HOTAIRM1 Promotes Malignant Progression of Transformed Fibroblasts in Glioma Stem-Like Cells Remodeled Microenvironment via Regulating miR-133b-3p/TGFβ Axis

Recent studies have reported that cancer associated fibroblasts (CAFs) and glioma stem-like cells (GSCs) played active roles in glioma progression in tumor microenvironment (TME). Long non-coding RNAs (lncRNAs) have been found to be closely associated with glioma development in recent years, however, their molecular regulatory mechanisms on CAFs in GSCs remodeled TME kept largely unelucidated. Our study found that GSCs could induce malignant transformation of fibroblasts (t-FBs) based on dual-color fluorescence tracing orthotopic model. Associated with poor prognosis, Lnc HOXA transcript antisense RNA, myeloid-specific 1 (HOTAIRM1) was highly expressed in high-grade gliomas and t-FBs. Depleting HOTAIRM1 inhibited the proliferation, invasion, migration, and even tumorigenicity of t-FB. Conversely, overexpression of HOTAIRM1 promoted malignancy phenotype of t-FB. Mechanistically, HOTAIRM1 directly bound with miR-133b-3p, and negatively regulated the latter. MiR-133b-3p partly decreased the promotion effect of HOTAIRM1 on t-FBs. Furthermore, transforming growth factor-β (TGFβ) was verified to be a direct target of miR-133b-3p. HOTAIRM1 can modulate TGFβ via competing with miR-133b-3p. Collectively, HOTAIRM1/miR-133b-3p/TGFβ axis was involved in modulating t-FBs malignancy in TME remodeled by GSCs, which had the potential to serve as a target against gliomas.


INTRODUCTION
Gliomas are the most prevalent primary intrinsic tumors in the central nervous system, and glioblastoma multiforme (GBM) accounts for nearly half of gliomas (1)(2)(3). Although the treatment is constantly optimized (4), GBM is regarded as one of the most malignant and aggressive cancer, which is characterized of poor prognosis and low overall survival. Recently, some studies have focused on the interactions between GSCs and stromal cells in TME, which is regarded as crucial element for glioma progression recently (5).
Among the stromal cells in TME, fibroblasts are key components involved in progression and metastasis of various malignancies (6,7). Chronic activation of peri-tumor fibroblasts by cancer cells leads to formation of a subgroup of cells generally known as cancer associated fibroblasts (CAFs) (8). As a significant crosstalk mediator between cancer and stroma in TME, fully exploring the roles of CAFs may help to develop new possible therapeutic approaches against gliomas (9). Local accumulation of CAFs in TME is related with poor prognosis (10,11), CAFs can promote the proliferation and metastasis of breast cancer (12), endometrial cancer (13), adenocarcinoma of esophagogastric junction (14), melanoma (15), gastric cancer (16), etc. However, the roles of fibroblasts in gliomas TME, and their interactions or crosstalk with GSCs have not been fully elucidated up to now.
LncRNAs are a group of non-coding RNAs with over 200 nucleotides in length (17), which could activate/silence target gene on the level of transcription and/or post-transcription level (18). Emerging studies have declared that aberrant expression of lncRNAs is ubiquitous and relevant with multiple malignant phenotypes of tumors (19)(20)(21). LncRNA HOTAIRM1 is located in the HOX gene cluster and regulates the family of HOXA genes (22), which is associated with tumor progression, including colorectal (23), gastric (24), lung cancers (25), and pancreatic cancer (26), et al. However, the role of HOTAIRM1 played in interactions between GSCs and stromal cells in glioma microenvironment has never been reported previously.
Based on the findings that in vivo malignant transformation of fibroblasts with HOTAIRM1 upregulation after interaction with GSCs, further exploration the roles of HOTAIRM1 initiated molecular regulatory pathway on proliferation, invasion, migration, and tumorigenicity of t-FBs, will provide new experimental references for potential treatment strategy targeting on t-FBs in glioma microenvironment.

