Targeting the lncRNA DUXAP8/miR-29a/PIK3CA Network Restores Doxorubicin Chemosensitivity via PI3K-AKT-mTOR Signaling and Synergizes With Inotuzumab Ozogamicin in Chemotherapy-Resistant B-Cell Acute Lymphoblastic Leukemia

Purpose This study aimed to determine the expression profiles of long non-coding RNA (lncRNA), microRNA (miRNA), and mRNA in chemotherapy-resistant B-cell acute lymphoblastic leukemia (B-ALL). Methods LncRNA, miRNA, and mRNA profiles were assessed by RNA-seq in diagnostic bone marrow samples from 6 chemotherapy-resistant and 6 chemotherapy-sensitive B-ALL patients. The lncRNA DUXAP8/miR-29a/PIK3CA signaling network was identified as the most dysregulated in chemoresistant patient samples, and its effect on cellular phenotypes, PI3K-AKT-mTOR signaling, and chemosensitivity of doxorubicin (Dox)-resistant Nalm-6 (N6/ADR), and Dox-resistant 697 (697/ADR) cells were assessed. Furthermore, its synergy with inotuzumab ozogamicin treatment was investigated. Results 1,338 lncRNAs, 75 miRNAs, and 1620 mRNAs were found to be dysregulated in chemotherapy-resistant B-ALL in comparison to chemotherapy-sensitive B-ALL patient samples. Through bioinformatics analyses and RT-qPCR validation, the lncRNA DUXAP8/miR-29a/PIK3CA network and PI3K-AKT-mTOR signaling were identified as significantly associated with B-ALL chemotherapy resistance. In N6/ADR and 697/ADR cells, LncRNA DUXAP8 overexpression and PIK3CA overexpression induced proliferation and inhibited apoptosis, and their respective knockdowns inhibited proliferation, facilitated apoptosis, and restored Dox chemosensitivity. MiR-29a was shown to affect the lncRNA DUXAP8/PIK3CA network, and luciferase reporter gene assay showed direct binding between lncRNA DUXAP8 and miR-29a, as well as between miR-29a and PIK3CA. Targeting lncRNA DUXAP8/miR-29a/PIK3CA network synergized with inotuzumab ozogamicin’s effect on N6/ADR and 697/ADR cells. Conclusion Targeting the lncRNA DUXAP8/miR-29a/PIK3CA network not only induced an apoptotic effect on Dox-resistant B-ALL and restored Dox chemosensitivity via PI3K-AKT-mTOR signaling but also showed synergism with inotuzumab ozogamicin treatment.


INTRODUCTION
Acute lymphoblastic leukemia (ALL), as a hematological malignancy with the unrestricted proliferation of abnormal, immature lymphocytes or their progenitors that cause bone marrow element dysregulation, affects 1.7 new cases per 100,000 populations annually and accounts for nearly 2% of all lymphoid neoplasms in the United States (1,2). ALL mainly occurs in children but less in adults, with the predominant type of B-cell ALL (B-ALL), which accounts for approximately 85% of total ALL cases (1,2). Along with the improvement of treatment strategies, precision medicine, drug innovations, and so on, the prognosis of ALL is much improved in recent decades (3)(4)(5)(6). However, long-term survival benefits are only achieved in the majority of childhood ALL, while only 30%-40% of adult ALL achieves long-term disease-free survival; especially, after relapse or refractoriness, the role of salvage chemotherapy is very limited, with less than 10% of long survival (7,8). Therefore, the efforts to uncover underlying pathogenesis then discover novel treatment targets for ALL are never ceased.
Long non-coding RNA (lncRNA), a kind of non-coding RNA recognized recently with a length of more than 200 bp, has been reported to regulate a huge amount of biological processes and to be involved in the pathogenesis of most diseases via its interaction with cellular compounds such as miRNA, mRNA, DNA, and proteins (9)(10)(11)(12)(13). Besides, lncRNA has been also observed to be deeply involved in the development, progression, and treatment response of hematological malignancies (14)(15)(16). In the aspect of B-ALL, a recent comprehensive bioinformatics analysis identifies 1,235 dysregulated lncRNAs engaged in B-ALL subtype classification, and 942 aberrant lncRNAs related to B-ALL relapse (17). In addition, some specific lncRNAs were identified to regulate B-ALL growth and apoptosis, including lncRNA TEX41, lncRNA CRNDE, and lncRNA TCL6 (18)(19)(20). However, the function of a large majority of lncRNAs in B-ALL remains obscure, and in particular, seldom studies have investigated the comprehensive lncRNA profile involved in treatment refractoriness, not to mention the detailed molecule mechanism of some key lncRNAs underlying chemotherapy resistance.
Therefore, the current study aimed to investigate lncRNA, miRNA, and mRNA expression profiles related to B-ALL chemotherapy resistance via RNA sequencing, then further to explore the effect and interaction of the lncRNA DUXAP8/miR-29a/PIK3CA network on drug-resistant B-ALL cell proliferation, apoptosis, and chemosensitivity as well as its synergy with inotuzumab ozogamicin.

