shuaishuai Liu, University of Maryland, Baltimore County, United States Wei Zhao, Chengdu Medical College, China Chang Zou, Shenzhen People's Hospital, China*

#### \*Correspondence:

*Jian-Qiang Zhao shenglee6871@sina.com*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *20 August 2019* Accepted: *17 December 2019* Published: *17 January 2020*

#### Citation:

*Tang D-R, Li C-L, Xu K-P, Wu Q-Q, Chen Q-Y, Lv J-J, Ji J, Zang B, Chen C, Gu B and Zhao J-Q (2020) Pigment Epithelium-Derived Factor Promotes the Growth and Migration of Human Esophageal Squamous Cell Carcinoma. Front. Oncol. 9:1520. doi: 10.3389/fonc.2019.01520* Keywords: pigment epithelium-derived factor (PEDF), esophageal carcinoma, tumorigenesis, proliferation, migration

#### INTRODUCTION

Esophageal carcinoma is a common gastrointestinal cancer that has two major subtypes: adenocarcinoma and squamous cell carcinoma (1). It is the sixth cancer-related deaths worldwide. There were about 16,940 newly diagnosed cases of esophageal carcinoma in the United States (accounting for 1% of all new-onset cancer cases) and 15,690 deaths (accounting for 2.6% of all cancer-related deaths) In 2016 (2). The incidence of esophageal squamous cell carcinoma (ESCC) is very high in some areas of China. There are ∼5,500 confirmed cases of disease in 5 million people each year in Huai'an city, the northern part of Jiangsu Province, China. Although progress has been made in the diagnosis and treatment of esophageal carcinoma, its 5-year survival rate is still dismal (1). Therefore, it is imperative to further explore the molecular basis of esophageal carcinoma and develop more effective treatment strategy.

Pigment epithelium-derived factor (PEDF), a member of the serine protease inhibitor superfamily, was initially isolated from the retinal pigment epithelial cells of human fetus (3). PEDF has neuroprotective and anti-angiogenic activity (4). PEDF is highly expressed in adipose tissue, liver, eye, heart, skeletal muscle, spleen, brain, and bone (5–10). PEDF plays significant biological

**227**

roles in many physiological and pathophysiological processes, including neuroprotection, fibrosis and inflammation (11). Previous studies have shown that PEDF plays important roles in cancer angiogenesis, tumor growth and metastasis (12, 13). Therefore, the potential use of PEDF as a target in the treatment of cancer has attracted much attention. However, the mechanisms of PEDF in cancer development remain controversial. PEDF shows antitumour effect in some tumors, including pancreatic, melanoma, and ovarian cancers (14–16). On the other hand, the level of PEDF is higher than that of normal tissues in some other cancers. For example, PEDF can promote stem cell growth and self-renewal of glioma stem cells (17, 18). PEDF was also found in hepatocellular carcinoma cells where PEDF levels were higher in HCC cases than in normal paracancerous tissues. The secretion of PEDF was higher in HCC patients than in normal controls, suggesting the potential of using serum PEDF as a biomarker for hepatocellular carcinoma (19). Therefore, the function of PEDF and its mechanism of action in various cancers need to be further investigated before it can be used as a biomarker for diagnosis and prognosis, or for cancer treatment.

In this study, we found that PEDF is overexpressed in tissues and cells of esophageal carcinoma. In addition, we observed that PEDF promotes esophageal cancer cell growth both in vivo and in vitro. This study will provide a rational for using PEDF as a prognostic biomarker and a potential therapeutic target for esophageal carcinoma.

#### MATERIALS AND METHODS

#### Tissue Samples

A total of 40 cases of esophageal cancer patients admitted to the Affiliated Huaian No.1 People's Hospital of Nanjing Medical University were enrolled in this study. Tumor and corresponding normal tissues were obtained. None of the patients received any radiation or chemotherapy before surgery. Surgical specimens were immediately frozen in liquid nitrogen and stored at −80◦ C for proteins assays. This study was carried out with the approval of the Ethics Committee of the Affiliated Huaian No.1 People's Hospital of Nanjing Medical University. A written informed consent was obtained from each patient.

#### Cell Lines and Cell Culture

Human esophageal cancer cell lines (EC9706, KYSE150) were purchased from Shanghai Cell Bank of Chinese Academy of Sciences. The cells were cultured in RPMI-1640 medium (Invitrogen, USA) containing 10% heat inactivated fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA) and 1% penicillin / streptomycin in an incubator with 5% CO<sup>2</sup> at 37◦C.

#### Cell Transfection

The short hairpin RNA was used to knockdown PEDF (shPEDF, 5′ -AGCGAACAGAATCCATCAT−3 ′ , shPEDF1, 5 ′ -GAAGCATGAGTATCATCTT−3 ′ , shPEDF2, 5′ -TGTTT GATTCACCAGACTT-3′ ), (Invitrogen, USA). Cells were transfected with pLL3.7-shPEDF, and 5 ug of packaging viral plasmid using Lipofectamine 2000 (Invitrogen).

#### Reverse Transcription-Quantitative Polymerase Chain Reaction (RT-PCR) Assay

Total RNA was extracted from tissues and cells using Trizol reagent (Invitrogen) according to the manufacturer's instructions. One microgram of RNA was then reversely transcribed into cDNA using M-MLV reverse transcriptase (Invitrogen). PEDF mRNA expression was detected using an Applied Biosystems 7500 Fast Real-Time PCR System (Applied Biosystems) using a fast SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA), with β-actin as an endogenous control. The primer sequences for PEDF and β-actin are as follows: PEDF, 5′ -ACT GAG TAA GAT GGC GGG TCG-3′ (forward) and 5′ -TTC TGG CGA AAG CGG GTA G-3′ β-actin, 5′ -AAA TCG TGC GTG ACA TCA AAG A-3′ (forward) and 5′ -GGC CAT CTC CTG CTC GAA-3′ (reverse).

#### Western Blot Analysis

Total protein was extracted from cells and patients' tissues using RIPA buffer (Beyotime, Shanghai, China) containing a protease inhibitor cocktail (Roche Diagnostics, Basel, Switzerland) and quantified using the PierceTM BCA Protein Assay Kit (Invitrogen; Thermo Scientific). The same amount of protein (50 µg) was then separated on 10% sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a nitrocellulose filter (NC membrane; Millipore, Billerica, MA, USA). Membranes were then blocked in 5% skim milk for 1 h at room temperature and blocked with primary antibodies against PEDF(ab10389), caspase 3(ab197202), caspase 9(ab219590) (Abcam, Cambridge, Mass., USA), overnight at 4◦C followed by incubation with horseradish peroxidase (HRP) conjugated goat anti-rabbit secondary antibody (ab6721, 1: 1,000, Abcam) for 1 h at room temperature. Finally, specific protein signals were visualized using ECL Western blotting substrates (Promega, Madison, WI, USA) and quantified by Image J software (National Institutes of Health, Bethesda, Maryland, USA).

#### Transwell Invasion Assay

Invasion chambers (BD Bioscience, San Diego, Calif., USA) were used to assess cell invasiveness using a membrane with a pore size of 8µm (BD Bioscience). Briefly, EC9076 and EC109 cells infected with shRNA were resuspended in serum-free RPMI-1640 medium (100 µl) and inoculated in 40 µl of Matrigel (BD Biosciences) with 700 µl RPMI-1640 medium and 10% FBS. After incubation at 37◦C for 48 h, cells on the upper side of the membrane were removed with a sterile swab. The cells on the lower side of the membrane were fixed with methanol for 30 min, stained with 0.1% crystal violet for 20 min, and counted in six randomly selected fields at 200 X magnification using an inverted microscope (Nikon Eclipse TE300, Tokyo, Japan).

# Flow Cytometry for Cell Cycle and Apoptosis Analysis

Cell cycle was determined according to the manufacturer's protocol with propidium iodide (PI) staining (Nanjing Kaiji Biotechnology Development Co. Ltd., Nanjing, China). Briefly, EC9076 and EC109 cells were transfected with shRNA. The cells were then harvested and resuspended in 500 µl 1X binding buffer at a concentration of 1 × 10<sup>6</sup> cells / ml, followed by the addition of 5 µl of PI. The treated cells were then incubated for 5 min at room temperature in the dark. Finally, the cell cycle and apoptotic rate of the cells was analyzed using a FACS Calibur flow cytometer (Beckman Coulter, Atlanta, GA, USA).

#### In vivo Experiments

BALB/c nude mice (6–7 weeks old, male) were obtained from Chinese Academy of Sciences (Shanghai, China) and grown under specific pathogen-free conditions. All animal studies are conducted in accordance with the University Laboratory Animal Management, and approved by the Ethics Committee of The Affiliated Huaian No.1 People's Hospital of Nanjing Medical University. To investigate the knock down effect of PEDF on tumor growth in vivo, stably transfected EC9706 cells were subcutaneously injected into the ventral region of nude mice at a concentration of 1 × 10<sup>7</sup> cells / ml. According to the formula of 0.5 × L × W<sup>2</sup> , the tumor volume was measured every 3 days with a caliper. On day 24 after injection, tumors from all mice were obtained, weighed, and fixed in formalin.

#### Statistical Analysis

All data from three independent experiments were obtained and expressed as mean ± SD. Student's t-test or one-way ANOVA was used to analyze differences between different groups. The difference was considered statistically significant when P <0.05.

### RESULTS

# PEDF Is Overexpressed in Esophageal Squamous Cell Carcinoma

To investigate the role of PEDF in ESCC, we compared the expression of PEDF in ESCC and adjacent normal tissues in 40 patients. Our results showed that the protein expression of PEDF in tumor samples is significantly higher than that in their corresponding normal tissues in those 40 patients (**Figure 1A**). These results indicated that PEDF is associated with the development of ESCC.

#### PEDF Enhances Cell Proliferation and Migration in Esophageal Squamous Cell Carcinoma

Because PEDF is overexpressed in esophageal carcinoma, we explored the role of PEDF in esophageal carcinoma by knocking down the expression of PEDF in two esophageal carcinoma cell lines EC9706 and KYSE150. In order to determine the best knock-down efficiency, we synthesized three shRNA. The results showed that shRNA-PEDF markedly suppressed the expression of PEDF proteins and mRNA (**Figure 1B**). Therefore, shRNA-PEDF was used in the following assays. Colony formation assay was used to determine the cell growth after knocking down

PEDF. The result showed significant reduction of the colony numbers of esophageal carcinoma cells at 7 days after transfection of shRNA (**Figure 2A**).

The effect of cell migration after knocking down PEDF in esophageal carcinoma cells was also investigated. The transwell assay revealed that shRNA-PEDF significantly attenuated cell

FIGURE 2 | Effect of PEDF knockdown on anchorage-independent growth of esophageal cancer cells. (A) Colony formation assay, and (B) invasion assay, of esophageal cancer cells after knocking down PEDF. \* < 0.05.

migration compared to control group. There were less esophageal carcinoma cells migrated in shRNA-PEDF transfection group than those in control group. This result indicated that PEDF promote esophageal carcinoma cell migration (**Figure 2B**). The above results suggested that suppression of PEDF could reduce proliferation and migration of esophageal carcinoma cells.

#### PEDF Promotes Cell Cycle and Reduces Cell Apoptosis in Esophageal Squamous Cell Carcinoma

Because PEDF enhances esophageal carcinoma cell growth, we further investigated whether PEDF affects cell cycle and cell apoptosis. To explore the cell cycle change after shRNA transfection, esophageal carcinoma cells were stained with propidium iodide (PI) and analyzed by Flow cytometry. As expected, knocking down PEDF increased cells in G0/G1 phase and decreased cells in S phase and G2/M phase compared to shRNA scramble group (**Figure 3A**).

Flow cytometry was used to determine cell apoptosis after shRNA transfection and Annexin-V/PI staining. The result demonstrated that knocking down PEDF increased early apoptotic cells, late apoptotic cells, and necrotic cells (**Figure 3B**), suggesting that knocking down PEDF increased apoptosis of esophageal carcinoma cells. Furthermore, Western blot shows that the levels of caspase 3 and caspase 9 in the shRNA-PEDF group were higher than in control group (**Figure 3C**).

#### PEDF Promotes Tumourigenesis of Esophageal Squamous Cell Carcinoma in vivo

To further test the effect of PEDF in esophageal carcinoma cell growth in vivo, we establish xenograft mouse models. Nude mice were injected with EC9706 cells transfected with shRNA control or shRNA-PEDF. Similar to in vitro results, the tumor volume and tumor weight of xenografts in mice inoculated with shRNA-PEDF cells were smaller than that with shRNA control cells, suggesting that PEDF promotes esophageal carcinoma growth in vivo (**Figures 4A–C**).

# DISCUSSION

PEDF is a 50 kDa secreted protein which is a versatile member of the widely expressed serpin family, also denoted as SERPINF1 (1). It plays important roles in angiogenesis, fibrogenesis, neuroprotection, bone matrix mineralization, and inflammation (4). Although PEDF was firstly identified as being produced by retinal pigment epithelial cells, it is now known to express in a variety of tissues and cell types, including chondrocytes, and synovial cells. PEDF is naturally present in serum (20, 21). In addition to the many beneficial effects that PEDF possesses, it is involved in the pathogenesis of diseases, such as chronic inflammatory diseases, atherosclerosis, type 2 diabetes, and some types of brain tumors (22, 23). Moreover, PEDF protects osteoblasts from glucocorticoid-induced apoptosis, and increases vascular permeability of triglycerides by ATGL degradation (24). Anti-inflammatory and antithrombotic effects of PEDF have been reported, and PEDF can also prevent the adhesion and invasion of liver cancer cells (25).

PEDF is associated with signaling pathways related to cancer development. PEDF can directly bind to PEDF receptor (PEDFR) and stimulate the activity of phospholipase (26). PEDFR is strongly linked to cell proliferation in cancers (27). Moreover, PEDF has an effect on cancer cell migration by activating MKK3 and MKK6. Tumor cell apoptosis is also regulated by PEDF acting on PPAR and NF-kB (28, 29). On the other hand, PEDF is a key regulatory factor in endothelial cells. Angiogenesis is inhibited by PEDF via cleaving VEGFR1 and VEGFR2 at the transmembrane region and VEGF-induced phosphorylation (30).

In this study, we demonstrate that PEDF is overexpressed in esophageal cancer tissues and cells compared to normal human counterparts. These results are consistent with studies of other cancers, such as prostate cancer (31). Previous studies have shown that the abnormal expression of PEDF is closely related to the pathological process of tumor development, including proliferation, migration, invasion, and apoptosis (32). shRNAmediated knockdown of PEDF was reported to be effective in inhibiting the growth of melanoma (33). We investigated the impact of knockdown of PEDF on the progression of esophageal cancer in vitro and in vivo. The results showed that shRNA-mediated reduction of PEDF significantly inhibited the proliferation and invasion of esophageal cancer cells and induced apoptosis. In vivo experiments further confirmed that shRNA-mediated PEDF knockdown significantly blocked xenograft esophageal tumor growth. Our results demonstrated that PEDF plays a role in the development of esophageal cancer. However, there was a report shown that PEDF may have potent antiangiogenic and antitumor effects in ESCC cells naturally not secreting endogenous PEDF, in the cell line secreting endogenous PEDF, there is no inhibition of angiogenesis and no subsequent antitumor properties (34).

Taken together, our study showed that PEDF expression is significantly increased in esophageal cancer tissues and cells. Knockdown of PEDF significantly inhibited esophageal cancer cell proliferation and tumourigenesis both in vitro and in vivo. Therefore, PEDF may serve as a prognostic biomarker and potential therapeutic target for esophageal cancer.

# DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

# ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the Ethics Committee of the Affiliated Huaian No.1 People's Hospital of Nanjing Medical University. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Ethics committee of The Affiliated Huaian No.1 People's Hospital of Nanjing Medical University.

#### AUTHOR CONTRIBUTIONS

D-RT performed experiments and drafted the manuscript. CL-L carried out the cell culture. K-PX participated in the design. Q-QW, Q-YC, and J-JL collected tissue

#### REFERENCES


specimens. JJ carried out the Western blot analysis. BZ participated in Transwell invasion study. BG processed specimens. J-QZ conceived the study, coordination, and edited the manuscript. All authors read and approved the study.

levels, in a rat model of ovarian hyperstimulation syndrome. Arch Gynecol Obstet. (2016) 293:1101–6. doi: 10.1007/s00404-015-3987-4


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Tang, Li, Xu, Wu, Chen, Lv, Ji, Zang, Chen, Gu and Zhao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Knockdown of Thymidine Kinase 1 Suppresses Cell Proliferation, Invasion, Migration, and Epithelial–Mesenchymal Transition in Thyroid Carcinoma Cells

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, Jinan University, China*

#### Reviewed by:

*Zhe-Sheng Chen, St. John's University, United States Yan Deng, The Chinese University of Hong Kong, China*

#### \*Correspondence:

*Weiqing Wu wweiqing007@sina.com Guangsuo Wang edward20111111@qq.com Fajin Dong dongfajin@szhospital.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *10 October 2019* Accepted: *09 December 2019* Published: *29 January 2020*

#### Citation:

*Liu C, Wang J, Zhao L, He H, Zhao P, Peng Z, Liu F, Chen J, Wu W, Wang G and Dong F (2020) Knockdown of Thymidine Kinase 1 Suppresses Cell Proliferation, Invasion, Migration, and Epithelial–Mesenchymal Transition in Thyroid Carcinoma Cells. Front. Oncol. 9:1475. doi: 10.3389/fonc.2019.01475* Chang Liu1,2†, Jian Wang3†, Li Zhao4†, Hui He<sup>4</sup> , Pan Zhao<sup>1</sup> , Zheng Peng<sup>1</sup> , Feiyuan Liu<sup>1</sup> , Juan Chen<sup>5</sup> , Weiqing Wu<sup>4</sup> \*, Guangsuo Wang<sup>3</sup> \* and Fajin Dong<sup>6</sup> \*

*<sup>1</sup> Clinical Medical Research Center, First Affiliated Hospital of Southern University of Science and Technology, Second Clinical College of Jinan University, Shenzhen, China, <sup>2</sup> Central Lab, Dalian Municipal Central Hospital, Dalian, China, <sup>3</sup> Department of Thoracic Surgery, First Affiliated Hospital of Southern University of Science and Technology, Second Clinical College of Jinan University, Shenzhen, China, <sup>4</sup> Department of Health Management, First Affiliated Hospital of Southern University of Science and Technology, Second Clinical College of Jinan University, Shenzhen, China, <sup>5</sup> Department of Medical Research, Shenzhen Shekou People's Hospital, Shenzhen, China, <sup>6</sup> Department of Ultrasound, First Affiliated Hospital of Southern University of Science and Technology, Second Clinical College of Jinan University, Shenzhen, China*

Patients with advanced thyroid carcinoma have poor prognosis with low overall survival. Unfortunately, the underlying mechanisms of thyroid carcinoma progression remain unclear. The elevated expression of thymidine kinase 1 (TK1) has been implicated in the progression of thyroid carcinoma, while the role of TK1 in thyroid carcinoma progression has not been explored. The present study aimed to determine the role TK1 in the progression of thyroid cancer and to explore the underlying molecular mechanisms. In this study, it was found that serum TK1 levels were markedly increased in the patients with thyroid nodules. Further online data mining showed that TK1 expression was upregulated in thyroid carcinoma tissues, and higher expression of TK1 was correlated with shorter disease-free survival of patients with thyroid carcinoma. Silencing of TK1 suppressed cell proliferation, invasion, migration, and epithelial–mesenchymal transition, and also induced cell apoptosis in the thyroid carcinoma cell lines. Animal studies showed that TK1 knockdown inhibited *in vivo* tumor growth of thyroid carcinoma cells. Importantly, miR-34a-5p was found to be downregulated in the thyroid carcinoma cells. Furthermore, miR-34a-5p targeted the 3′ untranslated region of TK1 and suppressed the expression of TK1 in thyroid carcinoma cell lines. In summary, first, these results demonstrated the upregulation of TK1 in thyroid nodules and thyroid carcinoma tissues; second, TK1 promoted thyroid carcinoma cell proliferation, invasion, and migration; lastly, TK1 was negatively regulated by miR-34a-5p. Our study may provide novel insights into the role of TK1 in regulating thyroid carcinoma progression.

Keywords: thyroid carcinoma, thymidine kinase 1, thyroid nodules, progression, miR-34a-5p

#### INTRODUCTION

Thyroid carcinoma is one of the most common human malignancies, and the incidence rate of thyroid carcinoma is expected to gradually increase (1). Based on the characteristics of histology, thyroid carcinoma can be divided into four types including anaplastic, follicular, medullary, and papillary thyroid carcinoma (2). Papillary thyroid cancer is the main type of this malignancy and accounts for more than 80% of all cases (2). The thyroid carcinoma at the early stage is commonly curable; however, patients with advanced thyroid carcinoma have poor prognosis with relatively low 5-years survival rates (3, 4). In this regard, it is important to elucidate the molecular mechanisms of thyroid carcinoma development and to develop novel diagnostic and therapeutic strategies for thyroid carcinoma.

Thymidine kinase 1 (TK1) is an important regulatory factor in modulating cell cycle. During different stages of cell cycle, the activities of TK1 are increased at the late G1 phase and reached maximum levels at the late S phase (5, 6). Dysregulation of TK1 has been shown to be associated with the progression of human malignancies. Yu et al. screened a total of 56,178 human subjects in the southeast of China and proposed that serum TK1 was a potential biomarker for the early discovery of human subjects having the risk to process into malignancy (7). Wei et al. showed that TK1 overexpression was correlated with the poor prognosis of patients with lung cancer (8). Wang et al. demonstrated that determination of TK1 expression using immunohistology could improve the overall prediction of prognosis of ovarian cancer patients (9). Knockdown of TK1 inhibited the progression of pancreatic cancer cell via targeting E2F1-TK1-P21 axis (10). In the thyroid cancer, high levels of TK1 were correlated with the advanced clinical stage of patients with thyroid carcinoma (11). Unfortunately, the molecular mechanisms of TK1 in regulating thyroid carcinoma progression have not been explored yet.

MicroRNAs (miRNAs) are an abundant class of short (18–24 nt), endogenous non-coding RNAs that act as regulators at the transcriptional or post-transcriptional level of gene expression. MiRNAs can affect essential cellular processes including cell growth, cell differentiation, and apoptosis, which are closely correlated to carcinogenesis. Aberrant expression of miRNAs has been reported in various types of cancers including thyroid carcinoma. For example, plasma miR-346, miR-10a-5p, and miR-34a-5p were elevated in papillary thyroid carcinoma, which could be used as important biomarkers for the progression of thyroid carcinoma (12).

In this study, we identified the upregulation of TK1 in the serum of patients with thyroid nodules. The in vitro functional studies showed that TK1 silencing suppressed thyroid cancer cell proliferation, invasion, migration, epithelial–mesenchymal transition (EMT) and induced cell apoptosis. Furthermore, the upregulation of TK1 in the thyroid cancer may be related to the downregulation the tumor-suppressive miR-34a-5p.

#### MATERIALS AND METHODS

#### Clinical Samples

The serum samples were collected from 1,112 subjects who underwent the physical examination at First Affiliated Hospital of Southern University of Science and Technology, Second Clinical College of Jinan University between 2015 and 2018. Among the subjects, 431 patients were positive for thyroid nodules by ultrasound examination, and 681 patients were negative for thyroid nodules. The protein levels of TK1 in the serum were detected using the enzyme-linked immunosorbent assay (ELISA) assay kit (#ab223595, Abcam, Cambridge, USA). All the experimental protocols were approved by the Ethics Committee of the First Affiliated Hospital of Southern University of Science and Technology, and all the patients signed the written informed consent.

#### Cell Lines and Cell Culture

The normal human primary thyroid follicular epithelial cells (Nthy-ori 3-1, #90011609) and thyroid carcinoma cell line (TPC-1, #SCC147) were obtained from Merck (Darmstadt, USA). The thyroid carcinoma cell lines (BC-PAP, #ACC273) were obtained from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). The cells were cultured in RMPI-1640 medium (Sigma-Aldrich, St. Louis, USA) supplemented with 10% fetal bovine serum (FBS; #10100154, Life Technologies, Waltham, USA) and were kept in a humid atmosphere of 5% (v/v) CO<sup>2</sup> and 95% (v/v) air at 37◦C.

#### Synthesis of Small Interfering RNAs and miRNAs, Cell Transfections

The small interfering RNAs (siRNAs) that silencing TK1 (TK1 siRNA#1 and #2) or the scrambled negative control (si-NC) were designed and synthesized by Ribobio (Guangzhou, China). The miRNA mimics for miR-34a-5p and the respective mimics NC were purchased from Thermo Fisher Scientific (Waltham, USA). The thyroid carcinoma cell transfections with these siRNAs or miRNAs were performed using Lipofectamine 2000 reagent (#11668030, Invitrogen, Carlsbad, USA) according to the manufacturer's protocol, and the transfected cells were collected after 24 h of transfection for further study.

#### Quantitative Real-Time PCR

Extraction of total RNAs from tissues or cells was performed using Trizol reagent (#15596018, Invitrogen) according to the manufacturer's protocol. For TK1 messenger RNA (mRNA) detection, RNAs were reversely transcribed into complementary DNA (cDNA) using the first-strand cDNA synthesis kit (#K1621, Thermo Fisher Scientific); for miR-34a-5p detection, RNAs were reversely transcribed into cDNA using the NCode miRNA firststrand cDNA synthesis kit (#MIRC10, Invitrogen). Real-time PCR was performed on an ABI7900 PCR system (Applied Biosystems, Foster City, USA) using SYBR Green PCR Master Mix (#RR820A, Takara, Dalian, China). GAPDH and U6 were used as internal control for TK1 and miR-34a-5p expression, respectively. The fold change between different groups was determined using comparative Ct method.

#### Western Blot

Extraction of proteins from tissues or cells was performed using radio-immunoprecipitation assay buffer (#P0013B, Beyotime, Beijing, China) supplied with the protease inhibitor cocktail (ST506, Beyotime). The protein concentrations were determined using the bicinchoninic acid method (#23252, Thermo Fisher Scientific). Equal amount of 30 µg proteins was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis followed by transferring to polyvinylidene fluoride membranes (#IPVH08100, Sigma-Aldrich). After incubating with 1.5% non-fat milk in the Tris-buffered saline Tween, the membranes were incubated with corresponding primary antibodies including TK1 (1:1,000; #8960, Cell Signaling Technology, Danvers, USA), active caspase-3 (1:1,000; #9661, Cell Signaling Technology), active caspase-9 (1:1,000; #7237, Cell Signaling Technology), vimentin (1:1,000) (5,741, Cell Signaling Technology), N-cadherin (1:1,000; #14215, Cell Signaling Technology), E-cadherin (1:1,000; #14472, Cell Signaling Technology), and β-actin (1:1,000; #3700, Cell Signaling Technology). After incubating with above primary antibodies overnight at 4◦C, membranes were then further probed against the respective horseradish-conjugated secondary antibodies (1:2,000; #7074, #7076, Cell Signaling Technology) at room temperature for 2 h. Protein detection was performed using the enhanced chemiluminescence kit (#15159, Thermo Fisher Scientific) according to the manufacturer's protocol. β-Actin served as the loading control.

#### Cell Counting Kit-8 and Colony Formation Assays

Cell proliferation was measured by cell counting kit-8 (CCK-8) assay (#C0039, Beyotime) in thyroid carcinoma cells at 0, 24, 48, and 72 h after transfection. Briefly, the transfected thyroid cells were incubated with 10 µl CCK-8 reagent for 1 h at 37◦C. The cell proliferation values were determined by measuring optical density values at 450 nm.

Cell growth was measured by colony formation assay. Briefly, the transfected thyroid carcinoma cells were seeded into a sixwell plate at a density of 500 cells per well, after growing for 10 days, the cells were fixed with methanol for 10 min and stained with 0.1% crystal violet (#C0121, Beyotime) for 10 min. The number of colonies was counted under a light microscope (TE2000, Nikon, Tokyo, Japan).

#### Transwell Invasion Assay

Thyroid carcinoma cell invasion was measured by Transwell invasion assay. Briefly, the Matri-gel (#E6909, Sigma-Aldrich) was coated on the 8-µm pore size membrane Transwell inserts (#140629, Thermo Fisher Scientific), and the coated inserted were placed into the wells of a 24-well plate (#142475, Thermo Fisher Scientific). The transfected thyroid carcinoma cells were seeded onto the upper chamber filled with FBS-free RMPI-1640 medium, while the bottom chamber was filled with RMPI-1640 medium supplied with 10% FBS. After incubation for 24 h, the non-invaded carcinoma cells were removed from the upper surface of the membrane, and cells in the lower surface of the membrane were fixed with methanol for 20 min and stained with 0.5% crystal violet for 15 min. The number of invaded cells was counted under a microscope by randomly selecting five fields.

#### Wound Healing Assay

Thyroid carcinoma cell migration was determined by wounding healing assay. Briefly, the transfected thyroid carcinoma cells were seeded onto the six-well plates (#140675, Thermo Fisher Scientific) and cultured for 24 h. The wound was created on the monolayer cells using a sterile 200-µl tip (#94410313, Thermo Fisher Scientific), and the cells were further cultured for 24 h. The wound width was measured at 0 and 24 h, respectively. The percentage of wound closure was calculated by (wound width at 24 h—wound width at 0 h)/wound width at 0 h.

#### Flow Cytometry

The cell apoptosis of thyroid carcinoma cells was detected using Annexin V-fluorescein isothiocyanate/propidium iodide

(PI) apoptosis detection kit (#V13241, Thermo Fisher Scientific). Briefly, the transfected thyroid carcinoma cells were trypsinized and collected, and the cells were incubated with Annexin Vfluorescein isothiocyanate and PI in binding buffer for 10 min. The stained cells were then analyzed using a BD FACSCanto II flow cytometer (BD Biosciences, San Jose, USA).

#### Caspase-3 Activity Assay

The caspase-3 activity of the transfected thyroid carcinoma cells was determined using a caspase-3 activity assay kit (#5723, Cell Signaling Technology) according to the manufacturer's protocol.

#### In vivo Tumor Growth Assay

A total of 12 male BALB/nude mice (6–8 weeks old) were obtained from Guangzhou Laboratory Animal Center (Guangzhou, China). All animal experiments were approved by the Animal Ethics Committee of First Affiliated Hospital of Southern University of Science and Technology. TPC-1 cells (5 × 10<sup>6</sup> cells) with stably expressing TK1 shRNA (sh\_TK1) or scrambled negative control shRNA (sh\_NC) were subcutaneously injected into the right flank of the nude mice and six animals in each group. After injection of carcinoma cells, the tumor volume of the nude mice was measured every 7 days for 42 days. At the end of the experiments, the mice were killed, and the tumor tissues were collected for further analysis.

#### Dual-Luciferase Reporter Assay

To construct the reporter vectors, the 3′ untranslated region (UTR) of TK1 containing the putative binding sites of miR-34a-5p was amplified by PCR and cloned into downstream of the luciferase gene of the pGL3 vector (#E1751, Promega,

qRT-PCR and Western blot analysis of TK1 mRNA and protein expression levels in TPC-1 and BC-PAP cells after scrambled siRNA (si-NC) or TK1 siRNAs (TK1 siRNA#1 or #2) transfections. (F,G) Cells proliferation by CCK-8 assay was determined in TPC-1 and BC-PAP cells, respectively. (H,I) Colony formation ability was assessed in TPC-1 and BC-PAP cells, respectively. (J,K) Flow cytometry analysis was used to detect the cell apoptotic rates in TPC-1 and BC-PAP cells, respectively. (L,M) Caspase-3 activity assay was used to determine the capsase-3 activity of TPC-1 and BC-PAP cells, respectively. (N,O) Protein expression levels of active caspase-3 and caspase-9 in TPC-1 and BC-PAP cells were detected by Western blot, respectively. *N* = 3; \**P* < 0.05, \*\**P* < 0.01, and \*\*\**P* < 0.001.

Madison, USA). The mutant reporter vectors were generated by mutating three nucleotides in the binding region. Thyroid carcinoma cells were cotransfected with reporter vectors and miRNAs using Lipofectamine 2000 reagent (Invitrogen). At 24 h after transfection, luciferase activity in the thyroid carcinoma cells was determined using the Dual-Luciferase Reporter Assay System (#E1910, Promega).

#### Statistical Analysis

All data analysis was performed using GraphPad Prism (Version 5.0; GraphPad Software, La Jolla, USA). Summary data are presented as the mean ± standard deviation. Significant differences between different groups were evaluated using Student's t test or one-way ANOVA followed by Bonferroni's post hoc test. Statistical significance was set at P < 0.05.

investigate the invasion ability of TPC-1 and BC-PAP cells transfected with scrambled siRNA (si-NC) or TK1 siRNAs (TK1 siRNA#1 or #2), respectively. (C,D) The migration ability of TPC-1 and BC-PAP cells transfected with scrambled siRNA (si-NC) or TK1 siRNAs (TK1 siRNA#1 or #2) was assessed by wound healing assay, respectively. (E,F) Protein expression levels of epithelial–mesenchymal transition (EMT)-related markers vimentin, N-cadherin, and E-cadherin in TPC-1 and BC-PAP cells transfected with scrambled siRNA (si-NC) or TK1 siRNAs (TK1 siRNA#1 or #2) were detected by Western blot. *N* = 3; \**P* < 0.05 and \*\**P* < 0.01.

#### RESULTS

# TK1 Was Upregulated in Serum From Patients With Thyroid Nodules and Was Upregulated in the Thyroid Carcinoma Tissues

We first analyzed the serum TK1 protein levels from the subjects who underwent physical examination in our hospital and found that serum TK1 levels were significantly higher in the subjects with thyroid nodules compared to the normal subjects (**Figure 1A**). A further analysis using data mining tool (http://gepia.cancer-pku.cn/) showed that TK1 was markedly upregulated in the thyroid carcinoma tissues when compared to normal thyroid tissues (**Figure 1B**). In addition, patients with higher expression of TK1 had poorer disease-free survival when compared to patients with lower expression of TK1 (**Figure 1C**).

#### TK1 Knockdown Suppressed Thyroid Carcinoma Cell Proliferation and Induced Cell Apoptosis

The mRNA expression of TK1 in carcinoma cells as well as in normal thyroid follicular epithelial cells was investigated. TK1 was upregulated in carcinoma cells when compared to normal thyroid follicular epithelial cells (**Figure 2A**). To determine whether the knockdown of TK1 could reserve the aggressiveness of carcinoma cells, TPC-1 and BC-PAP cells were transfected with scrambled siRNA or TK1 siRNAs. TK1 siRNA transfection significantly repressed mRNA and protein expression levels of TK1 in TPC-1 and BC-PAP cells when compared to scrambled siRNA transfection (**Figures 2B–E**). CCK-8 and colony formation assays showed that TK1 knockdown suppressed the TPC-1 and BC-PAP cell proliferation (**Figures 2F,G**) and colony formation ability (**Figures 2H,I**). Flow cytometry and caspase-3 activity analysis showed that TK1 knockdown increased cell apoptotic rates (**Figures 2J,K**) and caspase-3 activity of TPC-1 and BC-PAP cells (**Figures 2L,M**). The Western blot assay showed that TK1 knockdown increased the protein levels of active caspase-3 and caspase-9 in TPC-1 and BC-PAP cells (**Figures 2N,O**).

#### TK1 Knockdown Suppressed Thyroid Carcinoma Cell Invasion, Migration, and Epithelial–Mesenchymal Transition

A further analysis of cell invasion and migration using Transwell invasion assay and wound healing assay, respectively, showed that TK1 knockdown markedly suppressed TPC-1 and BC-PAP cell invasion (**Figures 3A,B**) and migration (**Figures 3C,D**). In addition, it was found that TK1 knockdown suppressed the protein levels of EMT-related markers vimentin and N-cadherin but increased the protein levels of E-cadherin (**Figures 3E,F**).

group and sh\_TK1 group. (B) Weight of the tumor tissues from sh\_NC and sh\_TK1 group. (C) qRT-PCR analysis of TK1 mRNA expression levels in tumor tissues from sh\_NC and sh\_TK1 groups. (D) Western blot analysis of active caspase-3, caspase-9, vimentin, N-cadherin, and E-cadherin protein expression levels of the tumor tissues from sh\_NC and sh\_TK1. *N* = 6; \**P* < 0.05 and \*\*\**P* < 0.001.

#### TK1 Knockdown Suppressed in vivo Tumor Growth of Thyroid Carcinoma Cells

The effects of TK1 knockdown on the in vivo tumor growth of TPC-1 were evaluated in nude mice xenografts model. As shown in **Figure 4A**, TK1 knockdown reduced the tumor volume in the nude mice; consistently, the weight of dissected tumor tissue was significantly lower in the h\_TK1 group when compared to the sh\_NC group (**Figure 4B**). The quantitative real-time PCR analysis showed that TK1 was downregulated in the tumor tissues from the sh\_TK1 group when compared to the sh\_NC group (**Figure 4C**). In addition, TK1 knockdown increased the protein levels of active caspase-3, caspase-9, and E-cadherin, but decreased the protein levels of vimentin and N-cadherin (**Figure 4D**).

#### TK1 Expression Was Regulated by miR-34a-5p in Thyroid Carcinoma Cells

To determine the factors that contribute to the upregulation of TK1 in thyroid carcinoma, the StarBase tool (http:// starbase.sysu.edu.cn/index.php) was used to predict miRNAs that could regulate TK1 expression. Among the predicted miRNAs, miR-34a-5p was selected for examination, and it was downregulated in the TPC-1 and BC-PAP cells when compared to normal thyroid follicular epithelial cells (**Figure 5A**). The overexpression of miR-34a-5p was achieved by transfecting TPC-1 and BC-PAP cells with miR-34a-5p mimics (**Figures 5B,C**). Furthermore, the luciferase reporter assay was performed to determine the interaction between miR-34a-5p and TK1. The putative binding sites between TK1 3′UTR and miR-34a-5p are shown in **Figure 5D**. Overexpression of miR-34a-5p significantly suppressed the luciferase activity of the wild-type luciferase constructs but not the mutant ones in TPC-1 and BC-PAP cells (**Figures 5E,F**). Consistently, miR-34a-5p overexpression markedly repressed the mRNA and protein expression levels of TK1 in both TPC-1 and BC-PAP cells (**Figures 5G–J**).

# DISCUSSION

Patients with advanced thyroid carcinoma had poor prognosis with low overall survival (13, 14). Unfortunately, the

cotransfection, respectively. (G–J) qRT-PCR and Western blot were used to detect TK1 mRNA and protein expression levels in TPC-1 and BC-PAP cells with mimics NC or miR mimics transfection, respectively. *N* = 3; \*\**P* < 0.01.

underlying mechanisms of thyroid carcinoma progression remain unclear. The elevated expression of TK1 has been implicated in the progression of thyroid carcinoma (11), while the role of TK1 in thyroid carcinoma progression has not been explored. In this study, we found that serum TK1 levels were markedly increased in the patients with thyroid nodules. Further online data mining showed that TK1 expression was upregulated in thyroid carcinoma tissues, and higher expression of TK1 was correlated with shorter DFS of patients with thyroid carcinoma. Silencing of TK1 suppressed cell proliferation, invasion, migration, and EMT and also induced cell apoptosis in the thyroid carcinoma cell lines. In vivo data showed that TK1 knockdown inhibited in vivo tumor growth of thyroid carcinoma cells. Moreover, the upregulated TK1 in the thyroid carcinoma cells was associated with the downregulation of miR-34a-5p.

The prognostic role of TK1 has been widely studied in the cancer studies (5, 6, 15, 16). The increased Tk1 mRNA levels in the plasma-derived exosomes are associated with clinical resistance to CDK4/6 inhibitors in metastatic breast cancer patients (17). Using immunohistochemistry, TK1 was found to be located on the cellular membrane of the colorectal, lung, and breast cancer cells. The upregulation of TK1 in these malignant tissues indicated the potential prognostic role of TK1 (18). Serum detection of TK1 is sensitive and specific for the prediction of early stage and advanced lung cancer (19). In this study, it was showed that serum TK1 levels were upregulated in patients with thyroid nodules, suggesting that TK1 might be involved in the development of thyroid cancer. Our findings were in accordance with the previous studies that thyroid carcinoma tissues were accompanied with higher expression levels of TK1 (20). Besides, the in vitro studies showed the consistent upregulation of TK1 in thyroid carcinoma cells.

In this study, it was the first time to show that TK1 knockdown suppressed in vitro and in vivo carcinoma cell progression. It was reported in pancreatic cancer the silencing of TK1 suppressed cancer cell proliferation via inducing S phase arrest by P21 upregulation. P21was an inhibitor of the cyclin-dependent kinase and could inhibit the activation of CDK1, CDK2, and CDK4. TK1 might interact with P21 by combining with the C-terminal domain of P21, and promote the proliferation of cancer cells (21). In addition, growth and differentiation factor 15 was considered to be the main downstream mediator of TK1 function, which induced the metastatic attributes of lung cancer cells (22). Taken together, the results in this study might imply that TK1 promoted thyroid carcinoma progression via P21 or growth and differentiation factor 15 by increasing cell proliferation, invasion, and migration.

The bioinformatic prediction results revealed that TK1 could be potentially regulated by miRNAs. MiR-34a-5p was found to be upregulated in the thyroid carcinoma cell lines. Further studies showed that miR-34a-5p repressed TK1 expression in thyroid carcinoma cells via targeting the 3 ′UTR of TK1. Previously, miR-34a-5p was shown to be downregulated in the thyroid carcinoma tissues and played a tumor-suppressive role in the thyroid carcinoma cells (23). In addition, miR-34a-5p also functioned as tumor-suppressive factor in esophageal squamous cell carcinoma (24), non-small cell lung cancer (25), and breast cancer (26). Collectively, the upregulation of TK1 may be related to downregulated miR-34a-5p expression, which contributed to the enhanced progression of thyroid carcinoma.

In summary, the results in this study demonstrated the upregulation of TK1 in thyroid nodules as well as thyroid carcinoma tissues. The downregulated miR-34a-5p in thyroid carcinoma resulted in higher TK1 level, which promoted thyroid carcinoma cell proliferation, invasion, and migration. This study may provide novel insights into the role TK1 played in regulating thyroid carcinoma progression. However, there are still some limitations in this study. For example, although higher level of TK1 was confirmed in thyroid cancer patients, the lower expression level of miR-34a-5p was not detected. In addition, the upregulated TK1 in thyroid cancer patient may be resulted from the downregulated miR-34a-5p, but the upstream and downstream molecular mechanisms were still unclear. To further investigate and elucidate the role of TK1 in thyroid carcinoma progression, future studies are necessary.

# DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

# ETHICS STATEMENT

The studies involving human participants were reviewed and approved, and all the experimental protocols were approved by the Ethics Committee of the First Affiliated Hospital of Southern University of Science and Technology. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by All the animal experiments were approved by the Animal Ethics Committee of First Affiliated Hospital of Southern University of Science and Technology.

# AUTHOR CONTRIBUTIONS

WW, GW, and FD designed and supervised the whole study. CL, JW, and LZ performed the experiments. HH, PZ, and ZP collected the serum samples. FL and JC performed the data analysis. WW wrote the manuscript. All authors approved the manuscript for submission.

# FUNDING

The study was supported by the First Affiliated Hospital of Southern University of Science and Technology, Second Clinical College of Jinan University.