MATERIAL AND METHODS
Tumor Specimens, Experimental Animals, and Cell Culture Surgical specimen from glioma patients were collected from the Department of Neurosurgery, the Second Affiliated Hospital of Soochow University with informed consent. The whole process obeyed the rules of the Ethics Committee of the Second Affiliated Hospital of Soochow University.
Balb/c nude mice expressing green fluorescent protein (GFP) were bred in specific pathogen free (SPF) experimental animal center in Soochow University (27). All of the animal studies adhered to the rules of the Ethics Committee of the Second Affiliated Hospital of Soochow University.

CCK8 Assay
Cells were seeded into 96-well plates at a density of 3,000 cells/ well in 100 µl DMEM. Ten µl CCK8 reagent (Dojindo, Japan) was added into each well every 24 h and incubated for another 2 h at 37°C. Absorbance value at 450 nm was recorded with a spectrophotometer (Tecan, Switzerland).

Immunocytochemical Staining
Transformed fibroblasts were fixed with methanol for 20 min, then permeabilized with 0.25% Triton X-100 (Beyotime, China), and incubated in blocking solution for 1 h. The primary antibodies of a-SMA, vimentin, and S100A4 (CST, USA) were applied for 1 h, respectively. After washing with PBS for three times, the second antibody (Beyotime, China) was applied for 30 min. Finally, the cells were stained with diaminobenzidine (DAB) and hematoxylin, observed under fluorescence microscope (Zeiss, Germany) at a magnification of 200×.

Vector Construction and Cell Transfection
The short hairpin RNA (shRNA) targeting HOTAIRM1 (shHOTAIRM1-1 and shHOTAIRM1-2), the overexpression vector of HOTAIRM1, TGF-b, and the corresponding negative control, the miR-133b-3p mimics, inhibitors, and corresponding negative control, were all designed by GenePharma (Shanghai, China). The corresponding vectors and their controls were transfected into transformed fibroblasts according to the manufacturer's protocol.

Quantitative Real-Time Reverse Transcription PCR (qRT-PCR)
Total RNA was extracted with TRIzol (Invitrogen, USA). qRT-PCR was performed to evaluate the expression levels of HOTAIRM1, miR-133b-3p, and TGF-b. The expression of U6 and GAPDH was applied as control. The expression level was analyzed with 2 −DDCt method.

Invasion Assay
Transwell chambers (Corning, USA) were coated with Matrigel (dilution 1:8; BD, USA). Then 120 ml serum-free medium containing 5 × 10 4 cells was added in the upper chamber, 600 ml complete medium containing 10% FBS was added in the lower chamber, then cultured for 24 h at 37°C. The cells on the upper surface were erased with cotton swabs. The cells on the lower surface were fixed with methanol for 30 min and stained with 0.1% crystal violet, cell images were captured with an inverted microscope (AMG, USA) at a magnification of 200×.

Wound Healing Assay
Cells were seeded on six-well plate and cultured at 37°C overnight. A 200 ml pipette tip was applied to make wounds on monolayer cells. Cells were washed by PBS and cultured in DMEM without FBS. The cells were imaged with an inverted microscope (AMG, USA) at a magnification of 100×. Images of the wound area were analyzed via Image J software after 24 h (NIH, Bethesda, USA).

Luciferase Reporter Assay
The HOTAIRM1 fragment with the miRNAs binding sequences were inserted into the pMIR-REPORT vectors. Similarly, the 3′-UTR fragments of TGF-b with the miRNAs binding sequence were also inserted into pMIR-REPORT vectors. Transformed fibroblasts were transfected with the corresponding miRNAs and the reporter vectors. The mutated vectors were used as control. The duration of transfection was 36 h. Luciferase activity was evaluated by the Dual Luciferase Reporter Assay System (Promega, USA).

Western Blot
Total cell proteins were extracted with RIPA buffer (Beyotime Biotechnology, China). Twenty mg total proteins were separated by 10% SDS-PAGE and transferred to PVDF membrane, then incubated with the primary antibodies against TGF-b (1:1,000, CST, USA) and GAPDH (1:5,000, CST, USA) overnight. Then the membrane was incubated with the secondary antibody for 1 h . E CL method was u sed for visualization for quantitative analysis.

