Silencing of hsa_circ_0009035 Suppresses Cervical Cancer Progression and Enhances Radiosensitivity through MicroRNA 889-3p-Dependent Regulation of HOXB7

ABSTRACT Circular RNAs (circRNAs), a novel type of endogenous noncoding RNAs, have been identified as critical regulators in human carcinogenesis. Here, we investigated the precise actions of hsa_circ_0009035 in the progression and radioresistance of cervical cancer (CC). The levels of hsa_circ_0009035, microRNA 889-3p (miR-889-3p), and homeobox B7 (HOXB7) were detected by quantitative real-time PCR (qRT-PCR) or Western blotting. RNase R and actinomycin D assays were used to assess the stability of hsa_circ_0009035. Cell proliferation, cell cycle progression, apoptosis, migration, and invasion were gauged with Cell Counting Kit-8 (CCK-8), flow cytometry, and transwell assays. Cell colony formation and survival were determined by the colony formation assay. Targeted correlations among hsa_circ_0009035, miR-889-3p, and HOXB7 were examined by the dual-luciferase reporter, RNA immunoprecipitation (RIP), or RNA pulldown assay. Animal studies were performed to evaluate the impact of hsa_circ_0009035 on tumor growth. We found that hsa_circ_0009035 was highly expressed in CC tissues and cells, and it was associated with the radioresistance of CC patients. Moreover, the silencing of hsa_circ_0009035 inhibited CC cell proliferation, migration, invasion, and it enhanced apoptosis and radiosensitivity in vitro and weakened tumor growth in vivo. Mechanistically, hsa_circ_0009035 directly targeted miR-889-3p by binding to miR-889-3p, and hsa_circ_0009035 modulated HOXB7 expression through miR-889-3p. HOXB7 was a functional target of miR-889-3p in regulating CC progression and radioresistance in vitro, and hsa_circ_0009035 modulated CC progression and radioresistance in vitro by miR-889-3p. Our current study first identified hsa_circ_0009035 as an important regulator of CC progression and radioresistance at least in part through targeting the miR-889-3p/HOXB7 axis, highlighting its significance as a potential therapeutic target for CC treatment.

hsa_circ_0000263 contributed to CC tumorigenesis by directly targeting miRNA 150-5p (miR-150-5p) (10). Song et al. highlighted the oncogenic property of hsa_circ_101996 in CC by reducing miR-8075 activity (11). Guo and colleagues identified the important involvement of hsa_circ_0023404 in CC carcinogenesis and chemoresistance via the regulation of miR-5047 (12). Interestingly, hsa_circ_0009035, engendered by the backsplicing of exons of Rac GTPase-activating protein 1 (RACGAP1), was found as one of the top four upregulated circRNAs in radioresistant HeLa cells (13). Nonetheless, the precise, critical actions of hsa_ circ_0009035 in CC progression and radioresistance remain underexplored.
Recently, aberrant up-and downregulation of miRNAs in CC carcinogenesis has been widely reported (14). Sun and colleagues identified that miR-889-3p (also called miR-889), a striking underexpressed miRNA in CC, functioned as a tumor suppressor in CC through targeting fibroblast growth factor receptor 2 (15). When we used computer algorithms to search a novel circRNA/miRNA/mRNA regulatory network mediated by hsa_circ_0009035, we found two targeted relationships among hsa_circ_0009035, miR-889-3p, and homeobox B7 (HOXB7), a critical player in CC development (16). For these reasons, we set out to identify how hsa_circ_0009035 regulated CC progression and radioresistance and whether the hsa_ circ_0009035/miR-889-3p/HOXB7 axis was an indeed functionally regulatory network.
Here, we provided evidence that hsa_circ_0009035 regulated CC progression and radiosensitivity. Furthermore, we identified the miR-889-3p/HOXB7 regulatory network as a novel molecular mechanism for the regulation of hsa_circ_0009035.

RESULTS
hsa_circ_0009035 was highly expressed in CC tissues and cells. Using the online software CircView (http://gb.whu.edu.cn/CircView/), we found that hsa_circ_0009035, a 375-bp circRNA, was generated by the backsplicing of exons 15,16, and 17 of RACGAP1 mRNA (Fig. 1A). The backsplicing junction was validated by Sanger sequencing (Fig. 1A). To investigate the involvement of hsa_circ_0009035 in CC development, we used quantitative real-time PCR (qRT-PCR) (primer details shown in Fig. 1A) to evaluate its expression level in CC tissues and adjacent normal tissues. In contrast, hsa_circ_0009035 expression was remarkably upregulated in CC tissues (Fig. 1B). Moreover, the level of hsa_circ_0009035 in the radiation-resistant group was higher than that of the sensitive group (Fig. 1C). In agreement with the CC tissues, hsa_circ_0009035 was significantly upregulated in CC cell lines compared with normal cervical Ect1/E6E7 cells (Fig. 1D). To elucidate the stability of hsa_ circ_0009035, we performed RNase R and actinomycin D assays. These data showed that hsa_circ_0009035 was resistant to RNase R, and RNase R treatment led to a striking reduction in the level of corresponding linear mRNA ( Fig. 2A and B). Furthermore, the incubation of cells with actinomycin D caused a significant downregulation of RACGAP1 linear mRNA, and hsa_circ_0009035 level did not decrease in the assayed time frame (Fig. 2C and D). Additionally, subcellular localization analysis showed that hsa_circ_0009035 was present mainly in the cytoplasm of HeLa and Siha cells ( Fig. 2E and F).