#### REFERENCES


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Liu, Wang, Zhao, He, Zhao, Peng, Liu, Chen, Wu, Wang and Dong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# miR-936 Suppresses Cell Proliferation, Invasion, and Drug Resistance of Laryngeal Squamous Cell Carcinoma and Targets GPR78

Xi-Jun Lin1†, Hui Liu2†, Pei Li 1†, Hai-Feng Wang1†, An-Kui Yang3,4,5, Jin-Ming Di <sup>6</sup> , Qi-Wei Jiang<sup>7</sup> , Yang Yang<sup>7</sup> , Jia-Rong Huang<sup>7</sup> , Meng-Ling Yuan<sup>7</sup> , Zi-Hao Xing<sup>7</sup> , Meng-Ning Wei <sup>7</sup> , Yao Li <sup>7</sup> , Zhi Shi <sup>7</sup> \* and Jin Ye<sup>1</sup> \*

*<sup>1</sup> Department of Otolaryngology-Head and Neck Surgery, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China, <sup>2</sup> Division of Pulmonary and Critical Care, Department of Internal Medicine, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China, <sup>3</sup> Department of Head and Neck, Sun Yat-sen University Cancer Center, Guangzhou, China, <sup>4</sup> State Key Laboratory of Oncology in South China, Guangzhou, China, <sup>5</sup> Collaborative Innovation Center for Cancer Medicine, Guangzhou, China, <sup>6</sup> Department of Urology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China, <sup>7</sup> Department of Cell Biology & Institute of Biomedicine, National Engineering Research Center of Genetic Medicine, Guangdong Provincial Key Laboratory of Bioengineering Medicine, College of Life Science and Technology, Jinan University, Guangzhou, China*

Edited by: *Chang Zou, Jinan University, China*

Reviewed by: *Xuerong Wang, Nanjing Medical University, China Hui Zhang, Shandong University, China*

#### \*Correspondence:

*Zhi Shi tshizhi@jnu.edu.cn Jin Ye yejin@mail.sysu.eud.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *27 June 2019* Accepted: *14 January 2020* Published: *04 February 2020*

#### Citation:

*Lin X-J, Liu H, Li P, Wang H-F, Yang A-K, Di J-M, Jiang Q-W, Yang Y, Huang J-R, Yuan M-L, Xing Z-H, Wei M-N, Li Y, Shi Z and Ye J (2020) miR-936 Suppresses Cell Proliferation, Invasion, and Drug Resistance of Laryngeal Squamous Cell Carcinoma and Targets GPR78. Front. Oncol. 10:60. doi: 10.3389/fonc.2020.00060* MicroRNAs (miRs) play important roles in tumor progression. miR-936 has been reported to suppress cell invasion and proliferation of glioma and non-small cell lung cancer. Nevertheless, the function of miR-936 in laryngeal squamous cell carcinoma (LSCC) remains undiscovered. Hence, our study was to investigate the role of miR-936 in LSCC. In our present research, we have testified that miR-936 was substantially downregulated in LSCC tissues compared with adjacent normal tissues. Furthermore, miR-936 could inhibit proliferation, migration and invasion, and improve the sensitivity to doxorubicin and cisplatin of LSCC cells. Additionally, luciferase reporter assays were performed to confirm that GPR78 was a novel target of miR-936, and the protein expression of GPR78 was obviously inhibited by miR-936 in LSCC cells. In summary, our study indicates that the miR-936/GPR78 axis could be both a diagnostic marker and a therapeutic target for LSCC.

Keywords: laryngeal squamous cell carcinoma, miR-936, GPR78, proliferation, invasion, drug resistance

# INTRODUCTION

Laryngeal cancer ranks at the fourteenth most universal type of cancer in the world (1). It is estimated there over 13,000 new laryngeal cancer cases and 3,000 death cases took place in the United States, while 26,000 new cases and 14,000 deaths in China (1–3). Laryngeal squamous cell carcinoma (LSCC) occupies 85–90% of total malignant tumors in the larynx (4, 5). Despite advances in diagnosis and treatment, the long-term prognosis of LSCC patients has not been satisfactory in the past 20 years (6–8). Accordingly, a profound apprehension of the molecular biological mechanisms involving in LSCC tumorigenesis and development is imminently needed.

MicroRNAs (miRs) are a classical non-coding small RNA playing crucial regulatory roles in diverse pathological and physiological progresses by a posttranscriptional mechanism through binding the 3′ -untranslated regions (3′ -UTR) of target genes (9–12). And miRs can act as tumor oncogenes or suppressors through regulating their target genes that are usually dysregulated in cancer (13–15). In our current study, we have proven that miR-936 expression profile is meaningfully reduced in LSCC specimens, and overexpression of miR-936 suppresses LSCC cells proliferation migration and invasion. Moreover, our results have further exhibited that GPR78 is a direct target of miR-936.

# MATERIALS AND METHODS

#### Patients and Specimens

Twenty-five cases of LSCC and matching normal tissues were acquired from patients at the Department of Otolaryngology-Head and Neck Surgery, the Third Affiliated Hospital, Sun Yat-sen University and the Department of Head and Neck, Cancer Center, Sun Yatsen University between December 2013 and February 2017. All patients were diagnosed as LSCC for the first time who underwent total or partial laryngectomy without chemical therapy or neoadjuvant radical before and after surgery. Signed informed approvals were acquired from patients, and the study was approved by the ethics committee of the Third Affiliated Hospital, Sun Yat-sen University.

FIGURE 1 | Downregulation of miR-936 in LSCC is correlated with T stages, differentiation and lymph node metastasis. (A) Expression of miR-936 in 25 pairs of LSCC tissues and adjacent normal tissues was detected using RT-qPCR. The relative miR-936 expression in two groups of LSCC tissues classified by age (B), T stage (D), and lymph node metastasis (F) were analyzed with Mann-Whitney *U*-test. The relative miR-936 expression in three groups of LSCC tissues classified by differentiation (C) and primary location (E) were analyzed with Kruskal-Wallis test. Data are presented as mean ± SD or median with the interquartile range. \**p* < 0.05; \*\**p* < 0.01; NS, no statistical significance.

#### Cell Culture and Reagents

The human LSCC cell line Hep-2, the normal bronchial epithelium cell line 16HBE and the HEK293T were ordered from China Center for Type Culture Collection (CCTCC). The human LSCC cell line KB-3-1 was kindly provided by Dr. Zhesheng Chen

TABLE 1 | Relationship between miR-936 expression level and clinicopathologic parameters.


*<sup>a</sup>Scores determined by qRT-PCR in mean* ± *SD.*

*<sup>b</sup>Student's T- test (for 2 groups) or one way ANOVA (for* > *2 groups).*

(St. John's University, USA). The cells were cultured at 37◦C with 5% CO<sup>2</sup> in a humidified atmosphere in Dulbecco's modified Eagle's medium (DMEM) containing 100 Unit/ml penicillin, 100 ng/ml streptomycin and 10% fetal bovine serum (FBS). Cell lines applied in this project were authenticated by short tandem repeat fingerprinting <3 months when this study was started. Anti-GPR78 (AB61731a) was from Sangon Biotech. The antibody of anti-GAPDH (KM9002) was purchased from Tianjin Sungene Biotech.

#### Plasmid

The synthesized precursor hsa-miR-936 was cloned into lentiviral vector pLKO.1-GFP to generate the hsa-miR-936 lentivirus construct. The GPR78 3′UTR fragment was cloned into a psiCHECK-2 dual luciferase reporter construct. Lentivirus was packaged with HEK293T cells and harvested from the supernatant of medium. Stable cell line was obtained through infecting lentivirus in Hep-2 or KB-3-1 cells and selecting with puromycin.

### RNA Extraction and Real-Time Quantitative PCR (RT-qPCR)

Total RNAs were extracted from cells and tissues by applying HiPure Total RNA Mini Kit (Magen). Reverse transcription was performed with HiFi-script cDNA kit (Cwbio) according to the instruction. The BestarTM Real time PCR Master Mix was applied for RT-qPCR by SYER Green Method. All reactions were carried out in triplicate and repeated at least three independent times. The results of RT-qPCR were normalized to U6 by applying the 2−11Ct method. The following primers were ordered from Sangon Biotech:

cells expressed vector control and miR-936. (B) Cell proliferation of the two LSCC cell lines was measured using the MTT assay. OD values were measured every 24 h for 96 h with or without miR-936 transfection. Data are presented as mean ± SD. Student's *t*-test was used for statistical analysis. \**p* < 0.05; \*\**p* < 0.01.

miR-936 forward: 5′ -AACGAGACGACGACAGAC-3′ ; miR-936 reverse: 5′ -ACAGTAGAGGGAGGAATCGCAG-3′ ; U6 forward: 5′ - GCGCGTCGTGAAGCGTTC-3′ ; U6 reverse: 5′ - GTGCAGGGTCCGAGGT-3′ (16).

#### Western Blot Analysis

Cells were washed with PBS, resuspended and lysed in RIPA buffer containing protease inhibitors (0.03% aprotinin, 10 ng/ml PMSF, 1µM sodium orthovanadate, 1% NP-40, 0.1% SDS and 0.5% sodium deoxycholate,) at 4◦C for 30 min. After centrifuging at 14,000 × g for 10 min, lysate supernatants were collected and stored at −80◦C. Proteins were isolated by 12% SDS-PAGE gels and transferred to the membrane of polyvinylidene difluoride. After that, membrane was blocked by 5% BSA for 1 h, then incubated with the primary antibody and secondary antibody, successively. According to instruction, signal was measured through the chemiluminescent gel imaging system of ChemiDoc XRS (Bio-RAD) (17, 18).

#### MTT Assay

Cells plated in 96-well plate were incubated for 0, 1, 2, 3 days, then add MTT 0.5 mg/ml. After 4 h incubation at 37◦C, carefully

absorb the culture medium in the well to prevent the cells from taking away formazan crystals, and dissolved crystals with 100 µl of DMSO. Multiscan Spectrum (Thermofisher) was used to measure the absorbance at 570 nm (19, 20).

# Wound Healing Assay

Cells were cultured in 6-well plate. Until the cells reached 80– 90% confluence, using a sterile 10 µl pipette tip to draw a straight mark on the cell monolayer. After drawing, the delineated cells were washed and incubated in serum-free medium. The gap of wounds were measured by microscopic photograph at a certain time (21).

#### Transwell Assay

Cells were plated in the upper compartment containing matrigelcoated polycarbonate membrane filter of a modified Boyden chamber (Corning), and the lower chamber was plated complete medium, and allowed to migrate for 24 h. Wiped the cells on the upper surface of membrane and fixed the cells on the lower surface of membrane by 4% paraformaldehyde and stained by 0.1% crystal violet staining solution (22).

#### Luciferase Reporter Assay

The GPR78 mutated and wild-type (WT) 3′ -UTR fragments were cloned into psiCHECK-2 reporter. HEK293T cells were plated in 24-well plates and co-transfected with mutated or WT 3′ -UTR luciferase reporter and pLKO.1-GFP-miR-936. After 24 h, according to a protocol, cell lysates were obtain, and Firefly/Renilla luciferases ratios were measured with the Dual Luciferase Reporter Assay Kit (Promega) (23).

# Statistical Analysis

All statistics were analyzed by SPSS 20.0, and the results were shown as mean ± SD or median with the interquartile range. The Student's t-test and Mann–Whitney U-test were applied to analyze comparisons of two groups, and oneway ANOVA and Kruskal-Wallis test were applied to analyze comparisons of multiple groups. The P < 0.05 were considered statistically significant.

# RESULTS

#### Downregulation of miR-936 in LSCC Is Correlated With Differentiation, Lymph Node Metastasis and T Stages

To explore miR-936 expression in the LSCC tissues, RT-qPCR was used to check with 25 pairs of laryngeal cancer and normal tissue. Results suggested that miR-936 expression was meaningfully downregulated in LSCC, with 72% (18/25) of

the tumor tissues showing reduced expression compared to matched normal controls (**Figure 1A**). Further, we found that miR-936 expression was correlated with tumor grade, lymph node metastasis and T Stages, but not correlated with tumor primary locations and age. The expression of miR-936 in negative lymph node metastasis, well-differentiation and T1-2 groups were higher than that in positive lymph node metastasis, poor differentiation, and T3-4 groups respectively (**Table 1** and **Figures 1B–F**). According to these data, the progression of LSCC may be associated with miR-936 expression.

#### Overexpression of miR-936 Suppresses the Proliferation of LSCC Cells

To investigate the function of miR-936 in LSCC, Hep-2 and KB-3-1 cells were infected with lentivirus expressing precursor miR-936, which successfully upregulated miR-936 in the cells (**Figure 2A**). The growth curves determined by MTT assay indicates that the proliferation abilities of Hep-2 and KB-3-1 cells were significantly attenuated when miR-936 was overexpressed (**Figure 2B**).

#### Overexpression of miR-936 Suppresses the Migration and Invasion of LSCC Cells

To further verify whether miR-936 has an influence on the migration and invasion of LSCC cells, we performed wound healing and transwell assays in Hep-2 and KB-3-1 cells with miR-936 overexpression. The outcomes revealed that the migration and invasion of miR-936 overexpressing cells were importantly decreased when compared with control cells (**Figures 3A–C**).

#### Overexpression of miR-936 Improves the Drug Sensitivity of LSCC Cells to Doxorubicin and Cisplatin

To verify the effect of miR-936 on LSCC cells treated with chemotherapy drugs, we treated indicated cells with doxorubicin or cisplatin in different concentrations. As shown in **Figures 4A–D**, the drug resistance to doxorubicin or cisplatin was significantly lower in cells overexpressing miR-936 in comparison with control groups in Hep-2 and KB-3-1 cells. These data suggested that increasing miR-936 expression could improve the drug sensitivity of LSCC cells to chemotherapeutic drugs.

### miR-936 Directly Targets GPR78

To understand the mechanism of miR-936 as a tumor suppressor in LSCC, we combined RNAhybird and PITA to search the new potential targets of miR-936. Both algorithms reveal that GPR78 was a downstream gene of miR-936. We then performed western blot analysis and found that overexpressing miR-936 in Hep-2 and KB-3-1 cells could decrease GPR78 protein levels notably (**Figure 5A**). The interaction between miR-936 and the 3′ -UTR of GPR78 was illustrated in **Figure 5B**. And luciferase reporter assays were used in HEK293T cells to test whether miR-936 could directly interact with the 3′ -UTR of GPR78. The ratio of

fluorescence activity indicates the inhibitory effect of miR-936 on GPR78 in wild or mutant 3′ -UTR. As exhibited in **Figure 5C**, overexpression of miR-936 markedly suppressed the luciferase activity of GPR78 WT 3′ -UTR compared to mutant 3′ -UTR. The result above indicated that miR-936 directly suppresses GPR78 expression through binding its 3′ -UTR (2214nt∼2222nt).

#### DISCUSSION

The development of tumors is a synergistic process of tumorassociated activation and inhibition genes. A growing number of studies have shown that abnormal expression of miRs happened in most types of human malignancies, including LSCC (24). In our present study, miR-936 was downregulated in LSCC tissues in comparison with matching normal tissues, and correlated with poor clinical features, suggesting that miR-936 might be associated with tumor progression in LSCC. Biological function experiments indicated that overexpression of miR-936 meaningfully inhibited the proliferation, migration, and invasion of LSCC cells. Furthermore, overexpression of miR-936 improved the drug sensitivity of LSCC cells to doxorubicin and cisplatin which currently are used for LSCC chemotherapy in clinic, indicating the significance of the findings that miR-936 sensitized the anticancer effects of doxorubicin and cisplatin in LSCC. Moreover, computerized algorithm predicted that GPR78 was a direct downstream target of miR-936, and that overexpression of miR-936 can effectively reduce GPR78 protein expression in LSCC cells. Previous researches have manifested that down-regulated miRs are often described as biomarkers or therapeutic targets in laryngeal squamous cell carcinoma. For instance, we recently reported t miR-194 works as a tumor suppressor in LSCC through targeting Wee1 (23). The expression of miR-375 is also reduced in LSCC tissues, and overexpression of miR-375 could reduce LSCC cell proliferation, motility and invasion by targeting IGF1R (25). In addition, miR-34a/c is also downregulated in LSCC tissues and inhibits LSCC cells proliferation by inducing cell cycle arrest through directly suppressing GALNT7 (26). While miR-27a is upregulated in LSCC specimens, and overexpression of miR-27a enhances LSCC cells proliferation by targeting PLK2 (27). MiR-93 is also upregulated in LSCC tissues and promotes the proliferation, migration, and invasion by inhibiting cyclin G2 (28).

Previous studies have been reported that miR-936 were down-regulated in glioma and induced cell cycle arrest via targeting CKS1 (29). Another report has also demonstrated that miR-936 could directly target E2F2 to inhibit the invasion and proliferation of non-small cell lung cancer cell (30). However, the biological functions of miR-936 have not been described in other tumors, including LSCC. Our study confirmed that miR-936 expression profile was downregulated in LSCC tissues, which indicates that miR-936 may be related to the progression

#### REFERENCES

1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. (2019) 69:7–34. doi: 10.3322/caac.21551

of LSCC. Searching miRNA target genes is fundamental to understanding its carcinogenic regulatory mechanism and the effective molecular therapeutic targets. By bioinformatics algorithm and experiment validation, we testified that GPR78 was a direct target of miR-936. GPR78, an orphan G-protein coupled receptor, is situated in an area of chromosome 4p where have been shown connection to schizophrenia and bipolar affective dysfunction (31). Recently, GPR78 was testified highly expressed in lung cancer cell, and knockdown of GPR78 prominently suppressed cell migration and metastasis (32). However, the expression and function of GPR78 in LSCC need to be further investigated in the future.

In conclusion, our results proved that miR-936 attenuated the proliferation, migration and invasion of LSCC cells and targeted GPR78. Moreover, high level of miR-936 could improve the sensitivity of LSCC cells to doxorubicin and cisplatin. The results manifests that the miR-936/GPR78 axis could play as a novel biomarker and a therapeutic target of LSCC.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the ethics committee of The Third Affiliated Hospital, Sun Yat-sen University. The patients/participants provided their written informed consent to participate in this study.

#### AUTHOR CONTRIBUTIONS

X-JL, HL, PL, H-FW, ZS, and JY designed the projects, conducted the experiments, analyzed the results, and wrote the manuscript. A-KY, J-MD, Q-WJ, YY, J-RH, M-LY, Z-HX, M-NW, and YL conducted the experiments. All authors read and agreed the final manuscript.

#### FUNDING

This work was funded by grants from the National Natural Science Foundation of China Nos. 81772540 (ZS), 81772752 (J-MD), 81472760 (HL), and 81902771 (PL), the Science and Technology Program of Guangdong Nos. 2019A050510023 (ZS), 2017A020215122 (J-MD), 2014A030313057 (JY), and 2014A020212078 (HL), the Medical Scientific Research Foundation of Guangdong Province No. 2018118215914106 (PL), the Science and Technology Program of Guangzhou No 201709010038 (J-MD).

2. Feng RM, Zong YN, Cao SM, Xu RH. Current cancer situation in China: good or bad news from the 2018 Global Cancer Statistics? Cancer Commun. (2019) 39:22. doi: 10.1186/s40880-019- 0368-6


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Lin, Liu, Li, Wang, Yang, Di, Jiang, Yang, Huang, Yuan, Xing, Wei, Li, Shi and Ye. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# 7-Methoxy-1-Tetralone Induces Apoptosis, Suppresses Cell Proliferation and Migration in Hepatocellular Carcinoma via Regulating c-Met, p-AKT, NF-κB, MMP2, and MMP9 Expression

Ying Wen1,2, Xiaoyan Cai <sup>2</sup> , Shaolian Chen<sup>3</sup> , Wei Fu1,2, Dong Chai 1,2, Huainian Zhang1,2 and Yongli Zhang1,2 \*

#### Edited by:

*Chang Zou, Shenzhen People's Hospital, Jinan University, China*

#### Reviewed by:

*Sabarish Ramachandran, Texas Tech University Health Sciences Center, United States Yuan Ye, Affiliated Hospital of Guilin Medical University, China Juan Cen, Henan University, China*

> \*Correspondence: *Yongli Zhang zyl28\_gdpu@163.com*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *27 October 2019* Accepted: *13 January 2020* Published: *07 February 2020*

#### Citation:

*Wen Y, Cai X, Chen S, Fu W, Chai D, Zhang H and Zhang Y (2020) 7-Methoxy-1-Tetralone Induces Apoptosis, Suppresses Cell Proliferation and Migration in Hepatocellular Carcinoma via Regulating c-Met, p-AKT, NF-*κ*B, MMP2, and MMP9 Expression. Front. Oncol. 10:58. doi: 10.3389/fonc.2020.00058* *<sup>1</sup> Guangzhou Key Laboratory of Construction and Application of New Drug Screening Model Systems, Guangdong Pharmaceutical University, Guangzhou, China, <sup>2</sup> Department of Cell Biology and Medical Genetics, School of Life Sciences and Biopharmaceutics, Guangdong Pharmaceutical University, Guangzhou, China, <sup>3</sup> Department of Clinical Laboratory, The First Affiliated Hospital, Guangdong Pharmaceutical University, Guangzhou, China*

This study aimed to determine the anti-proliferative and anti-migratory effects of 7-methoxy-1-tetralone (MT) in hepatocellular carcinoma (HCC) cells. MTT assay assessed HCC cell viability; cell apoptosis of HCC cells was determined by flow cytometry; wound healing assay evaluated HCC cell migratory ability; protein expression levels were assessed using western blot assay; the *in vivo* antitumor effects of MT were tested in BALB/c nude mice and the pathological changes within the tumor tissues were evaluated by immunohistochemistry. MT treatment significantly suppressed the cell proliferative and migratory potentials of HepG2 cells, and induced HepG2 cell apoptosis. The western blot assay showed that MT treatment caused a suppression on c-Met, phosphorylated AKT (p-AKT), NF-κB, matrix metallopeptidase 2 (MMP2)/MMP9 protein levels in HepG2 cells. Further *in vivo* animal studies deciphered that MT treatment suppressed tumor growth of HepG2 cells in the nude mice, but had no effect on the body weight and the organ index of liver and spleen. Further immunohistochemistry analysis of the dissected tumor tissues showed that MT treatment significantly suppressed the protein expression levels of NF-κB, MMP9, MMP2, and p-AKT. In summary, the present study demonstrated the anti-tumor effects of MT on the HCC, and MT suppressed HCC progression possibly via regulating proliferation- and migration-related mediators including c-Met, p-AKT, NF-κB, MMP2, and MMP9 in HepG2 cells.

Keywords: 7-methoxy-1-tetralone, hepatocellular carcinoma, cell proliferation, cell apoptosis, cell migration

# INTRODUCTION

Hepatocellular carcinoma (HCC) represents one of the most severe human malignancies with a high mortality (1). High risk factors such as hepatitis B virus (HBV) or hepatitis C virus (HCV) infection, alcohol abuse, hemochromatosis psychosis, and so on, may lead to a chronic liver disease, then with progression to cirrhosis, eventually leading to the occurrence of HCC (2–4).

**252**

Surgical interventions (surgical resection, liver transplantation, and locoregional therapies) have been playing important roles in the treatment of HCC (5, 6). However, only 15% patients are eligible for the potentially curative treatments since a majority of patients diagnosed with HCC already have liver dysfunction and/or are at advanced stages (III or IV), and these patients cannot benefit from these interventions (7, 8). Sorafenib is the only FDA-approved medication for the management of advanced HCC (9, 10). However, a large percentage of patients still experience disease progression after sorafenib treatment (11). Therefore, there is still a large unmet medical need to develop new therapeutic strategies to treat HCC.

7-methoxy-1-tetralone (MT) is mainly used as the organic material or intermediate in chemical engineering currently (12–15). Our research group revealed that wild Juglansshurica leather has strong anti-tumor effects through the long-term study, while MT may be one of the effective components. Sun et al. (16) revealed that the compounds extracted from green peel of Juglans mandshurica possessed the insecticidal activities, further investigation deciphered that MT is one of the major active components (the relative content: 6.81%). Recently, studies showed that extracts from green peel of J. mandshurica exhibited moderate inhibitory effects on the lung cancer cells (17) Nevertheless, systematic study of MT's potential to repress human hepatoma cell growth has not been documented.

This study was undertaken to gain deeper insights into the anti-hepatocellular carcinoma activities and anti-neoplastic molecular mechanisms of MT. Changes to cell proliferation, apoptosis and migration and AKT, phosphorylated AKT (p-AKT), NF-κB, and matrix metallopeptidase 2 (MMP2)/MMP9 protein expression following application of MT are defined in this study using in vitro cell culture and in vivo animal experiments, in order to provide the experimental basis for its future clinical application.

# MATERIALS AND METHODS

#### Cell Culture and Chemical Reagents

The two human hepatoma cell lines (HepG2 and LO2) were a generous gift from Sun Yat-sen University. HepG2 and LO2 cells were kept in DMEM (Thermo Fisher Scientific, Waltham, USA) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific) in a humidified incubator (Sanyong, Tokyo, Japan) with 5% CO<sup>2</sup> at 37◦C. Dimethyl sulfoxide (DMSO) was used to dissolve MT (purity > 98%; Sigma-Aldrich, St. Louis, USA) to prepare the stock solution, and the stock solution was diluted with cell culture medium as the respective working concentrations, and the concentration of DMSO in the working solution was <0.1% (18).

#### Cell Viability Assay

The anti-proliferative effects of MT were evaluated by MTT assay. LO2 and HepG2 (1 × 10<sup>5</sup> cells/well) were seeded at 96-well plates. The seeded cells were subjected to incubate with different concentrations of MT (31.25, 62.5, 125, 250, 500, and 1,000 µM) for 24, 48, and 72 h, respectively. Fluorouracil (5-FU, 50µM) served as a positive control. After 4 h incubation with MTT (5 mg/ml) at 37◦C. Cell viability was evaluated by measuring the absorbance at 570 nm.

#### Flow Cytometry Analysis of Cell Apoptosis

Apoptosis was determined using flow cytometer with a commercial Annexin V-FITC Apoptosis Detection Kit (KaiJi, Nanjing, China) by following the manufacture's protocol. In brief, HepG2 cells were subjected to treatment with different concentrations of drugs for 48 h after plating as a monolayer. Cells were rinsed twice with cold phosphate buffered saline (PBS) and trypsinized gently using the trypsin reagent, then cells were re-suspended in 1× binding buffer and were incubated FITC Annexin V and propidium iodide (PI) for 15 min at room temperature in the dark. A BD FACSCalibur flow cytometer (BD Biosciences, Franklin Lakes, USA) was used to identify the different subpopulations of apoptotic cells.

#### Wound Healing Assay

HepG2 cells after different treatments were allowed to grow in 6-well plates until ∼90% confluence. A sterile 200 µL pipette tip was used to create a wound in the HepG2 cell monolayer. HepG2 cells were rinsed twice with PBS to remove debris, and HepG2 cells were incubated with serum-free medium for indicated time durations. At indicated time points, images of the plates were acquired under a microscope and the migrating distances were analyzed by Image-Pro-Plus software (19).

#### Western Blot Analysis

Total proteins were obtained by lysing the cells or tissues using RIPA buffer (Beyotime). The BCA quantitative analysis kit was used to measure concentrations of protein samples (Beyotime). Equal aliquots of protein samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and the proteins were electro-transferred onto a PVDF membrane (Millipore, Burlington, USA). The PVDF membrane was probed with different primary antibodies, and protein bands were visualized by an enhanced chemiluminescence method (Thermo Fisher Scientific). Antibodies for c-Met (Cat. #4560), Akt (Cat. #4685) p-Akt (Cat. #4060), NF-κB (Cat. #8242), MMP2 (Cat. #40944), MMP9 (Cat. #13667), β-actin (Cat. #8457) and the corresponding secondary antibodies were obtained from Cell Signal Technology Inc. (Danvers, USA).

#### In vivo Animal Studies to Assess Tumor Growth

Five-week-old BALB/c nude mice (n = 25) used for in vivo experiments were purchased from the Guangdong Laboratory Animal Center. HepG2 cells (7 × 10<sup>6</sup> cells/ml) were subcutaneously administered into the right sub-axillary region of the mice. When the diameters reached a length >3 mm, mice were randomly assigned to five groups and administered with different drugs by the intraperitoneal injection. Mice from the control group mice received saline, mice from the positive control group mice received 5-FU (15 mg/kg/d) and mice from the experimental group received MT (80, 120, or 160 mg/kg/d). Subcutaneous tumors were measured using calipers, and tumor volume was assessed using the following formula: width<sup>2</sup> × Wen et al. 7-Methoxy-1-Tetralone and Hepatocellular Carcinoma

length × 0.5 (20). Body weight was measured every day. Mice received 19 doses in total and were sacrificed for harvesting tumors at 24 h after the last dose. The organ index for liver and spleen was determined as organ weight (g)/body weight (g). All the experimental procedures of the animal studies were under the approval by the Animal Ethics Committee of Guangdong Pharmaceutical University.

#### Hematoxylin and Eosin (HE) Staining Assay

HE staining assay was used to observe pathological changes within the tumor tissues. Tumor tissues were embedded in paraffin, sectioned into 4µm slices and deparaffinized in xylene for 15 min followed by re-hydration in the graded ethanol. After staining with hematoxylin for 5–10 min, the slides were incubated with 1% hydrochloride alcohol for 20 s. Following 5– 10 min running water washing, the slides were stained with 0.5% eosin for 3 min. The morphology of the HE-stained tumor tissues were evaluated using a light microscope.

#### Immunohistochemical Analysis of Tumor Tissues

The paraffin sections were deparaffinized using xylene and hydrated through graded ethanol. After antigen retrieval, 3% H2O<sup>2</sup> was used to quench endogenous peroxidase. After PBS washing, slides were incubated with 10% bovine serum albumin (BSA) for 10 min, and then primary antibodies including MMP2 (1:200), MMP9 (1:200), p-Akt (1:500) and NF-κB (1:100) were added to incubate with the slices overnight at 4◦C. After that, slides were incubated with secondary antibodies for 30 min at room temperature. 3,3′ - diaminobenzidine (DAB) as a chromogen was used to visualize the antigens. After counterstaining with hematoxylin, the slides were dehydrated and mounted for viewing. For the negative control, primary antibody was replaced by BSA. The percentage of positive staining area was measured by Image J.

#### Statistical Analysis

A SPSS statistical software package (IBM, Armonk, USA) was used for performing data analysis. Results data are presented as the mean ± standard deviation. Differences between the mean values for the different groups were determined using one-way ANOVA followed by Dunnett's multiple comparison test. A Pvalue less than 0.05 was defined to be statistically significant.

# RESULTS

#### MT Inhibited Proliferation of HepG2 Cells but Has Little Cytotoxic Effects on Normal LO2 Cells

The cytotoxic effects of MT on LO2 cells were determined by treating cells with different concentrations of MT (0–1,000µM) for 24, 48, 72 h, respectively. In addition, HepG2 cells were subjected to incubate with elevated MT concentrations (0– 500µM) for 48 h. 5-Fluorouracil (5-Fu, 50µM) was used as a positive control in this investigation. As shown in **Figure 1A**, increasing concentrations of MT exhibited anti-proliferative actions in LO2 cells in a time- and concentration-dependent

LO2 cells at after 48 h treatment with MT were determined by MTT assay. Data from at least three independent experiments performed in triplicates are presented; \**P* < 0.05 and \*\**P* < 0.01 compared to the corresponding control groups.

way. Nevertheless, the 48 h treatment duration was selected for our follow-up study because MT had the greatest dosedependent effect on LO2 cells at 48 h after treatment. The results in **Figure 1B** indicated that MT was more effective to inhibit the proliferation of HepG2 cells when compared with LO2 cells. For subsequent experiments three different concentrations for MT (40,100, 250µM) were used.

# Effects of MT on the Cell Apoptosis of HepG2 Cells

To validate the effects of MT on HepG2 cell apoptosis, flow cytometry was performed to analyze Annexin V-FITC/PI stained HepG2 cells. MT treatment mildly increased the number of apoptotic cells (AV+PI+ and AV+PI- cells) compared to the control group (**Figure 2**). The percentage of apoptotic HepG2

cells following treatment with 0, 40, 100, or 250µM MT for 48 h were 5.44 ± 0.84, 5.07 ± 1.22, 7.15 ±1.92, and 11.45 ±1.11%, respectively (**Figure 2**). On the other hand, 5-FU induced much more HepG2 apoptotic cells than that treated with MT (**Figure 2**).

#### MT Suppressed Migration in HepG2 Cells

Wound healing assay was undertaken to evaluate the effect of MT on HepG2 cell migration. As demonstrated in **Figure 3**, the cells of different concentration group were tightly connected at 0 h. The scratch of the control group, 40µM low dose group and 5- FU group was significantly prolonged at 12, 24, and 36 h after drug treatment, while 100 and 250µM groups were not obvious, indicated that the healing of the scratch was significantly reduced following treatment with 100 and 250µM MT (**Figure 3**).

#### Effects of MT on c-Met, NF-κB, p-Akt, Akt, and MMP2/MMP9 Protein Expression in HepG2 Cells

To assess the significance of the expression patterns of proliferation and migration-related proteins in response to MT, HepG2 cells were subjected to treatment with elevated MT concentrations (0, 40, 100, and 250µM) for 48 h, and expression levels of c-Met, AKT, p-AKT, NF-κB, and MMP2/9, which closely

involved in the regulation of proliferation and migration were demonstrated in **Figure 4**. The expression of c-Met, p-Akt, NFκB, and MMP2/9 significantly decreased compared with control group after 48 h of treatment with HepG2 cells with 40, 100, and 250µM MT (**Figure 5B**), while the expression change of Akt protein was not obvious. MT at concentration of 40µM significantly repressed c-Met and NF-κB protein expression levels (**Figure 4**); MT at a concentration of 100µM significantly downregulated p-AKT, NF-κB protein expression (**Figure 4**); MT at a concentration of 250µM inhibited c-Met, p-AKT, NF-κB, and MMP2/MMP9 protein expression levels (**Figure 4**); while the AKT protein levels were not affected by MT treatment in HepG2 cells (**Figure 4**).

#### MT Inhibits Tumor Growth in vivo

BALB/c nude mice were sacrificed after 19 days receiving MT, 5- FU, or saline by intraperitoneal injection. MT treatment failed to affect mice body weight, while 5-FU treatment reduce the mice body weight (**Figure 5A**). The photos of representative tumor images from different treatment groups are demonstrated in **Figure 5B**. Tumor volume and tumor weight were significantly reduced by MT and 5-FU treatment when compared to the control group (**Figures 5C,D**). The tumor inhibition rates were 40.57% (80 mg/kg MT group), 51.43% (120 mg/kg MT group), 79.43% (160 mg/kg MT group), and 89.71% (5-FU group). MT and 5-FU had no effects on the organ indexes of liver or spleen (**Figures 5E,F**).

# Pathological Morphology and Relative Protein Expressions in vivo

Pathological morphology of tumor tissues was detected in saline group, MT (80, 120, 160 mg/kg) groups and 5-FU group and were presented in **Figures 6A,B**. In the saline control group, tumor cells were mainly intact. While, in the MT and 5-FU group, some of tumor cells in the exposed area were characterized with necrosis and cell fragmentation. These observations suggested that the cells were in a vigorous growth stage. However, in different dose of MT and 5-Fu groups, the typical increased nuclei volume and shrunken intercellular space were considerably reduced, and some chromosome fractures and integrated nuclear membranes were observed. Immunohistochemistry staining (**Figure 6C**) presents the four proteins (p-Akt, NF- κB, MMP9, and MMP2) expressions in saline group, 160 mg/ml MT group and 5-FU group, respectively. The percent of positive staining in

saline group are: p-Akt (2.91 ± 0.96%), NF-κB (35.47 ± 1.78%), MMP9 (7.64 ± 1.27%), and MMP2 (9.69 ± 1.34%). The positive staining ratio of the four proteins in MT (160 mg/kg) group were all significantly lower; the positive expression of the four proteins in 5-FU group was also remarkably repressed (**Figures 6D–G**).

#### DISCUSSION

In the current investigation, we demonstrated the antitumor effects of MT against HCC progression and deciphered the underlying molecular mechanisms. Our results illustrated that MT treatment concentration-dependently repressed HCC cell viability. Mechanistically, MT downregulated NF-κB, Akt, p-Akt, MMP2/MMP9 expression levels in HCC cells. This inhibitory effect was further verified in a BALB/c nude mouse tumorhypodermic transplantation model. Thus, MT requires further assessment as an effective strategy for the HCC treatment.

HCC metastasis largely contributed to the recurrence of this human malignancy (21). The key process for tumor metastases involves the dissolved surrounding tumor matrix and basement membrane caused by tumor-associated proteases (22). Therefore, MTT assay and wound healing assay was carried out to detect the HepG2 cell growth and migratory abilities, respectively, after treatment with MT in our study. Our results elucidated that MT remarkably repressed the proliferation and migration of HepG2.

Potential molecular mediators that were involved in the inhibitory actions of MT on HepG2 cell proliferation and migration were analyzed. MMPs has been well-documented for their regulatory actions on tumor metastasis. Studies found that dysregulated MMP2/9 in the solid tumors largely contributed to the tumor metastasis including HCC (23, 24). Additionally, the cell growth inhibitory effect could also attributed to the inhibition of MMPs (25). C-Met is a high-affinity receptor and can be targeted activated by hepatocyte growth factor (HGF). The activation of c-Met involves various signaling pathways including PI3K/Akt and MAPK/Erk signaling pathways, which are key mediators in HCC cell proliferation and metastasis (26– 28). Moreover, studies also illustrated that deregulated c-Met is associated with the invasiveness and progression of HCC (29– 31) and anti-tumor mechanisms of many components extracted from Chinese herbal medicine and plants were related to the expression level of HGF/c-Met (31–38). MT showed moderate enhancing effects on the HepG2 cells apoptosis, but remarkably inhibited the migratory potential of HepG2 cells, suggesting that the inhibitory actions of MT may be more relevant to the tumor metastasis. In this study, we further investigated into the PI3K/AKT/NF-κB signaling pathway, which represents a key pathway in regulating HCC progression. Activation of PI3K/AKT/NF-κB signaling pathway is effective to enhance HCC cell proliferative, invasive and migratory potentials. Activation of NF-κB requires the Akt phosphorylation, which stimulates the IκB kinase complex, phosphorylates and inactivates IκB (39–42). There is evidence showing that NF-κB up-regulated MMP-9 expression (43–45), while the NF-κB inhibition could downregulate MMP-2 expression (46). In current investigation, MT treatment remarkably repressed Akt phosphorylation and NF-κB activation, which may lead to the suppressed MMP2/9 expression in HepG2 cells. Collectively, our results revealed a potential mechanism of MT-mediated tumor-suppressive actions in HCC.

From a clinical perspective, evaluation of chemotherapeutics should consider their antitumor effects as well as financial cost and adverse effects. In the current investigation, we revealed that MT did not affect mice body weight in comparison with 5-Fu treatment, which may suggest the potential application of MT in antitumor therapy with limited adverse effects. In addition, our data implied that MT treatment of HCC may be acted via repressing c-Met/PI3K/AKT pathway, thereby inhibited MMP2/9.

There are several study limitations for the current work. MT induced moderate apoptosis in the HepG2 cells, and the molecular mechanisms underlying MT-mediated HCC cell apoptosis require further examination. The anti-tumor effects of MT were determined in one HCC cell line, and other types of HCC cell lines should be employed to confirm the anti-tumor effects of HCC. The expression levels of p-AKT, NF-κB, MMP9/2 were only quantified by IHC in the tumor tissues, and further studies may employ western blot assay to determine the levels of these proteins in the tumor tissues.

In summary, the present study demonstrated the antitumor effects of MT on the HCC, and MT suppressed HCC progression possibly via regulating proliferation and migrationrelated mediators including c-Met, p-AKT, NF-κB, MMP2, and MMP9 in HepG2 cells. However, further examination is required to assess the therapeutic potential of MT in HCC.

# DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

#### ETHICS STATEMENT

The animal study was reviewed and approved by Animal Ethics Committee of Guangdong Pharmaceutical University.

#### AUTHOR CONTRIBUTIONS

YZ and YW designed the study and wrote the manuscript. YW, XC, and SC performed the experiments and collected data. WF and DC performed statistical analysis. HZ performed the animal studies. All authors read and revised the manuscript before submission.

#### REFERENCES


#### FUNDING

This study was supported by the projects of Guangzhou Key Laboratory of Construction and Application of New Drug Screening Model Systems (No. 201805010006) and Key Laboratory of New Drug Discovery and Evaluation of Ordinary Universities of Guangdong province (No. 2017KSYS002).

evaluation. Pharmacogn Mag. (2017) 13:222–5. doi: 10.4103/0973-1296.2 04566


carcinoma cells via PKC-cMET-ERK1/2-COX-2-PGE2 pathway. Int Immunopharmacol. (2016) 33:24–32. doi: 10.1016/j.intimp.2016.01.027


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Wen, Cai, Chen, Fu, Chai, Zhang and Zhang. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Change of Pathological Type to Metaplastic Squamous Cell Carcinoma of the Breast During Disease Recurrence: Case Report and Literature Review

Tianhui Guo<sup>1</sup> , Zhiying Chen<sup>2</sup> , Jinpeng Xu<sup>2</sup> and Yongchun Zhang<sup>2</sup> \*

<sup>1</sup> Department of Radiation Oncology, Laoshan Branch of Affiliated Hospital of Qingdao University, Qingdao, China, <sup>2</sup> Department of Radiation Oncology, The Affiliated Hospital of Qingdao University, Qingdao, China

Background: Metaplastic squamous cell carcinoma (SCC) of the breast is a rare and heterogeneous group of primary breast malignancies. The etiology, pathogenesis, and proper treatment for this kind rare breast cancer are still unclear.

#### Edited by:

Pascale Cohen, Université Claude Bernard Lyon 1, France

#### Reviewed by:

Daniele Vergara, University of Salento, Italy Shiyu Song, Virginia Commonwealth University Health System, United States

> \*Correspondence: Yongchun Zhang zyc18661805058@163.com

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 26 September 2019 Accepted: 09 January 2020 Published: 25 February 2020

#### Citation:

Guo T, Chen Z, Xu J and Zhang Y (2020) Change of Pathological Type to Metaplastic Squamous Cell Carcinoma of the Breast During Disease Recurrence: Case Report and Literature Review. Front. Oncol. 10:32. doi: 10.3389/fonc.2020.00032 Case presentation: We reported a case of a 55-year-old woman with a palpable lump in the inner quadrant of the right breast. She underwent a right breast mass resection and sentinel lymph node biopsy, which revealed that the tumor was an invasive ductal carcinoma, followed by four cycles of doxorubicin plus cyclophosphamide and four cycles of docetaxel as adjuvant chemotherapy, and then simultaneous integrated boost intensity modulated radiotherapy to the whole right breast. After 2 years' follow-up, she had biopsy-proven disease recurrence in the right breast, which revealed SCC, and a mammogram showed abnormalities in the lower inner quadrant of the right breast and left axillary lymph nodes. Then we performed bilateral breast modified radical mastectomy, which confirmed that the recurrent tumors were metaplastic SCC, followed by adjuvant chemotherapy and adjuvant radiotherapy of the left supraclavicular and apical axillary regions. There has been no recurrent or metastatic evidence in the 16 months' follow-up since the second surgery.

Conclusion: This case report shows that evolution of pathology type in recurrent breast cancer after initial treatment is possible. Detailed pathologic and immunohistochemical analyses are needed for identification of this change. Surgery and adjuvant radiation and chemotherapy are appropriate treatments for recurrent primary SCC of the breast.

Keywords: breast cancer, recurrence, metaplastic squamous cell carcinoma, treatment, case report

# BACKGROUND

Breast cancer is the most common cancer of women in China and the United States, for which the clinical manifestation and the biological behavior are heterogeneous (1, 2). The most common pathological types of the breast cancer are ductal carcinoma and lobular carcinoma, which account for more than 70% of breast carcinoma. Metaplastic breast carcinoma, however, is a rare and heterogeneous group of primary breast malignancies accounting for < 1% of invasive breast

**261**

cancer (3). These tumors including the type of squamous cell carcinoma (SCC) are non-glandular differentiation. The etiology and nosogenesis of metaplastic breast carcinoma are still uncertain. There are several hypotheses for the pathogenesis of breast SCC. One theory is that the lesion is adenocarcinoma with an excessive form of squamous metaplasia (4). Another one is that it is straightly developed from the epithelium of the mammary ducts; moreover, an alternate theory is that the tumor arises from foci of squamous metaplasia within a pre-existing adenocarcinoma of the breast (5).

Because primary breast SCC is rare, a detailed metastatic workup is needed to rule out the possibility of metastatic disease. There are no characteristic clinical and imaging manifestations for SCC of the breast. Hence, the nature of the lesion needs to be determined by the pathology (4). In this paper, we presented a report of recurrent breast cancer of which the pathological type changed from the ductal carcinoma to metaplastic SCC.

### CASE PRESENTATION

In 2015, a 55-year-old woman with a palpable lump in the inner quadrant of the right breast presented to the Breast Center in our hospital. She had no other clinical symptoms such as pain, skin change, nipple retraction, or nipple discharge. She was a non-smoker and denied having any systemic diseases or any family history of breast or ovary cancer, but her father died of gastric adenocarcinoma, and her mother died of lung

cancer. On physical examination, there was a 2 × 1 cm mass at 3 o'clock in the right breast, 2 cm away from the nipple. The lump was firm, border unclear, moveable, irregular, and not fixed to the skin or chest wall. No abnormality was found in the axillary or supraclavicular lymph nodes. A mammogram showed a lesion classified as Breast Imaging Reporting and Data System 4B in the inner quadrant of the right breast (**Figures 1A,B**). The ultrasound showed an irregular hypoechoic mass of 1.2 × 0.8 cm located at 3–4 o'clock, and no positive lymph node was detected (**Figures 1C,D**).