Enzyme Linked Immunosorbent Assay (ELISA)
Sample and the diluent buffer were added to the wells of the 96well plate which was pre-coated with anti-TGF-b antibody (Abnova, USA), then shake the plate gently to mix thoroughly, and incubate at 37°C for 30 min. Discard the solution in the plate and fill each well completely with Wash Buffer (1×) and vortex gently on the shaker for 2 min. Repeat this procedure four more times, 50 ml HRP conjugated anti-TGF-b antibody was added into each well. Incubate at 37°C for 30 min, then wash the plate for three times, following with adding 50 ml TMB chromogenic reagent A and B, vortexing gently the plate on the shaker for 30 s, and incubating in dark at 37°C for 15 min, finally adding 50 ml Stop solution into each well and mix thoroughly. The absorbance at 450 nm was detected with a spectrophotometer (Tecan, Switzerland).

Tumorigenicity Assay
Athymic Balb/c nude mice (4 weeks old, 15-20 g) were bred in the SPF animal center. Then 1 × 10 6 transformed fibroblasts with up/downregulation of HOTAIRM1 and the corresponding negative control were injected subcutaneously into the right flank of each mouse, respectively. After 5 weeks, all mice were sacrificed under general anesthesia, and the tumors were excised and weighed.

Overexpression of LncRNA HOTAIRM1 in Glioma Tissue/Cell Lines and Malignant Transformed Fibroblasts
HOTAIRM1 expression in gliomas was evaluated in The Cancer Genome Atlas (TCGA) database, indicating HOTAIRM1 upregulation in gliomas compared with adjacent normal brain tissue ( Figure 1A). TCGA survival curve analysis of gliomas showed that the survival of patients with high HOTAIRM1 expression decreased significantly ( Figure 1B). Clinical glioma samples (n = 10) were collected to further verify HOTAIRM1 expression, which was in accordance with the TCGA results ( Figure 1C). RNA fluorescence in situ hybridization (FISH) demonstrated that HOTAIRM1 expression was positive correlated with glioma malignancy grades in tissues sections of clinical specimens ( Figure 1D). In addition, HOTAIRM1 expression was detected in the malignant transformed fibroblasts (t-FB1 and t-FB2), normal fibroblasts, glioma cell lines (SNB19, A172, U343), GSCs-SU3, and normal human astrocytes (NHAs), which disclosed HOTAIRM1 upregulation in glioma cell lines, GSCs, and transformed fibroblasts, when compared with NHAs and normal fibroblasts ( Figure 1E).

HOTAIRM1 Downregulation Inhibited Proliferation, Invasion, Migration, and Tumorigenesis of Malignant Transformed Fibroblasts
After transfected with an shRNA targeting HOTAIRM1 (shHOTAIRM1-1 and shHOTAIRM1-2) in t-FB1 and t-FB2 cells, the transfection efficiency was validated by qRT-PCR ( Figure 2A). EdU assay indicated that HOTAIRM1 downregulation decreased the proliferation ability of t-FB1 and t-FB2 cells ( Figures 2B-D). Transwell assay showed that knockdown HOTAIRM1 expression resulted in significant decline of invasion ability of t-FB1 and t-FB2 cells ( Figures 2E, F). Downregulation of HOTAIRM1 also remarkably weakened migration ability of t-FB1 and t-FB2 cells ( Figures 2G-J), as shown in wound healing assay. Tumorigenicity assay indicated that HOTAIRM1 downregulation of transformed fibroblasts led to obvious decrease in both tumor volume and weight of subcutaneous implanted HOTAIRM1 knocking-down t-FB1/2 cells decreased obviously, compared with the control group ( Figures 2K-P). To further investigate the effect of HOTAIRM1 down-regulation in t-FB1 and t-FB2 cells on tumorigenicity ability of GSCs, in vivo combined inoculation of GSCs-SU3 and HOTAIRM1 downregulated t-FB1/2 cells was performed, the transplanted tumor volume and weight were significantly decreased, compared with the control group ( Supplementary Figures 2A-F).