Silencing of hsa_circ_0009035 weakened cell proliferation, migration, and invasion and enhanced apoptosis and radiosensitivity in vitro. To directly test the functional role of hsa_circ_0009035 in CC progression, we carried out loss-of-function analyses in vitro with hsa_circ_0009035-shRNA lentiviruses (sh-hsa_circ_0009035 and sh-hsa_circ_0009035#2; short hairpin RNA [shRNA] details are shown in Fig. 1A). The introduction of sh-hsa_circ_0009035 or sh-hsa_circ_0009035#2 significantly reduced the expression of hsa_circ_0009035 but did not affect RACGAP1 mRNA level ( Fig. 2G and H). Remarkably, the knockdown of hsa_circ_0009035 suppressed cell proliferation ( Fig. 3A and B), colony formation (Fig. 3C), and cell cycle progression (Fig. 3D). Conversely, the silencing of hsa_ circ_0009035 dramatically promoted cell apoptosis (Fig. 3E). Moreover, hsa_circ_0009035 silencing strongly repressed cell migration (Fig. 4A) and invasion (Fig. 4B). We also determined whether hsa_circ_0009035 could influence the radiosensitivity of CC cells in vitro. As expected, when cells were transduced with sh-hsa_circ_0009035 or sh-hsa_circ_0009035#2, the cell survival fraction was significantly reduced upon radiation exposure ( Fig. 4C and D), demonstrating that hsa_circ_0009035 silencing enhanced cell radiosensitivity. Additionally, hsa_circ_0009035 knockdown led to a remarkable increase in the level of apoptosis-related protein cleaved caspase 3 (c-caspase 3) and a striking decrease in the expression of the proliferating marker PCNA and the invasion-related protein MMP3 in the two CC cell lines ( Fig.  4E and F). Additionally, we overexpressed hsa_circ_0009035 in HeLa cells with an overexpression plasmid (Fig. 5A), and we found that the forced expression of hsa_circ_0009035 enhanced cell proliferation, migration, and invasion and suppressed apoptosis and radiosensitivity ( Fig. 5B to F).
hsa_circ_0009035 directly targeted miR-889-3p. To elucidate the mechanism by which hsa_circ_0009035 modulated CC progression and radiosensitivity, we used computer algorithms to help identify its targeted miRNAs. Interestingly, the prediction programs Circinteractome and Starbase showed that hsa_circ_0009035 harbored putative regions that matched the seed sequences of miR-1296, miR-182, miR-653, and miR-889-3p (Fig. 6A). In contrast, miR-889-3p was the most significantly upregulated miRNA in the sh-hsa_circ_0009035-transduced CC cells ( Fig. 6B and C), and a biotinylated antisense oligomer (Bio-anti-hsa_circ_0009035) spanning the junction of hsa_ circ_0009035 led to a remarkable increase in the enrichment level of miR-889-3p in  HeLa cells (Fig. 6D). Thus, we selected miR-889-3p for further analysis and found a binding sequence for miR-889-3p within hsa_circ_0009035 (Fig. 6E). We then determined whether hsa_circ_0009035 could influence miR-889-3p expression. As expected, the forced expression of hsa_circ_0009035 caused a significant downregulation of miR-889-3p level in both cell lines (Fig. 6F). However, the overexpression of mutant hsa_circ_0009035 lacking the miR-889-3p binding sites did not reduce the level of miR-889-3p (Fig. 6G), reinforcing that hsa_circ_0009035 regulated miR-889-3p via the binding site. RNA immunoprecipitation (RIP) assays revealed that the enrichment levels of hsa_ circ_0009035 and miR-889-3p were synchronously increased in the presence of AGO2 antibody (Fig. 6H). RNA pulldown assays showed that the hsa_circ_0009035 enrichment level was markedly augmented by a biotin-labeled miR-889-3p mimic (Bio-miR-889-3p) (Fig. 6I), whereas the endogenous expression of hsa_circ_0009035 was not affected by an miR-889-3p mimic or anti-miR-889-3p transfection ( Fig. 6J and K). Our data also showed that the silencing of hsa_circ_0009035 did not affect the level of pre-miR-889 in HeLa and Siha cells (Fig. 6L). In addition, miR-889-3p level was significantly downregulated in CC tissues compared with matched normal tissues, and miR-889-3p expression was lower in radiation-resistant tissues than that in sensitive tissues ( Fig. 6M and N). More intriguingly, a strong inverse correlation between miR-889-3p and hsa_circ_0009035 expression levels was found in CC tissues (Fig. 6O). Analysis of the absolute expression of hsa_circ_0009035 and miR-889-3p in CC cells and tissues showed that the copy number of hsa_circ_0009035 is higher than that of miR-889-3p in CC tissues and HeLa and Siha cells ( Fig. 6P and Q), suggesting that hsa_circ_0009035 was likely to bind to miR-889-3p, thereby liberating target mRNAs.