We performed an ultrasound-guided core needle biopsy, which confirmed the diagnosis of invasive ductal carcinoma (IDC). After detailed discussion with the surgeon, the patient chose to preserve her breast. So we performed a right breast mass resection and sentinel lymph node biopsy on December 14, 2015. An invasive carcinoma in the mammary gland was localized in the lower inner quadrant with a maximum diameter of 0.6 cm. Histologically, the tumor was predominantly grade 2 IDC (about 60%), partial invasive micropapillary carcinoma (about 20%), and partial ductal carcinoma in situ (about

#### TABLE 1 | The patient's characteristics and treatment process.


(–): Negative expression; (+): Positive expression.

4 × AC→ 4 × T: four cycles chemotherapy of doxorubicin and cyclophosphamide followed by four cycles of docetaxel.

SIB-IMRT: simultaneous integrated boost intensity modulated radiotherapy.

6 × TP: Six cycles chemotherapy of docetaxel and cisplatin.

DFS, disease-free survival.

20%) with focal dysplasia in the upper resection margin (**Figure 2**). There was no metastasis in the sentinel lymph nodes (0/6). Immunohistochemical staining showed that the tumor was ER-negative, PR-negative, CerbB-2–positive (intensity 1), CK5/6-negative, P53-positive 70%, EGFR-focal positive, and Ki-67–positive 80%, and a FISH test demonstrated negative Her-2 gene amplification (**Table 1**). Thus, the tumor was pT1N0M0, equivalent to stage Ia.

After surgery, the patient received adjuvant chemotherapy with four cycles of doxorubicin and cyclophosphamide followed by four cycles of docetaxel every 3 weeks. She then underwent adjuvant radiation with simultaneous integrated boost intensity modulated radiotherapy (SIB-IMRT) to the whole right breast to a total dose of 50.4 Gy in 28 fractions and to the high-risk area for recurrence to a total dose of 60.2 Gy in 28 fractions (**Supplementary Figure 1**). Since the surgery, there had been no evidence of recurrence or metastasis for 24 months' follow-up.

Then the patient felt a nodule in the right breast again, and the size of this nodule gradually increased. Four months later, she received an ultrasound examination, which found that there was an irregular non-homogeneous-echo mass in the lower inner quadrant of the right breast below the surgical incision and had biopsy-proven disease recurrence in the right breast which revealed SCC. A mammogram showed an abnormal density shadow in the lower inner quadrant of the right breast, which lead to suspicion of malignant lesions, and enlarged left axillary lymph nodes (**Figure 3**), which biopsy indicated to be invasive carcinoma. The multidisciplinary team (MDT) conference board recommended bilateral breast modified radical mastectomy on March 15, 2018. An invasive carcinoma in the right mammary gland was localized in the lower inner quadrant, measuring 1.6 × 1.5 × 1.3 cm, and 1 of the 15 left axillary lymph nodes removed was malignant. No abnormality was found in the left breast or right axillary lymph nodes. Microscopic examination of the tumors revealed metaplastic SCC with an ER-negative, PR-negative, Her-2–negative, CK5/6-positive, P53-positive 50%, EGFR-positive, and Ki-67–positive 70% phenotype detected by immunohistochemical staining (**Figure 4**; **Table 1**).

Following surgery, the case was then discussed in an MDT conference because of the change of pathological type. Adjuvant chemotherapy with six cycles of docetaxel and cisplatin was planned (TP chemotherapy). But after three cycles, the TP chemotherapy was stopped because of severe drug-related gastrointestinal adverse events. The patient then accepted docetaxel and capecitabine for another three cycles. From October 10, 2018, to November 15, 2018, she underwent adjuvant radiotherapy with intensity modulated radiotherapy (IMRT) to the left supraclavicular and apical axillary regions with a total dose of 50 Gy in 25 fractions (**Supplementary Figure 2**). There has been no recurrent or metastatic evidence in the 16 months' follow-up since the second surgery.

# DISCUSSION

axillary lymph nodes.

SCC of the breast was first reported by Troell in 1908 (6). Because SCC of the breast was a very rare type of malignancy, no large prospective randomized clinical trials have been performed to study its pathogenesis, specific radiologic characteristics, effective treatment management, and prognosis. As a metaplastic carcinoma, SCC also shows heterogeneous characteristic features. The question of the origin of metaplastic breast cancers is still unclear. Behranwala et al. proposed two theories explaining the development mechanism of SCC of the breast: (1) arising from benign breast disease and (2) arising from invasive duct carcinoma (7). However, van Deurzen et al. demonstrated that the phenotypic changes of breast cancer are the result of malignant transformation of breast cancer stem/progenitor cells (histogenesis) or specific genetic mutations taking place at early or late stages of carcinogenesis (dedifferentiation) (8). Furthermore, Avigdor et al. performed whole exome sequencing for eight patients containing both conventional in situ or IDC and metaplastic components, which showed that the genomic landscape of an intertwined metaplastic breast tumor may generally be the same as the non-metaplastic component, and the different histologies of these cancers may be driven mainly by epigenetic or non-coding changes (9). As for our patient, her recurrent pathological histology was a pure metaplastic SCC, which is classified as pure epithelial type according to the World Health Organization (WHO), (3) indicating that her phenotypic change was more likely the result of malignant transformation of breast cancer stem/progenitor cells.

SCC may have some common clinicopathological features, such as the larger tumor size and larger proportion of T3–4, grade III, and triple-negative (lack of expression of ER, PR, and HER2) tumors than IDC (10). It presents less frequent lymphovascular invasion, but there is no difference in occurrence of positive lymph nodes between SCC and IDC patients (10). According to previous studies, 3–4% of SCC patients develop distant metastases (7, 10). However, there are no specific imaging characteristics for the metaplastic SCC patients. The pathological criteria for the diagnosis of SCC of the breast are as follows: (a) nipple, skin, or its appendages are not the sources of the tumor; (b) over 90% of tumor areas must be squamous cells; (c) other invasive components (ductal or mesenchymal) do not exist in the whole tumor; and (d) other sites of primary SCC (PSCC) are excluded (11). The recurrent tumor in this case satisfied all of these conditions. Significantly, the pathological type of the first resected malignant tumor in the right breast was IDC, but after adjuvant chemotherapy and radiotherapy, the type of the recurrent mass changed to metaplastic SCC This indicated that pathological phenotypes of tumors could transform one type to another, especially for those with metaplastic components (6). Graziano et al. (12) reported that IDC of the breast could progress to metaplastic SCC after induction chemotherapy. Previous research revealed that long-standing breast implants and acute unilateral breast pain and enlargement might also cause secondary PSCC of the breast (13, 14). Furthermore, PSCC of the breast possibly arises from previous radiation (15). Above all, metaplastic SCC of the breast can be induced by a breast implant, chemotherapy, or radiation.

showed that the recurrent tumor was a metaplastic squamous cell carcinoma with prominent keratinization that exhibited an infiltrative growth pattern. (A) Original magnification (10×). (B) High magnification (40×). (C–H) IHC staining was used to detect the expressions of ER, PR, HER-2, CK5/6, Ki-67, and EGFR at original magnification (10×). (C) ER negative. (D) PR negative. (E) HER-2 negative. (F) CK5/6 positive. (G) Ki-67 positive (70%). (H) EGFR positive.

PSCC of the breast is usually a highly malignant and hormone receptor–negative tumor, which indicates that hormone-based therapy may not have an effect on these tumors. Besides, HER2 targeted therapy is not greatly effective either for negative Her-2 gene amplification. Treatment strategies of PSCC are not different from those of other triple-negative histologic types of breast tumors, which include neoadjuvant therapy, surgery, chemotherapy, hormonal therapy, and radiotherapy. Liu et al. (6) found that TP as neoadjuvant therapy may be effective for those breast PSCC patients. Several studies revealed that compared with IDC, the PSCC patients of the breast have more rapid progression, shorter overall survival, and poorer prognosis (6, 7, 10). For this patient, the second surgery was decided to perform bilateral breast modified radical mastectomy because we are not sure about the origin of the left metastatic axillary lymph node. It may be metastasized from the tumor of the right breast or from an occult cancer of the left breast. Then we used TP chemotherapy as the second adjuvant chemotherapy and adjuvant radiotherapy in the left supraclavicular and apical axillary regions. There has been no recurrent or metastatic evidence till now. Some researchers studied the landscape of somatic genetic alterations of breast PSCC patients, which demonstrated that TP53 and PI3KCA gene mutation might be caused by the activation of the Wnt and PI3K/AKT/mTOR pathway, (16) and this may provide us a new treatment strategy for breast SCC patients. Besides, epigenetic therapies may be effective treatments for metaplastic breast carcinomas (9). But the most therapeutic regimen for this rare disease still remains unclear. Therefore, knowing its pathogenesis, clinical features, and specific imaging characteristics is crucial to choose the optimal treatment for such a rare and aggressive disease.

In conclusion, this case report shows that evolution of pathology type in recurrent breast cancer after initial treatment is possible. Detailed pathologic and immunohistochemical analyses are needed for identification of this change. Surgery and adjuvant radiation and chemotherapy are appropriate treatments for recurrent PSCC of the breast.

#### DATA AVAILABILITY STATEMENT

All data sets generated for this study are included in the article/**Supplementary Material**.

#### ETHICS STATEMENT

The authors of this manuscript obtained patient consent for publication of clinical data and images. The patient's details

#### REFERENCES


were anonymized, and the patient signed the consent form for the publication. Due to the retrospective and non-interventional nature of the study, permission by the local ethics committee was not required.

# AUTHOR CONTRIBUTIONS

TG, ZC, JX, and YZ collected the patient's data and provided the figures. TG designed the study and finished the original manuscript. JX revised the manuscript and provided the immunohistochemical figures. YZ provided final approval for the version to be published. The final version of the manuscript was read and approved by all authors.

#### ACKNOWLEDGMENTS

We thank Girish Talakatta, Ph.D., a postdoc of Sun Yat-Sen University Cancer Center, for providing professional medical writing assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2020.00032/full#supplementary-material


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Guo, Chen, Xu and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Reversal Effect of ALK Inhibitor NVP-TAE684 on ABCG2-Overexpressing Cancer Cells

Jingqiu Wang1,2, Jing-Quan Wang<sup>1</sup> , Chao-Yun Cai <sup>1</sup> , Qingbin Cui 1,3, Yuqi Yang<sup>1</sup> , Zhuo-Xun Wu<sup>1</sup> , Xingduo Dong<sup>1</sup> , Leli Zeng1,4, Linguo Zhao<sup>2</sup> , Dong-Hua Yang<sup>1</sup> \* and Zhe-Sheng Chen<sup>1</sup> \*

*<sup>1</sup> Department of Pharmaceutical Sciences, College of Pharmacy and Health Sciences, St. John's University, Queens, NY, United States, <sup>2</sup> College of Chemical Engineering, Nanjing Forestry University, Nanjing, China, <sup>3</sup> School of Public Health, Guangzhou Medical University, Guangzhou, China, <sup>4</sup> Tomas Lindahl Nobel Laureate Laboratory, Research Centre, The Seventh Affiliated Hospital, Sun Yat-sen University, Shenzhen, China*

#### Edited by:

*Amit K. Tiwari, University of Toledo, United States*

#### Reviewed by:

*Hua Zhu, The Ohio State University, United States Shikha Kumari, University of Nebraska Medical Center, United States*

#### \*Correspondence:

*Dong-Hua Yang yangd1@stjohns.edu Zhe-Sheng Chen chenz@stjohns.edu*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *16 November 2019* Accepted: *10 February 2020* Published: *27 February 2020*

#### Citation:

*Wang J, Wang J-Q, Cai C-Y, Cui Q, Yang Y, Wu Z-X, Dong X, Zeng L, Zhao L, Yang D-H and Chen Z-S (2020) Reversal Effect of ALK Inhibitor NVP-TAE684 on ABCG2-Overexpressing Cancer Cells. Front. Oncol. 10:228. doi: 10.3389/fonc.2020.00228* Failure of cancer chemotherapy is mostly due to multidrug resistance (MDR). Overcoming MDR mediated by overexpression of ATP binding cassette (ABC) transporters in cancer cells remains a big challenge. In this study, we explore whether NVP-TAE684, a novel ALK inhibitor which has the potential to inhibit the function of ABC transport, could reverse ABC transporter-mediated MDR. MTT assay was carried out to determine cell viability and reversal effect of NVP-TAE684 in parental and drug resistant cells. Drug accumulation and efflux assay was performed to examine the effect of NVP-TAE684 on the cellular accumulation and efflux of chemotherapeutic drugs. The ATPase activity of ABCG2 transporter in the presence or absence of NVP-TAE684 was conducted to determine the impact of NVP-TAE684 on ATP hydrolysis. Western blot analysis and immunofluorescence assay were used to investigate protein molecules related to MDR. In addition, the interaction between NVP-TAE684 and ABCG2 transporter was investigated via *in silico* analysis. MTT assay showed that NVP-TAE684 significantly decreased MDR caused byABCG2-, but not ABCC1-transporter. Drug accumulation and efflux tests indicated that the effect of NVP-TAE684 in decreasing MDR was due to the inhibition of efflux function of ABCG2 transporter. However, NVP-TAE684 did not alter the expression or change the subcellular localization of ABCG2 protein. Furthermore, ATPase activity analysis indicated that NVP-TAE684 could stimulate ABCG2 ATPase activity. Molecular *in silico* analysis showed that NVP-TAE684 interacts with the substrate binding sites of the ABCG2 transporter. Taken together, our study indicates that NVP-TAE684 could reduce the resistance of MDR cells to chemotherapeutic agents, which provides a promising strategy to overcome MDR.

Keywords: NVP-TAE684, ATP-binding cassette (ABC) transporter, ABCG2, ALK inhibitor, multidrug resistance (MDR)

# INTRODUCTION

Antineoplastic drugs can induce cancer cells resistant to treatment which makes the therapeutic effect of anti-cancer drugs greatly reduced and leads to multidrug resistance (MDR) (1). Classical MDR are mainly involved in drug-resistant proteins, which include the permeability-glycoprotein (P-gp/ABCB1) (2), multidrug resistance proteins (MRPs/ABCCs) (3), and breast cancer resistance protein (BCRP/MXR/ABCP/ABCG2) (4, 5).

P-gp, also known as MDR1 or ABCB1, is one of the most representative protein of ABC transporters (6–8). ABCB1 can identify various anti-cancer chemotherapeutic drugs, such as taxanes, camptothecins, anthracyclines, et al. (9). ATP hydrolysis provides energy for ABC transporters to extrude substrates out of tumor cells and lower the intracellular concentration of anticancer drugs, which results in weakening of the efficacy of chemotherapeutic drugs and eventually produces MDR (10).

The sub-family of MRPs/ABCCs is the C subgroup of ABC transporters. At present, nine MRPs with transport function have been found, ranging from MRP1 to MRP9 (11–13). Some of their structures (MRPs 4, 5, 7, 8, and 9) are similar to that of ABCB1. They contain two transmembrane regions and two ATP binding domains (14). But some of them have three transmembrane domains such as MRPs 1, 2, 3, and 6. MRP1/ABCC1 mediates the transport of anticancer drugs, including anthracyclines, methotrexate and doxorubicin (15–17).

ABCG2 is the first ABC semi-transporter found on cell membrane (18–20). Overexpression of ABCG2 can lead cancer cells resistant to various chemotherapeutic drugs. There are many overlaps between ABCG2, ABCB1, and ABCC1 in chemotherapeutic substrates, such as doxorubicin, epirubicin and mitoxantrone (21–23). However, the efflux capacity of ABCG2 for some chemotherapeutic drugs, such as vincristine and paclitaxel, is significantly lower than that of the other two drug-resistant proteins. In addition to chemotherapeutic drugs, ABCG2 has strong efflux ability to tyrosine kinase inhibitors, which can cause drug resistance in molecular targeted therapy. For example, imatinib and nilotinib for leukemia, sorafenib for liver cancer, erlotinib for NSCLC and lapatinib for HER2-positive breast cancer are substrates of ABCG2 (24–26). Therefore, screening the inhibitors of these three drug-resistant proteins is one of the effective methods to reverse MDR and improve efficacy of chemotherapy.

NVP-TAE684 is a selective ALK inhibitor which inhibits different downstream signaling transduction molecules in cancer cells, thereby down-regulating cell cycle and cell proliferation regulatory genes, resulting in arresting cell cycle, inhibiting cell proliferation and inducing cell apoptosis. NVP-TAE684 shows good anti-tumor effects to some mutant cells that are resistant to other ALK inhibitors (27–29). It was reported that NVP-TAE684 reverses MDR in human osteosarcoma by inhibiting ABCB1 function (30). However, whether NVP-TAE684 could affect other ABC transports has not been reported. In this study, we evaluate whether NVP-TAE684 can improve anticancer efficacy of drugs in ABCG2 or ABCC1 overexpressing MDR cells.

# MATERIALS AND METHODS

#### Chemicals

NVP-TAE684 was acquired from Chemie Tek (Indianapolis, IN). Fetal bovine serum (FBS), penicillin/streptomycin (P/S), Dulbecco's modified Eagle's Medium (DMEM), 0.25% trypsin and bovine serum albumin (BSA) were obtained from Corning Incorporated (Corning, NY). The GAPDH loading control monoclonal antibody (GA1R) (1 mg/mL, Cat # MA5-15738, lot #: SA247966), Alexa Fluor 488 conjugated goat anti-mouse IgG cross-adsorbed secondary antibody (2 mg/mL, Cat # A32723) were obtained from Thermo Fisher Scientific Inc (Rockford, IL). The anti-ABCG2 antibody, clone BXP-21 (Cat # MAB4146, lot #: 3026758) was obtained from Millipore (Billerica, MA). Horseradish peroxidase (HRP)-conjugated rabbit antimouse IgG secondary antibody (Cat # 7076S, lot #: 32) was obtained from Cell Signaling Technology Inc (Danvers, MA). Mitoxantrone, SN-38, topotecan, cisplatin, dimethylsulfoxide (DMSO), 3-(4,5-dimethylthiazolyl)-2,5-diphenyltetrazolium bromide (MTT), 4′ ,6-diamidino-2-phenylindole (DAPI), paraformaldehyde, Triton X-100 and other chemicals were purchased from Sigma Chemical Co (St. Louis, MO). Ko143 was purchased from Enzo Life Sciences (Farmingdale, NY). [ <sup>3</sup>H]-Mitoxantrone (2.5 Ci/mmol) was purchased from Moravek Biochemicals, Inc (Brea, CA).

#### Cell Lines and Cell Culture

The non-small cell lung cancer (NSCLC) NCI-H460 and its mitoxantrone-selected NCI-H460/MX20 cell line withABCG2 overexpression were used. NCI-H460/MX20 cell line was maintained in medium with 20 ng/mL mitoxantrone (31). The ABCG2-transfected HEK293 cell lines (HEK293/ABCG2- 482-G2, HEK293/ABCG2-482-R2, and HEK293/ABCG2-482- T7) were transfected with full length ABCG2 coding arginine (R), glycine (G) or threonine (T) at 482 position, respectively. Its corresponding parental cell line, HEK293/pcDNA3.1, was transfected with an empty vector pcDNA3.1. All transfected cells were cultured with G418 at the concentration of 2 mg/mL. The human epidermal carcinoma cell line KB-3-1 and its ABCC1 overexpressing KB-CV60 cell line, were maintained in medium with 1 mg/mL of cepharanthine and 60 ng/mL of vincristine (32). Cells were grown in DMEM medium containing 10% FBS and 1% P/S, and kept in a 37◦C humidified incubator supplied with 5% CO2. All drug-resistant cells were cultured in drug-free medium for more than 2 weeks before use.

# Cell Viability Examined by MTT Assay

MTT assay was used to measure the cell viability for the ABCG2 and ABCC1 reversal study as previously described (33). NCI-H460 and NCI-H460/MX20 cells, HEK293/pcDNA3.1 and HEK293/ABCG2 cells, KB-3-1 and KB-CV60 cells were used for the study. A total of 5 × 10<sup>3</sup> cells were seeded into each well of a 96-well plate. On the next day, cells were treated with a serial concentrations of NVP-TAE684 for the toxicity test. For the reversal study, different concentrations of substrates were added 2 h after cells were pre-treated with NVP-TAE684, Ko143 (a positive reversal agent of ABCG2) or MK571(a positive reversal agent of ABCC1) at non-toxic concentrations. After 72 h treatment, MTT solution at 4 mg/mL was added and further incubated for 4 h. Finally, DMSO was added to each well after discarding the MTT solution. The OD values at 570 nm was determined with an accuSkanTM GO UV/Vis Microplate Spectrophotometer (Fisher Sci., Fair Lawn, NJ).

#### Western Blotting

NCI-H460/MX20 cells were incubated with or without NVP-TAE684 (0.5µM) for 0, 24, 48, and 72 h. Total protein was obtained by lysing cells on ice with lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mm β– glycerophosphate, 1 mM Na3VO4, and 1µg/mL leupeptin). The protein concentrations of the cell lysates were determined using PierceTM BCA Protein Assay Kit (Thermo Scientific, Rockford, IL) to ensure equal protein loading. SDS-polyacrylamide gel electrophoresis was used to separate the protein and then transferred onto the PVDF membrane. After 2 h blocking with non-fat milk, the membrane was incubated with primary antibody against ABCG2 or GAPDH (1:1000) at 4◦C. On the next day, the membrane was incubated with HRPlabeled secondary antibody (1:1000) at room temperature for 2 h and detected by electrochemiluminescence. Photographs were taken and the relative band density was analyzed by Image J.

#### Immunofluorescence Assay

NCI-H460 and NCI-H460/MX20 cells (2 × 10<sup>4</sup> cells per well) were seeded into 24-well plates. Then NCI-H460/MX20 cells were incubated with 0.5µM NVP-TAE684 for 0, 24, 48, and 72 h. After treatment with NVP-TAE684, cells were fixed in 4% polyformaldehyde, permeated with 0.25% Triton X-100. Then, 6% BSA was used to block the non-specific reaction. After incubation with antibody against ABCG2 (1:1000) overnight at 4◦C. Cells were incubated with fluorescent IgG antibody (1:1000) in the dark for 2 h. DAPI (1µg/mL) was used to stain nuclei of cells. A Nikon TE-2000S fluorescence microscopy (Nikon Instruments Inc., Melville, NY) was used to collect immunofluorescence images.

#### ATPase Assay

As previously described (34), the ABCG2 membrane vesicles that overexpressed ABCG2 were from the protein extraction kit (Qproteome Plasma Membrane Protein Kit, Qiagen). Briefly, 20 µg ABCG2 membrane vesicles were incubated in assay buffer (containing pH 6.8 50 mM MES, 50 mM KCl, 5 mM sodium azide, 2 mM EGTA, 2 mM DTT, 1 mM ouabain, and 10 mM MgCl2). Then 0-40µM NVP-TAE684 was incubated with these membrane vesicles for 3 min. The ATP hydrolysis was initialized by 5 mM of Mg-ATP, while 5% SDS solution was used to terminate the reaction. Subsequently, the light

TABLE 1 | NVP-TAE684 lowered the IC<sup>50</sup> values of anticancer agents in NCI-H460/MX20 cells.


*a IC<sup>50</sup> values represent the mean* ± *SD obtained from three independent experiments. <sup>b</sup>Resistance fold (RF) was calculated by dividing the IC<sup>50</sup> values of NCI-H460/MX20 cell by the IC<sup>50</sup> of NCI-H460 cell in the presence or absence of NVP-TAE684 or positive control inhibitor.* \* *Indicates p* < *0.05 vs. cells treated with antineoplastic drug only.*

absorption at 880 nm was measured by the accuSkan GO UV/Vis Microplate Spectrophotometer.

#### Accumulation and Efflux Assay

To determine the intracellular [3H]-mitoxantrone accumulation, 1 × 10<sup>5</sup> NCI-H460 and NCI-H460/MX20 cells were inoculated into 24-well plates. On the next day, 0.2 and 0.5µM NVP-TAE684 or 0.5µM Ko143 were added 2 h before adding [3H] mitoxantrone. After that, the cells were digested with trypsin to dissociate the cells into 5 mL scintillation solution. For [ <sup>3</sup>H]-mitoxantrone efflux determination, 0.2 and 0.5µM NVP-TAE684 or 0.5µM Ko143 were added 2 h before adding [3H] mitoxantrone. After that, the supernatant was discarded and medium was added with the absence or presence of inhibitor. Finally, cells were collected at 0, 30, 60, and 120 min. Packard TRICARB 1900CA liquid scintillation analyzer (Downers Grove, IL) was used to measure the radioactivity.

#### Molecular Docking Analysis

The molecular docking analysis was conducted in Maestro v11.1 (Schrödinger, LLC) by the default protocols (35). The ligand NVP-TAE684 was prepared, then ABCG2 protein (Protein Data Bank ID: 6FFC) (36) was prepared. The ABCG2 protein obtained was bound to a synthetic derivative of ABCG2 inhibitor Ko143 (36). The docking grid was generated based on the position of the Ko143 derivative with the default protocol. Subsequently, glide docking was performed and induce-fit docking was conducted based on the results of glide docking.

#### Statistical Analysis

All data were presented as the mean ± SD. All experiments were done independently at least three times. One-way ANOVA was used to analyze the difference between control and experimental group. It was considered as significant when p-value is < 0.05.

### RESULTS

#### Reversal Effect of NVP-TAE684 on the ABCG2-Mediated Drug-Resistant Cells

We examined the cytotoxicity of NVP-TAE684 on cells overexpressed ABCG2 or ABCC1. As shown in **Figure 1**, NVP-TAE684 showed non-toxicity at low concentrations. Therefore, based on this result, we selected non-toxic concentrations of NVP-TAE684, which are 0.2µM and 0.5µM, for the following studies.

As shown in **Table 1**, the IC<sup>50</sup> values of several known ABCG2 substrates (mitoxantrone, SN-38 and topotecan) in NCI-H460/MX20 cells was concentration-dependently decreased by NVP-TAE684 compared with their control cells. Also, the efficacy of above substrates in ABCG2-gene-transfected cells compared with that in empty vector transfectant cells was significantly increased after co-cultured with NVP-TAE684 (**Table 2**). However, the cytotoxicity of cisplatin, which is not a substrate of ABCG2, was not significantly affected by NVP-TAE684. These results indicated that NVP-TAE684 could antagonize MDR mediated by ABCG2-overexpression. As shown in **Table 3**, the IC<sup>50</sup> value of vincristine in KB-CV60 cells was not significantly reduced by NVP-TAE684, which indicated that NVP-TAE684 could not reverse MDR mediated by ABCC1.

#### NVP-TAE684 Does Not Change the Protein Expression or Localization of ABCG2

Western blotting and immunofluorescence analysis were performed to examine the expression and subcellular localization of ABCG2. According to the results in **Figure 2**, after incubation with 0.5µM NVP-TAE684 for 24, 48, and 72 h, the expression of ABCG2 (72 kDa) was not altered. In addition, the ABCG2 expression remained unchanged after incubation with 0.2, 0.5, and 1µM of NVP-TAE684 for up to 72 h. Furthermore, as shown in **Figure 3**, the localization of ABCG2 transporter was remained on cell membrane after treated with NVP-TAE684 at indicated concentration for 0, 24, 48, 72 h.

#### NVP-TAE684 Increased the [ <sup>3</sup>H]-Mitoxantrone Intracellular Accumulation in NCI-H460/MX20 Cells

To understand the mechanism of action of NVP-TAE684 for reversal activity, drug accumulation assay was conducted to evaluate the effect of NVP-TAE684 on the [3H]-mitoxantrone accumulation in sensitive and drug-resistant cells. It was found that NVP-TAE684 had the ability to significantly increase the intracellular concentration of [3H]-mitoxantrone in ABCG2 overexpression cells, while NVP-TAE684 did not have impact on TABLE 2 | NVP-TAE684 lowered the IC<sup>50</sup> values of anticancer drugs in ABCG2-gene-transfected cells.


*a IC<sup>50</sup> values represent the mean* ± *SD obtained from three independent experiments.*

*<sup>b</sup>Resistance fold (RF) was calculated from dividing the IC<sup>50</sup> values of HEK293/ABCG2 cells by the IC<sup>50</sup> of HEK293/pcDNA3.1 cells in the presence or absence of NVP-TAE684 or positive control inhibitor.* \* *Indicates p* < *0.05 vs. cells treated with antineoplastic drug only.*

the [3H]-mitoxantrone accumulation in its parental NCI-H460 cells (**Figure 4A**).

#### The Efflux Activity of ABCG2 Was Inhibited by NVP-TAE684 in NCI-H460/MX20 Cells

Since ABCG2 transporter can pump out drugs, drug efflux assay was used to evaluate whether NVP-TAE684 can affect the efflux function of ABCG2 transporter. It was found that NVP-TAE684 significantly reduced the extrusion of [3H]-mitoxantrone in NCI-H460/MX20 cells, but it had no significant effect on the efflux function mediated by ABCG2 in corresponding parental cells. These data demonstrated that NVP-TAE684 can impede the efflux activity of ABCG2 transporter which resulted in increasing the intracellular accumulation of anticancer drugs (**Figures 4B,C**).

#### NVP-TAE684 Stimulated the ABCG2 ATPase Activity

To determine the effect of NVP-TAE684 on ABCG2 ATPase activity, an ATPase assay kit was used to measure the ABCG2-mediated ATP hydrolysis in membrane vesicles after incubation with a serial concentrations of NVP-TAE684. According to **Figure 5**, the ATPase activity of ABCG2 transporter was stimulated by NVP-TAE684 in a concentration-dependent pattern. ATPase activity reached a maximum of 211.6% of the basal activity for ABCG2. The stimulatory effect of NVP-TAE684 reached 50% maximal (EC50) at 0.091µM for ABCG2.

TABLE 3 | NVP-TAE684 does not reverse MDR mediated by ABCC1.


*a IC*<sup>50</sup> *values represent the mean* ± *SD obtained from three independent experiments. <sup>b</sup>Resistance fold (RF) was calculated from dividing the IC<sup>50</sup> values of KB-CV60 cells by the IC<sup>50</sup> of KB-3-1 cell line in the presence or absence of NVP-TAE684 or positive control inhibitor.* \**Indicates p* < *0.05 vs. group treated with antineoplastic drug only.*

# Molecular Docking Analysis on the Interaction of NVP-TAE684 and ABCG2

To explore the interaction between NVP-TAE684 and ABCG2, a molecular docking analysis was performed. The docked position of NVP-TAE684 and ABCG2 protein with highest docking score (-12.929 kcal/mol) was shown in **Figure 6**. Both hydrogen bonds and π-π interaction are included in the interaction of NVP-TAE684 and ABCG2 protein: π-π interaction between the methoxy phenyl group of NVP-TAE684 and the residue Phe439; hydrogen bonds formed between the residue Asn436 and the sulfonylphenyl or methoxy groups of NVP-TAE684.

# DISCUSSION

ABCG2 protein is a member of ABC transporters (37). ABCG2 overexpression can lead to MDR. Substrates of ABCG2 include anthracyclines, camptothecins and methotrexate (38). Since ABCG2 is an important contributor to MDR, inhibiting ABCG2 activity may help improve the efficacy of chemotherapeutic drugs. At present, various specific and nonspecific inhibitors of ABCG2 have been found. FTC is a mycotoxin isolated from Aspergillus fumigatus (39, 40). It can specifically sensitize chemotherapeutic agents to MDR mediated by ABCG2. However, its use in vivo is limited due to its neurotoxic effect. FTC tetracyclic analog Ko143 could interrupt the efflux activity of ABCG2 transporter (41), which has no toxicity in mice with high oral dose. Ko143 is the first specific ABCG2 inhibitor suitable for use in vivo (42). However, Ko143 is rapidly metabolized into a compound that is ineffective for clinical use (35). Thus, it is important to find more effective and non-toxic ABCG2 reversal agents. Recent research reported that some ALK inhibitors could sensitize chemotherapeutic drugs to ABC-mediated MDR (7, 43, 44). NVP-TAE684 is an ALK inhibitor which was reported that it could inhibit ABCB1 transporter function, but whether NVP-TAE684 could inhibit ABCG2 or ABCC1 transporter has not been reported. In this study, we explore whether NVP-TAE684 could reverse ABCG2 or ABCC1-mediated MDR and the results showed that NVP-TAE684 had reversal effects on ABCG2-mediated drug-resistant cells but showed no reversal effect on ABCC1-mediated MDR.

First, we tested the cytotoxicity of NVP-TAE684 and we found that NVP-TAE684 showed low toxicity at low concentrations. We selected two non-toxic concentrations for reversal study. MTT results showed that IC<sup>50</sup> values of several known ABCG2 substrates, including mitoxantrone, SN-38 and topotecan, in mitoxantrone-selected NCI-H460/MX20 cells reduced upon the treatment by NVP-TAE684 at non-toxicity concentrations. To confirm that the reversal effect was related to ABCG2, we determined the effect of NVP-TAE684 on ABCG2-gene-transfected HEK293 cells. We found that NVP-TAE684 showed reversal effect on HEK293/ABCG2 cells. However, the cytotoxicity of cisplatin, which is a non-substrate of ABCG2 (45), was not altered by NVP-TAE684. Moreover, we evaluated the reversal effect of NVP-TAE684 on ABCC1-overexpressing KB-CV60 cells and found that NVP-TAE684 had no significant reversal effect on KB-CV60 cells. These results indicated that NVP-TAE684 was specific to the substrates of the ABCG2 transporter.

Downregulating ABCG2 expression may lead to reversing MDR. In order to understand whether NVP-TAE684 affects either protein expression or localization of ABCG2, immunofluorescence and Western blotting experiments were

carried out. Immunofluorescence results showed that the subcellular localization of ABCG2 transporter was unchanged when the cells were cultured with NVP-TAE684. In addition,

immunoblotting results indicated that NVP-TAE684 did not downregulate the expression of ABCG2 transporter after up to 72 h treatment. Therefore, NVP-TAE684 did not change the expression of ABCG2 transporter and its subcellular localization.

It has been found that cancer cells can pump antineoplastic drugs out of cells through a complicated efflux pump system, thereby reducing intracellular concentration of many structurally unrelated anticancer drugs and leading to MDR (46, 47). According to our results, the intracellular concentration of tritium-labeled antineoplastic drug in MDR cells was significantly increased by NVP-TAE684 treatment, while no change was found in corresponding parental cells. Further assay suggested that NVP-TAE684 significantly reduced the efflux of tritium-labeled chemotherapeutic agent in drug resistant cells. These results showed that NVP-TAE684 has inhibitory activity on efflux activity of ABCG2 transporter and it resulted in increasing the accumulation of antineoplastic agents. Since ABCG2 has an ATP-binding region that is essential for substrate transport, and the function of ABCG2 transporter relies on the energy from the hydrolysis of ATP by the transporter, which can be modulated by the presence of substrates or inhibitors. Thus, monitoring ATPase activity allows for identification of those compounds that interact with ABCG2. ABCG2 exhibits a drug-dependent ATP hydrolysis activity, and a variety of ABCG2 inhibitors, as well as ABCG2 substrates, can either stimulate or inhibit ATPase activity (48). The results showed that the activity of ABCG2 ATPase was stimulated by NVP-TAE684, suggesting that NVP-TAE684 could act as a substrate, which may competitively occupy the drug binding site of ABCG2 transporter. Molecular docking study showed that NVP-TAE684 can interact with the transmembrane domain of ABCG2 with a docking score of−12.929 kcal/mol, and the interaction between NVP-TAE684 and ABCG2 proteins includes π-π interaction and hydrogen bond. This indicates that NVP-TAE684 has a strong direct interaction with ABCG2. Therefore, the reversal effect of ALK

inhibitor NVP-TAE684 was related to its inhibition on drug efflux function probably by competitively occupy the substrate binding site of ABCG2 transporter.

# CONCLUSION

This study suggests that NVP-TAE684 reverses ABCG2 mediated MDR by inhibiting the efflux activity of cancer cells, therefore increasing intracellular concentration of chemotherapeutic drugs. Our study provides a rationale for the combinational use of NVP-TAE684 and ABCG2-substrate drugs to circumvent ABCG2-mediated MDR.

# DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

# AUTHOR CONTRIBUTIONS

Conceptualization: D-HY and Z-SC. Methodology: JW, C-YC, QC, J-QW, YY, Z-XW, XD, and LZe. Writing—original draft preparation: JW. Writing, review, and editing: JW, D-HY, Z-SC,

#### REFERENCES


and LZh. Supervision: Z-SC and D-HY. Funding acquisition: Z-SC and LZh.

# FUNDING

This work was supported by the College of Pharmacy and Health Sciences, St. John's University.

#### ACKNOWLEDGMENTS

The authors would like to thank Drs. Robert W. Robey and Susan E. Bates (NCI, NIH, Bethesda, MD) for providing the NCI-H460 and NCI-H460/MX20 cell lines. We thank Dr. Shin-Ichi Akiyama for providing KB-3-1 and KB-CV60 cell lines. Thanks are also given to Dr. Stephen Aller (The University of Alabama at Birmingham, Birmingham, AL) for the human ABCG2 homology model. The authors are thankful to Dr. Tanaji T. Talele (St. John's University, New York, NY) for providing the computing resources for the docking study. JW would like to thank the Project Funded by the National First-class Disciplines (PNFD), and the Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) for supporting his study at St. John's university.


Biochem Pharmacol. (2018) 147:201–10. doi: 10.1016/j.bcp.2017. 10.004


can be reversed by KO143. J Photochem Photobiol B Biol. (2018) 178:182– 91. doi: 10.1016/j.jphotobiol.2017.10.035


inner membrane multidrug transporter. Biochem Biophys Rep. (2018) 16:122– 9. doi: 10.1016/j.bbrep.2018.10.006


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Wang, Wang, Cai, Cui, Yang, Wu, Dong, Zeng, Zhao, Yang and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Glucose Metabolism on Tumor Plasticity, Diagnosis, and Treatment

Xiaoping Lin1,2,3 \*, Zizheng Xiao1,2, Tao Chen1,2, Steven H. Liang<sup>3</sup> and Huiqin Guo<sup>4</sup> \*

*<sup>1</sup> Department of Nuclear Medicine, Sun Yat-sen University Cancer Center, Guangzhou, China, <sup>2</sup> State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-sen University Cancer Center, Guangzhou, China, <sup>3</sup> Division of Nuclear Medicine and Molecular Imaging, Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Boston, MA, United States, <sup>4</sup> Department of Thoracic Surgery, Beijing Sijitan Hospital, Capital Medical University, Beijing, China*

Malignant cells support tumor proliferation and progression by adopting to metabolic changes. Tumor cells altered metabolism by increasing glucose uptake and fermentation of glucose to lactate, even in the aerobic state and the presence of functioning mitochondria. Glucose metabolism in tumor plasticity has attracted great interests by clinicians and scientists in the past decades. This review discusses the previous and emerging researches on the tumor plasticity altered by changing glucose metabolism in different cancer cells, including cancer stem cells (CSCs). In addition, we summarize the rising applications of glucose metabolism in tumor diagnosis and treatment. Our objective is to direct future investigation on this altered metabolic phenotype and its application in patient care.

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Yi Li, Fox Chase Cancer Center, United States Zhiyi Zhang, Jinan University, China*

#### \*Correspondence:

*Xiaoping Lin guohuiqin2@163.com Huiqin Guo linxp@sysucc.org.cn*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *27 December 2019* Accepted: *21 February 2020* Published: *06 March 2020*

#### Citation:

*Lin X, Xiao Z, Chen T, Liang SH and Guo H (2020) Glucose Metabolism on Tumor Plasticity, Diagnosis, and Treatment. Front. Oncol. 10:317. doi: 10.3389/fonc.2020.00317* Keywords: metabolism, tumor, plasticity, glucose, Warburg effect, diagnosis, treatment

# INTRODUCTION

The characteristics of malignant cells, including sustaining cell proliferation, escaping cell death, attaining immortality by inducing new blood vessel formation and promoting tumor cell invasion and metastasis, were summarized in the year 2000 (1). After one decade of conceptual advance, two emerging traits were added—altering energy metabolism and evading immune demolition (2). In recent years, increasing number of studies focus on the alteration of energy metabolism, that allows tumor cells to survive and spread even in challenging conditions. However, a paradigm shift has occurred adding to our knowledge of the function of glycolysis in glucose metabolism over the last decade (3). In this review, we attempt to provide a better understanding of glucose metabolism in tumor plasticity which may contribute to the design and outcome of novel diagnostics and treatment strategies.

#### GLUCOSE METABOLISM AND THE WARBURG EFFECT IN TUMORS

Normally, glucose is processed by glycolysis to generate ATP and pyruvate. Then the ribose 5-phosphate and NADPH were produced through the pentose phosphate pathway (PPP), or enter into the tricarboxylic acid (TCA) cycle in mitochondrion. Glucose-derived citrate is converted to acetyl-CoA, oxaloacetate (OAA), or a-ketoglutarate (a-KG). Glutamine is deaminated to form glutamate, which is processed to produce a-KG for use in the TCA cycle.

**278**

The main pathway of glucose metabolism in cancer cells is aerobic glycolysis, termed Warburg effect (4). In cancer cells, glucose uptake and the production of lactate was dramatically increased, even in the presence of oxygen and fully functioning mitochondria (5). This classic type of metabolic change provides substrates required for cancer cell proliferation and division, which is involved in tumor growth, metastatic progression and long-term survival (5–8). It must be emphasized that both glycolysis and mitochondrial metabolism are crucial to cancer cells in the Warburg Effect (5).

Glucose metabolism in tumor is governed by both oncogenes and cancer-producing factors (6). Metabolic reprogramming of cancer cells is regulated by transcription factors that include c-Myc, p53 and hypoxia-inducible factor (HIF) 1α (9). The reprogramming is a complex interaction of various signaling pathways, such as Notch, Akt, phosphoinositide-3-kinase (PI3K), PTEN, mammalian target of rapamycin (mTOR), and AMPactivated protein kinase (AMPK) (10, 11).

c-Myc can stimulate glycolysis, glutaminolysis, and nucleotide synthesis (12). c-Myc mediated glucose metabolic reprogramming primarily on mitochondrial aerobic metabolism (13). Glycolysis can be promoted by c-Myc through direct induction of glycolytic-associated enzymes (14). Besides, mitochondrial biogenesis can be promoted by c-Myc with stable function and the number of mitochondria in tumor cells.

p53 is the main adverse regulator during tumor metabolic reprogramming (15). p53 inhibits glycolysis by inducing glycolysis and apoptosis regulator (TIGAR), inhibiting phosphoglycerate mutase (PGM) to upregulate expression of TP53, and repressing glucose transporter (GLUT)-1 and GLUT -4 (6, 16–18). Also, p53 can alter oxygen consumption and the synthesis of cytochrome c oxidase 2 (SCO2) protein, which is critical for regulating the cytochrome c oxidase(COX) complex (19). Moreover, p53 promotes mitochondrial glutaminase (GLS2) and limits glutaminolysis in response to oxidative stress or DNA damage (20).

HIF-1 is a heterodimeric protein that could alter various genes coded for enzymes involved in glucose metabolism. The phosphatidylinositol 3-kinase (PI3K) and ERK mitogenactivated protein kinase (MAPK) pathways affect HIF-1α protein synthesis. In glucose metabolism, glyceraldehyde-3- P-dehydrogenase (GAPDH), GLUT-1, hexokinase (including HK1 and HK2), autocrine motility factor/ (AMF/GPI), enolase 1(ENO1), plasminogen activator receptor (TPI), Pyruvate kinase(PKM), 6-phosphofructo-2-kinase/fructose-2,6 biphosphatase-3(PFKBF3, PFKL, PGK1), and LDHA can be transcriptionally activated by HIF-1 (21).

#### THE IMPACT OF GLUCOSE METABOLISM ON TUMOR PLASTICITY

Tumor cells need to survive drastic changes in the microenvironment such as hypoxia, nutrient storage, and acidic pH (22). A huge number of cancer cells show remarkable plasticity in metabolic adaptation. The reprogrammed glucose metabolism allows cancer cells to satisfy high proliferation requests. In addition, it provides some survival and growth advantages, including high carbon source for anabolism, rapid ATP availability to supply the energy, abundant lactic acid to increase the redox status (NADPH) via the glycine–serine pathway (6–8). Lactic acid induces metabolic "dormancy" and is involved in EMT and tumor immune response by reducing pH in the tumor environment (5, 8, 23–25). To manage all the situations above, cancer cells must maintain a balance to deliver adequate energy with constrained resources and to meet the biosynthetic demands of proliferation. Though oxidative phosphorylation(OXPHOS) would be the best energy provider, the physiological reality is that both OXPHOS and glycolysis collaborate to produce ATP under the local oxygen concentration. Coordinate results are net increments in glucose utilization and lactic acid secretions. This process is known as the glycolytic switch, which is corresponding to uncoupling glycolysis from OXPHOS (26).