METHODS Samples
A total of 12 diagnostic bone marrow samples were collected from adult patients with B-ALL treated at our hospital. All bone marrows were sampled before the initiation of induction therapy. Of 12 bone marrow samples, 6 samples were from the adult B-ALL patients who achieved complete remission (CR) following the induction therapy with the VDP (vincristine, doxorubicin, prednisone) regimen, and these patients were considered as chemotherapy-sensitive patients (S-ALL) in the study; then, another 6 samples were from the adult B-ALL patients failing to achieve response at the end of the induction therapy with the VDP regimen, who exhibited a minimal residual disease (MRD) >1 × 10 -2 by flow cytometry at the end of the induction therapy, and these patients were considered as chemotherapy-resistant patients (R-ALL) in the study.
The inclusion criteria of patients were as follows: (1) newly diagnosed as B-ALL by ESMO criteria (21); (2) older than 18 years; (3) received VDP regimen induction treatment; (4) had available induction treatment response data; and (5) had accessible bone marrow samples before treatment. The exclusion criteria of patients were as follows: (1) other types of ALL; (2) complicated with or history of other hematological malignancies or solid tumors; and (3) history of chemotherapy. This study was approved by the Ethics Committee of our hospital. The written informed consents were offered by patients.
Workflow of RNA Sequencing Including lncRNA, mRNA, and MicroRNA Total RNA in R-ALL and S-ALL bone marrow samples was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and quantified with Agilent 2100 (Agilent, Santa Clara, CA, USA). The ribosomal RNA was depleted using the Ribo-Zero rRNA Removal Kit (Illumina, San Diego, CA, USA), and the library of mRNA and lncRNA was constructed using the Illumina TruSeq Stranded Total RNA Library Prep Kit (Illumina, USA) according to the kit's protocol. For the construction of the miRNA library, a miRNA library kit (Qiagen, Hilden, Germany) was applied. HiSeq 2500 (Illumina, USA) was applied to complete the sequencing of lncRNA, mRNA, and miRNA, respectively. After the acquisition of data, the R project (Version 3.6.3) was applied to complete quantile normalization, data processing, bioinformatics analysis, and graph plotting. Briefly, the Factoextra package and Pheatmap package were acquired to perform principal component analysis (PCA) and heatmap analysis, respectively; the DESeq2 and Limma package was adopted to analyze the mean gene expression, fold change (FC), and differentially expressed lncRNA, mRNA, or miRNA (DElncRNA, DEmRNA, or DEmiRNA). The DElncRNAs, DEmRNAs, and DEmiRNAs were shown with a volcano plot. Gene Ontology (GO) and Kyoko Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted via the DAVID web server. Pearson's correlation coefficient was used to calculate the correlation between the expression of DElncRNAs and DEmRNAs. The potential target of DEmiRNAs was investigated by miRanda. The competing endogenous RNA (ceRNA) network was displayed using the graph in R packages, which contained the following ceRNAs: (1) top 10 deregulated (5 upregulated and 5 downregulated) DElncRNAs according to log 2 FC; (2) DEmRNAs, which have a significant correlation with the DElncRNAs mentioned in (1) (Pearson's correlation coefficient >0.9 or <-0.9); and (3) DEmiRNAs, whose potential targets were DEmRNAs and DElncRNAs mentioned in (1) and (2). The expressions of the selected lncRNA, miRNA, and mRNA were evaluated by reverse transcription-quantitative polymerase chain reaction (RT-qPCR).
The plasmids were transfected into N6/ADR or 697/ADR cells in the presence of Lipofectamine ™ 2000 (Invitrogen, USA). After transfection, the cells were divided into oeCTL, oeLNC, shCTL, and shLNC groups. The N6/ADR and 697/ADR cells that were not transfected served as blank control.