HOXB7 in CC cells was directly targeted and inhibited by miR-889-3p. To further understand the role of miR-889-3p, we performed a detailed analysis for its molecular targets using Targetscan and Starbase online algorithms. Of the genes that overlapped between the two algorithms, we selected some that were identified as oncogenic The predicted data showed a putative complementary sequence for miR-889-3p within the HOXB7 39 untranslated region (UTR) (Fig. 7C). To confirm this, we constructed HOXB7 39 UTR wild-type (HOXB7 39UTR-wt) and mutant (HOXB7 39UTR-mut) luciferase reporters and analyzed them by luciferase activity. The transfection of the miR-889-3p mimic significantly reduced the luciferase activity of HOXB7 39UTR-wt but barely affected the luciferase activity of HOXB7 39UTR-mut (Fig. 7D). Moreover, RNA pulldown assays showed that the enrichment level of HOXB7 mRNA was dramatically augmented by a biotin-labeled miR-889-3p mimic (Bio-miR-889-3p) (Fig. 7E). We then elucidated whether miR-889-3p could modulate HOXB7 expression. The transfection efficiencies of the miR-889-3p mimic and anti-miR-889-3p were determined by qRT-PCR ( Fig. 7F and G). Remarkably, HOXB7 protein level was reduced by miR-889-3p overexpression Our data also showed that the overexpression of HOXB7 did not affect the levels of endogenous miR-889-3p and hsa_circ_0009035 in the two cell lines (Fig. 7J and K). Additionally, HOXB7 mRNA expression was significantly upregulated in CC tissues, and the radiation-resistant tissues had a higher HOXB7 level than the sensitive CC tissues (Fig. 7L and M). Importantly, a strong inverse correlation between HOXB7 mRNA and miR-889-3p expression levels in CC tissues was found (Fig. 7N). Furthermore, in line with mRNA expression, HOXB7 protein was strikingly overexpressed in CC tissues and radiation-resistant CC tissues compared to their counterparts ( Fig. 7O and P).
Forced expression of miR-889-3p regulated cell proliferation, migration, invasion, apoptosis, and radiosensitivity in vitro by downregulating HOXB7. To determine whether HOXB7 was a functional target of miR-889-3p in modulating CC progression and radiosensitivity, we upregulated the HOXB7 level in miR-889-3p-overexpressing CC cells. The transfection efficiency of a HOXB7-overexpressing plasmid was gauged by Western blotting (Fig. 8A). Remarkably, the transfection of the HOXB7-overexpressing plasmid abolished the reduction of miR-889-3p overexpression on HOXB7 protein level in both cell lines (Fig. 8B). In contrast, the forced level of miR-889-3p significantly weakened and G), and enhanced cell apoptosis (Fig. 8H), as well as suppressing cell migration ( Fig. 8I and J) and invasion ( Fig. 8K and L). Furthermore, miR-889-3p overexpression dramatically reduced cell survival fraction upon radiation exposure ( Fig. 8M and N). In addition, the forced miR-889-3p level resulted in increased c-caspase 3 expression and decreased levels of PCNA and MMP3 in the two CC cell lines ( Fig. 8O and P). However, these functional regulatory effects of miR-889-3p overexpression were significantly abrogated by the restored expression of HOXB7 in both cell lines ( Fig. 8C to P).
hsa_circ_0009035 modulated HOXB7 expression and CC progression and radiosensitivity in vitro by targeting miR-889-3p. We then determined whether hsa_circ_0009035 could modulate HOXB7 expression by miR-889-3p. In contrast, the transfection of anti-miR-889-3p prominently abrogated the increase of miR-899-3p expression of hsa_circ_0009035 silencing in both cell lines (Fig. 9A). As expected, the silencing of hsa_circ_0009035 led to a prominent reduction in the level of HOXB7 protein, and this effect was significantly abolished by miR-889-3p downregulation (Fig.  9B), suggesting that hsa_circ_0009035 modulated HOXB7 expression by regulating miR-889-3p. Interestingly, we found a strong positive correlation between HOXB7 mRNA and hsa_circ_0009035 levels in CC tissues (Fig. 9C).