# Glucose Metabolism and Cancer Cell Proliferation

Cell proliferation requires expanded uptake of supplements, lifted flux through biosynthetic pathways, support of metabolic intermediates, and proceeded recovery of cofactors required to supply energy or reducing equivalents for reactions. Cancer cells preferred aerobic glycolysis for cell proliferation. In addition, aerobic glycolysis produces metabolic precursors that are essential for rapid cell proliferation (25). As proliferation is the key feature of cancer cells, aerobic glycolysis allows cancer cells to meet the requirements of generating enough ATP and biosynthetic precursors. The goal of aerobic glycolysis is to preserve high levels of glycolytic intermediates to maintain anabolic reactions in cells instead of generating lactate and ATP. Thus, it may explain why increased glucose metabolism happens in proliferating cancer cells (26).

The biosynthesis in proliferating cells requires building blocks for the synthesis of nucleotides, lipids, and non-essential amino acids—those that glycolytic intermediates can supply (27). The PPP can produce the reducing equivalents in the form of NADPH molecules and generates nucleotide and lipid precursors. The TCA cycle can generate acetyl-CoA and glutamine and drive them into the cytosol. As a result, the anabolic metabolism of amino acids and lipids is supplied by both glycolysis and the TCA cycle within mitochondria (27). NAD+ is an essential cofactor of nucleotide and amino acid biosynthesis. The maintenance of biosynthesis in proliferating cells demands the regeneration of NAD+. The conversion of pyruvate to lactate can partially produce NAD+ (28). Because cells use as much as 10% of their entire proteome and half of all of their metabolic genes to produce proteins involved in glycolysis, the cost of using Warburg Effect in aerobic glycolysis as a tradeoff to promote biosynthesis is vast (29). Mitochondrial functions occur concomitantly with the aerobic glycolysis and limiting mitochondrial activity may not occur during the Warburg Effect (5). Under energy stress conditions, the apparent shift from glycolysis to OXPHOS by mitochondrial elongation contribute to tumor survival. Remodeling of mitochondrial morphology is a remarkable protection of tumor cells from stress (30).

#### Glucose Metabolism and EMT

EMT is a process that involves a high level of cellular plasticity. EMT is often activated during cancer cell invasion, systemic dissemination, and metastasis (31, 32). EMT is an important step preceding to invasion and metastasis in tumor cells. Epithelial cells lose the junctions among cells and their polarized organization during EMT. They change cytoskeletal organization, transform the shape, and acquire mesenchymal characteristics, such as fibroblast-like cell morphology and increased capability of invasion and migration (32, 33).

The plasticity of glucose metabolism is important in EMT (34). The genes and biochemical mechanisms impact glucose metabolism during EMT of cancer cells. Crosstalk network has been explored between EMT and cancer metabolism (35). PI3K-AKT-mTOR, EGFR-RAS-MAPKs, and JAK2-STAT3 signaling pathways can mediate EMT (36). HIF1α, Myc and FOXM1 regulate both metabolism and EMT (37). LKB1 / AMPK (Liver kinase B1/AMP-activated protein kinase) downregulates SNAIL and ZEB1and inhibits the invasion and migration of tumor cells, by regulating FOXO3, TGF- β, NF-κB, AKT, and mTOR signaling pathways (34).

Glucose transporters, especially GLUT-1 and GLUT-3, promote tumor progression by increasing glucose influx and activating downstream molecular pathways (38). GLUT-1 increases matrix metalloproteinase 2 (MMP-2) in vitro and in vivo, which contributes to EMT and cellular invasiveness (34, 39). GLUT-3 gene could be activated by ZEB1, which is an EMT marker (34). HK2 is a well-known hypoxia-inducible gene that can induce EMT (34). PFK increases glycolytic flux and EMT by maintaining this glycolytic phenotype in cancer cells in vitro. Pyruvate kinase M2 (PKM2) can prompt EMT both by metabolic and non-metabolic mechanisms (40). PKM2 increases glucose uptake and lactate production to support cell survival and invasion (41).

Besides the enzymes involved glucose metabolism, altered mitochondrial function also contributes to EMT induction (34). Tumor cell migration and metastasis was stimulated by abnormal TCA cycling coupled with mitochondrial superoxide production (42).

#### Glucose Metabolism and Cancer Stem Cells

Cancer consists of mainly stem cells (CSCs) and non-CSCs (2). CSCs have the potentials of self-renew and tumor initiation (43, 44). CSCs adapt to metabolic plasticity, which is determined by the factors present in the tumor microenvironment (TME). Metabolic plasticity allows these cancer stem cells to switch between OXPHOS and glycolysis (45). Only complete oxidation through the TCA cycle cannot supply enough anabolic precursors such as pyruvate and glutamine. CSCs prefer to rely on glycolysis and the PPP and devolve mitochondrial infrastructure and function. This predominantly glycolytic metabolism offers sufficient energy to support the basic needs of CSCs. The maturation of metabolic network matches the increasing energy demands of specialized progeny cells. The oxidative metabolism infrastructure contains mitochondrial biogenesis and maturation, and networks of the TCA cycle and electron transport chain. A concurrent rise in mitochondrial ROS may prime CSCs for lineage differentiation (46). The main difference between cancer cells and CSCs is the metabolic shift and mitochondrial resetting. CSCs display the metabolic change and mitochondrial resetting into precise bioenergetic states and lose the unique metabolic phenotypes after differentiation. Compared with differentiated neoplastic cells, CSCs exhibit a more prominent Warburg effect. Aerobic glycolysis may produce enough glycolytic intermediates into the PPP to supply molecules that are necessary for anabolic metabolism and growth of CSCs (47). In CSCs, the oxidative metabolism and mitochondrial structure are altered to commit glycolysis (48). Aerobic glycolysis is one of the important aspects in maintaining CSCs and inducing their differentiation. Specifically, aerobic glycolysis is critical in preserving the stemness of CSCs, while switching to oxidative metabolism is the characteristic of stem cell differentiation. Besides, aerobic glycolysis is essential to the properties of CSCs (46). Multiple regulatory factors including metabolic enzymes promoted the metabolic plasticity of stem cells in breast cancers. The metabolism-regulating genes and epigenetic factors that regulate glucose metabolism might also regulate the expression of EMT (49). An enhanced Warburg effect was observed in metastatic prostate cancers (44).

Moreover, metabolic flexibility diverge the fates CSCs, which include dormancy to minimize stress damage, proliferation and self-renewal to preserve progenitor pools, and pedigree specification for tissue regeneration (48). Depending on the metabolic characteristics of the tumor cells of origin, isogenic glioma stem cells (GSCs) exhibits heterogeneity in metabolic characteristics. They can be divided into mitochondrial and glycolytic phenotypes. Cells of the mitochondrial type consume more oxygen and maintain a higher ATP content; those of the glycolytic type consume more glucose and produce more lactate. Both metabolic phenotypes are independent and stable. They can coexist within a given tumor. The environmental factors further influence the metabolic preferences of these cells. For example, CSCs that rely on OXPHOS can switch to aerobic glycolysis in response to metabolic stress (50).

#### Glucose Metabolism and Tumor Microenvironment (TME)

In addition to the inherent alterations in the tumor cells, the metabolic competition and cooperation among the TME components support tumor proliferation, progression, and therapeutic resistance (9). TME consist of different types of cells, that include cancer associated fibroblasts (CAFs), noncancer cell stroma, immune cells, and endothelial cells (51). The TME forms a pro-tumorigenic cocoon around the tumor cells. Intrinsic traits (e.g., genetic programs in cancer cells) and extrinsic factors (e.g., nutrient availability, oxygen tension, pH) contribute to the deregulated metabolism in TME. TME enforces metabolic plasticity to adapt to hypoxic and acidic environment, nutrient deprivation and competition, oxidative stress, and immune surveillance (52). Reprogramming of the metabolism occurs in tumor and non-tumor cells. The interactions and competitions in TME components guarantee the steady supply of nutrients and molecules for tumor growth even under hypoxic conditions. Metabolic reprogramming also affects TME. Hypoxia inhibits this process by upregulating PDK1 and LDHA (53, 54). When HIF1α is activated in CAFs, the activity of mitochondria drops and lactate production increases. This is consistent with a glycolytic phenotype. It leads to slow down metabolism in the microenvironment (55). Lactate accumulation resulted from continuous activation of glycolytic and LDHA enzymes leads to a low-pH microenvironment during tumor progression. Nutrient deficiency is another microenvironment stress that cancer cell often encounters. Numerous studies have shown that reprogramming of glucose metabolism upon nutrient starvation in tumor cells occurs in order to use energy to support their growth (56–61).

The reverse Warburg Effect was proposed in the last decade (62). The energy-rich metabolites of aerobic glycolysis, such as lactate and pyruvate, are generated by cancer associated fibroblasts and taken up and used in the TCA cycle in mitochondria of epithelial cancer cells. Thus, efficient energy production such as ATP generation, which leads to increase cell proliferation and reduce cell death (62). The reverse Warburg effect can benefit tumor cells by cooperative utilization of oxygen between stromal cells and tumor cells (55). Cancer cells induce stromal cells to undergo "aerobic glycolysis." Their products are returned to the cancer cells to be used for mitochondrial OXPHOS (63). The metabolic heterogeneity in TME allows cancer and stromal cells to exchange metabolites between them to maintain maximal cellular growth (18).

The immune system plays an important role in TME. The immune cells in the TME can detect and eliminate the abnormal cells or tumor cells and protect the body from damage caused by tumor cells (64). Tumor cells activate immune cells, and the activated innate or adaptive immune cells can maintain homeostasis (9, 65). The immune system affects cancer survival and progression (23, 66). Tumor cells develop different mechanisms to escape the immune response. They include strategies at the genetic, epigenetic and metabolic levels. It implements to resist immune recognition and decrease apoptosis (67).

There is a complex interaction between malignant cells and immune cells in the tumor stroma (66). Metabolism regulates tumor cells to escape immune surveillance and to coexist with stroma cells. Tumor and other different types of cells in the TME compete for nutrient. Inflammation induced by oncogene in the tumors promotes adaptive metabolic changes in the surrounding non-tumor cells to secrete metabolites. Tumors use these metabolites as alternative nutrient sources to meet their increasing demands for anabolic function (68).

Metabolism is essential to lymphocyte for its development, function and inducing tolerance (69). T cells prefer glycolysis upon activation, though they had lower glycolytic flux when resting. Activated by TCR- and CD28-mediated co-stimulation, T cells switch to a rapid increase in glucose uptake and glycolysis (69–71), as well as glutaminolysis (69, 72, 73). In contrast, the activation of T cells and dendritic cells can be inhabited by lactate accumulation (74, 75). Lactate prevents cytokine release and monocyte migration, while promotes the formation of tumor-associated macrophage 2 (TAM2) phenotype. It leads to upregulate the expression of arginase 1, promote immune escape and tumor progression (25, 76–78). Lactate reduces the production of IFN-γ from T cells and NK cells, and decreases the level of nuclear factor of activated T cells (NFAT), which also contributes to immune escape and tumor progression (79, 80). As antigen presenting cells and a source of cytokines, B lymphocytes also have a critical role in antitumor immunity (81, 82). B cells upon activation use both glycolysis and OXPHOS, which is different from T cells. However, the deletion of GLUT-1 or inhibition of glycolysis in B cells suppresses antibody production in vivo (69, 70).

# CLINICAL APPLICATION OF TARGETING GLUCOSE METABOLISM IN TUMOR

Attempts to target the glucose metabolism, especially on Warburg effect, for cancer diagnosis and therapy emerges in the past decades and is still in developing.

# Application of Glucose Metabolism in Cancer Diagnosis

Many different types of human tumors dramatically enhanced uptake and use of glucose. Since 1976, positron emission tomography (PET) with a radiolabeled analog of glucose ( <sup>18</sup>F-fluorodeoxyglucose, FDG) was applied to non-invasively visualize glucose uptake in human body. This tracer is the most impressively clinical utility of the Warburg effect (2, 6). Increased FDG implies high glucose uptake which is an indication of the glycolytic switch. It provides information about the pathologic differentiation and precise stageing of tumors, predicts treatment response and gives an indication of overall prognosis (83). The positron-emitting radionuclide fluorine-18 replaces the normal hydroxyl group at the C-2 position in the glucose molecule. After injection into the body, the tracer is transported into cells by glucose transporters, especially GLUT-1 and GLUT-3, followed by phosphorylation with the hexokinase, especially HK2, to produce <sup>18</sup>F-FDG-6-phosphate (18F-FDG-6-p). <sup>18</sup>F-FDG-6-p cannot be released from the cell and is trapped in the cytoplasm. Because of the lack of the 2'hydroxyl group, <sup>18</sup>F-FDG-6-p cannot further proceed to glycolytic pathway (84). PET scanner detects the radioactive decay of <sup>18</sup>F-FDG-6-p and form the body images of the distribution of <sup>18</sup>F-FDG. Thus, the presence of living malignance can be identified by the accumulated amounts of <sup>18</sup>F-FDG-6-p (84, 85). The sites and the semi-quantitative analysis of high glucose uptake (e.g., standard uptake value, SUV) in the whole body can be identified. In the vast majority of malignance, glucose is trapped in cancer cells more than normal tissues with the exception of the brain and brown fat. It relates to the metabolic characteristics at the tumor site (42). Owing to the limitation of spatial resolution and some particular pathology subtypes(e.g., signet ring cell carcinoma, well-differentiated cancer), the sensitivity and specificity varied across different applications using18F-FDG–PET (26, 85).

PET imaging tracers are able to detect the PKM2, such as N,N-diarylsulfonamide (DASA) compounds bind to PKM2. <sup>11</sup>Clabeled analog of DASA-23 was applied to the orthotopic U87 and GBM39 patient-derived tumors in preclinical models of glioblastoma multiforme and to monitor the response of the PKM2 activator TEPP-46 in GBM39 tumors (86).

An analog of glutamine, 4-18F-(2S,4R)-fluoroglutamine (18F-FGln), is taken up by cancer cells in vitro and its specific uptake can be detected on PET imaging in mouse xenograft model in vivo. Thus, the glutamine metabolism in gliomas and its uptake can be evaluated (87). Other investigational PET agent, such as 5-11C-(2S)-glutamine (87, 88), was also exploited for its ability to take up and retain glutamine in some tumors (87).

As another device in molecular imaging, a few agents have been developed to detect the glucose metabolism in magnetic resonance (MR). The hyperpolarized agents can form the better imaging for remarkable enhancement compared to conventional MR imaging. Magnetic resonance spectroscopy(MRS) can image the conversion of <sup>13</sup>C-labeled pyruvate to lactate in patients (83). It is possible to identify malignance and monitor treatment response by evaluating the distribution of <sup>13</sup>Cpyruvate and the altered <sup>13</sup>C -lactate/13C -pyruvate in preclinical MRS study in prostate cancer and glioma (83, 89). Some other targetable processes in glucose metabolism, such as the detection of 2HG, can also be used for new imaging technology (90, 91).

#### Application of Glucose Metabolism in Cancer Treatment

To date, various agents involved in glucose metabolism are actively investigated as novel targets with therapeutic potential (**Table 1**). It helps to overcome drug resistance or increase the efficacy of current combination therapy (88, 109). Compared to diagnosis, targeting glucose metabolism in treatment seems faint due to their efficacy or safety concern. Several drugs with efficacy confirmed and multitargets, as well as some old non-chemotherapeutic drugs with new aspects of inhabiting tumor glucose metabolism will be discussed below.

GLUTs control the influx of glucose, especially GLUT-1. Several GLUT-1 inhibitory agents, including WZB117 and STF-31 have been tested (94, 105). STF-31 was effective in decreasing glucose uptake, inducing cell apoptosis and inhibiting tumor progression. However, STF-31 has a narrow therapeutic potential due to its molecular restriction (110, 111). WZB117 can effectively inhibit glucose uptake, cell proliferation, and tumor progression both in vivo and in vitro. However, its efficacy remains in doubt (105). A glucose-conjugated LDH inhibitor, glucose-conjugated methyl ester (NHI-Glc-2), is a promising compound. It is a weaker inhibitor than the N-OH methyl ester (NHI-2) on the isolated enzyme. It can increase the glucose uptake by exploiting the GLUT-1 overexpression, TABLE 1 | Metabolic modulators arising from the metabolic theory of cancer.


*(Continued)*

TABLE 1 | Continued


reduce lactate production and decrease proliferation of cancer cells (105).

Currently, inhibition of lactate transport is being tested as an alternative approach. But LDHA inhibitors had some limitations, such as high toxicity, low drug exposure or a lack of LDHA dependence in human tumor inhibitors (112). Nevertheless, an inhibitor of human LDH isoforms, Galloflavin is a natural phenol derivative and a product of gallic acid oxidation (105, 113). The analogs of Gossypol, a natural component extracted from the cotton seeds has been screened for small molecular inhibitors specific for LDHA. 3-dihydroxy-6-methyl-7-(phenylmethyl)-4-propylnaphthalene-1-carboxylic acid (FX11) was proved to effectively inhibit proliferation of cancer cells in vitro and in vivo by enhancing the levels of oxidative stress. Another isoform-specific inhibitors of LDHA, N-hydroxyindole-based compounds, can compete with its substrate pyruvate and the cofactor, NADH (92, 97).

The reversal reprogramming in IDH mutant tumors seems more successful (114, 115). The mechanism of 2HGmediated transformation may vary in different kinds of tumors. It may inhibit the dual metabolic flux of glycolysis and oxidative PPP (88, 116). In preclinical studies, the inhibition of mutant IDH has been shown to dramatically decrease the generation of 2HG and cause cancer cells to differentiate into normal cells (107, 108). An inhibitor of mutant IDH2, AG-221, has been put forward in early phase clinical trials (113).

New compounds also target dual inhibition, such as the inhibition of metabolic plasticity and metabolic rescue in cancer cells. Compounds targeting glucose, glutamine and lactate metabolism have been found to exert anticancer effects by inhibiting growth of tumor-associated endothelial cells (24). A new glucose uptake inhibitor, Glutor, targets GLUTs (GLUT-1, -2, and -3), diminishes glycolytic flux and selectively suppresses growth of a variety of cancer cells. Glutor combined with glutaminase inhibitor CB-839 synergistically inhibits the proliferation of colon cancer cells (95).

Compared to conventional cytotoxic therapy, modulation of particular targets with altered glycolytic metabolism would reduce treatment toxicity. A number of studies show that treatment combined with vitamin C leads to interfering with glycolysis and the TCA cycle and inhibits ATP and NADPH production. It can kill cancer cells by increasing oxidative stress and further inhibition in cancer cell survival and invasion. (117–122). Preclinical studies have shown that vitamin C at the concentration below 5 mM could prevent proliferation of cancer cells (123–125). Because of the anti-oxidant capacity, vitamin C can prevent the growth of circulating tumor cells (CTCs). Vitamin C also reduces pyruvate and glutathione (GSH). Extracellular matrix remodeling and cancer cell motility were reduced by vitamin C by boosting ROS levels. It can increase expression of E-cadherin, decrease expression of Snail and inhibit matrix metalloproteinases (MMPs) (105, 126, 127).

Metformin is a common drug for diabetes. The metformin/hypoglycemia combination has synergistic antineoplastic effects to decrease the pro-survival protein MCL-1 and cause cell death (128). Metformin combined with ritonavir targets OXPHOS, in particular, GLUT-4. It can effectively inhibit the AKT and mTORC1 phosphorylation and pro-survival mitochondrial complex I (MCL1) (116, 129).

Aspirin is a common pain reliever and anti-inflammatory drug. It also inhibits MCL1 activity (130). In the past two decades, its anti-neoplastic action has been investigated against different malignancies and tumor cell lines. Dalton's lymphoma (T-cell lymphoma) cells obtained from tumor-bearing mice treated by aspirin showed a change of expression of pH regulators MCT-1 and V-ATPase, as well as change in cell survival regulatory molecules including GLUT-1 (131). Aspirin can also modulate glucose uptake by depressing GLUT-1 through targeting NF-κB or NF κB/HIF1α signaling to inhibit proliferation (132).

#### CONCLUSION

Despite the extensive study on cancer metabolism with interesting results accumulated in the last decades, questions are still arising. The key process of balance among glycolysis, TCA and other pathways of the glucose metabolism in tumor remains unclear, as it is the essential mechanism of Warburg effect. In addition, it should be considered to develop more tumor-specific tracers and drugs based on the metabolic switch in tumor cells, cancer stem cells or the interaction with immune system. The ideal drugs should only applied in tumor by blocking specific pathway(s) for its metabolic plasticity but not in normal tissues. Despite the emerging of metabolic enzymes or transporters inhibitors, the efficiency of targeting tumor glucose metabolism is being challenged. To explore the metabolic plasticity in cancer under intrinsic and extrinsic influences, tumorous glucose metabolism should be addressed. Nevertheless, with technological advances, it is expected that we will uncover many other unknown aspects of glucose metabolism in cancer and use them to benefit patient care.

#### AUTHOR CONTRIBUTIONS

XL and HG conceived and designed the study. XL wrote the first draft. ZX, TC, and SL wrote some sections of the article. HG

#### REFERENCES


edited the article. All authors read and approved the final version of submission.

#### FUNDING

This work was supported by the Fund of Natural Science Foundation of Guangdong Province, China, No. 2018A030310239.


synergistically impairs tumor cell growth. Cell Chem Biol. (2019) 26:1214– 28.e25. doi: 10.1016/j.chembiol.2019.06.005


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Lin, Xiao, Chen, Liang and Guo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Role of Exosomal microRNA in Cancer Drug Resistance

Qiao-ru Guo1,2†, Hui Wang3†, Ying-da Yan1†, Yun Liu<sup>1</sup> , Chao-yue Su<sup>1</sup> , Hu-biao Chen<sup>4</sup> , Yan-yan Yan<sup>5</sup> , Rameshwar Adhikari <sup>6</sup> , Qiang Wu2,7 \* and Jian-ye Zhang1,2 \*

*<sup>1</sup> Guangdong Provincial Key Laboratory of Molecular Target & Clinical Pharmacology, School of Pharmaceutical Sciences and the Fifth Affiliated Hospital, Guangzhou Medical University, Guangzhou, China, <sup>2</sup> Key Laboratory of Tropical Translational Medicine of Ministry of Education, Hainan Medical University, Haikou, China, <sup>3</sup> Guangzhou Institute of Pediatrics/Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangzhou, China, <sup>4</sup> School of Chinese Medicine, Hong Kong Baptist University, Hong Kong, China, <sup>5</sup> Collaborative Innovation Center for Cancer, Institute of Respiratory and Occupational Diseases, Medical College, Shanxi Datong University, Datong, China, <sup>6</sup> Research Centre for Applied Science and Technology, Tribhuvan University, Kirtipur, Nepal, <sup>7</sup> Key Laboratory of Emergency and Trauma of Ministry of Education, School of Tropical Medicine and Laboratory Medicine, Hainan Medical University, Haikou, China*

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Xiumei Huang, Indiana University Bloomington, United States Bethany Hannafon, University of Oklahoma Health Sciences Center, United States*

#### \*Correspondence:

*Qiang Wu wuqiang001001@aliyun.com Jian-ye Zhang jianyez@163.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

> Received: *14 January 2020* Accepted: *16 March 2020* Published: *07 April 2020*

#### Citation:

*Guo Q, Wang H, Yan Y, Liu Y, Su C, Chen H, Yan Y, Adhikari R, Wu Q and Zhang J (2020) The Role of Exosomal microRNA in Cancer Drug Resistance. Front. Oncol. 10:472. doi: 10.3389/fonc.2020.00472* Exosomes affect the initiation and progression of cancers. In the tumor microenvironment, not only cancer cells, but also fibroblasts and immunocytes secrete exosomes. Exosomes act as a communicator between cells by transferring different cargos and microRNAs (miRNAs). Drug resistance is one of the critical factors affecting therapeutic effect in the course of cancer treatment. The currently known mechanisms of drug resistance include drug efflux, alterations in drug metabolism, DNA damage repair, alterations of energy programming, cancer stem cells and epigenetic changes. Many studies have shown that miRNA carried by exosomes is closely associated with the development of drug resistance mediated by the above-mentioned mechanisms. This review article will discuss how exosomal miRNAs regulate the drug resistance.

Keywords: exosome, miRNA, cancer, drug resistance, mechanism

# INTRODUCTION

Chemotherapy, radiotherapy, surgery, and targeted therapy are important modalities of cancer treatment. However, the emergence of drug resistance leads to dismal prognosis in cancer patients. As an emerging therapeutic target and diagnostic biomarker, exosomal miRNAs play vital roles in tumor invasion, metastasis and progression. Studies have found that the occurrence and development of drug resistance is closely related to miRNA carried by exosomes (1–5). In this review article we discuss several classic mechanisms that exosomal miRNA involves in drug resistance. Also, we summarize the role of exosomal miRNA mediated drug resistance in different types of cancers.

# THE BIOGENESIS OF EXOSOME AND miRNA

Exosomes are extracellular vesicles (EV) with the size ranges between 30 and 100 nm. Exosomes are released to the extracellular environment after the fusion of the multivesicular body (MVB) or late endosomes with the plasma membrane (6). It was first described in 1983 as "seems to be akin to reverse endocytosis" (7), and has gradually recognized an important factor in oncology research (8). Many studies have shown that exosome can promote the intercellular communication (9) by transferring varieties of cargos, such as nucleic acid, proteins and metabolites (10–16).

microRNA (miRNA) is an important cargo that delivered by exosomes (17). miRNAs usually consist of 19–25 nucleotides. It can regulate post-transcriptional silence of target genes. Following the transcription of miRNA gene, a small hairpinshaped RNA called pre-miRNA is generated. The pre-miRNA is exported into cytoplasm and processed by Dicer, a kind of RNase III-type endonuclease, and subsequently releases a small RNA duplex (18, 19). The RNA duplex will unwind after loading onto an Argonaute (AGO) protein and forming RNA-induced silencing complex (RISC) (20). Once binding to a RISC, the miRNA is complementary pairing with the mRNA. Depending on whether miRNA and mRNA are fully bind, two different mechanisms occur: (1) mRNA specifies cleavage if miRNA is sufficiently complementary to mRNA; (2) the productive translation is inhibited when miRNA is insufficiently complementary to mRNA (21, 22). Therefore, miRNA can regulate various physiological and pathological activities (**Figure 1**).

# THE CURRENT UNDERSTANDING OF CANCER DRUG RESISTANCE

Cancer drug resistance can be divided into intrinsic and acquired resistance. Intrinsic resistance occurs before receiving therapy, which limits the use of anticancer drugs. Acquired resistance may develop during treatment even if some drugs have anticancer effects at the early stage (23). Drug resistance seriously impacts the effectiveness of chemotherapy and molecular targeted therapies, ultimately leading dismal prognosis and tumor relapse.

Because of genomic instability, tumors may include a diverse collection of cells that possess different sensitivity to treatment (24). The positive selection of drug-resistant tumor subpopulation causes drug resistance. Therefore, accurate assessment of tumor heterogeneity is important to address drug resistance (25). The application of high-throughput screening technology facilitates the identification of genotype and helps predict drug response, providing convenience to individual therapy. In the past few years, microfluidic chips show tremendous promise in the study of tumor heterogeneity and the establishment of preclinical models (26). Grosselin et al. (27) set up a high-throughput droplet microfluidics platform. On this platform, the single cell chromatin landscapes of thousand cells can be profiled. They used the patient-derived xenograft models of acquired resistance to chemotherapy and target therapies in breast cancer and found a common chromatin signature between drug-sensitive and resistant cells (27). This technique paves the way to study the role of chromatin heterogeneity.

The limited cancer models hinder the clinical prediction of drug efficacy. Therefore, it is urgent to establish more reasonable, advanced and high-throughput cancer models to deal with drug resistance and explore the underlying factor of heterogeneous patientresponses. Gao et al. (28) established about 1,000 patient-derived tumor xenograft models (PDXs) with a diverse set of driver mutation using high-throughput screening technology. It has been demonstrated that these PDXs have the potential to predict patient response to targeted therapies and perform in vivo compound screens (28). Furthermore, a 3D model of tumor tissue made up of numerous different cell types can better mimic tumor microenvironment and provide the similar information about clinical response. Kather et al. developed a 3D model of tumor tissue which reproduced key features of colorectal cancer (CRC) and based on the individual patient data, yielding in silico tumor explant (29).

Combinations of drugs are also the effective way to overcome or bypass drug resistance (30). Epidermal growth factor receptor tyrosine kinase inhibitor (EGFR TKI) is beneficial for the treatment of non-small cell lung cancer with EGFR mutation (31). However, after treatment with EGFR TKI for 10–14 months, the efficacy declines (32), the primary and acquired drug resistance limits their clinical benefit (33). To combat resistance, in addition to developing new drugs, drugs combinations through a so-called bypass signaling mechanism, is an excellent choice (34). In addition, nanomedicine approach can be used to encapsulate and co-delivery drugs in specific materials to improve their bioavailability and thus overcome drug resistance (35, 36). The application of high-throughput drug screening can identify the effective drug combination regimens. Using highthroughput screening technology, researchers identified that potassium antimony tartrate in combination with topotecan can significantly enhance the sensitivity of non-small cell lung cancer and colorectal cancer to cis-diamminedichloroplatinum/cisplatin (CDDP). It was found that topotecan impairs the ability to repair CDDP-induced DNA damage (37). DNA damage repair is a classic mechanism by which cells develop drug resistance, as detailed later in this article.

Cancer biomarkers are present in tumor tissue or serum that help to detect cancers in their early stage, simplified the prognosis of cancer development (38). Cancer biomarkers help stratify patients to receive specific therapeutics. Biomarker could be DNA, mRNA, protein and various cellular metabolites (39). Over the past years, many advances have been made in the detection and evaluation of cancer biomarkers (40–44). In the next section, we will detail the development of exosomal miRNA as cancer biomarkers.

#### EXOSOMAL miRNA AS A REGULATOR AND BIOMARKER IN CANCER

Exosomal miRNA is involved in the proliferation, invasion, migration and drug resistance of various cancers. Therefore, exosomal miRNA has the potential to be the biomarkers for cancer diagnosis and treatment. In **Table 1**, we summarize some recent researches on various exosomal miRNA as regulators and biomarkers in various cancers.

Breast cancer (BC) is a highly prevalent cancer and the second leading cause of cancer-related death among women (95, 96). MiR-9 is a classic miRNA in cancer development. Baroni TABLE 1 | Exosomal miRNA as regulators and biomarkers in different cancers.


et al. indicated that exosomal miR-9 has the ability to induce human breast fibroblasts to have cancer associated fibroblasts (CAFs)-like properties (45). Exosome transferring miR-222 can promote BC cells migration and invasion by activating Nuclear factor-κB (NF-κB) (46). Exosomal miRNA also regulate tumor growth by influencing the metabolic reprogramming of BC cells. Yan et al. suggested that BC cells secrete exosomal miR-105 to promote tumor growth through the regulation of metabolic reprogramming in stromal cells (47).

Exosomal miRNA regulates tumor growth in other cancers as well. In colorectal cancer (CRC), transforming growth factorbeta (TGF-β) significantly contributes to the upregulation of exosome-meditated miR-200b, which promotes colorectal cancer cell proliferation by suppressing the expression of p27 in target cells. (51). Li et al. demonstrated that the absence of exosomal miR-148b derived from CAFs is the cause of invasion and metastasis in endometrial cancer (97). In lung cancer, Wu et al. indicated that exosomal miR-96 is associated with proliferation, migration and drug resistance by directly binding to wild-type LMO7 gene (60).

The expression of some exosomal miRNA in cancers is specific. The specificity enables exosomal miRNA to become cancer biomarkers. Sohn et al. proposed that serum exosomal miRNAs have the potential to become novel biomarkers for hepatocellular carcinoma (66). Moreover, Huang et al.

extracted serum-derived exosome from patients with gastric noncardia adenocarcinoma and detected the expression of miRNA. They identified the expression of miR-195-5p, miR-20a-3p, and miR-196-5p in exosomes and found that these miRNAs significantly increased. This finding provided a reference for clinical application and diagnosis using exosomal miRNAs (56).

# THE MECHANISM OF CANCER DRUG RESISTANCE WITH EXOSOMAL miRNA

The mechanism of drug resistance is complex, the currently known mechanisms of drug resistance include drug efflux (98), mutation of drug target (99, 100), alterations in drug metabolism (101), DNA damage repair (102, 103), alterations of energy programming, cancer stem cells and epigenetic changes (104, 105). Most of these processes were regulated by exosomal miRNA (**Figure 2**). Different treatments have been developed to circumvent these resistance mechanisms (106–112). In this section we associate several mechanisms of drug resistance with miRNAs.

#### Drug Efflux and Metabolism in Cancers

Drug resistance is always accompanied by the dysfunction of pharmacokinetic factors, that are absorption, distribution, metabolism and elimination (ADME) of drugs. Exosomal miRNA participate drug resistance by interfering drug efflux and metabolism as well.

The excessive drug efflux is a classic mechanism of drug resistance. The human ATP-binding cassette (ABC) transporter superfamily is closely associated with the excessive efflux of drug. In the ABC transporter superfamily, several ATP-driven efflux transporters are the classic regulator of drug efflux: ABCB1 (Pgp/MDR1), ABCC1 (MRP1), ABCG2 (BCRP), ABCC2, MDR4 and MDR5 (113–118). The efficacy of drugs is closely related to the concentration of the drug inside the cells. In drug resistant-tumor cells, overexpression of these ABC transporters pumps anticancer drug out of cells, decreasing the concentration of drugs.

Tumor-derived exosomal miRNA cargo regulates the expression of ABC transporters and facilitates drug resistance in tumor cells. ABCB1 is one of the ABC transporters, some researchers reported that ABCB1 enriched in microvesicles and exosomes shed by drug-resistant cells (119). These EVs transfer ABCB1 to drug-sensitive cells, making the recipient cells express functional ABCB1 and acquiring drug resistance. However, the half-life of ABCB1 is shorter than 24 h and the transfer of ABCB1 is unstable (120). The resistant mechanism of drug-sensitive cells cannot be merely explained by the transfer of ABCB1. Sousa conjectured in his review that ABCB1 may co-transport with miRNA so that ABCB1 can be expressed stably for a long time (121). After that, exosomal miRNA modulates transcripts in recipient cells to acquire resistance phenotype (122). For example, exosomal miR-1246 secreted by ovarian cancer (OC) cells inhibits the expression of Cav1 and upregulates ABCB1 expression to induce tumorpromoting phenotype and drug resistance. Based on the preclinical experiments in vivo, miR-1246 inhibitor treatment in combination with chemotherapy shows great potential in the treatment of OC (12). Exosomal miRNA is a double-edged sword in the occurrence and development of drug resistance. Some miRNAs have positive effects in drug resistance, while some can enhance the chemosensitivity of cancer cells. Liu et al. found that exosome-transmitted miR-128-3p downregulates the expression of MDR5, decreasing oxaliplatin efflux and improving chemosensitivity of oxaliplatin-resist cells in colorectal cancer (123).

The activation of drug is related to the corresponding enzymes in body. The prodrug is activated by enzyme action, or the drug is metabolized into an inactive form due to some enzymes in vivo. For example, gemcitabine is metabolized by deoxycytidine kinase (dCK) and incorporate with nucleosides in DNA and RNA, preventing DNA from replicating properly. Cytidine deaminase (CDA) is an enzyme that metabolizes gemcitabine to become an inactive form. When the tumor emerges drug resistance, it is often accompanied by the inactivation of dCK or activation of CDA, leading to gemcitabine degradation or inactivation and eventually causing drug resistance (78, 124, 125). Exosomal miRNA is involved in drug metabolism. In pancreatic adenocarcinoma, tumor associated macrophages secrete exosomes that transferring miR-365 to induce gemcitabine-resistance (79). The specific mechanism is that macrophage-derived exosomes (MDE) transfer miR-365 into pancreatic ductal adenocarcinoma (PDAC) cells. Once miR-365 entered PDAC cells, the concentration of triphosphatenucleotide (NTP) is increasing. NTP can compete with phosphorylated gemcitabine for DNA incorporation, so that it prevents activation of gemcitabine. Moreover, exosomal miR-365 upregulates the expression of CDA and promotes the inactivation of gemcitabine leading to gemcitabine resistance (79).

#### Metabolic Reprogramming and TME Acidosis

Energy reprogramming has been accepted as a hallmark of cancer (126). In order to maintain survival, proliferation and dissemination, cancer cells need to reprogram their metabolism to ensure the increasing energy demand (127– 129). Mitochondrial oxidative phosphorylation (OXPHOS) and glycolysis are two major metabolic pathways to generate adenosine triphosphate (ATP) to support physiological activities in our daily life. A common characteristic in primary and metastatic cancer is the upregulation of glycolysis (130). Glycolysis usually occurs in an anoxic condition. However, even in aerobic conditions, cancer cells undergo aerobic glycolysis by reprogramming the glucose metabolism and glycolysis is still widespread in TME. This phenomenon is called Warburg Effect. An important reason of this effect is that during glycolysis, glucose is metabolized into pyruvate and lactate. In cancer cells, excessive production of lactate leads to TME acidosis (131). The acidic TME largely contributes to the immunologic escape, because the decrease in extracellular pH leads to the reduction of cytotoxic T- cell function, thus the cancer cells can acquire a strong survival advantage which promotes cancer metastasis, invasion and drug resistance (132–134).

Regulation of glycolysis is one of the ways to inhibit cancer drug resistance (134). The GLUT family is closely related to glucose transport into cells. GLUT1, one of the family members in GLUT, is upregulated in many malignant tumors (135). The upregulation of GLUT1 is associated with mammalian target of rapamycin (mTOR) and the activation of mTOR increases glycolysis and promotes drug resistance (132). The decreased expression of miR-100 is involved in drug resistance in several cancer. mTOR is a target gene of miR-100-5p which binds to the 3′UTR directly and decreases the expression of mTOR and enhances chemo-sensitive of cancer cells. Qin et al. (136) indicated in their study that the expressing of miR-100-5p is not only related to the cell itself, but also to the extracellular microenvironment. Exosome as a messenger for intracellular communication, the concentration of miR-100-5p in exosomes is reflected the content in surrounding microenvironment. The downregulation of miR-100-5p in microenvironment leads to cisplatin resistance in lung cancer cells (136). In addition, TP53INP1 is also a stress protein, which has been indicated to play a tumor suppressive role by regulating metabolic homeostasis (137). Fang et al. showed that CAF derives exosomal miR-106b, which promotes gemcitabine resistance by directly targeting TP53INP1 (138).

# DNA Damage Repair

As a target of anticancer drugs, DNA damage induces cancer cell death. Genotoxic agents are designed for damaging DNA or preventing the synthesis of new DNA to inhibit cell proliferation. Genotoxic agents are classified as direct damage, such as cisplatin; and indirect damage, such as topoisomerase inhibitors (24). However, in addition to cell death, DNA damage response (DDR) includes the DNA damage repair (139).

DNA damage repair is originally a way to maintain genomic stability in cells. However, DNA damage repair has also been found to be a resistance mechanism because of the widespread use of genotoxic agents (140). DNA repair mechanisms can be briefly divided into the following four categories: (a) Nucleotide excision repair (NER): NER works in a way that is suitable for repairing bulky DNA lesions by using DNA ligase to attach repair patch to the damage DNA regions, which is associated with platinum agent resistance. (b) Base excision repair (BER): BER works through repairing a small number of bases and performing some modification, such as alkylation and oxidative lesions. This repair mechanism is related to the resistance of genotoxic agents nitrosoureas. (c) Mismatch repair (MMR): MMR participates in the modification of oxidation and methylation by bypassing the lesions to replicate. (d) DNA double-strand break repair: Double-strand break (DSB): DSB is the most toxic form of DNA damage. Two main repair pathways of DSB are non-homologous end joining (NHEJ) and homologous recombination (HR) (139, 141). Briefly, these two repair pathways have their own characteristics. NHEJ is more rapid, while HR is more complex and accurate. This mechanism is applicable to the damage induced by topoisomerase inhibitors, temozolomide (TMZ) and some alkylating agents (142).

Exosomal miRNA is a regulator to inhibit DNA damage repair. XRCC4 is a major participator of NHEJ, which forms a heterodimer with DNA ligase IV and covalently joins the broken DNA (143). There have been reports of XRCC4 linked to TMZ resistance in earlier years. XRCC4 is a direct target of miR-151a, the low expression of which leads to the upregulation of XRCC4 and triggers the DNA repair that makes cell resistant to TMZ. To investigate the effects of exosomal miR-151a on cancer cells, researchers incubated glioblastoma multiforme (GBM) receptor cells with exosomes secreted by TMZ-resistant cells and TMZsensitive cells. The result shows that GBM receptor cells cocultured with TMZ-resistant exosomes have stronger resistance to TMZ. However, when researchers restore miR-151a in TMZresistant exosomes, the TMZ resistance of GBM recipient cells is significantly decreases (144). This study shows that exosomes have the ability to transfer chemoresistance to sensitive cancer cells and exosomal miR-151a has the potential to become a prognostic factor in GBM treatment.

#### Deregulation of Apoptosis

Resisting cell death is a characteristic of cancers, which leads the unlimited proliferation of cancer cells and the development of drug resistance (145, 146). Drug resistant-cancer cells are often accompanied by downregulation of intracellular apoptotic proteins or up-regulation of anti-apoptotic proteins.

Exosome secreted by drug-resistant cells can transmit the resistance to neighboring cells. Zhang et al. (147) indicated that exosomal miR-214 mediates gefitinib resistance in nonsmall cell lung cancer (NSCLC). Compared with sensitive cancer cells, the miR-214 in exosomes secreted by gefitinib resistant-cells is significantly increased. Gefitinib resistant-cells secreted exosomal miR-214 could confer gefitinib resistance in NSCLC by suppressing cell apoptosis (147). In addition to cancer cells, exosomes secreted by stroma cells also act on resistant targets by transferring miRNA, making cancer cells to acquire drug resistance. Paclitaxel is a common agent for the treatment of ovarian cancer. However, the efficacy of paclitaxel treatment is greatly reduced if the ovarian cancer cells develop resistance to paclitaxel. By using sequencing technology, Au Yeung et al. (148) identified that miR-21 isomiRNAs have higher expression level in the exosomes of cancer-associated adipocytes (CAAs) and CAFs than in those from ovarian cancer cells. After exosomal miR-21 transship to ovarian cancer cells, miR-21 binds to apoptotic protease activating factor 1 (APAF1) and the expression of APAF1 is downregulated (148). APAF1 combined with cytochrome c (Cyt-c) and dATP to form apoptosomes, increasing caspase-9 and caspase-3, leading to massive mitochondrial damage and finally inducing cell apoptosis (149). Therefore, the decrease of APAF1 has the ability to suppress apoptosis and eventually cause drug resistance in cancer cells. This result showed that in omental tumor microenvironment, cancer cells have a negative effect on neighboring stromal-derived exosomal miR-21 and acquire malignant phenotype, including drug resistance (148). Moreover, exosomal miR-196a derived from CAFs confers cisplatin resistance in head and neck cancer (HNC). In order to explore the mechanism of exosomal miR-196a in HNC cells, Qin et al. (150) used miRecords algorithm and finally found the target of exosomal miR-196a: CDKN1B and ING5. CDKN1B and ING5 exhibit different functions in miR-196a-mediated cisplatin resistance. ING5 gene is a major gene to regulate apoptosis. Therefore, they proposed that exosomal miR-196a promote cisplatin-resistance in HNC cells by suppressing apoptosis of cancer cells (150).

#### Epithelial-to-Mesenchymal Transition (EMT) and Cancer Stem Cells (CSCs)

In the process of cancer growth, genetic and non-genetic factors induce biological heterogeneity, resulting in phenotypic difference of tumors. The phenotypic diversity of malignant cancers is considered as a significant driver that induces drug resistance.