PIK3CA Plasmid Transfection
The pEX-2 vector and pGCMV/miR vector (GenePharma, China) were applied to construct the control mRNA overexpression, PIK3CA overexpression, control miR overexpression, and miR-29a overexpression plasmids by Shanghai GenePharma Co., Ltd. (Shanghai, China). To perform the co-transfection of control mRNA overexpression plasmid, PIK3CA overexpression plasmid, control miR overexpression plasmid, and miR-29a overexpression plasmid, Lipofectamine ™ 2000 was adopted with the manufacturer's instruction followed. The following groups were generated after transfection: (a) blank group, cells without transfection; (b) oeCTL group, cells were transfected with control mRNA overexpression plasmid and control miR overexpression plasmid; (c) oeMIR group, cells were transfected with miR-29a overexpression plasmid and control mRNA overexpression plasmid; (d) oePIK3CA group, cells were transfected with control miR overexpression plasmid and PIK3CA overexpression plasmid; and (e) oeMIR + oePIK3CA group, cells were transfected with miR-29a overexpression plasmid and PIK3CA overexpression plasmid.

PI3K-AKT-mTOR Pathway Activation
The PIK3CA overexpression plasmid or mTOR activator (MHY1485) (2 mM, MCE, China) was cocultured with lncRNA DUXAP8 knockdown plasmid in N6/ADR and 697/ADR cells, followed by the detection of cell viability via CCK-8 assay, to explore the role of the PI3K-AKT-mTOR pathway in the restored chemosensitivity by lncRNA DUXAP8 knockdown.

Cell Proliferation and Cell Apoptosis
The Cell Counting Kit-8 (Beyotime, China) was applied to assess cell proliferation. 3 × 10 3 cells were loaded into a 96-well plate. 10 ml reagent was added and incubated with cells for 2 h after transfection. A microplate reader was used to read optical density (OD) values. The TUNEL Apoptosis Assay Kit (Beyotime, China) was used to assess the cell apoptosis rate according to the instructions, with 2.5 × 10 4 cells seeded into a 24-well plate.

Chemosensitivity to Dox Assay
For the confirmation of Dox resistance, un-transfected Nalm-6, N6/ADR, 697, and 697/ADR were seeded in a 96-well plate with the number of 1 × 10 4 and incubated with a range concentration (0, 5, 10, 20, 40, and 80 nM) of Dox for 24 h referring to a previous paper (23). For chemosensitivity detection of transfected cells, the cells were cultured with a range concentration (0, 5, 20, 80 nM) of Dox for 24 h. After the incubation, the cell viability was detected with Cell Counting Kit-8 as described in Cell Proliferation.

Luciferase Reporter Gene Assay
To detect the binding of lncRNA DUXAP8 and miR-29a, the wild type and mutant type of lncRNA DUXAP8 were cloned into the pGL6 vector (Beyotime, China). The wild-type and mutanttype lncRNA DUXAP8 plasmids, miR-29a mimic, and control mimic were co-transfected into 293T cells (ATCC, USA) by Lipofectamine ™ 2000. After 48 h of incubation, the cells were lysed and the fluorescence value was detected using Luciferase Reporter Gene Assay Kit (Beyotime, China). To detect the binding of PIK3CA and miR-29a, the PIK3CA wild-type and mutant-type plasmids were constructed with a pGL6 vector. With the application of Lipofectamine ™ 2000, the PIK3CA wildtype plasmid, PIK3CA mutant-type plasmid, miR-29a mimic, and control mimic were co-transfected into 293T cells. At 48 h, the fluorescence value was detected as mentioned above.