HOTAIRM1 Upregulation Promoted Proliferation, Invasion, Migration, and Tumorigenesis of Malignant Transformed Fibroblasts
To further validate the biological function of HOTAIRM1 in the malignant transformed fibroblasts, overexpression of HOTAIRM1in t-FB1 and t-FB2 cells was achieved via transfection of HOTAIRM1, and verified by qRT-PCR ( Figure  3A). EdU assay showed HOTAIRM1 upregulation enhanced the proliferation capacity of t-FB1 and t-FB2 cells in vitro ( Figures  3B, C). The results of Transwell assays indicated that overexpression of HOTAIRM1promoted the invasion of t-FB1 and t-FB2 cells (Figures 3D, E). Wound healing assay suggested that overexpression of HOTAIRM1 significantly increased the migration of t-FB1 and t-FB2 cells (Figures 3F, G). In vivo tumorigenicity assay also showed that overexpression of HOTAIRM1 in t-FB1/2 cells led to higher tumor volume and weight, compared with the control group ( Figures 3H-M). To further investigate the effect of HOTAIRM1 overexpression of transformed fibroblasts on tumorigenicity of GSCs, in vivo combined inoculation of GSCs-SU3 and HOTAIRM1 upregulated t-FB1/2 cells was performed, tumor volume and weight were increased, compared with the control group ( Supplementary Figures 2G-L).

MiR-133b-3p Was a Direct Target of HOTAIRM1
To explore the role of competing endogenous RNA (ceRNA) on regulating biological activities of malignant transformed fibroblasts, the online database starBase (http://starbase.sysu. edu.cn/) was applied to identify the possible miRNA targets of HOTAIRM1. Potential binding sites between HOTAIRM1 and miR-133b-3p ( Figure 4A) were predicted with bioinformatic analysis via starBase. Further verification with qRT-PCR found dramatic downregulation of miR-133b-3p expression in t-FB1/2 and glioma cell lines, compared with normal fibroblasts and NHAs ( Figure 4B). Specifically, miR-133b-3p was negatively regulated by HOTAIRM1 in t-FB1/2 cells (Figures 4C, D). In order to clarify the direct interaction between miR-133b-3p and HOTAIRM1, wild type (WT) and mutant type (MUT) vector of HOTAIRM1 were constructed for luciferase activity assay, which showed that miR-133b-3p significantly inhibited the luciferase activity of HOTAIRM1-WT ( Figures 4E, F), indicating that miR-133b-3p was the direct target of HOTAIRM1.

MiR-133b-3p Inhibited Proliferation, Invasion, and Migration of Malignant Transformed Fibroblasts by Targeting TGFb
TGFb was one of the predicted downstream targets of miR-133b-3p according to bioinformatic analysis ( Figures 5A). To further verify the direct bonding between miR-133b-3p and TGFb, TGFb-WT and TGFb-MUT vectors constructed for further luciferase activity assay ( Figure 5A), which revealed that miR-133b-3p significantly weakened luciferase activity of TGFb-WT, compared with TGFb-MUT ( Figures 5B, C). TGFb expression level was also detected with qRT-PCR in both glioma cell lines and t-FB1/2 cells, which showed TGFb was overexpressed in both glioma cell lines and transformed fibroblasts ( Figure 5D).