The cancer stem cells (CSCs) concept provides a good explanation for the association between heterogeneity and the resistance of cancer cells. Because of the renewal properties and genomic instability, CSCs are closely related to the proliferation, metastasis, and recurrence of cancer (151). Epigenetic regulation has a great contribution to the behaviors of cancer cells. Epigenetic differences between CSCs and non-CSCs have a great possibility that caused by epithelial-tomesenchymal transition (EMT) (152). When epithelial cells transform into mesenchymal cells, cancer cells acquire the properties of migration and invasion and even drug resistance (153–155). What's interesting lies on the study that shows that EMT only occurs in tumors with CSCs (156–158). In tumor microenvironment, CSCs comprise a small proportion of total cells in tumor, most of the cancer cells are non-CSCs (157). However, the traditional cancer treatment merely kills most of the non-CSCs and the CSCs are retained. These residual CSCs eventually induce tumor recurrence and drug resistance through differentiation (159–162).

In recent years, targeting therapy of CSCs by inhibiting EMT has become an effective way to treat cancers and prevent drug resistance (163–167). EMT is an effective target to affect drug resistance. Exosomal miR-32-5p is proved to induce multidrug resistance in hepatocellular carcinoma via the PI3K/AKT pathway to promote EMT and angiogenesis (168). CSCs themselves also secret exosomes to induce drug resistance. MiR-155 is a classic and multifunctional modulating miRNA which is overexpressed in multiple malignant cancers (169). Santos et al. (170) carried out a study which supports a putative mechanism of exosomal miRNA transmission between cancer cells: miR-155 is enriched in exosomes secreted by CSCs and drug resistant cells. In addition, they observed the downregulation of E-Cadherin (E-Cad) and upregulation of mesenchymal biomarkers, which demonstrated that CSCs and drug resistant cells have the ability to trigger the EMT process in recipient cells by transferring exosomal miR-155 and eventually lead to the recipient cells possess resistance (170). In pancreatic cancer cells, the gemcitabine-resistant CSCs can secret miR-210 enriched

TABLE 2 | Summary of common anticancer drugs and exosomal miRNA involved in drug resistance.


exosomes. Gemcitabine-resistant CSCs enhance drug resistant by transferring exosomal miR-210 to gemcitabine-sensitive cells (171).

During these years, more and more studies have revealed that different types of cells secrete exosomal miRNA in tumor microenvironment and participate in the process of drug resistance. The drug resistant mechanism of exosomal miRNA on several common anticancer chemotherapeutic agents and molecular targeted agents are summarized in **Table 2**.

#### CONCLUSIONS

Drug resistance is an eternal topic in cancer treatment. In this article, we discussed the role of exosomal miRNA in different mechanisms of drug resistance. Some of them act as "communicators" and some of them "biomarkers" that facilitate communication between cancer cells with other cancer cells or cancer cells with tumor microenvironment, enriching the knowledge background about the diagnosis of cancer. However, drug resistance in cancer is not caused by only one or several mechanisms, it is the combined action of the intrinsic (such as mutation) and extrinsic (such as drug inactivation) factors. Although progress has been made in suppressing the emergence of drug resistance, there is still a long way to go to eradicate the problem of drug resistance. Nevertheless, the knowledge of exosomal miRNA will provide some clues to help exploring the secret of cancer drug resistance.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

QG, QW, and JZ conceived the review. QG, YL, and CS searched the literature and drafted the manuscript. YaY, RA, and HC critically appraised the literature. YiY, HW, and CS edited the manuscript. All authors approved the final version of the manuscript.

#### FUNDING

This work was supported by National Natural Science Foundation of China (81773888, U1903126 and 81902152), Guangdong Basic and Applied Basic Research Foundation (2020A1515010005, 2020A1515010605), Fund from Guangzhou Institute of Pediatrics/Guangzhou Women and Children's Medical Center (No: IP-2018-012).


premetastatic niche to promote metastasis. Nat Cell Biol. (2015) 17:183–94. doi: 10.1038/ncb3094


to foster lung metastasis of liver cancer. Nat Commun. (2018) 9:191. doi: 10.1038/s41467-017-02583-0


Rev Cancer. (2018) 18:669–80. doi: 10.1038/s41568-018- 0056-x


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Guo, Wang, Yan, Liu, Su, Chen, Yan, Adhikari, Wu and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Circulating Tumor DNA Is Capable of Monitoring the Therapeutic Response and Resistance in Advanced Colorectal Cancer Patients Undergoing Combined Target and Chemotherapy

Hua Cao1†, Xinyi Liu2†, Yixin Chen<sup>1</sup> , Pan Yang<sup>2</sup> , Tanxiao Huang<sup>2</sup> , Lele Song2† and Ruilian Xu<sup>1</sup> \*

*<sup>1</sup> Department of Oncology, Shenzhen People's Hospital, The 2nd Clinical Medical School of Ji'nan University, The First Affiliated Hospital of Southern University of Science and Technology, Shenzhen, China, <sup>2</sup> HaploX Biotechnology, Co., Ltd., Shenzhen, China*

Colorectal cancer (CRC) is a highly lethal disease worldwide. The majority of patients receiving targeted therapy or chemotherapy develop drug resistance, while its molecular mechanism remains to be elucidated. The plasma circulating tumor DNA (ctDNA) exhibited the potential in identifying gene variations and monitoring drug resistance in CRC treatment. In this study, we monitored the ctDNA mutational changes in advanced CRC patients underwent first-line therapy with bevacizumab and cetuximab combined with chemotherapy. The mutation spectrum of 43 patients was established by a 605 gene next-generation sequencing (NGS) panel. The baseline measurement shows that genes with the highest mutation frequency were TP53 (74%), APC (58%), KRAS (40%), SYNE1 (33%), LRP1B (23%), TOP1 (23%), and PIK3CA (21%). Mutations in TP53, APC, and KRAS were detected in 29 paired plasma and tissue samples with the consistency of 81, 67, and 42%, respectively. Clinically targetable gene mutations, such as APC, RNF43, SMAD4, BRAD1, KRAS, RAF1, and TP53, were also identified in ctDNA. The overall consistency between ctDNA and tissue samples was 54.6%. Alleviation of mutational burden in BRAF, KRAS, AMER1, and other major driving genes was observed following the first-line therapy. Patients with KRAS and TP53 mutations in tissues appeared to benefit more than the wild-type counterpart. The dynamic change of plasma mutation status was consistent with the tissue tumor burden and was closely correlated with disease progression. In conclusion, ctDNA monitoring is a useful method for molecular genotyping of colorectal cancer patients. Dynamic changes in resistance can be sensitively monitored by gene variation status, which potentially helps to develop treatment strategy.

Keywords: colorectal cancer, ctDNA, NGS, sequencing, bevacizumab, cetuximab, monitoring, resistance

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Jigang Wang, China Academy of Chinese Medical Sciences, China Francesco Grignani, University of Perugia, Italy*

> \*Correspondence: *Ruilian Xu xuruilian@126.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *30 November 2019* Accepted: *16 March 2020* Published: *07 April 2020*

#### Citation:

*Cao H, Liu X, Chen Y, Yang P, Huang T, Song L and Xu R (2020) Circulating Tumor DNA Is Capable of Monitoring the Therapeutic Response and Resistance in Advanced Colorectal Cancer Patients Undergoing Combined Target and Chemotherapy. Front. Oncol. 10:466. doi: 10.3389/fonc.2020.00466*

# INTRODUCTION

Colorectal cancer is one of the most common malignant tumors worldwide, responsible for the second and fourth highest mortality among men and women, respectively (1, 2). Standard therapies, including chemotherapy, radiotherapy, targeted therapy, and surgery, are frequently used for colorectal cancer treatment. However, tumor cells can evolve under the pressure of treatment selectivity. Therefore, elucidating the genetic alterations that may potentially drive tumor cell resilience is crucial to advance the knowledge in cancer therapy (2).

Liquid biopsy has been widely recognized as a real time monitoring method to detect tumor-related genetic alterations (3). In fact, liquid biopsy has been largely used to analyze circulating tumor DNA (ctDNA), circulating tumor cells (CTC) and exosomes isolated from peripheral blood. This method can potentially recognize, at the genomic level, tumors associated with lower invasiveness (4). CTC, exosomes, and ctDNAs are widely used in clinical diagnosis and in treatment monitoring, however, there are still some limitations that deserve improvement. Among these three types of biological materials, ctDNA has unique advantage in monitoring the tumor genotype with the development of next generation sequencing (NGS). It is recognized as a method to reveal tumor genome information and can be used to discover cancer evolution and clonal heterogeneity. Therefore, it can be used as a tumor biomarker to evaluate treatment efficiency and drug resistance (5–9).

The NGS technology has been used to monitor changes on ctDNA levels in peripheral blood and in the dynamic change of drug-resistant genes, therefore allowing the selection of novel therapeutic approaches and drug strategies. Due to its effectiveness and fast accessibility, NGS has been widely used in liquid biopsies to analyze genomic alterations in the peripheral blood of cancer patients. In the present study, we used an established sequencing workflow to detect genomic alterations in tumor tissues and peripheral blood from CRC patients who have received first-line therapy with bevacizumab and cetuximab combined with chemotherapy. The NGS panel containing 605 tumor-associated genes was used to detect the main driver gene mutations, resistance-related mutations, and to track dynamic changes of tumor ctDNA during treatment. Our observation suggests that it is vital to identify CRC patients with drug resistance during treatment and adapt the treatment strategy accordingly.

#### METHODS AND MATERIALS

#### Patients and Samples

A prospective cohort study was designed and implemented in the Shenzhen People's Hospital (Shenzhen, China). Blood samples and intestinal tumor tissues were collected at the Shenzhen People's Hospital. This research was approved by the Shenzhen People's Hospital Ethics Committee and conducted in accordance with its guiding principles. All patients received written informed consent for the use of clinical samples. Patient information was kept anonymous for confidentiality. The main inclusion criteria include adults over 18 years old and those have complete clinicopathological information and confirmed diagnosis of CRC by imaging examination (including endoscopy, ultrasound, MRI, CT, etc.) and/or subsequent pathological examination. Subjects were included for those who have the indications for adjuvant chemotherapy, neoadjuvant chemotherapy, or chemotherapy combined with target therapy (target therapy includes but is not limited to cetuximab, apatinib, bevacizumab, and trastuzumab) based on current condition. Patients were also included for those with blood samples and/or tissue samples available before the start of the current therapy, and those who can be followed up and agree to provide the subsequent blood samples during and after therapy. The main exclusion criteria include pregnant woman, and those who have history of cancers other than CRC, or history of therapy on other cancers. Subjects were excluded for those with no blood and/or tissue samples available before the start of therapy or those not available for follow-up or cannot provide blood samples during and after therapy. Patients with incomplete information were also excluded. As a result, a cohort of 41 patients with advanced CRC (excluding two patients with stage I CRC), treated with cetuximab, apatinib, bevacizumab, trastuzumab, neoadjuvant, adjuvant chemotherapy, or any combination, was enrolled into the study. Tissue samples were prepared from formalin-fixed and paraffin-embedded (FFPE) samples and 10 ml peripheral blood samples were collected with anticoagulant tubes. The patient clinical information related to each sample is shown in **Table 1**. Progression free survival (PFS) was used to assess the effectiveness of therapy.

#### DNA Extraction and Quantification

For the FFPE samples, ten 5µm tumor slices were used for DNA extraction using the QIAamp DNA FFPE Kit (QIAGEN, Valencia, CA, USA) following the manufacturer's instructions. DNA from fresh tissue samples was extracted using the EasyPure <sup>R</sup> Genomic DNA Kit (Beijing TransGen Biotech, Beijing, China). Blood samples from patients were collected in Ethylene Diamine Tetraacetic Acid (EDTA) tubes and centrifuged at 1,600 g for 10 min and at 4◦C. The supernatants were further centrifuged at 10,000 × g for 10 min at 4◦C, and plasma was harvested and stored at −80◦C until further use. ctDNA was extracted from 3 to 3.5 ml plasma using the QIAamp Circulating Nucleic Acid kit (Qiagen, Inc., Valencia, CA, USA) according to the manufacturers' instructions. Blood cell fragments (including peripheral blood lymphocytes and red cells) were preserved at −20◦C for further study. We applied the RelaxGene blood DNA system (Tiangen Biotech) to extract genomic DNA from peripheral blood lymphocytes (PBLs) as the normal control for mutation calling from cancer tissues and ctDNA. The quality control of the DNA was achieved using Qubit 2.0 (Thermo Fisher Scientific), in accordance with manufacturer's instructions.

#### Library Construction and Sequencing

DNA from blood samples was cleaved using a doublestranded DNA Fragmentase (Roche Sequencing and Life Science, Indianapolis, IL 46250, USA). The construction of the ctDNA library was performed using a KAPA Library preparation kit TABLE 1 | Demographic and clinical characteristics of study participants.


(KAPA Biosystems, Wilmington, MA 01887, USA). Sequencing was performed to an average depth of 5,000 × on Illumina Novaseq6000. A sequencing panel of 605 genes targeting the exome regions was used to identify mutations. The logarithmic ratio of each gene region was properly computed. WESPlus gene panel (an upgraded version of the standard wholeexome sequencing (WES) (HaploX Biotechnology) for cancer tissue sequencing. Seven to eight polymerase chain reaction (PCR) cycles, depending on the amount of DNA input, were performed on Pre-LM-PCR Oligos (Kapa Biosystems, Inc.) in 50 µl reactions. DNA sequencing was then performed on the Illumina Novaseq 6000 system according to the manufacturer's instructions. Data which meet the following criteria were chosen for subsequent analysis: the ratio of remaining data filtered by fastq in raw data is ≥85%; the proportion of Q30 bases is ≥85%; the ratio of reads on the reference genome is ≥85%; target region coverage ≥98%; average sequencing depth in tissues is ≥500×; average sequencing depth in blood cfDNA is ≥1,500×. The called somatic variants need to meet the following criteria: the read depth at a position is ≥20×; the variant allele frequency (VAF) is ≥2% for tissue DNA and ≥0.05% for cfDNA from blood; somatic-P ≤ 0.01; strand filter ≥1. Allele frequencies were calculated for Q30 bases. The copy number variation was detected by CNVkit version 0.9.3 (https://github. com/etal/cnvkit). Further analyses of genomic alterations were also performed, including single nucleotide variants (SNVs), copy number variations (CNVs), insertion/deletion (Indels), fusions, and structural variation. Tumor mutation burden (TMB) was referred as the total number of incorrect coding, base substitution, insertion, and deletion in somatic cells per million bases.

#### Statistical Analysis

All charts and data analyses were performed using R statistical software package (https://www.r-project.org/). The significant difference of TMB in tumor tissues was determined by Student's t-test. According to the type of KRAS and TP53 mutation identified (i.e., common or non-common). Survival curves were compared by Log-rank (Mantel-Cox) test. P < 0.05 was considered statistically significant. Data was represented with 95% confidence interval. Several packages of the R software were used to plot some figures, including the "ComplexHeatmaps" package (**Figures 1**, **2B**, **3**, **4**), the "ggplot2" package (**Figures 2C,D**), and the "survival" package (**Figure 5**).

# RESULTS

#### Mutation Profiling of CRC Patients Before Therapy in Tissue and Blood

In order to verify the feasibility of ctDNA in peripheral blood by NGS, we first recruited 43 CRC patients (including 41 stage III-IV and 2 stage I patients) undergoing chemotherapy combined with target therapy agents (cetuximab, apatinib, trastuzumab) in neoadjuvant and/or adjuvant therapy. The baseline clinical characteristics of these patients are shown in **Table 1**. Their median age was 53 years old (ranged from 25 to 84 years old). The majority of the CRC patients were male (n = 31, 72.1%). The most frequent site of metastasis was the liver (n = 11, 25.58%), followed by lymph node (n = 7, 16.28%), lungs (n = 6, 13.95%), and peritoneum (n = 3, 6.98%). Among all patients, a total of 26 (60.47%) presented wild-type RAS. The number of patients

FIGURE 1 | Mutation spectrum of baseline plasma and tissue samples from 43 patients CRC. The genes with a high mutation frequency among all patients are listed on the left and individual patients are represented by the columns. Mutation types of non-synonymous single nucleotide variant, SNV (single nucleotide variant), indel (insertion-deletion), gain(stop gain, non-sense mutation), and loss(stop loss, missense mutation) are represented by blue, orange, and dark, respectively.

28 patients. (D) Comparison of tumor mutation burden (TMB) in low consistency (<75%) and high consistency (>75%) groups.

with stage I, III, and IV was 2 (4.65%), 6 (13.96%), and 35 (81.4%), respectively. Two patients were treated with adjuvant chemotherapy (4.65%), while 8 were treated with neoadjuvant chemotherapy (18.6%), 5 patients were at the first-line therapy (11.63%), and 2 patients at the second-line therapy (4.65%), and 2 patients at the third-line therapy (4.65%). A total of 24 patients were solely treated with chemotherapy (55.81%).

Among 41 advanced colorectal cancers, 78 genes with high frequency mutations were identified. The top ten highly mutated genes were TP53 (74%), APC (58%),

KRAS (40%), SYNE1 (33%), LRP1B (23%), TOP1 (23%), PIK3CA (21%), SRC (21%), BRCA2 (16%), and SMAD4 (19%) (**Figure 1**). The most frequent SNV mutations were observed in KRAS (n = 32, 74.4%), TP53 (n = 18, 41.8%), and SYNE1 (n = 23, 53.5%). The most frequent Indel mutations were observed in APC3 gene in 25 of 43 patients (58.1%). The most frequent CNVs occurred in TOP1 and SRC (**Figure 1**).

Blood and tissue samples were collected from all patients before receiving any therapy. By comparing the mutational rates (and their consistency) in tissues and plasma, we found that 249 out of 506 mutations detected in tissue samples were also found in the corresponding ctDNA samples (**Figure 2A**). The consistency calculated on mutational sites between tissue and blood samples was 32%. Our sequencing strategy enabled the detection of SNVs, indels, CNVs, and gene fusions in DNA

first line group) indicate the high-frequency mutational status for baseline (before the first-line therapy) and after the first-line therapy. Annotations to the right of the panel show the exact mutations and the percentage to the left of the panel shows the frequency for a certain mutation. The three lanes in both group illustrates the mutational status for PD, PR, and SD groups, respectively. Blue squares indicates cases with bevacizumab/chemotherapy that carried corresponding labeled mutations and red squares indicates cases with cetuximab/chemotherapy that carried corresponding labeled mutations.

derived from tumor tissues and ctDNA. In patients with paired tissue and blood samples, a total of 206 SNVs and 43 short Indels were detected (**Figure 2B**). The frequency of mutations and the consistency between tissue and blood samples are shown in **Figure 2B**. The most consistent gene was TP53 (81%), followed by APC (67%) and KRAS (42%).

The consistency at individual CRC stages was determined by comparing the detection of mutations in ctDNA and corresponding tumor tissues for each patient (**Figure 2C**). It can be observed that the consistency for stage IV patients distributed in a wide range, and 24% of patients exhibited a consistency >0.75 (7/29), while 76% of patients exhibited a consistency <0.75 (22/29). The tTMB (tissue TMB) between those with a consistency >0.75 and those <0.75 was not statistically different (**Figure 2D**), although the trend showed that the group with lower consistency (< 0.75) had a lower tTMB.

# Treatment Significantly Altered Mutational Landscape in ctDNA

A NGS panel containing 605 genes was used to monitor the genetic alterations following the therapy. Here we compared the alterations of gene mutations at different time points following treatment to assess the potential effect of therapy and to find any potential instruction on therapeutic strategy. Four specific cases were presented to illustrate the significance of NGS assay in clinical treatment.

Patient 00601 first presented adenocarcinoma (stage IV), and bilateral lung metastasis were found 6 months after surgery. The patient was treated with Bevacizumab and Nivolumab, and blood samples were collected 1 year after treatment. BRAF V600E and SMAD4 R361H were still detected after a series of therapies compared with previously surgical resected samples. Although mutations in some genes (such as CCNE1, DICER1, MSH2, and PIK3CD) were altered, new mutations (ATM p.L2541P, ATM

p.V2540L, and NF1 p.P2742L) were identified (**Figure 3**). We speculate that these new mutations, combined with those that still existed, may suggest the development of drug resistance or metastasis.

Patient 00603 presented sigmoid adenocarcinoma (T4aN0M1), and metastasis was found in the liver, lung, and lymph nodes. A combined treatment of Bevacizumab and Apatinib with FOLFIRI was applied but did not substantially alter the key driver gene mutations in APC, ASXL1, BARD1, and KRAS (**Figure 3**). These mutations could correlate with disease progression and poor overall prognosis of the patient, as he only survived 194 days after confirmation of diagnosis.

Patient 00606 presented rectal cancer (stage IV) with lung and brain metastasis. Blood samples were collected after 1 week treatment with Bevacizumab and FOLFIRI (second-line treatment). However, gene mutations of APC, ERBB2, IKZF1, KRAS, and RAF1 were still detected after treatment compared with pre-therapeutic results. Moreover, new mutations of APC, IRS2, NR4A3, NTRK1, PRDM1, and VEGFA were detected. Discover of these new mutation normally suggest disease progression or new metastasis, which was proved by the clinical status of PD of the patient (**Figure 3**).

Patient 00608 presented rectal cancer (stage IV) and lung metastasis. Blood sampling was also performed to investigate the disease progression following Bevacizumab and FOLFOX6 treatment. The genetic variation from needle biopsy samples of pulmonary metastases greatly differed from that of the blood detection (**Figure 3**). Novel RNF43 mutation sites in ctDNA were identified, supporting previous observation showing that RNF43 frameshift mutation may contribute to tumorigenesis (6).

#### Comparison of Key Mutations in Baseline and After First-Line Therapy

We monitored key mutation changes of 13 patients between baseline and after the first-line therapy of bevacizumab, cetuximab with chemotherapy and grouped the patients by response [progressed diseases (PD), partial response (PR), and stable disease (SD)]. The key mutations were different between baseline tissues and progressive ctDNA samples in all three groups (**Figure 4**). Most of the genes were reported to contribute to the carcinogenesis of CRC. Compared with baseline, a new mutation (PDGFRB p.Q443R) was observed during disease progression. More importantly, the combination of bevacizumab/chemotherapy or cetuximab/chemotherapy was able to alleviate the mutation of the driving gene, such as BRAF, KRAS, AMER1. Mutations in PPIAL4D p.S99F and SPATA31A5 p.T1139R were not observed after therapy. The above observations suggest that mutational profile of ctDNA may help to determine the response of patients to treatment.

#### TP53/KRAS Mutations in Tumor Tissue Are Potential Predictive Factors for Treatment Response

In this study, 18 patients were treated with bevacizumab combined with chemotherapy, while 4 patients were treated with cetuximab combined with chemotherapy, and 2 with chemotherapy alone. We divided these patients into two groups by TP53/KRAS co-mutation. It can be seen from the survival analysis in **Figure 5** that PFS appeared to be shorter for patients with no TP53/KRAS co-mutation, with a median PFS of 381 days for no co-mutation group compared with 460 days in TP53/KRAS co-mutation group (p = 0.13). Patients with KRAS and TP53 co-mutations in tissues exhibited a potentially better response to treatment than those containing the respective wild type genes. The relationship between the prognosis of first-line therapy and the TP53/KRAS co-mutations is worth more investigation.

#### DISCUSSION

In this study, we explored the practicability and clinical value of ctDNA in CRC therapy using paired blood and tissue biopsy samples. To clarify the correlation between drug efficacy and mutations, we built up mutational profiles for 43 CRC patients. We dynamically monitored the mutation status of each patient and investigated its relationship with therapeutic response, including drug resistance (10). Our study revealed many gene mutations with the 605-gene panel. Apart from previously reported mutations in APC, TP53, KRAS, SYNE1, PI3KCA, SMAD4, and BRAF, some other mutations may also potentially be used as biomarkers for CRC prognosis. We also showed that ctDNA may be used to analyze TMB, which is an effective method to monitor the burden of mutations following therapy. The consistency between ctDNA and tissue biopsy was 32%.

We collected plasma ctDNA and tumor tissues in CRC patients following cetuximab, apatinib, trastuzumab, neoadjuvant, adjuvant chemotherapy, or any combination of them to track tumor dynamics, including therapeutic response, metastasis, and drug resistance. Despite the limited cohort (n = 41), the samples represented patients of advanced CRC with integrated clinical information. Four representative patients with treatment plans and mutational changes were presented and analyzed in detail (**Figure 3**). Although the mutational profile of each patient was distinct, our results indicated that the mutation of driver genes dynamically changed in different patients and treatment plans. Therefore, it is vital to detect these gene mutations in an individualized manner. We believe this strategy can support the establishment of reasonable treatment plan for each patient (11–14).

After comparing key mutation profiles between baseline and after first-line therapy (**Figure 4**), a new mutation (PDGFRB p.Q443R) was observed during disease progression. Consistent with previously observations, PDGFRB appeared to promote the development of CRC (15, 16). Moreover, combinations including bevacizumab-chemotherapy and cetuximab-chemotherapy could diminish the mutations of driving genes, such as BRAF, KRAS, AMER1, and therefore potentially prevent disease progression. BRAF and KRAS are driver genes of CRC, while AMER1 is a frequently mutated gene in this condition (17). Mutations in two other genes (PPIAL4D p.S99F and SPATA31A5 p.T1139R) were previously rarely reported, which may constitute alternate mutation sites that deserve more indepth investigation. Since gene mutations typically accumulate over time, CRC with distinct heterogeneity may exhibit different genetic characteristics. The mutational information from a single tissue biopsy is limited by space and time, and may be biased in accessing the therapeutic effect or monitoring cancer progression. Ideally, multiple biopsies may be obtained to avoid this bias. Although we did not achieve ideal condition, our study improved the understanding in the roles of ctDNA in CRC monitoring and response assessment in the practice of precision medicine.

TP53 mutations in early CRC have been considered a poor prognostic factor. However, the function of mutated TP53 has not been fully characterized (18). In our study, TP53 and KRAS mutations were considered favorable factors for overall survival and disease progression of CRC. PFS was potentially shorter for patients without TP53/KRAS co-mutations compared with TP53/KRAS mutated patients and therefore might have more treatment benefit (**Figure 5**). Indeed, these two genes were often found co-mutated in our sequencing results. Therefore, although some potential interesting findings were revealed in the current study, it requires further validation using a larger number of patients. In this study, we confirmed the roles of ctDNA in dynamic monitoring of CRC therapy, which supports more extensive use of the method in future therapy.

In this study, we monitored the mutational changes and therapeutic response of late-stage CRC patients following chemotherapy combined with bevacizumab and/or cetuximab. It appeared that high frequency mutations of key driver gene, such as APC, TP53, and KRAS, were the markers that sensitively reflected the therapeutic response, while alleviation of mutations in BRAF, AMER1, and other major driver genes was also observed following the therapy. The dynamic change of plasma mutation status following the combined therapy was consistent with the tissue tumor burden and was closely correlated with disease progression. Therefore, ctDNA detection appeared to be a useful method for the molecular genotyping and sensitive for monitoring mutational status, which potentially helps to develop treatment strategy targeting actionable variations. Our observations were supported by several previous reports focusing on the monitoring capability of NGS-based liquid biopsy in late-stage CRC therapy (4, 19–24). However, the regimes used in these studies varied according to different situation. Some studies focused on the monitoring of cetuximab-based therapy for RAS wild type patients (19–21), while others emphasize the monitoring of multiple targets, including EGFR, HER2, SMAD4, and NF1 (4, 21–23). Interestingly, one study investigated the ability of both mutation and methylation markers in monitoring (24). Although the therapeutic regimes and targets varied, the observations from these studies all support the use of NGS-based liquid biopsy for therapeutic response monitoring and target identification, which endorsed our conclusions.

There were some limitations in this study. Firstly, the sample size was still small. Although 43 patients were included, only 35 patients had paired baseline plasma and corresponding tumor tissues, and key mutation changes between baseline and after the first-line therapy were obtained from 13 patients. Therefore, incomplete paired samples and loss of follow-up were key issues in the study, which increased the difficulties in analysis and making solid conclusion. These issues can be solved by increasing the number of total subjects and patients with complete information may increase accordingly. Meanwhile, strict fulfillment of inclusion and exclusion criteria may also increase the ratio of patients with complete information. Secondly, the therapeutic strategy in this study varied among different patients. Therefore, studies on mutational changes may be differentially affected by various chemotherapy drugs, and conclusions based on mixed therapies may be compromised. It would be nice to study the mutational changes with patients from homogeneous treatment, however, this was difficult operationally, as late-stage CRC patients from multiple lines of therapy generally have diversified conditions and will adopt different therapies in the real world.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study can be found in the Genome Sequence Archive for Human (GSA-Human) (Accession: PRJCA002282).

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by Shenzhen People's Hospital. The

#### REFERENCES


patients/participants provided their written informed consent to participate in this study.

#### AUTHOR CONTRIBUTIONS

RX designed the study. HC and XL performed the sample collection, data collection, and manuscript writing. YC, PY, and TH performed the sequencing and data analysis. XL, PY, and LS wrote the manuscript. LS and RX proof read the manuscript.

# FUNDING

This study was supported by the project Monitoring the efficacy of chemotherapy and evaluation of drug resistance in digestive tract tumors based on liquid biopsy and project Head-to-head comparison of apatinib mesylate combined with chemotherapy: a randomized study of bevacizumab combined with chemotherapy as second-line therapy for advanced intestinal cancer patients (SYLY201725), supported by Sanming project of Shenzhen People's Hospital.

guidelines in oncology. J Natl Comprehens Cancer Netw. (2017) 15:370–98. doi: 10.6004/jnccn.2017.0036


patients with metastatic colorectal cancer. ESMO Open. (2018) 3:e000353. doi: 10.1136/esmoopen-2018-000353


patients. Clin Colorectal Cancer. (2018) 17:e369–79. doi: 10.1016/j.clcc.2018. 02.006

**Conflict of Interest:** XL, PY, TH, and LS are currently employed by HaploX Biotechnology. HaploX provided the next generation sequencing service for this study.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Cao, Liu, Chen, Yang, Huang, Song and Xu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Recent Advances in Cancer Plasticity: Cellular Mechanisms, Surveillance Strategies, and Therapeutic Optimization

Giuseppe Nicolò Fanelli, Antonio Giuseppe Naccarato and Cristian Scatena\*

*Division of Pathology, Department of Translational Research and New Technologies in Medicine and Surgery, University of Pisa, Pisa, Italy*

The processes of recurrence and metastasis, through which cancer relapses locally or spreads to distant sites in the body, accounts for more than 90% of cancer-related deaths. At present there are very few treatment options for patients at this stage of their disease. The main obstacle to successfully treat advanced cancer is the cells' ability to change in ways that make them resistant to treatment. Understanding the cellular mechanisms that mediate this cancer cell plasticity may lead to improved patient survival. Epigenetic reprogramming, together with tumor microenvironment, drives such dynamic mechanisms favoring tumor heterogeneity, and cancer cell plasticity. In addition, the development of new approaches that can report on cancer plasticity in their native environment have profound implications for studying cancer biology and monitoring tumor progression. We herein provide an overview of recent advancements in understanding the mechanisms regulating cell plasticity and current strategies for their monitoring and therapy management.

Keywords: cancer plasticity, stem cell, heterogeneity, recurrence, liquid biopsy

# CANCER CELL PLASTICITY: A NEW LEVEL OF HETEROGENEITY IN A TUMOR

Tumor heterogeneity can be inter-tumoral, if genetic variations are found among different patients with tumors of the same type, or intra-tumoral, involving different cancer cells in the same tumor. In particular, intra-tumor heterogeneity can be caused by genetic variation, modulation in the expression of a gene, transition among cellular states or environmental changes (1). Thus, it is easy to understand that intra-tumor heterogeneity drives cancer progression and represents the main cause of treatment failure (2).

Initially, two models were proposed to justify intra-tumor heterogeneity: the "clonal evolution" model and the "cancer stem-like cell" (CSC) model. The first contemplates differences among cancer cells due to stochastic alterations in genes; according to this theory, clones which gain a growth advantage are selected over time (3, 4). The second involves CSCs, a minority population of cancer cells with self-renewing capacity that initiates and maintains tumor growth, in contrast with the majority of the cancer cells which show a more differentiated phenotype (5–7). Lately, a third model has been proposed: the "CSC plasticity" model, where CSCs possess the capacity to move between stem and differentiated states. This shift may be caused by intrinsic cues such as

#### Edited by:

*Dong-Hua Yang, St. John's University, United States*

#### Reviewed by:

*Andreas Stylianou, University of Cyprus, Cyprus Pei-Wen Hsiao, Academia Sinica, Taiwan*

#### \*Correspondence:

*Cristian Scatena cristian.scatena@unipi.it*

#### Specialty section:

*This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology*

Received: *30 November 2019* Accepted: *30 March 2020* Published: *22 April 2020*

#### Citation:

*Fanelli GN, Naccarato AG and Scatena C (2020) Recent Advances in Cancer Plasticity: Cellular Mechanisms, Surveillance Strategies, and Therapeutic Optimization. Front. Oncol. 10:569. doi: 10.3389/fonc.2020.00569*

**311**

genetic mutations and/or epigenetic modifications but also by extrinsic cues from the tumor microenvironment (inflammation, injury, senescence). In addition, the tumor-initiating potential is enhanced by the overexpression of transcription factors involved in the process of epithelial-to-mesenchymal transition (EMT) (8–10) and CSCs exhibit an induced EMT program (11). These data suggest that EMT is strictly linked to CSC features. Indeed, CSCs switch between epithelial and mesenchymal states and this process depends on both genetic mutations, epigenetic modifications and transcriptional modulation of cancer cells and signals provided by the tumor microenvironment through the mediation of growth factors, cytokines, cancer-associated fibroblasts (CAFs), tumor associated macrophages (TAMs) and hypoxia (12, 13) (**Figure 1**). These transitions promote metastasis at distant sites as well as drug resistance and, therefore, disease recurrence (14–16). In breast cancer cells co-expression of epithelial and mesenchymal genes promotes stemness inducing the formation of 3D-spheroid structures named "tumor-spheres" (17). Moreover, cells with intermediate state of EMT showed similar tumor-initiating potential when compared with fully differentiated mesenchymal cells in a mouse model of prostate cancer (18). Thus, we suppose that cancer cell stemness may be associated with a partial EMT phenotype and, indeed, cells which exhibit this intermediate EMT state possess a much more pronounced plasticity (19). According to this definition, the CSC plasticity model suggests that the two historical models of cancer heterogeneity, i.e., the clonal evolution model and the CSC model, are not mutually exclusive (1, 20–22). We believe that this third model suggests a new level of complexity in tumor heterogeneity concept.

# NOVEL APPROACHES FOR MONITORING TUMOR CELL PLASTICITY AND PROGRESSION

Solid tumors are consisted of several sub-clonal cells populations, which compete in a Darwinian manner under the selective pressures of endogenous and exogenous factors, leading to the

non-CSC/differentiated state. Tumor cell modifications as genetic and epigenetic alterations and microenvironment perturbations as inflammation, injury, and senescence represent the major causes of cancer cells plasticity. Moreover, CSCs exhibit an induced epithelial-to-mesenchymal transition (EMT) program and, particularly, they display an intermediate state of EMT. This process depends on both genetic mutations, epigenetic modifications and transcriptional modulation of cancer cells and signals provided by the tumor microenvironment (i.e., growth factors, cytokines, CAFs or TAMs). Created with BioRender.com.

clonal evolution of a dominant subclone that will characterize the tumor's molecular landscape. Hence, it will be highly heterogeneous and will dynamically change during the disease progression, so longitudinal sampling is essential to define therapeutic strategies.

Currently, cancer molecular profile is evaluated through "solid biopsies" from primary tumor or metastatic nodule; however, this approach has several issues: (i) biopsies are not representative of the whole tumor mass; (ii) often tumor site is not accessible; (iii) frequently, biopsies cannot be serially performed; (iv) each metastasis could have a different genomic landscape; (v) finally, therapeutic selective pressure has to be considered too (23–28). To overcome these limitations the novel approach of "liquid biopsy" is gaining attention.

The rapid turnover of cancer cells results in the constant release into the bloodstream of: (i) cell-free circulating tumor DNA (ctDNA); (ii) tumor derived RNA (predominantly micro-RNAs and long-non-coding-RNA) (29); (iii) circulating tumor cells (CTCs); and (iv) extracellular vesicles (EVs) (sub-cellular structures with a membrane that contain nucleic acids and/or proteins) (30–33). This enables clinicians to repeatedly and noninvasively interrogate the dynamic evolution of human cancers.

CTCs are (probably) intravasated or passively spread from the primary and/or secondary tumor sites into the bloodstream, and could be responsible for the beginning of distant metastases.

In cancer patients CTCs can be isolated single or in clusters with other CTCs or with endothelial cells, platelets, leukocytes and fibroblasts, conferring them resistance to oxidative stress, and protection from the immune system (34). Their absolute number is really low (∼1 CTC per 1 × 10<sup>9</sup> blood cells), especially in early cancer stage, and can vary between cancer types (34–36).

CTCs detection and isolation challenges are related to the high sensitivity and specificity required, and several factors still hamper standardized clinical application. Different approaches have been extensively investigated to isolate CTCs: (i) technologies such as density gradient stratification, membrane filtration, photoacoustic detection, dielectric mobility, and microfluidic separation are based on CTCs physical properties (density, size, mechanical plasticity, and dielectric mobility) that are different from those of other blood cells (37, 38). However, these techniques have low specificity (39), so new antibody-based functional assays have been developed: (ii) cytometric high-throughput imaging which provides the scanning of cells on slides; (iii) negative depletion of leucocytes and erythrocytes (Batch cell lysis, Microfluidic CTC-iChip, Immunomagnetic separation) (40) using specific antigens such as CD45 for leucocytes and glycophorin for erythrocytes; (iv) positive CTC enrichment by specific markers expressed on the cell surface (CellSearch, Magsweeper, Microfluidic CTC-Chip) such as epithelial cell adhesion molecule (EpCAM) (41) cytokeratins (CK8, CK18, CK19) (42) or tumor specific markers (TTF-1, PSA, HER-2 etc.) (43, 44). Nevertheless, no agreement has been reached on the specific antibodies to test. Indeed, EpCAM is usually lost during EMT, that sustains CTC migration, extravasation and apoptosis/anoikis resistance (45). Additionally, CTCs may develop a stem-like phenotype (46, 47). Hence, it is possible to find "commingling" CTCs that express epithelial, EMT or cancer stem cells phenotype; these CTCs have the highest plasticity potential and thus may represent CSCs (48). Different expression levels of stem cell markers such as CD24, CD44, CD133, ALDH, NANOG, OCT4, were found in ovarian (49), breast (50), and prostate CTCs (51). Remarkably, CTCs differentially express genes involved in oncogenic signaling pathways depending on their plasticity or stemness levels (52–54).

Finally, innovative developed approaches to CTCs/CSCs isolation are based on: (v) CTCs functional features such as protein secretion and cell migration (Epispot assay, Invasion assay) that allow the attachment of these cells to synthetic substrates co-treated with specific matching molecules (55); (vi) nanotechnology (Immunomagnetic nanobeads, Nanostructures substrates in microchip) (56, 57); (vii) the combination of surface/cytoplasmic markers, size and dielectrophoretic migration properties (DEPArray) (58).

Despite different approaches, in our opinion none of them completely satisfy the necessary requirements since low purity, loss of CTCs, and a narrow detection range still need to be tackled.

Finally, an additional central aspect to consider in the cancer plasticity is the complex network of epithelial-stromal cells interactions. Stroma undergoes, in parallel with the epithelial compartment, in a dynamic remodeling that may predict and explain several clinico-pathological features (59–63). To date, several in-vitro and in in-vivo models have been created and novel approaches have been used to study this interaction and its remodeling (64–66): genomic (scRNA-seq); protein translation and secretion (serial analysis of gene expression, antibody arrays and bead-based arrays, mass spectrometry and yeast, bacterial and mammalian secretion traps); autocrine, paracrine and long distance (cells co-culture, proximal culture); and directly in human tissue (multispectral imaging analysis). However, stroma characterization is still incomplete and fragmentary, also because of the difficulty to perform an "evolution tracking" of the whole stromal compartment.

Since malignancies development and progression are the result of these complex interactions, we believe that the treatment with chemotherapeutic agents against the cancer epithelial compartment combined with novel stroma-targeted therapies, may efficiently reduce cancer recurrence, also thank to the targeting and eradication of CSCs.

#### CLINICAL RELEVANCE OF CANCER CELL PLASTICITY: LIMITATIONS AND NEW OPPORTUNITIES

Though the presence of CTCs has been known since the 1869 (67), their clinical relevance was demonstrated only in 1994 (68). Despite their low number in the blood stream, they are related to clinical outcomes (34–36). In our opinion CTCs and CSCs may represent the key for early diagnosis, better prognostic stratification and a more accurate therapeutic response prediction; in addition, their concentration and pheno/genotyping could be easily measured and repeated over time. To date, however, only few authors tried to demonstrate advantages of liquid biopsy over the solid biopsies in cancer surveillance and follow-up (69, 70); this is also due to the important technical issues still to be overcome. In addition, according to recent insights, CSCs do not constitute an autonomous compartment; rather, they play an active role in the microsystem, constituted both by the epithelial and the stromal compartments; indeed several authors have demonstrated the mutual influences between CSCs and their microenvironment (71–74).

We think that one promising approach to eradicate CSCs may be to target the EMT (75): inhibitors of TGFβ-induced EMT as well as SRC, MEK, or ALK5 inhibitors have been tested (76, 77). Interestingly, also inflammatory cytokines—IL6 and IL8 in particular—may represent potential therapeutic targets of EMT: IL-6 acts as a direct regulator of breast CSCs (BCSCs) self-renewal (78) and high levels of IL-6 are demonstrated to be associated to poor clinical outcome (79); on the other hand, BCSCs have been successfully eradicated both in vitro and in animal models by blocking the IL-8 receptor CXCR1 (80). In addition, in patients with HER2 positive breast cancer, treatment with HER2 inhibitors decreased the content of BCSCs (81), suggesting that combination therapies that include HER2 targeting agents may overcome BCSCs resistance. Based on this knowledge, we believe that therapies targeting BCSCs represent an urgent need to prevent recurrence. Other authors have suggested to target also Notch, Hedgehog, Wnt and PI3K/Akt/mTOR pathways (82). Intriguingly recent evidences demonstrate that CSCs rely on mitochondrial biogenesis for their propagation (83). Lamb et al. previously demonstrated that the antibiotic doxycycline, in a known inhibitor of the 28S mitochondrial ribosome subunit, inhibits CSC propagation in vitro (84). In 2018 we performed a pilot clinical trial and demonstrated that doxycycline treatment decreases the expression of CSC markers in breast cancer tumor samples (85). We thus propose that selected antibiotics, in monotherapy or in combination, may be further studied as interesting drugs for the eradication of CSCs.

From now on, this review concentrates on specific issues concerning cancer cell plasticity in breast cancer, glioblastoma, and melanoma, which represent our expertise and, in our opinion, the most challenging models in this field. A detailed table is then provided reporting the latest knowledge in other tumor models.

#### CSC PLASTICITY IN BREAST CANCER

Breast cancer has been largely investigated in terms of its etiology (86–89) and still little is known on the mechanisms of its progression. Breast cancer cells commonly gain genetic and epigenetic modifications in their genome (90), contributing to its characteristic intra-tumor heterogeneity (91–96). Intratumor heterogeneity is strongly influenced by numerous factors from the tumor microenvironment: breast cancer cells are indeed under continuous selective pressure due to attacks by the immune system or administered therapies (97, 98). This supports breast cancer progression, conferring a competitive advantage to specific subclones (92).

In recent decades, a hierarchical organization has been proposed, where cancer cells with self-renew capacity, the socalled BCSCs, are postulated to be at the top of the tumor pyramid. Al-Hajj et al. in 2003 first isolated a population of BCSCs expressing high levels of CD44 and low levels of CD24 (CD44+CD24−/low) and capable to form tumors when injected into immune deficient mice (99). Since then, numerous studies have tested other biomarkers to sort BCSCs: among all, aldehyde dehydrogenase 1 (ALDH1) resulted to be a potentially useful alternative or complement to the CD44+CD24−/low phenotype, particularly in high grade and HER2 positive tumors (100). BCSCs not only possess high tumorigenic properties but represent the cells that mediate tumor metastasis. Indeed, the CD44+/CD24−/low phenotype is highly expressed in triple negative breast cancers (101, 102) and is associated to poor overall survival (103, 104); moreover, it has been reported among cancer cells spread into the bone marrow (105) or to the lung (106) of patients with breast cancer. At present, BCSCs are believed to enter the circulation and become CTCs: indeed, high expression levels of BCSC markers have been found in CTCs (107). Thanks to their capacity of anoikis resistance, CTCs with BCSC phenotype have the potential to seed metastatic lesions (108). Studies from liquid biopsy samples demonstrate that CTCs with a BCSC phenotype are enriched in the group with clinical disease progression (107).