LncRNA DUXAP8 Modification and Inotuzumab Ozogamicin (CMC-544) Treatment
The control lncRNA overexpression plasmid, lncRNA DUXAP8 overexpression plasmid, control lncRNA shRNA plasmid, and lncRNA DUXAP8 shRNA plasmid were transfected into N6/ ADR and 697/ADR cells according to the methods described in the LncRNA DUXAP8 Plasmid Transfection. The N6/ADR and 697/ADR cells that were not transfected served as blank control. Then, CMC-544 (Pfizer Inc., New York, NY, USA) was added into cells with various concentrations of 0, 0.5, 1, 2, 4, and 8 ng/ ml (24). After incubation for 48 h, cell viability was evaluated with Cell Counting Kit-8, and the half-maximal inhibitory concentration (IC 50 ) was calculated using Probit regression analysis by SPSS 23.0 (IBM, Armonk, NY, USA). At last, the transfected and blank control cells were incubated with CMC-544 at an amount of 2.670 or 2.794 ng/ml for 48 h, with the apoptosis rate determined by TUNEL Apoptosis Assay Kit. The cell viability and apoptosis detection were carried out as described in Cell Proliferation and Cell Apoptosis.

Statistical Analysis
All data in this study were expressed as mean ± standard deviation. The difference among groups and between two groups was assessed by one-way ANOVA followed by Dunnett's test or t-test. GraphPad Prism 7.02 (GraphPad Software Inc., La Jolla, CA, USA) was applied to analyze data and make graphs. The statistical significance was defined as p < 0.05. No significance: NS (p>0.05); *p<0.05, **p<0.01, ***p<0.001.

Study Flow
The current study consisted of two main parts: the clinical part and the experimental part (Supplementary Figure 1). In the clinical part, 6 R-ALL patients and S-ALL patients were enrolled to detect the lncRNA, miRNA, and mRNA profiles via RNA-seq and RT-qPCR validation to sort key networks related to chemoresistance. Then, experiments were performed via modifying lncRNA DUXAP8, miR-29a, and PIK3CA to explore their effect (alone or in combination) on Dox-resistant B-ALL cell proliferation, apoptosis, Dox chemosensitivity, etc.

Characteristics
The enrolled B-ALL patients were aged 40 ± 9 years, with 50% males and 50% females. Meanwhile, the mean WBC was 59 ± 45 × 10 9 /l, and the mean bone marrow blast percentage was (83 ± 10)%. The detailed information of their characteristics is shown in Table 1. It was notable that by comparison, S-ALL patients showed a trend of lower WBC (p = 0.064) and CD19 percentage (p = 0.065) compared to R-ALL patients but lacked statistical significance. Furthermore, other characteristics were of no difference between them ( Table 1).

LncRNA, miRNA, and mRNA Expression Profiles
The lncRNA expression profile could clearly distinguish R-ALL patients from S-ALL patients via PCA plot analysis ( Figure 1A) and showed a good inter-consistent trend in R-ALL patients and S-ALL patients, respectively, by heatmap analysis ( Figure 1B). Then, a total of 759 upregulated and 579 downregulated lncRNAs were identified in R-ALL patients compared with S-ALL patients ( Figure 1C).
Besides, the miRNA expression profile also distinguished R-ALL patients from S-ALL patients via PCA plot analysis ( Figure 1D) and exhibited an acceptable inter-consistent trend by heatmap analysis (Figure 1E), with 46 upregulated/29 downregulated miRNAs discovered in R-ALL patients compared to S-ALL patients ( Figure 1F). Furthermore, the mRNA expression profile was also able to differentiate R-ALL patients from S-ALL patients ( Figure 1G), and disclosed an inter-consistent trend by heatmap analysis ( Figure 1H), with 676 upregulated/944 downregulated mRNAs uncovered in R-ALL patients compared to S-ALL patients ( Figure 1I). In addition, the detailed information of each dysregulated lncRNA, miRNA, and mRNA is exhibited in Supplementary Tables 1-3, respectively.
Identification of lncRNA DUXAP8/miR-29a/ PIK3CA Network and Downstream PI3K-AKT Pathway KEGG enrichment analyses of dysregulated lncRNA (Figure 2A), miRNA ( Figure 2B), and mRNA ( Figure 2C) expression profiles were carried out, which observed that they were all closely related to the PI3K-AKT pathway. Inspiringly, a comprehensive ceRNA network was established based on dysregulated lncRNA, miRNA, and mRNA expression profiles, which discovered that the lncRNA DUXAP8/miR-29a/PIK3CA network was a key component ( Figure 2D). What is more, PIK3CA is the key component involved in the PI3K-AKT pathway, which is closely engaged in ALL pathogenesis and treatment sensitivity (25)(26)(27). Subsequently, RT-qPCR was performed to verify their expressions, which also demonstrated that lncRNA DUXAP8 was upregulated ( Figure 2E), miR-29a was downregulated ( Figure 2F), and PIK3CA was overexpressed ( Figure 2G) in R-ALL patients compared to S-ALL patients (all p < 0.05). Based on the abovementioned information,