DISCUSSION
In this study, glioma associated fibroblasts were found undergoing malignant transformation in GSCs remodeled tumor microenvironment. The role of lncHOTAIRM1 was highly associated with the pro-tumor abilities of transformed fibroblasts, including promoting proliferation, invasion, migration, and tumorigenesis on t-FBs. Further mechanistical studies suggested that HOTAIRM1 competitively bound miR-133b-3p to regulate the expression of TGF-b, thus producing obvious pro-tumor effect.
Most studies attributed treatment resistance and high recurrence of glioma to GSCs (29)(30)(31), which based on their infinite self-renewal capacity, persistent proliferation, and strong remodeling of TME (32)(33)(34). Despite the key roles of GSCs in tumor initiation, glioma microenvironment, including the relevant various stromal cells, interacted actively with GSCs, and played vital roles on GSCs initiated tissue remodeling processes. Various glioma biological processes were largely affected by TME (35), and targeting on TME has been the potential treatment strategy in recent years (36). Among stromal cells in glioma microenvironment, fibroblasts were essential for regulating glioma development (37)(38)(39). Normal fibroblasts maintained homeostasis, which was characterized of inertness and low levels of synthetic properties if they were not under any local microenvironment stress (40). When the intrinsic crosstalk between normal fibroblasts (NFs) and stroma changed due to education by cancer stem cells, NFs may acquire modified phenotypes and can converted into CAFs (41). Recent studies revealed that CAFs were main cancerpromoting components in TME, which were involved in carcinogenesis, proliferation, invasion, and chemoresistance (42)(43)(44)(45), and the existence of CAFs implied poor prognosis in ovarian and breast cancer (43,44). Yes-associated protein 1 (YAP1) was responsible for converting NFs into CAFs, thus facilitating prostate cancer development (46). However, the role of fibroblasts in glioma microenvironment and how they interacted with GSCs kept largely unknown. In vitro studies showed CAFs were associated with glioma cell migration (38). Fibroblasts can differentiate from mesenchymal cells (47,48), which can be recruited into glioma microenvironment. Our results verified malignant transformation of fibroblasts in glioma microenvironment induced by GSCs, and further disclosed that HOTAIRM1 played active roles on malignancy phenotype of t-FBs, which could be a potential target against gliomas.
Long non-coding RNAs were pivotal in molecular dysregulation of gliomas, including the aberrant expression profile of gliomas (49)(50)(51). Previous studies reported that Lnc HOTAIRM1 can act as promoter or suppressor in tumor development: HOTAIRM1 exerted its tumor inhibitory effect through competitive combination with endogenous RNA in gastric cancer (24), head and neck tumor (52), it also can suppress tumor cell proliferation and promote cell apoptosis in hepatocellular carcinoma via inhibiting Wnt pathway (53). Besides, HOTAIRM1 can promote glioma progression through certain ceRNA networks (54)(55)(56)(57). The role of HOTAIRM1 in transformation and malignant phenotype of fibroblasts in glioma microenvironment was explored in the current studies, which also disclosed that miR-133b-3p was the target of HOTAIRM1 and can inhibit biological behaviors of t-FBs. MiR-133b played tumor suppressing roles in several malignancies through regulating in gastric cancer (58), targeting Sox9 in breast cancer (59), negatively regulating EMP2 in glioma (60), targeting methyltransferase DOT1L in colorectal cancer (61), and targeting EGFR in esophageal squamous cell carcinoma (62). Our results supported that TGFb was the functional target of miR-133b-3p, miR-133b-3p can inhibit t-FBs via targeting TGFb.
In summary, the current studies showed that HOTAIRM1 was upregulated in t-FB and gliomas, it can promote malignant biological phenotype of t-FBs by regulating TGFb via miR-133b-3p. Therefore, the HOTAIRM1/miR-133b-3p/TGFb signal axis may serve as the potential therapeutic target against t-FBs in glioma microenvironment. However, our results were mainly based on an orthotopic xenograft tumor model, which may not fully reflect the real TME of glioma patients, which needs to be verified through high throughput single cell sequencing of glioma surgical specimens to confirm the molecular mechanisms regulating CAFs in glioma microenvironment.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by the ethics committee of Second Affiliated Hospital of Soochow University.

AUTHOR CONTRIBUTIONS
HW and JD conceived and designed the experiment. HW performed the experiments and wrote the manuscript. HL performed the statistical analyses and generated the figures. QJ and XD collected the public data. SL, SC, JS and LL collected the patient samples. ZQ and JD revised the manuscript. All authors contributed to the article and approved the submitted version.