A large number of studies also suggest that BCSCs display resistance to traditional cancer therapies (109–116). Cytotoxic chemotherapies target the bulk of the tumor composed of highly proliferative breast cancer cells and does not affect BCSCs that, over time, cause tumor relapse (81). In addition, genetic alterations may confer to BCSCs intrinsic chemoresistance, including modifications in proteins involved in the detoxification of chemotherapy agents (117). As reported above, BCSCs express high levels of ALDH1, that metabolizes cyclophosphamide, thus minimize its toxic effects (101). Also, tumor microenvironment plays a crucial role in BCSC chemoresistance: in hypoxic conditions, activation of hypoxia induced factors not only promotes the formation of new blood vessel but also a BCSCs quiescent phenotype (118, 119).

#### CSC PLASTICITY IN GLIOBLASTOMA

Glioblastoma (GBM) is the most frequent and deadly glial tumor (120); it is morphologically (121) and molecularly (97, 122, 123) characterized by high intra- and inter- tumor heterogeneity, which may play a pivotal role in recurrence and therapy resistance (124, 125).

The Cancer Genome Atlas has identified four GBM molecular subtypes: proneural, neural, classical, and mesenchymal (126). However, it has been demonstrated how multiple molecular subtypes may co-exist in the same tumor mass (122) or how GBM presents hybrid states with the expression of a peculiar signature overlapping two molecular subtypes (127). The establishment and the constant evolution of this heterogeneity equilibrium are due to glioma stem cells (GSCs) (128) and can be influenced by cytotoxic therapies and other endogenous factors (129, 130). However, how GSC heterogeneity is determined still remains unclear; in-vitro studies have shown that GSCs preserve their capability for recapitulating their primary heterogeneity also after many cell divisions, and temozolomide (TMZ) does not influence this capacity (131, 132); though, the same cytotoxic drug is able to drive GSCs heterogeneity and further drug resistance (133).

GSCs' isolation and characterization are based on stem markers expression; therefore, their choice is fundamental. One of the first discovered marker was CD133 (134); however, its expression is highly variable (∼20–60%) (135), and also CD133– cells have a clonogenic potential. Indeed, Chen et al. (136) divided GSCs into three subtypes based on malignant potential (MP): type 1 (high MP) and type 3 (mild MP) were CD133−; whereas, type 2 GSCs (moderate MP) were CD133+. An additional marker is CD15, which is more frequently expressed in GBM than CD133; CD15+ GSCs are more clonogenic, proliferative and tumorigenic (137). CD44 represents another reliable marker: indeed, CD44+ GSCs present high tumor-sphere forming and tumorigenic potential, and have the capability to restore the heterogeneity of the parental GBM (138). Furthermore, ALDH1A3+ GSCs, besides having the above mentioned features, express other stem cell markers, such as musashi and nestin, and are able to differentiate into several neural lineages (139, 140), and promote TMZ resistance (141).

Nevertheless, a clear-cut segregation of GBM cells between CSCs and non-CSCs is not possible yet; instead, it is more conceivable the ability of GBM cells to transit among states or the acquisition of intermediate or metastable cellular state, exhibiting a wide and continuous range of CSC signature (142, 143).

# CSC PLASTICITY IN MELANOMA

Melanoma represents a significant challenge, with low curative rates (<10%) and poor prognosis (median survival: 6–9 months) in the metastatic stage (144–146). Aggressive melanoma has revealed to co-express specific genes and proteins of multiple cellular types, including embryonic stem cells and endothelial cells, underlying cell plasticity.

3D in vitro models demonstrated that melanoma cells are able to form perfusable, vasculogenic-like channels, a biological phenomenon called vasculogenic mimicry (VM) (147). The treatment with endostatin has proved no effect on the inhibition of melanoma VM (148), thus portraying aggressive melanoma as being able to survive by its own perfusion network (149).

On the other hand, a large number of molecular studies jointly revealed a strong stem signature in aggressive melanoma, with still unknown practical significance (150–152). In particular, Nodal, a signaling pathway active in embryonic development, was notably upregulated in more aggressive melanoma (153). The nodal family of proteins, are a subset of the TGFβ superfamily and cooperate to the pluripotency of human embryonic stem cells (154). This observation led researchers to recognize a commonality in the phenotype of aggressive melanoma, linking vascular, embryonic and cancer stem cell properties.

#### CSC PLASTICITY IN OTHER SOLID TUMORS

Several authors have demonstrated how it is possible to isolate CSCs in most solid malignancies. However, several aspects and molecular features regarding cell stemness still remain uncovered; this means that even if most markers across different cancer are the same (**Table 1**), a common and reliable signature is still lacking, due to technical issues mostly. Nevertheless, in our opinion, a change in clinical trials approach may be of help to overcome this limitation. Indeed, the implementation of biobanks of fresh tissues and biological fluids may represent a precious source for the next future when new techniques and novel approaches will be introduced.



*Most stemness markers are the same but a universal signature is still lacking.*

#### CONCLUSIONS AND FUTURE DIRECTIONS

Future research studies will be needed in order to improve our understanding of the complex phenomenon of cancer cell plasticity. The recent insights on the role of plasticity in cancer progression and relapse highlights the need to develop new and combinatorial therapies, that aim to: (i)

#### REFERENCES


inhibit specific cell markers; (ii) interfere with stemness and EMT signaling pathways; (iii) affect also components of the tumor microenvironment.

#### AUTHOR CONTRIBUTIONS

GF and AN wrote the paper. CS conceived the idea, supervised, and edited the manuscript. All authors discussed and commented on the manuscript.


oesophageal mucosa. World J Gastroenterol. (2017) 23:7653–6. doi: 10.3748/wjg.v23.i43.7653


xenograft model of glioblastoma. Neuroscience. (2017) 356:35–43. doi: 10.1016/j.neuroscience.2017.05.010


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Fanelli, Naccarato and Scatena. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Long-Term Exposure of Early-Transformed Human Mammary Cells to Low Doses of Benzo[a]pyrene and/or Bisphenol A Enhances Their Cancerous Phenotype via an AhR/GPR30 Interplay

#### Edited by:

Hamid Morjani, Université de Reims Champagne-Ardenne, France

#### Reviewed by:

Christos Roussakis, Université de Nantes, France Fabrice Fleury, UMR6286 Unité de fonctionnalité et Ingénierie des Protéines (UFIP), France

\*Correspondence:

Pascale A. Cohen pascale.cohen@univ-lyon1.fr

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 29 November 2019 Accepted: 15 April 2020 Published: 29 May 2020

#### Citation:

Donini CF, El Helou M, Wierinckx A, Gyorffy B, Aires S, Escande A, ˝ Croze S, Clezardin P, Lachuer J, Diab-Assaf M, Ghayad SE, Fervers B, Cavaillès V, Maguer-Satta V and Cohen PA (2020) Long-Term Exposure of Early-Transformed Human Mammary Cells to Low Doses of Benzo[a]pyrene and/or Bisphenol A Enhances Their Cancerous Phenotype via an AhR/GPR30 Interplay. Front. Oncol. 10:712. doi: 10.3389/fonc.2020.00712 Caterina F. Donini 1,2,3†, Myriam El Helou1,2,4†, Anne Wierinckx 1,2,5, Balázs Gyorffy ˝ 6 , Sophie Aires 1,2, Aurélie Escande<sup>7</sup> , Séverine Croze1,5, Philippe Clezardin1,8 , Joël Lachuer 1,2,5, Mona Diab-Assaf <sup>4</sup> , Sandra E. Ghayad<sup>4</sup> , Béatrice Fervers 1,2,3† , Vincent Cavaillès 9†, Véronique Maguer-Satta2† and Pascale A. Cohen1,2,3,5,8 \*

<sup>1</sup> Université Lyon 1, Lyon, France, <sup>2</sup> CRCL-Centre de Recherche en Cancérologie de Lyon-Inserm U1052-CNRS U5286, Lyon, France, <sup>3</sup> Département Cancer et Environnement, Centre Léon Bérard, Lyon, France, <sup>4</sup> Faculty of sciences II, Lebanese University, Fanar, Lebanon, <sup>5</sup> ProfileXpert, SFR-Est, CNRS UMR-S3453, INSERM US7, Lyon, France, <sup>6</sup> Department of Bioinformatics, Semmelweis University and TTK Lendület Cancer Biomarker Research Group, Budapest, Hungary, <sup>7</sup> UMR5569 (UM IRD CNRS), Montpellier, France, <sup>8</sup> INSERM, UMR1033 LYOS, Lyon, France, <sup>9</sup> IRCM - Institut de Recherche en Cancérologie de Montpellier, INSERM U1194, Université de Montpellier, Institut régional du Cancer de Montpellier, CNRS, Montpellier, France

It is of utmost importance to decipher the role of chronic exposure to low doses of environmental carcinogens on breast cancer progression. The early-transformed triple-negative human mammary MCF10AT1 cells were chronically (60 days) exposed to low doses (10−<sup>10</sup> M) of Benzo[a]pyrene (B[a]P), a genotoxic agent, and/or Bisphenol A (BPA), an endocrine disruptor. Our study revealed that exposed MCF10AT1 cells developed, in a time-dependent manner, an acquired phenotype characterized by an increase in cancerous properties (anchorage independent growth and stem-like phenotype). Co-exposure of MCF10AT1 cells to B[a]P and BPA led to a significantly greater aggressive phenotype compared to B[a]P or BPA alone. This study provided new insights into the existence of a functional interplay between the aryl hydrocarbon receptor (AhR) and the G protein-coupled receptor 30 (GPR30) by which chronic and low-dose exposure of B[a]P and/or BPA fosters the progression of MCF10AT1 cells into a more aggressive substage. Experiments using AhR or GPR30 antagonists, siRNA strategies, and RNAseq analysis led us to propose a model in which AhR signaling plays a "driver role" in the AhR/GPR30 cross-talk in mediating long-term and low-dose exposure of B[a]P and/or BPA. Retrospective analysis of two independent breast cancer cohorts revealed that the AhR/GPR30 mRNA expression signature resulted in poor breast cancer prognosis, in particular in the ER-negative and the triple-negative subtypes. Finally, the study identified targeting AhR and/or GPR30 with specific antagonists as a strategy capable of inhibiting carcinogenesis associated with chronic exposure to low doses of B[a]P and BPA in MCF10AT1 cells. Altogether, our results indicate that the engagement of both AhR and GPR30 functions, in particular in an ER-negative/triple-negative context of breast cells, favors tumor progression and leads to poor prognosis.

Keywords: environmental factors, Benzo[a]pyrene, Bisphenol-A, breast cancer, tumor progression, AhR, GPR30

#### INTRODUCTION

Progression of human breast epithelial cells from non-cancerous to pre-malignant and of early-transformed mammary cells to malignant stages is a multiyear, multistep, multiscale, and multipath disease process. More than 85% of breast cancers are sporadic and potentially attributable to long-term exposure to environmental factors, such as chemical carcinogens (1– 5). Given the increasing evidence that common environmental carcinogens play a significant role in breast cancer, increased attention has been paid to molecular mechanisms through which pollutants affect breast tumor formation, progression, and/or invasion (6–8). The identification of such molecular mechanisms could have several societal and environmental consequences, and may lead to the discovery of human biomarkers of exposure to environmental carcinogens exploitable for breast cancer prevention.

Previous in vitro investigations have mainly been conducted on human mammary epithelial cells or on human breast cancer cells, reflecting the impact of environmental factors on the earlier and later stages of carcinogenesis (9–13). However, little is known on the impact of exposure to pollutants on the breast earlytransformed stage. Short-term exposure of cells to carcinogens at micro- to millimolar concentrations was previously typically investigated (1, 2, 14–16) which, while informative, is not optimal in mimicking natural chronic exposure to low doses of environmental carcinogens and to reflect physiologicallyachievable levels of environmental mammary carcinogens. Additionally, few studies have attempted to mimic natural environmental exposure by assessing the impact of exposure to a combination of several pollutants with distinct mechanisms of action that may interact or induce a greater adverse effect than the use of individual compounds.

Benzo[a]pyrene (B[a]P), a family member of poly-cyclic aromatic hydrocarbons, is considered to be a tobacco, environmental, and dietary chemical carcinogen classified as Group 1 carcinogen by the IARC (17). B[a]P is a tumor initiator that binds and forms a complex with the aryl hydrocarbon receptor (AhR) (18–20). Upon such activation, the AhRtranscriptional complex activates specific DNA-recognition elements, such as xenobiotic response elements (XREs), and upregulates the expression of genes such as cytochrome P450 isoforms (including CYP1A1). These latter are involved in the metabolic activation of B[a]P in genotoxic metabolites forming DNA adducts relevant for carcinogenesis [for review, (21, 22)]. A growing body of evidence is accumulating implicating the B[a]P/AhR/CYP1A1 pathway in carcinogenesis (23–25). At early stages of carcinogenesis, short-term and millimolar B[a]P doses were shown to induce aggressiveness and transformation of non-cancerous human mammary epithelial cells (26–28). The impact of chronic and low-dose exposure of these cells to B[a]P was scarcely investigated, but seems to foster progression toward the early-transformed stage by favoring increased mesenchymal, stem-like, and anchorage-independent growth properties (9–13, 29).

Bisphenol-A (BPA) is a monomer of polycarbonate plastics and human exposure to BPA mainly occurs through the oral route due to the leaching of BPA in food and beverage containers, but non-dietary sources such as dust, air and cosmetics are also relevant (30). BPA has been the focus of widespread concern due to the fact that it interferes with endocrine signaling pathways even at extremely low doses, and thus belongs to the endocrinedisrupting compounds (EDC) (31, 32). BPA is known to bind to estrogen receptors alpha and beta (ERα and ERβ), to the G protein-coupled receptor 30 (GPR30) but also to the pregnane X receptor (PXR) (31, 33). Although several in vivo studies reported a carcinogenic potential of BPA [reviewed in (32)], the World Health Organization (WHO) indicated that there is currently insufficient evidence on which to base this carcinogenic potential (34). In vitro studies have however revealed that BPA causes adverse effects in non-cancerous mammary epithelial cells or in breast cancer cell lines, including increased cell proliferation, cell stemness, oxidative stress, and alterations of cell signaling pathways involved in carcinogenesis (13, 29, 35–38).

The MCF10 unique model of breast cancer progression comprises a series of isogenic triple-negative cell lines derived from MCF10A cells (MCF10A, MCF10AT1 and MCF10CA1a.cl1 cells). The parental cell line (MCF10A) having been originally isolated from a woman with fibrocystic change (39), the members of the MCF10 series belong to the triple negative/basallike subtype (ER-negative, progesterone receptor (PR)-negative, HER2-negative) (40–42). These cell lines thus recapitulate the stages of mammary carcinogenesis (43), making this a valuable in vitro model for studying the progression of triple-negative breast cancer (44–46). In the present study, we used MCF10AT1 breast cells which represent the transformed early stage in the MCF10 unique model of breast cancer progression (43, 44) to further characterize the carcinogenic potential of B[a]P and BPA. To our knowledge, these cells have never been used to test the impact of chronic and low-dose exposure to environmental pollutants.

**Abbreviations:** B[a]P, benzo[a]pyrene; AhR, aryl hydrocarbon receptor; XRE, xenobiotic response element; BPA, bisphenol-A; EDC, endocrine-disrupting compound; ERα, estrogen receptor alpha; ERβ, estrogen receptor beta; GPR30, G protein-coupled receptor 30; PXR, pregnane X receptor; PR, progesterone receptor; AIG, anchorage-independent growth; MFE, mammosphere formation efficiency; RT-qPCR, real-time quantitative polymerase chain reaction; TCDD, 2,3,7,8-Tetrachlorodibenzo-p-dioxin; OS, overall survival.

The main objectives of this work were to newly investigate: (i) whether long-term and low-dose exposure to B[a]P and/or BPA triggers the progression of early-transformed mammary cells to a more aggressive stage; (ii) whether their combination enhances the effect of each compound tested individually, in particular whether BPA facilitates the pro-carcinogenic activity of B[a]P; and (iii) to identify candidate strategies capable of inhibiting mammary carcinogenesis linked to chronic exposure to the environmental pollutants B[a]P and/or BPA.

Our data reveal that long-term and low-dose exposure to B[a]P and BPA increases cancerous properties of the MCF10AT1 cell line. Importantly exposure to the two pollutants leads to a greater deleterious impact than the compounds tested individually, and our data highlight the existence of a unique functional cross-talk between AhR and GPR30 in mediating those effects. The clinical relevance of the AhR/GPR30 interplay is validated by the observation of high mRNA expression levels of these two receptors in breast cancer patients as markers of poor prognosis. Finally, this study identified AhR and GPR30 as novel targets for strategies inhibiting the development of cancerassociated properties (AIG and MFE) in early-transformed human mammary cells following long-term and low-dose exposure to B[a]P and BPA.

# MATERIALS AND METHODS

#### Cell Culture

Early-transformed human mammary MCF10AT1 cells and the MCF10AT1-derived cancerous MCF10CA1a.cl1 cells (Karmanos Institute, Detroit, USA) were purchased from Karmanos Institute (Detroit, USA) and maintained in complete DMEM/Ham's F12 medium with 5% horse serum (Thermo Fisher Scientific, Waltham, USA) and additional supplements: 100 ng/mL cholera enterotoxin, 10 mg/mL insulin, 0.5 mg/mL hydrocortisol, 20 ng/mL epidermal growth factor (Sigma, Saint Louis, USA) 100 units/mL penicillin and 100 mg/mL streptomycin. MCF-7 cells were purchased from ATCC (Teddington, UK) and grown according to the manufacturer's recommendations. HG5LN PXR cells stably expressing PXR (47), were grown in DMEM containing phenol red and 1 g/L glucose with 5% fetal calf serum (FCS) and additional supplements: 1 mg/mL G418 and 0.5 µg/mL puromycin.

#### Reagents

B[a]P and BPA were purchased from Sigma (Saint Louis, USA). 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD), used as a control, was purchased from LGC Standard (Molsheim, France). The GPR30 agonist G1, the GPR30 antagonist G15, and the AhR agonist ITE were from TOCRIS Bioscience (Bristol, UK); the AhR antagonist GNF351 from Calbiochem (Billerica, USA).

#### Establishment and Maintenance of the Chronically Exposed Cellular Model

MCF10AT1 cells were chronically exposed or not to 10−<sup>10</sup> M of B[a]P or to 10−<sup>10</sup> M BPA, alone or in combination, during 60 days (≈20 passages) in phenol red-free DMEM/Ham's F12 medium with 5% steroid-depleted, dextran-coated and charcoal-treated horse serum, containing the above-mentioned supplements (further referred to as DCC medium). Unexposed MCF10CA1a.cl1 cells were grown concomitantly in the same medium for 60 days and named MCF10CA1a.cl160d. Media and treatments were changed every 2 days. Cells were frozen every 2 weeks.

#### Anchorage-Independent Growth (AIG)

Anchorage-independent growth was assessed by soft agar assay as previously described (48). Single-cell suspensions (75 × 10 3 ) were seeded onto soft agar, and colonies were counted after 3 weeks of incubation.

#### Mammosphere Formation Efficiency (MFE)

Single-cell suspensions were seeded using non-adherent mammosphere culture conditions (49). After 7 days, primary mammospheres (first generation) were counted, collected, trypsinized, and replated for 10 days in non-adherent culture conditions to generate second-generation mammospheres. The culture media were replenished every 2–3 days.

#### RNA Extraction and Real-Time Quantitative Polymerase Chain Reaction (RT-qPCR)

Total RNA extraction, reverse transcription and RT-qPCR measurements were performed as described previously (29, 48). RNA lysates were extracted after exposure to the different molecules tested, untreated cells were used as controls in the presence of the corresponding volume of solvent. One microgram of total RNA from each sample was reverse-transcribed as previously described (48). RTqPCR measurements were performed using a CFX96 with the SsoAdvanced Universal SYBR green supermix (BioRad, Hercules, USA), according to the manufacturer's recommendations. The primers used to explore the expression of the ERα, ERβ, PXR, AhR, GPR30, CYP1A1, and 28S genes are listed in the **Supplementary Table 1**.

#### Western Blot

Western blot experiments were performed as previously described (48) using the following antibodies: AhR (1:1,000, ab2770; Abcam, Paris, France), GPR30 (1:1,000, NBP1-31239; Novus Biologicals, Littleton, USA), α-tubulin (1:10,000, T5168; Sigma), phospho-p42/44 MAPK (1:1,000, 9106; Cell Signaling), p42/44 MAPK (1:1,000, 9102; Cell Signaling, Danvers, USA).

#### GPR30 or AhR Silencing

StealthTM siRNAs siRNA-GPR30 and StealthTM siRNAs siRNA-AhR and their corresponding scrambled control RNA (scrambled) were obtained from Ambion (Carlsbad, USA,4390825) and Invitrogen (Carlsbad, USA, AHRHSS100337/336), respectively. Fifty nM of siRNA-GPR30, 5 nM of siRNA-AhR or corresponding scrambled RNA were transfected into MCF10AT1 cells with lipofectamine RNAimax (Invitrogen). Transfections were performed directly at the time of cell seeding. Western blots were performed 48 h posttransfection. Exposure to G1, TCDD, or ITE for RNA collection or luciferase assay was performed 24 h post-transfection.

#### Luciferase Assay

Cells were plated and then transfected with 150 ng XREfirefly luciferase reporter plasmid (XRE-luc) (50) and 10 ng Renilla luciferase plasmid (pTK-RL). Twenty-four hours after transfection cells were grown for 8 h in the presence of the indicated treatment, and luciferase activity was then assessed as previously described (48).

#### Short-Term Exposure Experiments

In AIG experiments, MCF10AT1 cells were exposed to BPA and/or B[a]P 10−<sup>10</sup> M, ITE 10−<sup>10</sup> M, or G1 10−<sup>10</sup> M for 72 h in the presence or the absence of a 2 h pre-treatment with the AhR antagonist GNF351 10−<sup>7</sup> M or the GPR30 antagonist G15 10−<sup>8</sup> M. Exposure was maintained throughout the course of the experiments. In MFE assays, MCF10AT1 cells were exposed to BPA and/or B[a]P 10−<sup>10</sup> M, ITE 10−<sup>10</sup> M, or G1 10−<sup>10</sup> M in the presence or the absence of GNF351 10−<sup>7</sup> M or G15 10−<sup>8</sup> M only during the time-course of the experiments.

#### RNA-Seq Experiments and Analyses

RNA isolation, library preparation, and RNA-Seq were performed by the core facility ProfileXpert (Lyon, France) from three independent cell-culture replicates of each tested cell line (unexposed MCF10AT160d cells, B[a]P 10−<sup>10</sup> M exposed MCF10AT160d, BPA 10−<sup>10</sup> M exposed MCF10AT160d cells, B[a]P+BPA 10−<sup>10</sup> M exposed MCF10AT160d). The resulting RNA were isolated using the RNeasy mini kit (Qiagen) according to the manufacturer's protocol and ribosomal depletion was performed with the Ribo-zero gold kit (Epicentre). Libraries were performed from 20 ng ribosomal depleted RNA with the NEXTFLEX <sup>R</sup> Rapid Directional RNA-Seq Library Prep Kit (BIOO-Scientific). Libraries were sequenced using an Illumina NextSeq 500 platform (flow cell highoutput V2) and a 75 bp paired-end sequencing with ∼30–35 million reads per sample. After trimming, reads were aligned to the human genome (hg19) using TopHat-2 v. 2.1.0 and data normalization (FPKM) was performed with Cufflinks software v.2.1.1. Data were logged on the NCBI Gene Expression Omnibus (GEO) website (http:// www.ncbi.nlm.nih.gov/geo/) and are available as a GSE142073 dataset. Transcripts were considered as differentially expressed when the p-value of a student non-parametric t-test was ≤ 0.05. The Aryl Hydrocarbon Receptor Signaling Canonical Pathway was evaluated with a functional analysis created with Ingenuity Pathway Analysis software (IPA <sup>R</sup> , QIAGEN Redwood City, www.qiagen.com/ingenuity). The GPR30 gene expression signature described by Pandey and collaborators (51) was introduced in the Ingenuity Pathway Analysis software to assess the GPR30 signaling pathway.

#### Cell Proliferation

A total of 4 × 10<sup>4</sup> cells/well were seeded onto and cultured in 24 well plates. Proliferating cells were analyzed using the ScepterTM 2.0 Cell Counter (Merck Millipore, Billerica, USA).

#### Cell Viability Assay

A total of 10<sup>4</sup> cells/well were plated onto a 96-well plate and treated for 4 days as indicated. Cell viability was assessed as previously described (52).

#### Breast Tumor Cohorts

Women with primary breast tumors (n = 113) and known clinical follow-up who had not received any therapy before surgery and who relapsed, or not, while receiving endocrine therapy and/or chemotherapy were recruited from the BB-0033-00050, Biological Resources Center (CRB) Centre Léon Bérard, Lyon France (CLB cohort, **Supplementary Table 2**) (53). This study has been approved by the local ethics committee (CRB Centre Léon Bérard, France). The CRB Centre Léon Bérard is quality certified according NFS96-900 French standard and, ISO 9001 for clinical trials, ensuring scientific rigor for sample conservation, traceability and quality, as well as ethical rules observance and defined rules for transferring samples for research purposes (Ministry of Health for activities authorization n ◦ AC-2019-3426 and DC-2008-99). The material used in the study has been collected in agreement with all applicable laws, rules, and requests of French and European government authorities, including the patients' informed written consents. Extraction of total RNA from frozen tumor samples and RTqPCR measurements were performed as previously described (52, 53). Univariate analyses were performed using the SPSSTM Software (IBM, USA). The IBM SPSS software (IBM) was used for all statistical analyses in which the prognostic value of AhR and GPR30 mRNA levels was analyzed. The data were divided at the median value of AhR or GPR30 mRNA expression into two groups with either high or low expression levels. The Kaplan-Meier plotter (KMP) cohort was established from a meta-analysis of the gene-expression profiles of 1,877 primary breast cancer samples from patients who had not received any therapy prior to surgery (54). A p < 0.05 was considered statistically significant.

#### Long-Term Inhibitory Strategies

MCF10AT1 cells were chronically (60 days) exposed or not to (B[a]P + BPA) 10−<sup>10</sup> M, alone or in combination with GNF351 10−<sup>7</sup> M and/or G15 10−<sup>8</sup> M. Control experiments were performed in MCF10AT1 cells exposed for 60 days to GNF351 10−<sup>7</sup> M and/or G15 10−<sup>8</sup> M. The resulting established cells were then tested for AIG and MFE as described above.

# RESULTS

#### Chronic and Low-Dose Exposure to B[a]P and/or BPA of Early-Transformed MCF10AT1 Cells Leads to an Enhanced and Acquired Aggressive Phenotype

The MCF10AT1 cells and the MCF10AT1-derived MCF10CA1a.cl1 respectively represent the early-transformed and cancerous stages in the unique MCF10 model of triple negative breast cancer progression (43, 44). Validation of progression to malignancy of MCF10AT1 cells was investigated by assessing: (i) anchorage-independent growth (AIG), a hallmark of carcinogenesis associated with aggressiveness and metastasis in malignant cells; (ii) cancer stem-like and selfrenewing properties by assessing first and second generation mammosphere-forming efficiency (MFE), as a growing body of evidence suggests that cancer stem-like cells are involved in generating and maintaining pre-malignant and malignant lesions (55–57). Consistent with the substage of breast cancer progression displayed by each cell line, MCF10AT1 cells formed significantly fewer and smaller colonies in soft agar (AIG) and significantly fewer mammospheres than the cancerous MCF10CA1a.cl1 cells (**Supplementary Figures 1A,B**).

In order to identify the impact of B[a]P and BPA on progression to malignancy in conditions mimicking environmental exposure, the MCF10AT1 cells were chronically (60 days) exposed or not to physiologically-relevant concentrations (10−<sup>10</sup> M) of B[a]P and/or BPA (58–62) (MCF10AT160d). Unexposed MCF10AT160d and MCF10CA1a.cl160d cells displayed unmodified AIG and MFE phenotypes (**Figures 1A,B** and **Supplementary Figure 1C**) compared to the corresponding parental cells (**Supplementary Figures 1A,B**). MCF10AT160d cells exposed to the carcinogenic B[a]P pollutant at a concentration as low as 10−<sup>10</sup> M exhibited significantly increased AIG (**Figure 1A**), increased colony size (**Supplementary Figure 1C**) and higher MFE (**Figure 1B**) compared to the unexposed MCF10AT160d cells. The EDC BPA (10−<sup>10</sup> M, 60 days exposure) also enhanced the cancerous properties of the exposed MCF10AT160d cells (significant increase in AIG, in colony size and in MFE), while that impact was always significantly lower than that of 60-days exposure to 10−<sup>10</sup> M B[a]P (**Figures 1A,B**, **Supplementary Figure 1C**). Of utmost interest, the combination of B[a]P with BPA (10−<sup>10</sup> M) had a significantly greater effect than B[a]P (10−<sup>10</sup> M) tested individually (**Figures 1A,B**, **Supplementary Figure 1C**). The cancerous features displayed by the MCF10AT160d cells exposed to the B[a]P + BPA (10−<sup>10</sup> M) combination was at least equivalent (**Figure 1A** and **Supplementary Figure 1C**) if not greater (second generation of mammospheres, **Figure 1B**) than those displayed by the unexposed cancerous MCF10CA1a.cl160d cells. **Supplementary Figure 2** eliminated the possibility that the aggressive phenotypes observed in the different exposed MCF10AT160d cells were the consequence of an increase in cell proliferation (compared to unexposed MF10AT160d).

Unexposed or exposed MCF10AT160d cells were grown for a further 30 days in the absence of any treatment (MCF10AT160+30d cells), and then tested for AIG and MFE. The number and size of colonies in soft agar (**Figure 1C** and **Supplementary Figure 1D**) and the number of mammospheres (**Figure 1D**) were similar between MCF10AT160d and MCF10AT160+30d cells, thus demonstrating that the aggressive phenotype induced by chronic exposure of MCF10AT1 cells to low-doses of B[a]P and/or BPA is an acquired phenotype.

Taking advantage of the systematic freezing of exposed cells over the experimental time-course, AIG and MFE were retrospectively tested after 23 and 48 days of exposure and compared to 60 days of exposure (**Figures 1E,F**). The number of colonies in soft agar and mammospheres significantly increased between 23–48 and 48–60 days of chronic exposure, thus demonstrating that the longer MCF10AT1 cells are exposed to low doses of B[a]P and/or BPA, the more aggressive is their resulting phenotype. Once again, the combined exposure to B[a]P + BPA (10−<sup>10</sup> M) caused more deleterious effects than those of B[a]P or BPA individually, irrespective of the exposure time.

Collectively, our results suggest that: (i) chronic, low-dose exposure to B[a]P or BPA enhances the cancerous properties of MCF10AT1 cells, with exposure to B[a]P being the most effective; (ii) the resulting aggressive phenotype was acquired and not reversed or softened when exposure was stopped; (iii) the duration of the exposure to pollutants impacted the magnitude of the aggressiveness developed by the exposed cells; (iv) combining BPA to B[a]P had more impact than exposure to B[a]P alone and was at least equivalent or higher than that of cancerous MCF10Ca1a.cl1 cells, supporting progression toward the cancerous substage.

#### AhR and GPR30 are Both Expressed and Functional in MCF10AT1 Cells

As AhR is the main target of B[a]P and BPA is known to bind to ERα, ERβ, GPR30, and PXR, we investigated the presence of these receptors in MCF10AT1 cells. **Supplementary Figure 3** validated that ERα, ERβ, and PXR were not or scarcely expressed in MCF10AT1 and MCF10CA1a.cl1 cells. Conversely, we newly reported that AhR and GPR30 receptors are concomitantly expressed at the mRNA and protein levels in MCF10AT1 and MCF10CA1.cl1 cells (**Figures 2A,B**). Of interest, AhR expression levels are higher both at the mRNA and protein levels in the MCF10AT1 cells compared to MCF10ACA1a.cl1.

Gene reporter experiments performed in MCF10AT1 cells demonstrated that TCDD, a well-known AhR ligand and activator of AhR-direct transcriptional activity (50), led to an increase in XRE-luciferase activity (**Figure 2C**). The selective AhR agonist ITE (63) also led to a stronger and dose-dependent activation of AhR-driven transcriptional activity in MCF10AT1 cells (**Figure 2C**). A concentration as low as 10−<sup>11</sup> M of ITE was sufficient to give rise to a significant increase in XRE-luc activity (**Figure 2C**) and the effect of ITE was prevented (**Figure 2D**) by a 100-fold excess of the AhR antagonist GNF351 (64).

Rapid phosphorylation of MAPK is one of the main downstream effects of GPR30 activation (65). In MCF10AT1 cells, the MAPK pathway is rapidly activated (increase in the phospho-p42/p44 MAPK/MAPK ratio) in the presence of the selective GPR30 agonist G1 (66) and this activation is impaired in the presence of the GPR30 antagonist G15 (67) (**Figure 2E**), newly revealing the presence of functional GPR30 in MCF10AT1 cells.

#### The MFE and AIG Responses Triggered by B[a]P and/or BPA Occur Through a Functional Cross-Talk Between AhR and GPR30 in MCF10AT1 Cells

We then aimed at investigating the possible involvement of the AhR and/or GPR30 receptors in mediating the

FIGURE 1 | Cumulative and low-dose exposure of BPA and/or B[a]P increases MCF10AT1 cancerous properties (AIG and MFE assays). (A) Average number of MCF10AT160d colonies in soft agar after 60 days of chronic exposure to B[a]P and/or BPA (10−<sup>10</sup> M). (B) Average number of primary and secondary MCF10AT160d (Continued) FIGURE 1 | mammospheres formed after 60 days of chronic exposure to B[a]P and/or BPA (10−<sup>10</sup> M). In (A,B), exposure was maintained throughout the course of the experiments. (C,D) After 60 days of chronic exposure to B[a]P and/or BPA (10−<sup>10</sup> M), MCF10AT1 cells were grown for a further 30 days without any exposure, and single-cell suspensions of cells were seeded in soft agar (C) or in non-adherent mammosphere culturing conditions (D). (E,F) Time-dependent acquisition of the aggressive phenotype: average number of colonies in soft agar (E) and of secondary mammospheres (F) following 23, 48, or 60 days of chronic exposure to B[a]P and/or BPA (10−<sup>10</sup> M). Data illustrated in (A–F) represent mean ± SD of at least three independent experiments, in triplicate. \*\*\*p < 0.001, \*\*p < 0.01, \*p < 0.05 or NS (not significant) in Student t-test.

tumorigenic effects of exposure to B[a]P and/or BPA 10−<sup>10</sup> M in short-term experiments. This was tested in the presence or absence of selective antagonists of AhR or GPR30 (GNF351 and G15, respectively) at non-toxic concentrations (**Supplementary Figures 4A,B**).

**Figure 3A** demonstrates that short-term (72 h) exposure to 10−<sup>10</sup> M BPA and/or 10−<sup>10</sup> M B[a]P were sufficient to generate increased colony numbers in AIG assays. GNF351 and G15 were able to completely inhibit the effects of B[a]P and/or BPA on AIG (**Figure 3A**). This result was, at least to some extent, expected with the B[a]P/GNF351 and the BPA/G15 combinations, but was totally surprising concerning the BPA/GNF351 and the B[a]P/G15 combinations. In MFE assays conducted with MCF10AT1 cells exposed only during the time-course of the experiments (**Figure 3B**), the increased number of mammospheres induced by 10−<sup>10</sup> M BPA was totally blocked in the presence of G15, as expected, but also in the presence of GNF351. Conversely, B[a]Pinduced MFE was completely inhibited by GNF351, while the GPR30 antagonist G15, surprisingly resulted in a partial but significant inhibition. The impact of the G15 or GNF351 on the combination B[a]P+BPA was similar to that observed with B[a]P alone.

We then performed the same experiments using G1 and ITE agonists, at 10−<sup>10</sup> M non-toxic concentrations (**Supplementary Figures 4C,D**). **Figures 3C,D** show similar data as those obtained with B[a]P and/or BPA in **Figures 3A,B**, namely that: (i) low-dose exposure of MCF10AT1 cells to 10−<sup>10</sup> M G1 or 10−<sup>10</sup> M ITE gives rise to significantly increased colony number (AIG) or increased MFE in the same range as that observed with 10−<sup>10</sup> M BPA or 10−<sup>10</sup> M B[a]P, respectively; (ii) G15 and GNF351 were both able to totally inhibit G1 dependent AIG or MFE; (iii) the impact of ITE on MFE was totally blocked in the presence of GNF351, while only partially inhibited by G15.

To confirm the above data supporting the involvement of both GPR30 and AhR receptors in mediating the effects of low dose (10−<sup>10</sup> M) BPA, B[a]P, G1, or ITE, we used a siRNA strategy silencing either AhR or GPR30 (**Figures 4A,B**). We verified that siRNA-AhR or siRNA-GPR30 had, however, no impact on GPR30 or AhR expression, respectively (**Figures 4A,B**), nor on MCF10AT1 cell viability (**Supplementary Figures 4E,F**).

Findings from AIG and MFE experiments conducted in MCF10AT1 exposed to BPA, B[a]P, G1, or ITE 10−<sup>10</sup> M in the presence of siRNA-AhR or of siRNA-GPR30 (**Figures 4C–F**) corroborated those obtained using the AhR- or GPR30 antagonists (GNF351 and G15, respectively) (**Figures 3A–D**). Indeed, AhR knock-out resulted in a total inhibition of BPA-, B[a]P-, G1-, or ITE-mediated effects (**Figures 4C,E**). GPR30 silencing, while totally inhibiting BPA- or G1-mediated effects (**Figures 4D,F**), was slightly less effective in inhibiting MFE or AIG assays when cells were exposed to B[a]P (**Figures 4D,F**) or ITE (**Figure 4D**). The impact of siRNA-AhR or siRNA-GPR30 on the combination B[a]P+BPA was similar to that observed with B[a]P alone (**Figures 4C–F**).

Altogether these results support the idea that both AhR and GPR30 receptors play a role in mediating the deleterious effects exerted by BPA, G1, B[a]P, or ITE on the MCF10AT1 early-transformed cells. More importantly, our data highlight a functional cross-talk between GPR30 and AhR.

#### GPR30-Dependent Mechanisms are Correlated With Enhanced AhR Transcriptional Activity

To further decipher the interplay between AhR and GPR30, we performed XRE-luciferase reporter assays. As anticipated, AhRdriven transcriptional activity in MCF10AT1 cells was fostered by TCDD 10−<sup>7</sup> M or ITE 10−<sup>10</sup> M and lost following AhR silencing (**Figure 5A**). Very interestingly, the GPR30 agonist G1 was also able to significantly increase XRE-luciferase activity in a dose-dependent manner, as illustrated by its effect at 10−<sup>7</sup> M and 10−<sup>6</sup> M, which is, to our knowledge, the first such report in the literature (**Figure 5B**). The G1-induced XREluciferase signal was totally inhibited in the presence of GNF351 (**Figure 5C**) and also in the presence of siRNA-AhR (**Figure 5D**). Silencing GPR30 by a siRNA-GPR30 strategy gave rise to a partial but significant decrease in G1- (10−<sup>6</sup> M) (**Figure 5E**), TCDD- (10−<sup>7</sup> M), and ITE- (10−<sup>10</sup> M) (**Figure 5F**) -dependent activation of XRE-luciferase activity. Altogether, these data recapitulate the cross-talk between AhR and GPR30 on a simplified XREluciferase response.

As CYP1A1 is amongst the canonical genes regulated by AhR with a promoter containing typical XRE, we further investigated endogenous CYP1A1 mRNA levels in MCF10AT1 cells upon ITE and G1 treatment. Consistent with **Figures 5**, **6** revealed that: (i) CYP1A1 mRNA levels significantly increased upon treatment with TCDD 10−<sup>7</sup> M and ITE 10−<sup>7</sup> M (**Figure 6A**); (ii) the impact of ITE 10−<sup>7</sup> M was completely impaired in the presence of siRNA-AhR (**Figure 6A**) or GNF351 10−<sup>6</sup> M (**Figure 6B**), and significantly decreased in the presence of siRNA-GPR30 (**Figure 6C**). **Figures 6D–F** highlight that G1 10−<sup>6</sup> M resulted in a significant increase in CYP1A1 mRNA levels, and that this was totally impeded in the presence of siRNA-AhR (**Figure 6D**), GNF351 10−<sup>6</sup> M (**Figure 6E**) and significantly decreased in the presence of siRNA-GPR30 (**Figure 6F**). Altogether, our XRE-luciferase experiments and RT-qPCR experiments confirmed, at the transcriptional

FIGURE 2 | independent experiments conducted in triplicate. (B) Representative Western blot analyses from three independent experiments of AhR and GPR30 protein expression in MCF10AT1 and MCF10CA1a.cl1 cells. MCF-7 cells were used as a control. (C) XRE-luciferase activity following 8 h exposure of MCF10AT1 cells to ITE at the indicated concentrations. TCDD 10−<sup>7</sup> M was used as a control and results were expressed as % of TCDD 10−<sup>7</sup> M activity. \*\*\*p < 0.001 in Student t-test. (D) XRE-luciferase activity upon 8 h of exposure to ITE 10−<sup>10</sup> M alone or in combination with GNF351 at the indicated concentrations. TCDD 10−<sup>7</sup> M was used as a control, and results were expressed as % of TCDD 10−<sup>7</sup> M activity. Student t-tests revealed the statistically significant differences between unexposed and exposed cells: \*\*\*p < 0.001; and between ITE and ITE+GNF351: ###p < 0.001. Values in (C,D) represent mean ± SD of three independent experiments. (E) Representative Western blot analyses from three independent experiments of the phospho-MAPK/MAPK ratio upon exposure of MCF10AT1 cells to G1 (GPR30 agonist) for the times indicated, in the presence or absence of a 2 h pre-treatment with G15 (GPR30 antagonist).

FIGURE 3 | Effects of short-term exposure of MCF10AT1 cells by BPA, B[a]P, ITE, and G1 on AIG and MFE are inhibited by GPR30 and AhR antagonists. Average number of colonies in soft agar (A) or of secondary mammospheres (B) upon exposure to B[a]P and/or BPA 10−<sup>10</sup> M for 72 h in the presence or absence of a 2 h pre-treatment with GNF351 10−<sup>7</sup> M or G15 10−<sup>8</sup> M. Exposure was maintained throughout the course of the experiments. Unexposed MCF10CA1a.cl1 cells were used as a control. (C) Average number of colonies in soft agar or (D) secondary mammospheres formed after 72 h exposure with ITE 10−<sup>10</sup> M or G1 10−<sup>10</sup> M in the presence or the absence of a 2 h pre-treatment with GNF351 10−<sup>7</sup> M or G15 10−<sup>8</sup> M. Exposure to BPA 10−<sup>10</sup> M or B[a]P 10−<sup>10</sup> M was used as controls. Exposure was maintained throughout the course of the experiments. Data illustrated in (A–D) represent mean ± SD of at least three independent experiments, in triplicate. \*\*\*p < 0.001, \*\*p < 0.01 in Student t-test.

level, that GPR30-dependent mechanisms significantly impact and favor, directly or indirectly, AhR-driven transcriptionaldependent events.

Having verified that protein expression levels of AhR and of GPR30 were similar in both the unexposed and the pollutantsexposed MCF10AT160d cells (**Supplementary Figure 5A**), the

Western blot analysis from three independent experiments of AhR and GPR30 expression in transfected MCF10AT1 cells with (A) siRNA-AhR, (B) siRNA-GPR30 or their scrambled controls. Quantification of protein expression levels was normalized against tubulin expression. (C,D) Secondary mammospheres formation and (E,F) average number of colonies in soft agar, with the following treatments: BPA and/or B[a]P, G1, or ITE, 10−<sup>10</sup> M. Cells were transfected with either siRNA-AhR, siRNA-GPR30 or their scrambled controls before being subjected to the treatments. Treatments were maintained throughout the course of experiments. (mean ± SD of 2 independent experiments, in triplicate). \*\*\*p < 0.001, \*p < 0.05 vs. their respective unexposed; ###p < 0.001 siRNA vs. scrambled in Student t-test.