Interaction Between lncRNA DUXAP8 and miR-29a
It was then observed that lncRNA DUXAP8 negatively regulated miR-29a expression in both N6/ADR ( Figure 4A) and 697/ADR ( Figure 4B) cells (all p < 0.05). By the Luciferase reporter gene  assay with the designed binding sequence shown in Figure 4C, lncRNA DUXAP8 was found to be directly bound to miR-29a to regulate its expression (p < 0.01, Figure 4D).
In addition, PIK3CA overexpression plasmid or mTOR activator (MHY1485) was cocultured with lncRNA DUXAP8 knockdown plasmid to explore the role of the PI3K-AKT-mTOR pathway in the restored chemosensitivity by lncRNA DUXAP8 knockdown. Then, it was observed that PI3K-AKT-mTOR pathway activation decreased Dox chemosensitivity in N6/ ADR (Supplementary Figure 3A) and 697/ADR cells (Supplementary Figure 3B) and also attenuated the effect of lncRNA DUXAP8 knockdown.

LncRNA DUXAP8 Modification and Inotuzumab Ozogamicin on Dox-Resistant B-ALL Cells
Inotuzumab ozogamicin is a novel drug recommended for the treatment of refractory B-ALL, so we further explored whether modification of lncRNA DUXAP8 affected inotuzumab ozogamicin treatment in Dox-resistant B-ALL cells. It was