RNAseq data of each cell line were analyzed using the Ingenuity Pathway Analysis software. AhR core signaling was identified as significantly dysregulated in the MCF10AT160d exposed cells compared to unexposed control MCF10AT160d cells. Indeed, the significant enrichment of this canonical pathway was observed both in the B[a]P-exposed MCF10AT160d cells (p = 9.39 10−<sup>3</sup> ), in the B[a]P + BPA exposed MCF10AT160d cells (p = 7.56 10−<sup>3</sup> ), but also in the BPA-exposed MCF10AT160d cells (p = 1.26 10−<sup>2</sup> ) (data not shown). Conversely, assessing the GPR30 signature previously described by Pandey and

as a control and results are expressed as % of TCDD 10−<sup>7</sup> M activity. (C) CYP1A1 mRNA expression levels in cells transfected with either siRNA-GPR30 or scrambled RNA and exposed to ITE 10−<sup>7</sup> M for 4 h. TCDD 10−<sup>7</sup> M was used as a control and results are expressed as % of TCDD 10−<sup>7</sup> M activity. (D) CYP1A1 mRNA expression levels in cells transfected with either siRNA-AhR or scrambled RNA and exposed to G1 10−<sup>6</sup> M or TCDD 10−<sup>7</sup> M for 24 h. TCDD 10−<sup>7</sup> M was used as a control. (E) CYP1A1 mRNA expression levels in cells treated with G1 10−<sup>6</sup> M in the presence or absence of GNF351 10−<sup>6</sup> M. (F) CYP1A1 mRNA expression levels in cells transfected with either siRNA-GPR30 or scrambled RNA and exposed to G1 10−<sup>6</sup> M for 24 h. Data represent mean ± SD of 3 independent experiments conducted in triplicate. (A–F) The Student t-test was applied to reveal statistically significant differences between treatments: \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001.

collaborators (51) in the Ingenuity Pathway Analysis software did not reveal any dysregulation in GPR30 signaling when the exposed MCF10AT160d cells were compared with the unexposed control MCF10AT160d cells (data not shown). As rapid phosphorylation of MAPK is one of the main downstream effects of GPR30 activation (65), we assessed the activation status of the MAPK pathway in the unexposed and exposed MCF10AT160d cells. **Supplementary Figure 5B** demonstrated that chronic and low-dose-exposure of MCF10AT1 cells to B[a]P and/or BPA at 10−<sup>10</sup> M did not lead to any activation of p42/p44MAPK (MCF1Ca1.cl160d cells were used as controls). Altogether, our data suggest that AhR signaling is constitutively


TABLE 1 | Univariate analysis of the GPR30 mRNA expression levels, the AhR mRNA expression levels and the GPR30/AhR mRNA expression signature with regards to overall survival (OS) in different subclasses of the 113 breast cancer samples of the CLB cohort.

<sup>a</sup>HR, Hazard ratio.

<sup>b</sup>95% CI, 95% confidence interval.

<sup>c</sup>p was considered significant when p < 0.05 (bold values). NS, not significant.

d subclasses of breast cancer were determined using immunohistology (ER, PR, HER2) according to the St Gallen recommendation (68).

<sup>e</sup>N/A, not applicable as all the cases are censored in the high GPR30 mRNA level group.

<sup>f</sup> N/A, not applicable as all the cases are censored in the high GPR30 and/or AhR mRNA level group.

activated in the MCF10AT160d exposed to pollutants vs. unexposed cells.

#### The GPR30/AhR Gene Expression Signature Indicates Poor Prognosis

To investigate the clinical relevance of AhR and GPR30, we performed RT-qPCR analyses to explore GPR30 and AhR mRNA expression levels in a cohort of 113 human primary breast tumor samples (**Table 1**) (53). The resulting Kaplan-Meier curves are shown in **Figure 7**. By univariate analysis, we found that neither GPR30 nor AhR mRNA levels were informative (p = 0.09 and p = 0.41, **Figures 7A,B**, respectively), while an AhR/GPR30 gene expression signature based on high expression levels of both GPR30 and AhR was significantly associated with shorter overall survival (OS) (p = 0.01, **Table 1** and **Figure 7C**). Regarding breast cancer subclasses, the AhR/GPR30 gene expression signature was more informative in the ER-negative (p = 0.001) than in the ER-positive (p = 0.41) (**Table 1**, **Figures 7D,E**) or luminal subclasses and (p = 0.50) (**Table 1**). The situation was less clear for the HER2-enriched and triple-negative subclasses, considering the limited size of available samples (n =16 and n = 21, respectively). We thus performed retrospective analysis of gene-expression array data using the KMP cohort, which contains a sizeable number of breast cancer patients (1,308 ERpositive, 569 ER-negative, 1,055 luminal, 419 HER2-enriched, and 403 triple-negative/basal-like). The most striking results (**Table 2**, univariate analysis) indicated that the AhR/GPR30 signature was again more informative than GPR30 mRNA or AhR mRNA levels alone and was associated with shorter OS in the ER-negative subclass (p = 0.005), but not in the ERpositive or the luminal subclasses (**Table 2**). The "all breast tumor samples" univariate analysis of the KMP cohort did not validate what was observed in the CLB cohort, but this discrepancy might reflect the difference in proportion of the ER-positive and ER-negative subclasses in the two cohorts. Finally, the KMP cohort revealed that the AhR/GPR30 signature was associated with shorter OS in the triple-negative subclass (p = 0.033). Altogether, our data reveal a new/original signature based on a combination of high AhR mRNA expression levels and high GPR30 mRNA expression levels that represents a novel marker for poor prognosis in breast cancer, especially in ER-negative or triple-negative subclasses.

#### Strategies Inhibiting the Impact of Chronic and Low-Dose Exposure to B[a]P and BPA in Early-Transformed MCF10AT1 Cells

To identify candidate strategies capable of blocking mammary carcinogenesis associated with chronic exposure to low doses (10−10M) of the environmental pollutants B[a]P and BPA, MCF10AT1 cells were exposed for 60 days, in the presence or the absence of the AhR antagonist GNF351 10−<sup>7</sup> M and/or the GPR30 antagonist G15 10−<sup>8</sup> M (**Figures 8A,B**). Exposure for 60 days to the two antagonists, alone or in combination, had no impact on MFE and AIG. Strikingly, co-exposure for 60 days to B[a]P + BPA 10−<sup>10</sup> M with GNF351 10−<sup>7</sup> M and/or G15 10−<sup>8</sup> M was sufficient to inhibit the development of pollutantsdriven enhancement of cancerous properties (AIG and MFE) in MCF10AT1 cells (**Figures 8A,B**).

#### DISCUSSION

A growing body of in vitro and in vivo experimental evidence suggests the implication of environmental factors in the development and progression of breast cancer. People are chronically exposed to a mixture of environmental factors, usually present at low-dose concentrations, constituting a complex exposome. B[a]P is detected at picomolar concentrations in body fluids and tissues of cancer patients (58–60). BPA is detected at nanomolar concentrations in human samples such as serum, urine and maternal milk (61, 62). The major obstacles in studying the impact of these two pollutants in vitro on breast tumorigenesis are thus the choice

mRNA expression signature in the ER-negative subclass. NS, not significant.


TABLE 2 | Univariate analysis of the GPR30 mRNA expression levels, the AhR mRNA expression levels and the GPR30/AhR mRNA expression signature with regards to overall survival (OS) in different subclasses of the 1,877 breast cancer samples of the Kaplan-Meier plotter (KMP) cohort.

<sup>a</sup>HR, Hazard ratio.

<sup>b</sup>95% CI, 95% confidence interval.

<sup>c</sup>p was considered significant when p < 0.05 (bold values). NS, not significant.

<sup>d</sup>breast cancer subclasses were based on the St Gallen recommendation (68) according to Gyorffy et al. (69).

of a relevant cellular model and the use of relevant exposure conditions (long-term and low-dose exposure) mimicking natural exposure. The few studies having investigated the impact of chronic and low-dose exposure of B[a]P or BAP on tumor progression have focused on the early stage of carcinogenesis (non-transformed epithelial cells) (9–13, 29) MCF10AT1 breast cells (representing the early-transformed stage of the unique MCF10 model of mammary progression from normal epithelium to triple negative breast cancer (43, 44) have, to our knowledge, never been used to test the impact of chronic and low-dose exposure to environmental pollutants. While the genotoxic and pro-carcinogenic B[a]P and the EDC BPA are two of the most studied pollutants, no previous studies, to our knowledge, have investigated whether the combination of BPA and B[a]P, each compound possessing distinct mechanisms of action, induces potentiating effects on tumor progression.

The present study aimed at addressing: (i) the effects of long-term, cumulative exposure to low doses of B[a]P or BPA on the mammary early-transformed substage; (ii) whether the combination of BPA with B[a]P, each compound possessing distinct mechanisms of action, impacts breast tumor progression differently from what might be observed with each compound tested alone. Our study reveals that long term and low dose (10−<sup>10</sup> M) exposure to B[a]P and/or BPA increases the cancerous properties (AIG and cancer stem-like properties) of the earlytransformed MCF10AT1 cells, and the longer the cells were exposed, the greater was the impact. Co-exposure of MCF10AT1 cells with the B[a]P and BPA led to a significantly greater aggressive phenotype compared to B[a]P alone, suggesting that BPA facilitates the pro-carcinogenic activity of B[a]P and supporting the potentiating effects of distinct pollutants present in the exposome. Importantly, the aggressiveness developed by the exposed MCF10AT160d cells was acquired and not softened or reverted after stopping B[a]P and/or BPA exposure. Altogether, our data reveal that long-term and low dose exposure to B[a]P and/or BPA irreversibly favors the evolution of early-transformed human mammary cells toward breast tumor progression.

Mechanistically, while B[a]P and BPA are well-known activators of AhR and GPR30, respectively, our short-term exposure experiments highlight that the MFE and AIG response triggered by exposure of MCF10AT1 cells to B[a]P and/or BPA occurs through a functional cross-talk between AhR and GPR30. Regarding the impact of inactivating AhR or GPR30 by two different strategies (using an antagonist molecule or a siRNA strategy), the BPA, BPA+B[a]P, or G1 impact on AIG and MFE was totally reversed when AhR was inhibited, while the GPR30 inactivation globally seemed to lead to a significant but not total, reversion of B[a]P, BPA+B[a]P, or ITE impact (in particular in MFE experiments). Finally, this study provided evidence that long-term (60 days) co-exposure to B[a]P + BPA 10−<sup>10</sup> M with the AhR antagonist GNF351 10−<sup>7</sup> M and/or the GPR30 antagonist G15 10−<sup>8</sup> M inhibited the development of pollutants-driven enhancement of cancerous properties in MCF10AT1 cells.

The AhR/CYP1A1 pathway participates in carcinogenesis by mediating stem properties, the formation of mammospheres, expansion of breast cancer stem cells, and the transcriptional activity of AhR is mainly implicated in such effects (24). Our study also reveals that GPR30 participates and favors AhR-dependent transcriptional activity in early-transformed breast cells. Indeed, the GPR30 agonist G1 was able to stimulate AhR-dependent driven activity and CYP1A1 gene transcription. GPR30 silencing by a siRNA strategy led to a significant, yet moderate, decrease in ITE-, G1-, or TCDD-mediated activation of AhR-driven transcriptional activity or CYP1A1 transcription. This suggests that GPR30 dependent mechanisms other than those influencing the AhR-driven transcriptional mechanisms might be involved in the AhR/GPR30 cross-talk impacting AIG and MFE. Altogether, our in vitro data support a model in which GPR30 is involved in an AhR-dependent network leading to increased cancerous properties of early-transformed mammary cells.

We thus propose a model (**Figure 8C**) in which AhR signaling plays a "driving role" in the AhR/GPR30 cross-talk

in mediating the effects of long-term and low dose exposure of B[a]P and/or BPA on AIG and MFE in MCF10AT1 cells. The relevance of our model was supported by our RNAseq data demonstrating that the canonical AhR signaling pathway was significantly enriched in the B[a]P-exposed MCF10AT160d cells, in the B[a]P + BPA exposed MCF10AT160d cells, but also in the BPA-exposed MCF10AT160d cells. Strengthening our findings, previous studies support a role for AhR and breast tumor progression: (i) AhR expression levels were significantly up-regulated in human breast ductal carcinoma in situ and breast cancer tissues compared to normal/benign breast tissues (70, 71); (ii) an in vivo model of breast tumorigenesis suggests that AhR is constitutively activated at early stages of mammary tumorigenesis (72); (iii) the prognostic value of AhR seems to be dependent on the activation/inactivation of metastatic processes (73).

Diagram summarizing our findings.

While this research was ongoing, a study was published supporting our data by demonstrating in the ERα-negative SKBR3 breast cancer cells that the environmental pollutant 3-methylcholantrene, mainly known to exert its carcinogenic effects through AhR, stimulates cell growth response through a functional interaction between AhR and GPR30 (74). However, in the ERα-positive MCF-7 breast cancer cell line, possessing the well-identified close cross-talk between AhR and ERα [for review (75)], 10−<sup>5</sup> M G1 was demonstrated to increase transcription of CYP1A1 mRNA in AhR-dependent, but GPR30-independent, mechanisms (76). As MCF10AT1 cells used in this study are triple-negative cells and the SKBR3 cells are ERα-negative, one cannot exclude that the detrimental functional cross-talk between AhR and GPR30 might be exacerbated in such a specific cellular breast context. Supporting previous data highlighted that in the ERα-negative/triple negative context, an AhR-signaling reinforces cell aggressiveness and induces breast cancer stem cells (77–80).

Altogether, our in vitro data thus demonstrated the role of the AhR/GPR30 cross-talk in favoring tumor progression, at least in a triple-negative breast context. Previous controversial studies emerged regarding the prognostic value of AhR or GPR30 expression levels in breast cancers (71, 73, 80–84), and these discrepancies have been suggested as possibly related to the breast cancer subgroup or the substage considered. In the present study, the clinical relevance of our in vitro findings was further reinforced by retrospective analysis of two independent breast cancer cohorts, showing that in only ERnegative or triple-negative breast cancer subclasses, the gene signature involving both AhR and GPR30 mRNA levels were of poor prognosis.

Hence, we have provided novel insights into the progression of early-transformed human mammary cells upon long-term and low-dose exposure to B[a]P and/or BPA and deciphered the involvement of the functional crosstalk occurring between AhR and GPR30 leading to an exacerbated AhR-driven network. Our in vitro and retrospective data analyses further support the idea that the deleterious impact of this cross-talk might be of utmost importance in the progression of ER-negative or triple-negative breast cancers. More importantly, strategies targeting AhR and/or GPR30 were demonstrated to be efficient in inhibiting the deleterious impact of cumulative and lowdose exposure to B[a]P and/or BPA in early-transformed MCF10AT1 cells. The identification of such molecular mechanisms may help in the discovery of human biomarkers of environmental carcinogen exposure and the development of preventive strategies.

#### REFERENCES


# DATA AVAILABILITY STATEMENT

The datasets generated from this study can be found in the NCBI Gene Expression Omnibus (GEO) [http://www.ncbi.nlm. nih.gov/geo/ (GSE142073)].

# ETHICS STATEMENT

This study has been approved by the local ethics committee (CRB Centre Léon Bérard, France). All subjects gave written informed consent.

# AUTHOR CONTRIBUTIONS

PAC, VM-S, VC, and BF participated in the design of the study. CD, ME, SG, and SA performed the in vitro data. PAC and CD performed the biostatistical analysis on the CLB cohort. BG performed the biostatistical analysis on the KMP cohort. AW, SC, and JL performed and analyzed the RNAseq data. SG, MD-A, AE, and PC participated in scientific discussions. PAC, CD, and ME wrote the manuscript. BG, AW, PC, JL, VC, VM-S, BF, CD, SG, and ME participated in the scientific revisions of the manuscript. PAC supervised the study.

#### FUNDING

CD was funded by the ANR (2011 ANR-CESA-018-01, Agence Nationale de la Recherche, France) and the University of Lyon 1. ME was funded by a PhD grant from the Lebanese University. This work was funded by the ANR (2011 ANR-CESA-018-01), Lyon Biopôle, and supported by grants from the Ligue nationale contre le cancer LNCC (committees 42 and 71).

#### ACKNOWLEDGMENTS

The authors thank the Centre de Ressources Biologiques (CRB) of the Centre Léon Bérard (BB-0033-00050). We are grateful to B. Manship for editing. We thank Dr. Balaguer for providing the HG5LN PXR cell line and the XRE-luc plasmid. Some of the work performed by ME during her thesis has been partly included in this article (85).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2020.00712/full#supplementary-material


Toxicol Appl Pharmacol. (2015) 288:131–42. doi: 10.1016/j.taap.2015. 01.009


85. El Helou M. Exploration of the Impact of Chronic and Low Doses Exposure of Environmental Factors on the Pre-cancerous Mammary Cells MCF10AT1. (dissertation/PhD's thesis). Lyon, University Claude Bernard Lyon 1, Beirut, Lebanese University (2017).

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Donini, El Helou, Wierinckx, Gyorffy, Aires, Escande, ˝ Croze, Clezardin, Lachuer, Diab-Assaf, Ghayad, Fervers, Cavaillès, Maguer-Satta and Cohen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteoglycans as Therapeutic Targets in Brain Cancer

#### Zoya Yan<sup>1</sup> and Shanzhi Wang<sup>2</sup> \*

<sup>1</sup> Horace Greeley High School, Chappaqua, NY, United States, <sup>2</sup> Chemistry Department, University of Arkansas at Little Rock, Little Rock, AR, United States

Proteoglycans (PGs) are heavily glycosylated diverse proteins consisting of a "core protein" covalently attached to glycosaminoglycans (GAGs) and present on the cell surface, extracellular matrix, and intracellular milieu. Extracellular proteoglycans play crucial roles in facilitating cell signaling and migration, interacting with growth factor receptors, intracellular enzymes, extracellular ligands, and matrix components, as well as structural proteins and promoting significant tumor-microenvironment interactions in cancerous settings. As a result of their highly regulated expression patterns, recent research has focused on the role of proteoglycans in the development of nervous tissue, such as their effect on neurite outgrowth, participation in the development of precursor cell types, and regulation of cell behaviors. The present review summarizes current progress for the studies of proteoglycan function in brain cancer and explains recent research involving brain glycoproteins as modulators of migration, cell adhesion, glial tumor invasion, and neurite outgrowth. Furthermore, we highlight the correlations between specific proteoglycan alterations and the suggested cancer-associated proteoglycans as novel biomarkers for therapeutic targets.

#### Edited by:

Chang Zou, Shenzhen People's Hospital, Jinan University, China

#### Reviewed by:

Joseph Zaia, Boston University, United States Yongwei Zhang, Albert Einstein College of Medicine, United States

#### \*Correspondence:

Shanzhi Wang sxwang2@ualr.edu

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

Received: 01 December 2019 Accepted: 29 June 2020 Published: 05 August 2020

#### Citation:

Yan Z and Wang S (2020) Proteoglycans as Therapeutic Targets in Brain Cancer. Front. Oncol. 10:1358. doi: 10.3389/fonc.2020.01358 Keywords: proteoglycan, glycosaminoglycan, glioblastoma, brain cancer, biomarker, inflammation

#### INTRODUCTION OF BRAIN CANCER

Brain tumors are abnormal cell growths in the brain, though only malignant tumors are cancerous. There are two types of brain tumors: the primary brain tumor, which originates from and resides within the brain, and the secondary (metastatic) brain tumor, which originates from cancer outside of the central nervous system (CNS) then spreads into the brain. While primary tumors are the more frequent solid tumors in children, metastatic tumors are more commonly diagnosed in adult patients (1).

In recent decades, the worldwide brain tumor incidence rate has increased across all ages. The standardization of age in varying countries is between 0.01–12.7 in males and 0.01–10.7 in females per 100,000 people, with the highest incidence in northern Europe and the lowest in Africa (2). According to the Central Brain Tumor Registry of the United States (CBTRUS), the incidence rate of CNS tumors in the United States (23.03 per 100,000 cases for a total count of 392,982) incident tumors, of which 121,277 cases

**341**

are malignant and 271,705 non-malignant) is lower in males (20.59 per 100,000 for 165,148 total cases) than in females (25.31 per 100,000 for 227,834 total cases) (3). In addition, in the United States, the 5-years survival rate following the diagnosis of a primary malignant CNS tumor is about 35% (2008–2014 data)<sup>1</sup> . The mortality rate of CNS cancers is estimated at about 3.4 per 100,000 in the world (2).

Across all age groups, the most common brain tumors develop from glial cells, which are gliomas that encompass a large scope of tumors and can be classified into four grades as follows: grade I (pilocytic astrocytoma), grade II (diffuse astrocytoma), grade III (anaplastic astrocytoma), and grade IV (glioblastoma multiforme) (4). Grade III and IV are categorized as high-grade or malignant gliomas with extremely poor prognosis, with grade IV diagnoses (which account for half of primary brain tumors) seeing a 5-years survival rate of <10% (5). The most common intracranial tumors in adults are brain metastases, with over 150,000 cases in the United States alone. Despite the considerable effect of varying primary tumor types on the incidence of metastases, 8–10% of adults diagnosed with cancer will develop brain metastases (6).

Developing treatment for CNS cancer is one of the most exigent branches of study in oncology. Although therapeutic approaches that exploit the immune system are a promising alternative strategy to surgery, radiotherapy, and anticancer drug therapy, multidrug resistance is a substantial obstacle restricting the success of conventional chemotherapy (7). The innate chemoresistance of many primary brain tumors and insufficient penetration of cytotoxic drugs across the blood-brain barrier (BBB) are also both responsible for the unsuccessful response of brain tumors to chemotherapy (8). Due to the urgency for novel therapies that combat these occurrences, researchers have emphasized and prioritized the development of anticancer drugs.

Using in vitro methods, Kwon et al. directly tested the effects of sialic acid glycan and glycosylation on BBB influx and efflux of IgG, specifically the influx and efflux processes for BBB endothelial cells, and facilitated "direct measurement of the Permeability Coefficient in each direction" (9, 10). In the study, BBB pharmacokinetics were found to be considerably affected by modest changes of IgG glycan profiles with sialylated glycans, suggesting that modifying IgG glycan could become an effective technique in increasing the concentration of the brain's therapeutic antibodies. Because immune pathways induced by sialylated IgG cause little inflammation, sialylation may therefore suggest beneficial clinical possibilities and further implications for patients with other CNS diseases such as Alzheimer's disease (11–13).

# ROLES OF PROTEOGLYCANS IN BRAIN CANCER

Proteoglycans (PGs) are heavily glycosylated proteins and present on the cell surface, extracellular matrix (ECM), and intracellular milieu (14). The basic PG unit consists of a "core protein" covalently attached to glycosaminoglycans (GAGs), which are long chains consisting of linear or branched carbohydrate polymers that are negatively charged under certain physiological conditions, expressed on most mammalian cells. The six known GAGs are heparin (HP), hyaluronic acid (HA), heparan sulfate (HS), chondroitin sulfate (CS), keratan sulfate (KS), and dermatan sulfate (DS). These GAGs' respective disaccharide units contain different uronic and amino sugars (15). Their structure is demonstrated in **Figure 1**. GAGs are considerably linked to several diseases, including cancer, inflammation, bacterial infections, multiple sclerosis, viral infections, and Alzheimer's disease, as they interact with ligands to modulate physiological and pathological processes. Given such active involvement, GAG-based drugs are of considerable interest to researchers and have yielded promising outcomes in both animal and clinical trials, suggesting prospective development in therapeutics (16).

GAG to PG linkage is crucial to establishing and maintaining the fundamental functions of CNS, such as migration, cellular proliferation, specification, plasticity, synaptogenesis, and regeneration. Their mechanisms and functions have been summarized in many reviews (17–20). PG diversity depends on differential expression of protein sequences, variations in the length, and profile of GAG modifications. PGs regulate growth factors that affect cell adhesion, neurite outgrowth (21), ECM assembly, and tumor cell invasion (22, 23). Syndecans and glypicans are the two main transmembrane PGs containing HS chains in the CNS. Heparan sulfate PGs directly influence the aggregation and activity of AMPA receptors, which hinders cognitive functions by inducing or maintaining long-term potentiation (LTP) (24). Chondroitin sulfate PG, which is expressed abundantly in the cerebellum and hippocampus but decreases significantly postnatally, affects the stabilization of synapses and axonal sprouting (25). **Figure 2** demonstrates the selected cellular localization and significance for tumor development of the PGs discussed.

#### Chondroitin Sulfate Proteoglycan

Chondroitin sulfate PG 4, commonly referred to as neuronglial antigen 2 (NG2), contributes to the stabilization of interaction between cell and substratum on endothelial basement membranes, especially at the early spreading stage of melanoma cells (26) In addition, CS PG 4 is a suggested biomarker in glioblastoma (GBM) (27). NG2-expressing oligodendrocyte precursor cells support neurons and synaptic signaling physiologically and carry out these functions in both brains that are healthy and those in the process of injury repair and regeneration. Moreover, NG2 protein facilitates tumorigenesis and tumor progression (28). In the previous study of human GBM cells, more cells survived under NG2-mediated

**Abbreviations:** BBB, blood-brain barrier; CS, chondroitin sulfate; CNS, central nervous system; DS, dermatan sulfate; ECM, extracellular matrix; GAGs, glycosaminoglycans; GBM, glioblastoma; HS, heparan sulfate; HP, heparin; HA, hyaluronic acid; KS, keratan sulfate; NG2, neuron-glial antigen 2; PG, proteoglycan.

<sup>1</sup>Available online at: https://www.cancer.org/content/dam/cancer-org/research/ cancer-facts-and-statistics/annual-cancer-facts-and-figures/2019/cancer-factsand-figures-2019.pdf

activation (29) and chemoresistance through integrin-dependent PI3K/Akt signaling (30).

Gliomagenesis is induced by unusual expression of neuronglial antigen 2 endocytosis in vivo murine oligodendrocyte precursor cells, which provides another mechanism through which benign precursor cells can be converted into cancer stem cells (31). Additionally, NG2-expressing precursor cells demonstrate significant developmental plasticity. For instance, activating Notch signaling induced pericyte-like differentiation in NG2-positive GBM cancer stem cells, which during tumor angiogenesis contributed to vessel stabilization (32). Although these results suggest that NG2 plays a role in cancer stem cells, it is still uncertain whether the GAG moieties or the PG's other functional domains are responsible for the stemnessrelated functions.

Lam et al. reported an efficient and effective "glial progenitor cell-based therapy" for congenital myelin CNS disorders (33). From bone marrow stromal cells, they produced glial progenitor cells in a 14-days CS PG 4-based induction protocol. The generated cells were highly enriched in oligodendrocyte precursor cell marker expression. After being transplanted into the myelin-deficient mice, the cells differentiated successfully into myelinogenic oligodendrocytes. Both lifespan and motor function were improved significantly by remyelination of the shiverer mouse. Their study demonstrated the feasibility of human bone marrow stromal cells as a source of glial progenitor cells for attaining such myelinogenic oligodendrocytes (33). The novel induction protocol overcame existing hurdles of cell source restriction and timeframe requirements, providing a method for efficient myelin disorder glial therapy.

In addition, lecticans were also investigated as a group of chondroitin sulfate PGs due to their role in linking ECM molecules (34). The unique composition of brain ECM causes brain tissue to resist invasion by non-neuronal tumors (35). Due to its moderate plasticity, CNS has a considerable capacity for regeneration, although changes in ECM have been observed after trauma and throughout the development of CNS disease. The modification of PGs in ECM is shown as one of the factors leading to change in CNS plasticity. Through control of neurotransmission and synaptic connections, the scaffold of proteins and sugars in the ECM changes the functionality of surrounding tissue (36).

To activate immune cells, CS PGs generally collect the microenvironment's signals and bind immunological receptors,

thus boosting inflammatory responses. CS PGs also stimulate matrix-degrading enzymes and bind signaling molecules in immune cells such as chemokines and cytokines (37).

#### Heparan Sulfate Proteoglycans

Heparin has anticoagulant activity and can only be produced by mast cells. Heparan sulfate also functions as an anticoagulant, though on a lower level than heparin, and is generated by nearly all cell types. Present both in the ECM and on the cell surface, heparan sulfate PGs (HSPG) facilitate cellmicroenvironment interactions and cell signaling pathways. In GBM, HS glycosaminoglycans expression and their regulating enzymes are changed, but the structure and content of the HS itself remain unknown. For example, glypicans of the HS PG families [See review: Wang et al. (38)] are proteins that are membrane-bound and that modulate morphogen gradient formation and extracellular growth signals to engage in organ development (39). Crucially, some studies reported an increased level of glypicans in the peripheral blood of patients, holding glypicans as a promising new biomarker in the cancer field (40). Tran et al. used LC-MS analysis to portray the differences in both HS disaccharide content and structure. As a result, they suggested inter-tumoral differences in PG expression and function have potential implications for therapeutic stratification (41).

Spyrou studied the inhibition of heparanase in brain tumor cells of children and subsequently reduced their invasive capacity, proliferation, and tumor growth in vivo. The results suggest that heparanase affects both tumor cells and their ECM in cases of malignant brain tumors. However, the inhibitor (PG545) failed to pass the BBB due to its size, and thus direct injection or a new drug delivery system is required (42).

#### Hyaluronic Acid

Hyaluronan (or hyaluronic acid) is a "multifunctional GAG synthesized as a large negatively charged linear polymer by distinct hyaluronan synthases" (43). While aggrecan-related components generally result in clear regional distribution patterns, hyaluronan is widely distributed in the white and gray matter (44). HA interacts with several cell membrane receptors, including CD44 and Lymphatic vessel endothelial hyaluronan receptor 1, the former being the more thoroughly studied receptor for HA-mediated motility in cancer progression. In addition, certain PGs use link modules to form supramolecular complexes with HA. Generally, high levels of HA and HA receptors are correlated with poor prognosis in cancer patients.

Using pluripotent stem cells-derived and primary brain microvascular endothelial cells, Al-Ahmad et al. tested the effect of HA on BBB properties. The impact of HA signaling on developmental and mature brain microvascular endothelial cells was assessed by measuring changes in transendothelial electrical resistance, permeability, brain microvascular endothelial cells markers localization, and expression, CD44 expression, and hyaluronan levels. HA treatment generally reduced barrier function and P-glycoprotein activity. The effects were more evident with treatment using oligomeric forms of HA and exacerbated when the treatment was applied during the brain microvascular endothelial cells differentiation phase (considered developmental BBB). The hyaluronidase activity, as well as an increase in CD44 expression during prolonged oxygenglucose deprivation stress, were also observed. Inhibiting HA signaling by the antibody blockade of CD44 reversed the treatment's adverse effects, thus conveying the significance of HA signaling through CD44 on BBB properties (45). Moreover, Hartheimer et al. determined how hyaluronidases can sensitize GBM stem cells to chemotherapy drugs by disrupting the HA-CD44 signaling, with which they further developed a combined treatment of hyaluronidases and chemotherapy drugs by disrupting the stemness-promoting HA to target GBM stem cells. This combination therapy shows promise even when temozolomide treatment alone causes resistance (46).

#### Dermatan Sulfate PG—Endocan: A New Biomarker and Therapeutic Target

Endocan is a novel endothelial cell-specific molecule with 50 kDa molecular weight and high solubility in water. As a proteoglycan, endocan is secreted into the blood and formed in the presence of CS. In normal tissues, CS and DS PGs are expressed in endothelial cells but are overexpressed in certain tumor endothelial cells. Unsurprisingly, abnormal expression levels of endocan were observed in tumor prognosis, angiogenesis, and metastasis. Researchers believe that the role of endocan is to regulate the tumor by tumor-related angiogenesis, cell inflammation, lymphangiogenesis, and other aspects (47). Accordingly, Kijima et al. studied surface marks from patient derived xenografts and cell lines based on array comparative genomic hybridization to investigate the early stages of GBM tumorigenesis (48). In additional to research that found raised levels of systemic inflammatory markers to be correlated with cardiovascular disease (49), a recent study revealed that the specific sulfation level of DS is crucial in synaptic plasticity and is related to changes in the expression of glutamate receptors and other associated synaptic proteins (50). As such, endocan became a valuable target for GBM diagnosis and therapy.

As another interesting target, dermatan sulfate epimerase 1 is overexpressed in many types of cancer as a tumor-rejection antigen. The CS/DS chains mediate several growth factor signals. However, investigating their roles in gliomas involves less work. Liao et al. examined the expression of Dermatan sulfate epimerase 1 in gliomas by utilizing a public database and conducting immunohistochemistry on a tissue array. Their investigation revealed that Dermatan sulfate epimerase 1 regulates the HB-EGF/ErbB pathway, which participates in GBM cells' malignant behavior. Treating epidermal growth factor receptor and ErbB2 with selected inhibitors thus suppressed malignant phenotypes, demonstrating that the upregulation of Dermatan sulfate epimerase in gliomas contributes to controlling malignant behavior in cancer cells (51).

# Keratan Sulfate (KS)

Keratan sulfate (KS) is a sulfated GAG, which contains structurally unique characteristics of diversity in the linker oligosaccharides connecting to the core protein. The repeating disaccharide unit in KS contains one galactose and one Nacetylglucosamine and is linked to core proteins via either N-linked or O-linked glycosylation of the PGs. KS is most abundant in the cornea, and second abundant in the brain (52). Negatively charged KS modifications of synaptic vesicle protein 2 interacted with both Ca2<sup>+</sup> ions and other neurotransmitters such as dopamine, establishing the PG delivery complex (53). Furthermore, high sulfation level KS PGs are commonly found in the brain. For example, synaptic vesicle proteins 2 played significant neuronal and synaptic regulatory roles (54).

An earlier study reported that highly sulfated KS was overexpressed in malignant astrocytic tumors (55). It has also been found that the interruption of Synaptic vesicle proteins 2 functionality is associated with epilepsy (56). These results were confirmed by the subsequent research of the interactivity of KS with nerve growth factor and receptor proteins, neuroregulatory proteins, synaptic proteins and neurotransmitters (57). In addition, abnormal sulfation degrees of KS are observed in the brains of Alzheimer's patients (58). Tsidulko's work demonstrated that PG composition and ECM structure in normal brain tissue were affected during temozolomide induced chemotherapy. These changes were believed to participate in the development of the tumorigenic niche for the expansion of the residual glioma cells and the disease progression (59). Recently, researchers have reviewed the influences of KS sulfation on electrosensory tissues and neuronal regulation. KS with overexpressed sulfation level interacts with neuroregulatory proteins. Hence, actin and tubulin cytoskeletal development was stabilized by KS PG microtubuleassociated proteins during neuritogenesis (60).

#### THE ROLE OF PGs IN BRAIN INFLAMMATION AND PLASTICITY

In general, PGs have influenced several aspects of tumor biology such as tumor cell adhesion and migration, cell proliferation, angiogenesis, and inflammation. Up- and downregulated expression in PG core proteins is observed in many cancers and usually related to changes in cell signaling and invasion (19, 61). Jang et al. found that the interaction between the intracellular domain of some transmembrane PGs with the cytoplasmic domain of proteins promoted the signaling (62). In an earlier review, the cytoplasmic domain of syndecan-1 was found to have interaction with talin to modulate integrin signaling via a syndecan-1-integrin-insulin-like complex (63). Likewise, the cytoplasmic tail of transmembrane heparan sulfate PG syndecan-4 interacts with α-actinin regulating cytoskeletal organization. Fröhling et al.' recent research suggested that the loss of syndecan-4 expression is correlated with the increase if intestinal inflammation. While primarily expressed in the colonic epithelium, syndecan-4 accumulated the deficiency during the growth of susceptibility regarding the intestinal inflammation. Mechanisms were proposed that syndecan-4 played a role in protecting against inflammation, keeping the epithelial gut barrier's unity and regeneration (64). When using anti-syndecan-4 antibodies as a therapeutic approach to treat patients with inflammatory disorders, researchers must carefully evaluate patients who have inflammatory diseases associated with an epithelial barrier function.

Many PGs were proposed as markers for therapy evaluation. For example, syndecan-4 mRNA expression was specified as the unique marker to predict the GBM multiforme patient's response during the WT1 peptide vaccine treatment (65). Letoha et al. reported that syndecan-4 bound and mediated the transfer of a cell-penetrating short peptide with 17 amino acids into the cells (66). Roy et al. then demonstrated a positive correlation between glioma grade and serglycin expression level in GBM progression (67).

There is significant evidence showing that the sulfate composition of CS GAG chains changes with age. As a result of aging and aggregation of proteins, the deposition of HS PGs and CS PGs results in the injury of protective perineuronal nets with increased cell death (60). Dying neurons then induce inflammation, ECM degrades through the proteolytic activity of enzymes, inducing responses that amplify neuronal death and neuroinflammation (68). Simultaneously, overexpression of chondroitin 6-O-sulfotransferase 1 may decrease the ratio of 4– 6S in perineuronal nets and increase seizure susceptibility (69). This is supported by Foscarin's work that the age-associated rise in the ratio of 4–6S GAG in perineuronal nets may decrease synaptic plasticity (70). Their studies highlighted the necessity for genetic manipulation of other enzymes such as chondroitin

#### REFERENCES


sulfotransferases to discover their biological functions and generate the profile of sulfation's role in development and aging.

Recent preclinical research demonstrated the antitumoral effects of chondroitin sulfate PG 4. The NG2-directed chimeric antigen receptor T-cells were proved to efficiently target GBM cancer stem cells (71). The combination of anti-NG2 antibodies was induced in chemotherapy in B-cell acute lymphoblastic leukemia (72).

#### CONCLUSION

In summary, adjustments in PG core proteins, biosynthetic enzymes, and extracellular regulating enzymes are correlated with many developmental anomalies and overgrowth or tumor predisposition syndromes. PGs facilitate the activity of various signaling pathways and stimulate cell-microenvironment interactions in tumors. Due to such a diverse range of functions, PGs and their modifying enzymes are an imperative area of study that may potentially uncover therapeutic targets and biomarkers of GBM. In the damaged CNS, PGs accumulate during traumatic brain injuries, multiple sclerosis, and spinal cord injuries, driving pathogenesis and neuroinflammation. It should be noted that compared to the in vitro examination, more complex factors may interfere with the regulated expression of PGs in ECM in vivo. When assessing experiment data for therapeutic targets and treatment strategies, researchers should carefully consider adverse side effects that can be avoided in advance.

#### AUTHOR CONTRIBUTIONS

All authors participated in writing and revising the manuscript.

#### FUNDING

This work has been supported by UA Little Rock Startup Grant.

#### ACKNOWLEDGMENTS

We wish to acknowledge the support of UA Little Rock for the Startup Package.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Yan and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effects of the Oncoprotein PAX3-FOXO1 on Modulation of Exosomes Function and Protein Content: Implications on Oxidative Stress Protection and Enhanced Plasticity

Assil Fahs1,2, Farah Ramadan<sup>1</sup> , Farah Ghamloush<sup>3</sup> , Abeer J. Ayoub1,2 , Fatima Ali Ahmad<sup>1</sup> , Firas Kobeissy<sup>4</sup> , Yehia Mechref<sup>5</sup> , Jingfu Zhao<sup>5</sup> , Rui Zhu<sup>5</sup> , Nader Hussein<sup>6</sup> , Raya Saab2,3 \* and Sandra E. Ghayad<sup>1</sup> \*

<sup>1</sup> Department of Biology, Faculty of Science II, Lebanese University, Fanar, Lebanon, <sup>2</sup> Department of Anatomy, Cell Biology and Physiology, American University of Beirut, Beirut, Lebanon, <sup>3</sup> Department of Pediatrics and Adolescent Medicine, Children's Cancer Institute, American University of Beirut, Beirut, Lebanon, <sup>4</sup> Department of Biochemistry and Molecular Genetics, Faculty of Medicine, American University of Beirut, Beirut, Lebanon, <sup>5</sup> Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, TX, United States, <sup>6</sup> Cancer Biology Stem Cells and Molecular Immunology Laboratory, Faculty of Sciences, Lebanese University, Beirut, Lebanon

Rhabdomyosarcoma (RMS) is a highly malignant soft tissue sarcoma classified into two major histologic subtypes: embryonal (ERMS) and alveolar (ARMS). ARMS subtype is clinically more aggressive, and characterized by an oncogenic fusion protein PAX3-FOXO1 (P3F) that drives oncogenic cellular properties. To understand the role of the fusion oncoprotein in paracrine signaling, we focused on secreted exosomes, which have been demonstrated to contribute to metastasis in multiple tumor types. Advanced Proteomics-bioinformatics analysis of the protein cargo of exosomes isolated from C2C12 myoblasts transduced with P3F fusion gene revealed 52 deregulated proteins compared to control cells, with 26 enriched and 26 depleted proteins. Using both PANTHER gene classification and Ingenuity Pathway Analysis (IPA) software, we found that the main biological processes in which the 52 deregulated proteins are involved, include "catalytic activity," "binding," "metabolic process," and "cellular process." The pathways engaging the 26 enriched proteins include the "14-3-3 mediated signaling," "cell cycle," and "ERK5, VEGF, IGF1,and p70S6K signaling." Furthermore, the main nodes in which deregulated exosome proteins and miRNAs intersected revealed pathways conferring protection from stress and promoting plasticity. Based on the bioinformatics analysis and the altered exosome proteome profile, we performed biochemical functional analysis to study the diverse properties of these exosomes where angiogenesis, stemness, and anti-oxidative stress properties were validated using different platforms. P3F-modulated exosomes activated ERK, 4-EBP1, and MMP-2 in recipient cells, and enhanced angiogenesis and stemness. In addition, P3F led to lower

#### Edited by:

Boris Zhivotovsky, Karolinska Institutet (KI), Sweden

#### Reviewed by:

Francesco Marampon, Sapienza University of Rome, Italy Luisa Lanfrancone, European Institute of Oncology (IEO), Italy

#### \*Correspondence:

Sandra E. Ghayad sandra.ghayad@ul.edu.lb Raya Saab rs88@aub.edu.lb

#### Specialty section:

This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology

> Received: 15 February 2020 Accepted: 11 August 2020 Published: 01 October 2020

#### Citation:

Fahs A, Ramadan F, Ghamloush F, Ayoub AJ, Ahmad FA, Kobeissy F, Mechref Y, Zhao J, Zhu R, Hussein N, Saab R and Ghayad SE (2020) Effects of the Oncoprotein PAX3-FOXO1 on Modulation of Exosomes Function and Protein Content: Implications on Oxidative Stress Protection and Enhanced Plasticity. Front. Oncol. 10:1784. doi: 10.3389/fonc.2020.01784

**349**

cellular reactive oxygen species levels and enhanced resistance against oxidative stress; and treatment of stromal cells with P3F-modulated exosomes also conferred protection against exogenous oxidative stress. Our findings highlight the role of P3F fusion protein in modulating exosome cargo to confer a protective effect on recipient cells against oxidative stress and to promote plasticity and survival, potentially contributing to the known aggressive phenotype of the fusion gene-positive subtype of RMS.

Keywords: Rhabdomyosarcoma, PAX3-FOXO1, exosomes, oxidative stress, plasticity

#### INTRODUCTION

Rhabdomyosarcoma (RMS) is an aggressive soft tissue sarcoma of the skeletal muscle generally affecting children and adolescents (1). It is currently classified, according to the World Health Organization (WHO), into four histological subtypes with embryonal (ERMS), and alveolar (ARMS) being the two most frequent subtypes (2). ARMS has been described as more aggressive having a prognostic disadvantage compared to ERMS, specifically in the majority subset expressing the fusion oncoprotein PAX3-FOXO1 (or less commonly PAX7- FOXO1) (3). In fact, fusion status which divides RMS cases into fusion positive (FPRMS) and fusion negative (FNRMS) allows a better classification and determination of prognostic features (4). Common translocations involve the DNA binding domain of either PAX3 or PAX7 with the transactivation domain of FOXO1 resulting in PAX3/7-FOXO1 fusion oncoprotein that enhances RMS growth and metastasis by acting at target genes involved in proliferation, migration and invasion (5). RMS cells can modulate the tumor microenvironment to enhance both cancer cell and recipient cell motility favoring metastatic disease (6). For instance, MMP (matrix metalloproteinase) production by RMS cells, and to a higher propensity by ARMS cells, has been found to induce extracellular matrix remodeling and enhance invasiveness (7).

Exosomes are extracellular vesicles with a diameter ranging between 20–150 nm secreted under both physiological and pathological conditions and their cargo modulates signaling pathways in recipient cells (8). They have been demonstrated to promote tumor growth in a variety of cancer types including Ewing sarcoma, osteosarcoma and melanoma by promoting metastasis (9–11). In fact, different proteins and nucleic acids are enriched or depleted within cancer cell-derived exosomes, an indication that their sorting into exosomes is a selective process and that cancer cells can influence recipient cell behavior through the release of exosomes with modulated cargo. Functional changes induced by cancer cell-derived exosomes include the activation of angiogenesis and enhancement of recipient cell migration and invasion (6, 12–15). Proteomic profiling of RMSderived exosomes revealed the presence of proteins such as integrins and annexins in both FPRMS and FNRMS-derived exosomes which have been implicated in tumor metastasis (12, 13, 16, 17). Selective miRNA enrichment has also been demonstrated in RMS-derived exosomes and in P3F transduced myoblasts which were found to enhance recipient fibroblast migration and invasion (6, 18).

Cellular plasticity involves adaptation to changes in the microenvironment and physiological conditions that would allow cell survival and growth (19). In cancer, plasticity involves the acquisition of characteristics favoring tumor progression and cancer cell survival under stress conditions. For instance, Ewing sarcoma and melanoma cells were found to express the TGFβ co-receptor endoglin correlating with cell plasticity, enhanced survival and invasion (20). Plasticity also involves the attainment of stem cell characteristics that include an ability to invade and disseminate into surrounding tissue (21). Cancer stem cell (CSC) features have been observed in ERMS cells that express markers such as CD133, exhibit chemoresistance and can form rhabdomyospheres, all of which are stem cell characteristics (22, 23). Moreover, aspects of plasticity that sustain cancer cell survival include acquiring resistance to and protection from oxidative stress. Oxidative stress results from an imbalance between pro-oxidative and anti-oxidative cellular pathways. Reactive oxygen species (ROS) including hydrogen peroxide (H2O2), are required for physiological signal transduction, but their levels are elevated in cancer cells as a result of augmented metabolic rates (24, 25). While ROS can enhance cancer proliferation and metastasis, an over-accumulation of ROS creates oxidative stress that can threaten cell survival. Therefore, cancer cells exhibit resistance to oxidative damage by the activation of alternate signaling pathways with lower ROS production and increased induction of antioxidant enzymes (26). Of note, oxidative stress can also enhance exosomes production with altered nucleic acid content that can confer changes in recipient cells and sustain their survival against stress (27, 28). In this study, we demonstrate that the P3F fusion gene modulates exosomes to promote recipient cell plasticity and response to oxidative stress that may favor tumor growth and metastasis.

#### MATERIALS AND METHODS

#### Cell Lines and Viruses

C2C12 mouse myoblasts were purchased from the ATCC (Virginia, United States) and cultured in Dulbecco's modified Eagle's medium AQ (DMEM AQ, Sigma, Setagaya, Japan) supplemented with 20% fetal bovine serum (FBS, Sigma) and 1% penicillin (100 units/ml) – streptomycin (100 µg/ml) antibiotics (Sigma). MSCV-IRES-GFP-PAX3-FOXO1 (MSCV-P3F) and MSCV-IRES-GFP (MSCV-GFP) plasmids were a kind gift from Dr. Gerard Grosveld (St. Jude Children's Research Hospital, Memphis, United States). The human ERMS cell lines

JR1 and the ARMS cell lines Rh30, were generously donated by Dr. Peter Houghton (Columbus, OH, United States), and have been previously described [reviewed in (29)]. HUVEC (human umbilical vein endothelial cells) and HEK293T (human embryonic kidney) cells were also purchased from ATCC. As described previously, viral supernatants were produced by transfecting 293T cells with MSCV-P3F or MSCV-GFP vectors using calcium phosphate (18). C2C12 cells were then transduced in suspension with either MSCV-GFP viruses (forming Ctrl-C2C12 cells) or MSCV-P3F (forming P3F-C2C12 cells) at 32◦C, 1250 × g for 1 h with 8 µg/ml Polybrene (hexadimethrine bromide; Sigma), and sorted using FACS Aria SORP cell sorter (BD, New Jersey, United States) after selection with 2 µg/ml Puromycin (Abcam). Wild-type mouse embryonic fibroblasts (MEFs) and p53−/− (FVB.129-Trp53tm1Brn) MEFs were isolated from E13.5 embryos of mixed C57BL/6 × 129/Sv 77 background (Jackson Laboratory, Maine, United States) using the procedure approved by the Institutional Care and Use Committee (IACUC) at the American University of Beirut, and following the IACUCapproved guidelines. MEFs are cultured in DMEM AQ, media supplemented with 10% fetal bovine serum (FBS, Sigma), 1% penicillin (100 units/ml) – streptomycin (100 µg/ml) antibiotics (Sigma), 1% sodium pyruvate, and 1% non-essential amino acids. All cells were maintained under standard conditions (humidified atmosphere, 95% air, 5% CO2, 37◦C).

#### Exosome Isolation

Exosomes were isolated as previously described (6). Briefly, Ctrl-C2C12 cells, P3F-C2C12 cells, JR1 and Rh30 cells were incubated in exosome-free medium prepared by ultracentrifugation at 100,000 × g, overnight at 4◦C. Exosomes were then isolated from the conditioned media by differential ultracentrifugations (18): 300 × g for 10-min 2,000 × g for 20-min centrifugation, and then 10,000 × g for 30 min. Finally, ExoQuick Exosome Precipitation Solution (System Biosciences, mountainite, CA, United States) was added to the resulting supernatant, and stored overnight at 4◦C to allow exosome precipitation. The following day, the condensed medium was ultracentrifuged at 100,000 × g for 70 min, the pellet resuspended in PBS 1X, and ultracentrifuged at 110,000 × g for 70 min to remove any contaminating element. The final pellet was resuspended in 300 µL of PBS for functional assays and stored at −80◦C. All centrifugations were done at 4 ◦C. Recipient cells were treated with exosomes at 1X and 10X concentration, where 1X corresponds to exosomes isolated from an equivalent number of cells to those treated and 10X is 10 times this value.

# Protein Extraction, and Western Blot

Proteins were extracted from 3 independently transduced cells and their respective exosomes using CHAPS lysis buffer (30 mM Tris–Cl, pH 7.5; 150 mM NaCl; and 1% CHAPS) mixed with 25X protease inhibitor (Roche, Basel, Switzerland). After adding the appropriate volume of lysis buffer, the mixture was sonicated for 20 min, centrifuged for 30 min at 13,000 × g at 4◦C, then the supernatant containing the proteins was collected (12) and quantified using Bradford protein assay. Equal amounts of proteins were loaded and separated using 10% or 12% acrylamide gels, then transfered to a polyvinylidene difluoride (PVDF, Bio-Rad, CA, United States) membrane or nitrocellulose membrane (Santa Cruz, Heidelberg, Germany) in TGS1X-10% methanol transfer buffer. Membranes were blocked to prevent non-specific binding with 3% BSA-TBS1X-0.001% Tween [Tris (hydroxymethyl); NaCl; KCl and Tween 20; pH = 7.5] or 5% milk-TBS1X-0.001% Tween for 1 h, and then probed with primary antibodies (listed below) at 4◦C overnight where they were washed 3 times by TBS1X-0.001% Tween for 5 min before adding the corresponding species-specific Horseradish Peroxidase (HRP)-conjugated secondary antibodies (Santa Cruz) for 2 h. Results were then detected on ChemiDoc machine (Bio-Rad) using Clarity Western ECL reagent (Bio-Rad) as a substrate.

The primary antibodies used were: anti-Hsc70, anti-GAPDH, anti-Vimentin, anti-Fibronectin, anti-Actin, anti-Basigin, anti-Calnexin (Santa Cruz), anti-phospho-ERK, anti-phospho-4EBP1, anti-ERK, anti-4EBP1 (Cell signaling, Danvers, United States), anti-TSG101 (Abcam, Cambridge, United Kingdom), anti-gp47, and anti-gp91, and anti-NOX1 (Upstate, NY, United States).

# LC-MS/MS Analysis

Extracted proteins from Ctrl-C2C12- and P3F-C2C12-derived exosomes were subjected to sample clean-up and tryptic digestion as described previously (12). The LC-MS analysis of the tryptic digested proteins was performed using a 3000 Ultimate nano-LC system (Dionex, CA, United States) interfaced to an electrospray ionization (ESI) source equipped LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, CA, United States). Aliquots of 1 µg of tryptic digests were injected and initially desalted online using an Acclaim PepMap100 C18 pre-column (75 µm I.D. 20 mm length, 3 µm particle sizes, 100Å particle size, Thermo Scientific) at a flow rate of 3 µL/min. The purified peptides were separated using an Acclaim PepMap100 C18 capillary column (75 µm I.D., 150 mm length, 2 µm, 100Å, Thermo Scientific) at a flow rate of 0.35 µL/min. The gradient applied to achieve separation was as follow: solvent B (99.9% ACN with 0.1% formic acid) was maintained at 5% from 0 to 10 min, increased to 20% over 55 min, 20–30% over 25 min, 30–50% over 20 min, 50– 80% over 1 min, 80% over 4 min, decreased to 5% over 1 min, and finally was maintained at 5% for 4 min. Solvent A was 98% HPLC water, 1.9% ACN, and 0.1% FA. The LTQ Orbitrap Velos was operated in a data-dependent acquisition mode with 2 scan events. The first scan was a full MS scan of m/z 400-2000 with a mass resolution of 15,000. The second scan event was a CID MS/MS repeated on top 10 intense ions selected from the previous scan event with an isolation window of m/z 3.0. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository.

# LC-MS Data Processing

Proteome Discover version m1.2 (Thermo Scientific) was used to convert raw files acquired from LC-MS analysis to mascot generic format (<sup>∗</sup> .mgf) files. The <sup>∗</sup> .mgf files were subjected to MASCOT version 2.4 (Matrix Science Inc., Boston, United States) to search against protein database. The carbamidomethylation of cysteine was a fixed modification and the oxidation of methionine was considered as a variable modification. The mass shift tolerance for matching peptide precursors and fragments was 6 ppm and 0.8 Da, respectively. Enzyme was specified as trypsin. A maximum of 2 miscleavage was allowed. MASCOT searching results were then exported to Scaffold (version 3.6.3, Proteome Software, Portland, OR, United States) for further processing and quantification. Identified proteins were filtered using the following criteria: Peptide threshold = 95%, protein thresholds = 99%, with a minimum of 2 peptides identified. The quantification was based on spectral counting. The quantitative proteomic results were further subjected to statistical analysis.

#### Bioinformatic Analysis

fonc-10-01784 October 1, 2020 Time: 12:27 # 4

For gene ontology analysis, including differential molecular function and biological processes, PANTHER software (Protein Analysis Through Evolutionary Relationships; http://www.pantherdb.org/genes/batchIdSearch.jsp) was used to classify genes into distinct categories of molecular functions and biological processes. Genes were classified into families and subfamilies of shared function, which were then categorized using a highly controlled vocabulary (ontology terms) by the biological process and molecular function. Functional pathways and protein association were examined using the computational bioinformatics program, Pathway Studio software (v.11.4.0.8; Elsevier) as previously described (30, 31). This bioinformatics platform enables the analysis and visualization of altered pathways and protein network association needed to construct and recognize altered cellular processes and involved molecular functional pathways.

The identified proteins were analyzed using Ingenuity Pathway Analysis (IPA) software (Ingenuity <sup>R</sup> Systems, www.ingenuity.com), to find diseases, functions, and networks in which the identified proteins are implicated; the analysis was done for the deregulated proteins with their fold changes. The analysis was also done by PANTHER Software<sup>1</sup> , which was used to identify the molecular processes and biological pathways linked to these proteins.

# Gelatin Zymography Assay

Proteins were extracted from culture supernatant of MEF cells treated with 10X exosomes for 72 h and run along with FBS (positive control) on 0.75 mm SDS-PAGE gel, containing gelatin (Bio-Rad) as a substrate, washed twice and incubated overnight in substrate buffer at 37◦C with shaking. Gels were then stained with coomassie blue (Thermo Scientific) for 1 h at RT, de-stained with ethanol, acetic acid and water solution. Results were detected on ChemiDoc machine. Band staining intensity was determined by densitometry, using ImageJ software (32).

# Tube Formation Assay

Matrigel (Corning, NY, United States) was diluted 1:1 by HUVEC cell medium and added very carefully into a 96-well plate, avoiding bubbles. The plate was placed in the incubator at 37◦C for 30 min to solidify, after which HUVEC cells (15,000 cells/well) were plated on top and treated with either Ctrl-C2C12 derived exosomes or P3F-C2C12-derived exosomes at1X and 10X concentrations then left for 2 h to form tubes. Images were taken by light transmission microscopy and analyzed by ImageJ software (angiogenesis analyzer) as previously performed (6).

### Sphere Formation Assay

p53−/− MEFs and C2C12 cells (2 × 10<sup>3</sup> /well) suspended in MatrigelTM:serum-free media (1:1) were uniformly plated around the bottom rim of 24-well plates with 3% serum-containing media supplemented with 20 ng/mL epidermal growth factor (EGF, R&D Systems, Minnesota, United States) and 2% B27 (Invitrogen, CA, United States). The medium was replenished every 72 h. Micrographs of the spheres formed after 5 days were taken and counted (33).

#### MTT Proliferation Assay

Transduced C2C12 cells (8000 cell/well) were plated in 96 well plates. The following day, they were treated with different concentrations of H2O<sup>2</sup> (Sigma) ranging from 0 to 500 µM. MTT (Roche) was added according to the manufacturer's instructions after 24 h. The absorbance was measured by ELISA reader (at a 595 nm wavelength). Results were computed as the mean percent absorbance of H2O2-treated condition relative to control (untreated). Each experiment was repeated at least 3 times, each performed in triplicate.

# Trypan Blue Exclusion Assay

C2C12 cells were seeded in 24-well plates at a density of 10,000 cells/well. After 5 h of incubation at 37◦C, Ctrl-C2C12-, or P3F-C2C12-derived exosomes (at 10X concentration) were added to the cells. Then 40 h later, cells were treated simultaneously with 60 µM H2O<sup>2</sup> and Ctrl-C2C12 or P3F-C2C12 exosomes at 10X concentration. Cells were counted using trypan blue (Sigma) at 8 and 24 h following H2O<sup>2</sup> treatment. Each experiment was repeated at least 3 times, each performed in triplicate.

### Flow Cytometry for ROS Production Evaluation

Reactive oxygen species levels were evaluated for Ctrl-C2C12 and P3F-C2C12, or C2C12 cells treated with Ctrl-C2C12 derived exosomes or P3F-C2C12-derived exosomes at a 10X concentration. In summary, 30,000 cells were seeded in 6-well plates, and incubated at 37◦C for 48 h. When needed, exosomes were added 5 h post plating. On the day of the experiment, control wells were incubated for 1 h with either H2O<sup>2</sup> (positive control) or 10 mM N-Acetyl-Cysteine (NAC), negative control, which is a ROS inhibitor (Sigma). ROS was detected by incubating cells for 30 min with the cell permanent reagent 2', 7'–dichlorofluorescin diacetate (DCF-DA; 1 mg/ml, Sigma), a fluorogenic dye that measures hydroxyl, peroxyl, and other ROS activity within the cell. After diffusion into the cell, DCF-DA is deacetylated by cellular esterases to a non-fluorescent compound, which is later oxidized by ROS into 2', 7' –dichlorofluorescein (DCF). DCF is a highly fluorescent compound, which can be

<sup>1</sup>http://pantherdb.org

detected by fluorescence spectroscopy with maximum excitation and emission spectra of 495 nm and 529 nm, respectively. Cells were then trypsinized, centrifuged at 1500 rpm for 7 min, and the pellet was resuspended with 300 µL of PBS. Finally, cells were analyzed by the Guava EasyCyte8 flow cytometer.

### Statistical Analysis

fonc-10-01784 October 1, 2020 Time: 12:27 # 5

Statistical comparisons were made between the control and treatment groups using Student's t-test. A p-value < 0.05 was considered to indicate a statistically significant difference. The PCA analysis for LC-MS results was performed by MarkerView 1.1 (AB Sciex, Framingham, United States).

#### RESULTS

### PAX3-FOXO1 Alters Protein Cargo of C2C12 Derived Exosomes

The myogenic background of the murine C2C12 myoblasts is the ideal system to evaluate cellular effects of P3F (18, 34). As previously published, C2C12 cells transduced with P3Fexpressing vector (P3F-C2C12) show features of transformation, including enhanced anchorage-independent growth when compared to cells transduced with empty vector (Ctrl-C2C12) (18). Analysis of this system previously showed that P3F alters the miRNA content of exosomes, leading to over-expression of miR-486-5p, in turn promoting pro-tumorigenic cellular properties (18). However, the effect of possible changes in the protein content of exosomes in response to P3F has not been characterized to date. In this study, we interrogated the proteomic content of C2C12 exosomes in response to P3F expression. We extracted exosomal proteins and subjected them to LC-MS/MS proteomics analysis (**Supplementary Figure 1**). The results showed excellent reproducibility within each cell line (**Figure 1A**). Principal components analysis (PCA) shows that the protein content of 3 independent extractions from Ctrl-C2C12 exosomes clustered together and separately from the 3 independent protein extractions from P3F-C2C12 exosomes, meaning that there is a difference between the protein cargo of Ctrl-C2C12 and P3F-C2C12 exosomes (**Figure 1A**). LC-MS/MS results were validated by western blot for a subset of proteins identified to be common among Ctrl-C2C12 and P3F-C2C12 exosomes, showing excellent concordance of results. Along with the common exosomal protein markers, TSG101, Hsc70, and GAPDH, we detected the presence of Vimentin, Basigin, Fibronectin, and Actin. The negative control Calnexin, a cellular protein that is localized to the endoplasmic reticulum was, as expected, excluded from exosomes (**Figure 1B**).

By comparing the expression levels of the proteins found within P3F-C2C12-derived exosomes to those within Ctrl-C2C12-derived exosomes revealed by LC-MS/MS analysis, we identified 52 proteins differentially expressed; 26 proteins were significantly enriched in P3F-C2C12-derived exosomes among which 7 were exclusively present within these exosomes and 26 were significantly down-regulated proteins among which 13 were excluded from these exosomes (**Table 1**).

#### PAX3-FOXO1 Fusion Protein Modifies the Exosomal Protein Cargo in Favor of Processes and Pathways Implicated in Cancer Progression

Bioinformatics analysis was performed in order to unravel the possible effects that the 52 deregulated proteins can impose on recipient cells, using the two software: Panther classification system and IPA. Panther software highlighted the molecular functions and the biological processes in which these 52 deregulated proteins were implicated (**Figures 2A,B**). "Catalytic activity" and "binding" were the top molecular functions for both enriched and down-regulated proteins (**Figure 2A**). It is important to note that "antioxidant activity" is the only molecular function exclusively found in the exosomes derived from P3F-C2C12 cells with 4% of the enriched proteins implicated (**Figure 2A**). Furthermore, "metabolic process," and "cellular process" were the top biological processes in which both enriched and down-regulated proteins are implicated (**Figure 2B**). "Apoptotic process" appears to be the only biological process eliminated from the exosomes after P3F transduction (**Figure 2B**). These processes underline the implication of the P3F-C2C12-derived exosomes in tumor progression.

Ingenuity Pathway Analysis software was used to assess the canonical pathways implicated and the downstream diseases and functions in which the 26 significantly enriched proteins are involved (**Figures 2C,D**). The enrichment of the identified proteins seem to implicate "14-3-3 mediated signaling," "cell cycle: G2/M DNA damage checkpoint regulation," and "ERK5, VEGF, IGF1, and p70S6K signaling," all of which are important in cell plasticity and cancer progression (**Figure 2C**). Moreover, "cellular movement" and "cell death and survival" appear in the top 10 diseases and functions where these deregulated proteins are implicated (**Figure 2D**). IPA software allows for identification of common networks in which the deregulated proteins may be intersecting. Three networks resulted from this analysis with all three having more than one identified deregulated focus molecule (**Figures 2E–G**). In these networks, we could identify important nodes previously implicated in RMS biology and progression such as: FAK, PKC, MEK, PI3K/AKT/ERK/MAPK, NFkB, TP53, HRAS, as well as proteins involved in cellular signaling and angiogenesis such as VEGF.

#### PAX3-FOXO1-Deregulated Proteins and miRNA in Exosomes Are Involved in Cell Survival and Cellular Plasticity

We had previously characterized the miRNA cargo modulated by P3F in C2C12-derived exosomes (18). We now paired both the deregulated 111 miRNA and the 52 proteins using IPA, and found largely common functions where these molecules are implicated, specifically down-regulating apoptosis, cell death and necrosis, and inducing cell proliferation, invasion, and metastasis (**Figure 3A**). This result is in line with the functional effect of P3F-C2C12 exosomes presented in our previous work (18), where these exosomes were found to promote viability, invasion, and migration of recipient cells. Comparing the

generated networks by IPA from miRNA and proteins separately, we identified 3 common nodes focusing on Insulin, Folliclestimulating hormone (FSH), and TP53. **Figures 3B–D** clarify the upstream and downstream deregulated miRNA and proteins identified within the exosomes that are regulated by or regulate these three nodal proteins.

### PAX3-FOXO1-Modulated Exosomes Induce Cell Survival Pathways, Angiogenesis and Stemness in Recipient Cells

We have previously shown that exosomes isolated from P3F-C2C12 cells were able to induce proliferation, invasion and migration of recipient cells (MEFs and C2C12) compared to cells treated with Ctrl-C2C12-derived exosomes (18). In line with the pathways identified by network analysis in **Figures 2C,E,F**, we verified that treatment with P3F-C2C12-derived exosomes leads to increased phosphorylation (and therefore activation) of ERK and 4-EBP1 (**Figure 4A**), which are essential mediators of cancer cell survival and proliferation (35, 36).

In view of the nodal proteins identified to be involved in cytoskeletal organization and matrix degradation, such as Basigin, Filamin A, and others (**Figure 2E**), we examined the functional effects of the modulated exosomes on collagenase activity using zymography assay. Indeed, P3F-modulated exosomes led to higher MMP-2 activity in treated MEFs, as compared to cells treated with Ctrl-C2C12-derived exosomes evident by the increase in gelatinolytic bands (digested regions; **Figure 4B**).

Since the identified networks also involved angiogenic proteins such as VEGF and TNF (**Figures 2C,G**), we investigated the effect of P3F-C2C12-derived exosomes on angiogenesis by conducting a matrigel tube formation assay that evaluates the ability of HUVECs to differentiate into capillary-like structures when plated on matrigel (6). Counting the nodes, which are defined as the intersection of 3 or more branches of tubular connections, the total length which is the sum of length of segments, isolated elements and branches in the analyzed area and the total mesh area, which represents the area occupied by the primary capillary-like tubes, we found a significant increase when HUVECs were treated with P3F-C2C12-derived exosomes compared to control, at 10X but not 1X concentration (**Figure 4C**).

Exosomes can reprogram transcription within recipient cancer cells allowing their dedifferentiation into CSC with sphere-forming capacities (37). To examine whether P3Fmodulated exosomes would enhance their ability to confer stem-like properties to recipient cells, we examined their effects on sphere-forming capacity of MEFs and C2C12 cells. We used p53−/− MEFs for this assay, as wild-type MEFs were not prone to forming spheres in culture. For both p53−/− MEFs and C2C12 cells, we found that treatment with P3F-modulated exosomes led to a significant increase in the number of formed spheres 5 days post treatment, relative to the control-derived exosomes (**Figures 4D,E**).

Expression level Exo-P3F versus Exo-Ctrl p-value Enriched proteins 1 Leukocyte elastase inhibitor (Serpinb1a) Infinity 6.3.E-06 2 Lymphocyte antigen 6C1 (Ly6c1) Infinity 7.5.E-04 3 Lymphocyte antigen 6D (Ly6d) Infinity 2.6.E-03 4 Histone H1.5 (Hist1h1b) Infinity 7.0.E-03 5 Calreticulin (Calr) Infinity 8.9.E-03 6 Retinoic acid-induced protein 3 (Gprc5a) Infinity 2.8.E-02 7 Rho GDP-dissociation inhibitor 2 (Arhgdib) Infinity 3.7.E-02 8 Prostaglandin reductase 1 (Ptgr1) 28.2 2.9.E-03 9 Embigin (Emb) 15.47 2.0.E-02 10 Basigin (Bsg) 5.88 3.3.E-02 11 Hsc70-interacting protein (St13) 3.74 2.3.E-02 12 Glutathione S-transferase P 1 (Gstp1) 3.65 2.2.E-02 13 Nucleoside diphosphate kinase B (Nme2) 3.56 3.8.E-02 14 Annexin A1 (Anxa1) 3.49 4.6.E-03 15 Superoxide dismutase (Sod1) 3.45 3.4.E-04 16 Monocarboxylate transporter 1 (Slc16a1) 3.41 2.3.E-02 17 Granulins (Grn) 2.94 3.3.E-02 18 Fatty acid-binding protein, epidermal (Fabp5) 2.76 7.4.E-03 19 Alpha-actinin-1 (Actn1) 2.68 1.7.E-02 20 14-3-3 protein zeta/delta (Ywhaz) 2.2 9.1.E-04 21 Vimentin (Vim) 2.18 3.5.E-02 22 CD82 antigen (Cd82) 2.06 9.7.E-03 23 Major vault protein (Mvp) 2.01 9.0.E-03 24 Malate dehydrogenase, cytoplasmic (Mdh1) 1.84 3.9.E-02 25 14-3-3 protein epsilon (Ywhae) 1.82 4.7.E-02 26 Sodium/potassium-transporting ATPase subunit alpha-1 (Atp1a1) 1.7 3.2.E-02 Down-regulated proteins 1 Alpha-enolase (Eno1) 0.67 4.5.E-02 2 Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) 0.32 1.5.E-02 3 Adenosylhomocysteinase (Ahcy) 0.28 4.8.E-02 4 T-complex protein 1 subunit theta (Cct8) 0.23 4.9.E-02 5 T-complex protein 1 subunit zeta (Cct6a) 0.21 2.7.E-02 6 Integrin alpha-5 (Itga5) 0.2 2.2.E-02 7 Serine protease HTRA1 (Htra1) 0.19 4.0.E-02 8 Myosin-9 (Myh9) 0.18 2.8.E-02 9 Voltage-dependent calcium channel subunit alpha-2/delta-1 (Cacna2d1) 0.14 3.9.E-02 10 Transitional endoplasmic reticulum ATPase (Vcp) 0.13 2.8.E-02 11 Talin-1 (Tln1) 0.13 5.7.E-03 12 Peptidyl-prolyl cis-trans isomerase C (Ppic) 0.08 7.2.E-03 13 Filamin-A (Flna) 0.02 1.0.E-03 14 Sorcin (Sri) 0 4.3.E-06 15 Serine protease 23 (Prss23) 0 4.7.E-04 16 Ras-related protein Rab-5C (Rab5c) 0 4.7.E-04 17 N(G),N(G)-dimethylarginine dimethylaminohydrolase 2 (Ddah2) 0 9.5.E-04 18 Mimecan (Ogn) 0 4.7.E-03 19 Band 4.1-like protein 2 (Epb41l2) 0 1.1.E-02 20 Ras-related protein R-Ras2 (Rras2) 0 1.4.E-02 21 EGF-like repeat and discoidin I-like domain-containing protein 3 (Edil3) 0 2.4.E-02 22 Inter-alpha-trypsin inhibitor heavy chain H3 (Itih3) 0 2.4.E-02 23 DnaJ homolog subfamily A member 2 (Dnaja2) 0 2.9.E-02 24 LIM and senescent cell antigen-like-containing domain protein 1 (Lims1) 0 3.4.E-02 25 Large neutral amino acids transporter small subunit 1 (Slc7a5) 0 3.4.E-02 26 Integrin-linked protein kinase (Ilk) 0 3.8.E-02

TABLE 1 | List of the 26 significantly enriched and 26 significantly down-regulated proteins in P3F-C2C12-derived exosomes compared to those in Ctrl-C2C12-derived exosomes with their fold changes and p-value.

The expression levels presented are the ratio of the level of the indicated proteins found by LC-MS/MS within the P3F-C2C12 exosomes compared to Ctrl-C2C12 exosomes.

presenting the molecular functions (A), and the biological processes (B) for the 26 enriched and 26 down-regulated proteins. (C,D) histograms from IPA analysis of the 26 enriched proteins showing the top 10 canonical pathways (C) and the top 10 diseases and functions (D). (E–G) Generated networks by IPA analysis of the deregulated proteins in C2C12-P3F-derived exosomes compared to Ctrl-C2C12-derived exosomes. Proteins colored in red are enriched and those colored in green are down-regulated. The higher the level of enrichment or down-regulation, the more intense the color is presented. Edges (lines and arrows between nodes) represent direct (solid lines) and indirect (dashed lines) interactions between molecules as supported by information in the Ingenuity knowledge base.

These results highlight the ability of the P3F fusion protein to promote cellular plasticity and tumorigenic aspects through exosomes.

#### PAX3-FOXO1 Fusion Protein Is Implicated in Redox Homeostasis and Tolerance/Protection From Oxidative Stress by Decreasing ROS Levels in Both Transduced and Recipient Cells

Of the 26 enriched proteins in P3F-C2C12-derived exosomes, we found that 4 proteins are implicated in the ROS pathway. These proteins include prostaglandin reductase 1 (PtgR1), which has an oxidoreductase activity that can reduce cytotoxic unsaturated aldehydes and ketones produced by lipid peroxidation during oxidative stress (38); glutathione S-transferase (GST), which detoxifies secondary metabolites produced by interaction between ROS and cellular components (39); superoxide dismutase (SOD1), which is an antioxidant enzyme that reduces superoxide (O<sup>−</sup> 2 ) and prevents its accumulation (40); and malate dehydrogenase, which is activated by ROS and seems to act as a redox regulator in response to environmental changes (41). These 4 enzymes mainly act by decreasing ROS levels. Since ROS contributes to oxidative stress, we hypothesized that the P3F fusion protein may be influencing redox homeostasis and response to oxidative stress. Indeed, the expression levels of the 3 main subunits of the NADPH oxidase complex, which is a main enzymatic producer of ROS: gp91, gp47, and Nox1 were decreased in P3F-transduced C2C12 cells compared to control cells (**Figure 5A**). Several bands appear for Nox1 protein due to different glycosylation sites (42). This result was confirmed by flow cytometry analysis using the DCF-DA dye to detect ROS levels, where we found that P3F-C2C12 cells have markedly decreased

proportion of DCF-positive cells as compared to Ctrl-C2C12 cells, demonstrating lower ROS levels in response to P3F expression (**Figure 5B**).

Next, we subjected the transduced cells to oxidative stress in order to test the protective effect of the fusion protein. After treating the cells with different concentrations of H2O<sup>2</sup> (0 to 500 µM) for 24 h, P3F-C2C12 cells showed a higher IC<sup>50</sup> (68 µM) than Ctrl-C2C12 cells (36.8 µM; **Figure 5C**). Thus, P3F-C2C12 cells are more tolerant to endogenous and exogenous oxidative stress than control cells. These results show that the PAX3-FOXO1 fusion protein can decrease ROS levels in host cells, likely contributing to cell survival.

Since P3F-C2C12 cells were found to be tolerant to H2O2 induced oxidative stress, we next investigated whether this resistance is provided to recipient cells through exosome cargo. Treating C2C12 myoblasts with a combination of H2O<sup>2</sup> and P3F-C2C12-derived exosomes at 10X concentration showed significantly higher viability in comparison to cells treated with H2O<sup>2</sup> and Ctrl-C2C12-derived exosomes (1.88 versus 1.05 at 8 h; and 1.27 versus 0.88 at 24 h**; Figures 6A,B**). Interestingly, this protective effect was also observed when treating human ERMS JR1 cells with a combination of H2O<sup>2</sup> and ARMS Rh30-derived exosomes showing a significantly higher viability when compared to cells treated with H2O<sup>2</sup> and JR1-derived exosomes (**Figure 6C**). This confirms a protection conferred by FPRMS-derived exosomes to recipient cells in conditions of oxidative stress.

To clarify whether this protection was related to decreased levels of ROS in cells treated with P3F-modulated exosomes as opposed to control exosomes, we pretreated C2C12 cells with exosomes for 48 h, and then exposed them to H2O2, and evaluated ROS levels by DCF-DA.

FIGURE 4 | C2C12-P3F-derived exosomes promote angiogenesis and induce stemness of recipient cells. (A) Western blot for the indicated proteins in MEFs treated with Exo-free (EF) media, 10X Ctrl-C2C12-derived exosomes, and 10X P3F-C2C12-derived exosomes. GAPDH is used as a loading control. Histograms presenting the levels of phosphorylation of ERK and 4EBP1 compared to Exo-free condition, quantified by ImageJ software from n = 3 different western blots. (B) Zymography image of digested regions showing MMP2 activity in MEFs treated with either Ctrl-C2C12-derived exosomes or P3F-C2C12-derived exosomes for 72 h. Relative MMP-2 activity from 3 independent experiments was quantified by ImageJ software and presented by a histogram. (C) Representative phase contrast photomicrographs of endothelial tube formation of HUVECs cultured on Matrigel with either Exo-Free medium (EF), 1X, or 10X C2C12-MSCV-derived exosomes or C2C12-P3F-derived exosomes, as specified. Quantitation of the total number of nodes, total length and total mesh area of the different conditions are shown in the histogram, which represent the mean of 3 independent experiments, each performed in triplicate. (D,E) Representative phase contrast photomicrographs and histograms showing the ratio of the spheres formed by p53−/− MEFs (D) or C2C12 cells (E) treated with either Exo-Free medium (EF) or 10X C2C12-MSCV-derived exosomes or C2C12-P3F-derived exosomes, as specified. All data are reported as means of at least three independent experiments ± SD. Asterisks (\*) denote a statistically significant difference (p-value < 0.05).

numbers shown are average of three independent experiments. (C) MTT assay determining the IC50 of Ctrl-C2C12 and P3F-C2C12 cells treated with ascending concentrations of H2O<sup>2</sup> for 24 h. Data are reported as mean of three independent experiments ± SD. Asterisks denote a statistically significant difference (p-value < 0.05); (\*) compared to control Exo-free treated cells.

Treatment with exosomes, whether control or PAX3- FOXO1-modulated, both had an inhibitory effect on ROS production (**Figure 6D**). The induction of ROS observed by treating the recipient cells with H2O<sup>2</sup> was significantly decreased when co-treated with either exosome type, but to a much higher extent when treated with P3F-C2C12-derived exosomes. Indeed, P3F-C2C12-derived exosomes were able to decrease ROS levels by 6.6 fold whereas Ctrl-C2C12-derived exosomes decreased it only by 2.4 fold when compared to H2O<sup>2</sup> treatment alone (**Figure 6D**). These observations define a clear role of P3F-modulated exosomes in protecting recipient cells, even in the presence of oxidative stress, from elevated intracellular ROS levels.

# DISCUSSION

Rhabdomyosarcoma is an aggressive pediatric cancer with limited improvement in overall survival outcomes in patients with aggressive subtypes and advanced stages (1). Presence of the P3F or the less common PAX7-FOXO1 fusion oncoprotein is associated with increased incidence of metastasis and disease aggressiveness (43, 44). There is therefore a pressing need for identifying effective therapeutic targets and biomarkers for these fusion gene positive (FP) RMS.

Exosomes can act via paracrine signaling on neighboring stromal cells altering their behavior and enhancing tumor growth (9, 15, 45). Both FPRMS and FNRMS cells secrete quantifiable amounts of exosomes with specific cargo that can influence

cancer-related processes including cell motility and angiogenesis (6), and fusion status plays a role in modulating exosome cargo to favor tumor invasion (6, 12, 18).

In this study, we investigated the effects of P3F on myoblasts using C2C12 cells, and studied its role in modulation of exosome protein content and paracrine signaling. Both the previously

identified deregulated miRNA revealed by microarray profiling (18) and the deregulated exosomal proteins identified in this study were associated with pathways important in cell plasticity including cellular motility, cell survival, and protection from oxidative stress. The networks identified centered on insulin, FSH and TP53, all of which have previously been associated with the expression of P3F or implicated in RMS tumor progression. For instance, both insulin, and insulin-like growth factors (IGF) regulate skeletal muscle differentiation, and their pathways are deregulated in RMS (46, 47). Using C2C12 myoblasts, Wang et al. showed that P3F and IGF2 cooperate to induce cell proliferation and invasion (48). Additionally, P3F has been shown to up-regulate the expression of IGF1 receptor and IGF2, both contributing to FPRMS proliferation (49). P3F also induces aberrant and elevated expression of GLUT4, enhancing cellular response to insulin (50). As for FSH, its receptors are expressed in both FPRMS and FNRMS human cell lines, as well as in primary tumor samples. Stimulation of RMS cells with FSH results in enhanced proliferation and chemotaxis, indicating a role in promoting RMS progression (51). FSH down-regulates FOXO1 mRNA levels, and FSH receptor activation results in FOXO1 phosphorylation, which inhibits FOXO1-mediated repression of apoptosis and promotes cell survival (52, 53). However, whether P3F contributes to FSH-mediated repression of wild type FOXO1 is yet to be determined. Finally, p53 pathway deregulation has been observed in RMS and contributes to tumor progression and relapse (54–56). While p53 pathway deregulation in FNRMS is mainly via inactivation mutations in TP53, in FPRMS its deregulation is likely mediated by the fusion oncoprotein, as PAX proteins have been shown to transcriptionally inhibit p53 (57). Moreover, TP53 loss markedly increases the progression of P3F-driven tumors in mice (58). Indeed, 85% of human tumors harboring P3F fusion were associated with p53 inactivation, suggesting a cooperativity between the fusion protein and p53 pathway deregulation in driving tumorigenesis (59). In concordance, our results further reveal that P3F expression leads to exosomal-mediated deregulation of these signaling pathways, likely contributing to their deregulation in recipient cells.

Functionally, exosomes of P3F-transduced cells modulated recipient cell plasticity to favor proliferation, invasion, angiogenesis and stemness, as well as protect them from the hostile conditions of the tumor microenvironment (TME). Angiogenesis and neovascularization are indicative of cell plasticity that allows tumor growth and cell survival in hypoxic conditions of the TME (19, 60). Ploeger et al. showed that fibroblasts exhibit a high dynamic plasticity in wound healing resulting from a combination of recruitment and cell proliferation, a plasticity that seems to be regulated by the micro-environment (61). P3F-modulated exosomes were able to induce tube formation and migration of recipient HUVECs, as well as induce degradation of matrix via activation of MMPs, indicating a role for these exosomes in inducing angiogenesis and alteration of the stromal matrix.

Escaping cell death is another indication of cellular plasticity in response to change in physiological conditions (21, 62). Pathway analysis by IPA of the proteins and miRNA modulated in exosomes by P3F showed a common impact on reducing apoptosis and increasing cell viability (see **Figure 3A**: 10 out of 25 pathways are implicated in death pathways). This is further evidenced by the observed resistance of cells treated with P3Fmodulated exosomes to apoptosis in response to exogenous oxidative stress, which may be impacted by both modulation of oxidative stress pathway, as well as resistance to apoptosis in general. Of note, elevated ROS levels are detected in almost all cancers, due at least in part to the high metabolic activity and mitochondrial dysfunction of cancer cells. In comparison, normal cells exhibit moderate levels of ROS, enough for them to act as second messengers in signal transduction, and contribute to physiological processes within these cells (63). To cope with the ever-present state of oxidative stress, cancer cells upregulate their antioxidant pathways to escape cell death (64). For instance, in RMS, elevated ROS levels are compensated by an up-regulation of ROS-scavenging pathways, thus conferring resistance to ROS-induced stress (65). Moreover, up-regulation of ROS production in RMS cells following radiation therapy was rapidly compensated by an up-regulation of nuclear factor erythroid 2-related factor (NRF2) levels which, in turn, promotes the expression of antioxidant enzymes and miRNA that protected RMS cells from ROS-induced DNA damage (66). This has been suggested as a possible Achilles' heel of cancer cells, as they may be susceptible to treatments that can alter their redox homeostasis, either by further increasing intracellular ROS levels or by targeting antioxidant abilities (64, 67). RMS might be particularly vulnerable to this pathway, as demonstrated by Chen et al. using a high-throughput screen. They reasoned that this sensitivity may be especially applicable to RMS, as it derives from a skeletal muscle cell lineage, since skeletal muscle cells have unique metabolic properties including a high aerobic activity and a robust antioxidant defense system against excessive ROS, which make them particularly sensitive to redox-altering agents (55). In fact, analysis of RMS samples revealed oxidative stressinduced genomic mutations that enhance RMS cell survival against oxidative damage whereas combinational targeting of the major antioxidative pathways up-regulated in RMS, the thioredoxin (TRX) and glutathione (GSH) synthesis pathways, using GSH-depleting agents and TRX-reductase inhibitors seems to induce cancer cell death (55, 64). Therapeutic interventions using agents that can increase ROS production also offer promising therapeutic results. For instance, carfilzomib and alvocidib, in addition to synthetic statins, where shown to act in synergism with HDAC inhibitors by increasing oxidative stress showing effective activity against ERMS xenografts (55). Erastin, another GSH-depleting agent that subsequently up-regulates ROS production, also induces RMS cell death (67). Notably, following ROS-inducing radiation therapy, FPRMS cells were found to produce higher levels of antioxidant miRNA including miR-22, miR-210, and miR-375 compared to FNRMS, where the expression of miR-375 was only observed in FPRMS, and certain antioxidant enzymes including catalase and glutathione peroxidase 4 (66). Moreover, Martin et al. demonstrated that FPRMS treatment with fenretinide induces ROS production accompanied by a reduction in P3F transcriptional activity. ROS-induction in FPRMS using Triterpenoid down-regulated the expression of transcription factors including P3F suggesting

the need to determine the link between oxidative stress and P3F in RMS (68, 69). Altogether, RMS cells have been shown to modulate intracellular signaling pathways to maintain redox homeostasis, an aspect of cellular plasticity that protects against oxidative damage and highlights the importance of using redoxbased therapeutic strategies in combination with conventional chemotherapeutic treatments against RMS. Our results further underline this finding, and specifically implicate the P3F protein in modulating oxidative stress response, not only in an autocrine manner in RMS cells, but also through paracrine effects on stromal cells via exosomes. We observed lower ROS levels within the P3F transduced C2C12 cells, with enhanced survival upon treatment with H2O2. We identified an enrichment of 4 REDOX proteins in exosomes in response to P3F that may be contributing to the observed decrease in intracellular ROS levels in recipient cells, confirming paracrine protective effects of the fusion oncoprotein via exosomes. Indeed, previous studies have implicated exosomes in modulating response to oxidative stress and conferring antioxidant resistance in several settings (27, 70), and our results now demonstrate the additional effect of oncoproteins such as P3F in further augmenting this response.

Altogether, our results indicate that the P3F oncoprotein down-regulates ROS levels, exerting a protective effect against oxidative stress-induced cell death in an autocrine manner. This effect can also be transferred to stromal and other cells through secreted exosomes that also affect recipient cell invasion, migration, and pro-angiogenic properties. We have shown that the P3F fusion protein is able to modulate the content of exosomes secreted by fusion-positive cells and that these exosomes seem to confer cellular plasticity and invasive properties, including augmentation of pro-angiogenic signaling. Detailed understanding of the pathways through which P3F alters the metabolic response of cells may uncover novel therapeutic targets for enhancing cell death in response to metabolic stress in this setting, and should be the focus of future studies.

# REFERENCES


# DATA AVAILABILITY STATEMENT

The original contributions presented in the study are publicly available. This data can be found here: ProteomeXchange Consortium via the PRIDE partner repository (Accession Number: PXD017543).

#### AUTHOR CONTRIBUTIONS

AF, FG, AA, and FA conducted most of the experiments and data analysis. AF, FR, and FG helped in drafting the manuscript. FK, YM, JZ, and RZ performed and analyzed the LC-MS/MS experiments. NH helped in the design of the flow cytometry experiments and their analysis. SG and RS oversaw the design and coordination of the studies, and drafted the manuscript. All authors read and approved the final manuscript.

# FUNDING

This work was funded by an MPP grant from the American University of Beirut Medical Center, Faculty of Medicine and with support from the Lebanese University.

#### ACKNOWLEDGMENTS

The authors thank shared core facilities at the American University of Beirut Faculty of Medicine (AUBFM).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc. 2020.01784/full#supplementary-material

of alterations affecting a common genetic axis in fusion-positive and fusionnegative tumors. Cancer Discov. (2014) 4:216–31. doi: 10.1158/2159-8290.CD-13-0639




**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Fahs, Ramadan, Ghamloush, Ayoub, Ahmad, Kobeissy, Mechref, Zhao, Zhu, Hussein, Saab and Ghayad. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.