DISCUSSION
A previous secondary bioinformatics analysis involving The Cancer Genome Atlas (TCGA) data identifies 469 upregulated and 286 downregulated lncRNAs in ALL patients compared to controls (28). Another study analyzes the GSE67684 dataset from the Gene Expression Omnibus then validates the candidate lncRNAs by RT-qPCR, which observes 21 dysregulated lncRNAs in accordance with ALL patients compared to controls (29). However, no study has explored the aberrant pattern of the lncRNA profile between chemotherapyresistant B-ALL patients and chemotherapy-sensitive patients. Our current study filled this gap, which observed that 1,338 lncRNAs, 75 miRNAs, and 1,620 mRNAs were dysregulated between chemotherapy-resistant B-ALL patients and chemotherapy-sensitive B-ALL patients. These data would propose new information on the interaction between genetics and chemotherapy resistance in B-ALL.
Since the concept of lncRNA is just initiated in a short time, in-depth studies of its mechanisms underlying ALL are very limited, and only a few studies have reported several specific lncRNAs that are involved in ALL pathogenesis, such as lncRNA MALAT1, lncRNA VPS9D1-AS1, and lncRNA TEX41  (18,30,31). However, few studies on lncRNA have been disclosed in the context of B-ALL chemotherapy resistance. LncRNA DUXAP8 is previously reported as an oncogene in several cancers. For instance, lncRNA DUXAP8 promotes neuroblastoma progression by regulating miR-29 mediated NOL4L and downstream Wnt/b-actin pathway (32); it also enhances colorectal cancer proliferation, migration, and invasion by regulating EZH2 and LSD1 (33). In terms of leukemia, only a related paper is available, which observes that lncRNA DUXAP8 regulates acute myeloid leukemia glycolysis and apoptosis through the Wnt/b-actin pathway (34). Furthermore, lncRNA DUXAP8 is also revealed as a regulator of an anticancer agent, such as a PARP inhibitor (35). However, no data in the aspect of lncRNA DUXAP8 in ALL are reported.  In our present study, further bioinformatic analyses and RT-qPCR validation observed that lncRNA DUXAP8 was closely related to chemotherapy resistance in B-ALL patients; subsequent experiments also revealed that lncRNA DUXAP8 promotes cell proliferation and inhibits cell apoptosis, then targeting it restored the Dox chemosensitivity in chemoresistant B-ALL cell lines. The possible explanations were as follows: (1) lncRNA DUXAP8 regulated the miR-29a/PIK3CA network and downstream PI3K-AKT pathway, as shown by subsequent molecule mechanism experiments, to modify cell proliferation, apoptosis, and Dox chemosensitivity in B-ALL, and (2) lncRNA DUXAP8 activated several key carcinogenetic pathways such as Wnt/b-actin, AKT/mTOR, and WTAP/Fak, to regulate these cell functions (34,36,37). MiR-29a is a well-known anti-oncogene in both solid tumor and hematological malignancies (38,39). In leukemias, MiR-29a deficiency activates CD40 signaling/T-cell interaction to engage in chronic lymphocytic leukemia pathogenesis (40); meanwhile, it is greatly dysregulated in acute myeloid leukemia, chronic lymphocytic leukemia, and ALL patients (41)(42)(43). In the aspect of PI3K-AKT, it is a documented pathway involved in ALL development, progression, and even treatment sensitivity (25,44,45). In our present study, we observed that the lncRNA DUXAP8/miR-29a/PIK3CA network and downstream PI3K-AKT pathway were closely related to chemotherapy resistance in B-ALL patients. Then, comprehensive experiments were performed to demonstrate that the lncRNA DUXAP8/miR-29a/PIK3CA network regulates cell proliferation and apoptosis and targeting the network could restore the Dox chemosensitivity in chemoresistant B-ALL cell lines. The most possible explanation was further excavated, which observed that the effect of the lncRNA DUXAP8/miR-29a/PIK3CA network in chemoresistant B-ALL resulted from activation of downstream PI3K-AKT-mTOR signaling. This finding provided a novel target network for chemoresistant B-ALL, which was exciting.
For chemotherapy-refractory or relapsed B-ALL patients, salvage treatment options are lacking and urgently need investigation. Currently, inotuzumab ozogamicin, an anti-CD22 antibody-drug conjugate, achieves encouraging efficacy and acceptable tolerance as a salvage treatment regimen in refractory or relapsed B-ALL patients (46)(47)(48)(49)(50). In the current study, it was further discovered that targeting lncRNA DUXAP8 and inotuzumab ozogamicin had synergistic effects in killing chemoresistant B-ALL cells. These would provide a new train of thought regarding refractory B-ALL treatment.
Despite the interesting findings initiated in our present study, some limitations existed as well: firstly, the validation of expression of the lncRNA DUXAP8/miR-29a/PIK3CA network could be further conducted in a larger sample size of B-ALL patients. Secondly, targeting lncRNA DUXAP8 restored the Dox chemosensitivity in Dox-resistant B-ALL cell lines via miR-29a/ PIK3CA and subsequent PI3K-AKT-mTOR signaling, while whether other potential mechanisms beyond this could be further investigated. Thirdly, overall changes induced by modification in lncRNA DUXAP8 were relatively minor, then the synergistic effect of modifying lnc DUXAP8 on inotuzumab ozogamicin was relatively minor and observed only in one tested cell line, so further validation experiments were needed in the future.
In conclusion, targeting the lncRNA DUXAP8/miR-29a/ PIK3CA network not only exhibits a good killing effect on Dox-resistant B-ALL and restores Dox chemosensitivity via PI3K-AKT-mTOR signaling but also synergizes with inotuzumab ozogamicin. These imply the effectiveness of this network as a treatment target for chemotherapy-refractory B-ALL.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are publicly available. These data can be accessed using the following link/accession number: https://www.biosino.org/node/ search; OEP003016.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Ethics Committee of The Affiliated Hospital of Southwest Medical University. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
LZ and JT conceived and designed the study. LZ, SZ, XL, and JT collected and analyzed the data. LZ, TZ, and JT prepared the figures. LZ, SZ, and XL wrote the manuscript. TZ and JT edited the manuscript. All authors contributed to the article and approved the submitted version.