A crucial question was whether miR-889-3p could work as a downstream effector of hsa_circ_0009035 in regulating CC progression and radiosensitivity. To address this, we carried out rescue experiments in the two CC cell lines. These results revealed that the reduced level of miR-889-3p remarkably abrogated sh-hsa_circ_0009035-mediated antiproliferation ( Fig. 9D and E), anti-colony formation (Fig. 9F), cell cycle arrest (Fig. 9G and H),  (Fig. 9I), antimigration (Fig. 10A), and anti-invasion (Fig. 10B). Moreover, the downregulation of miR-889-3p significantly reversed the reduction of survival fraction of hsa_circ_0009035 silencing upon radiation exposure ( Fig. 10C and D). Additionally, the reduced level of miR-889-3p abolished the impact of hsa_circ_0009035 silencing on PCNA, c-caspase 3, and MMP3 levels in the two CC cell lines (Fig. 10E to H).
hsa_circ_0009035 regulated tumor growth in vivo. We next asked whether hsa_ circ_0009035 could modulate tumor growth in vivo. To address this, we established the xenograft mouse model using sh-hsa_circ_0009035-transduced or lenti-hsa_circ_0009035infected HeLa cells by subcutaneous injection. In contrast, sh-hsa_circ_0009035 transduction significantly weakened tumor growth (Fig. 11A to C). qRT-PCR analysis showed that Identification of hsa_circ_0009035 in CC Molecular and Cellular Biology hsa_circ_0009035 was downregulated in sh-hsa_circ_0009035-transduced HeLa tumors (Fig. 11D). Moreover, the knockdown of hsa_circ_0009035 led to an increase in the expression of miR-889-3p and a decrease in the level of HOXB7 protein (Fig. 11E and F). Conversely, the transduction of lenti-hsa_circ_0009035 remarkably enhanced tumor growth (Fig. 11G to I). hsa_circ_0009035 was highly expressed in lenti-hsa_circ_0009035-transduced xenograft tumors (Fig. 11J). Furthermore, the overexpression of hsa_circ_0009035 led to a reduction in the expression of miR-889-3p and a clear increase in the level of HOXB7 protein ( Fig. 11K and L).

DISCUSSION
Recently, the altered expression of circRNAs has been demonstrated to influence CC carcinogenesis and chemoresistance (6,12,17). Although a large number of circRNAs were discovered to be differentially expressed in radioresistant HeLa cells (13), the precise actions of these circRNAs in CC radioresistance are still elusive. Here, we report, for the first time, the functional actions and molecular determinant of hsa_ circ_0009035 in CC progression and radioresistance (Fig. 12).
A recent study uncovered the promotional impact of circRACGAP1 on gastric cancer cell sensitivity to apatinib through the modulation of the miR-3657/ATG7 axis (18). As a member of circRACGAP1, hsa_circ_0009035, an upregulated circRNA in radioresistant HeLa cells, was illuminated the crucial involvement in CC radioresistance (13). Our results indicated that hsa_circ_0009035 was upregulated in CC and associated with the radioresistance of CC patients, implying its potential clinical significance as a diagnostic biomarker. MMP3, a key member of the MMP family, has been implicated in CC cell migration and invasion (19). Our results showed that hsa_circ_0009035 regulated CC progression and radiosensitivity in vitro, as well as modulating tumor growth in vivo. Additionally, as previously reported for other circRNAs (20,21), hsa_circ_0009035 were unusually stable and resistant to RNase R, which was attributed to their covalently closed loop structures with neither 59 caps nor 39 polyadenylation tails (22).
Several previous reports had demonstrated the conflicting roles of miR-889-3p in human carcinogenesis (23)(24)(25). These contradictory conclusions might be partially due to the different tumor types in these reports, where miR-889-3p exhibited an antitumor property in non-small cell lung cancer (25) and functioned as a tumor promoter in esophageal squamous cell carcinoma (23) and osteosarcoma (24). Interestingly, miR-889-3p was identified as a strong tumor suppressor in CC (15,26). Our results extended this by determining the regulation of miR-889-3p on CC radiosensitivity. Moreover, we first discovered that hsa_circ_0009035 directly bound to miR-889-3p, and miR-889-3p was a functionally important mediator of hsa_circ_0009035 in modulating CC FIG 12 Schematic model of the hsa_circ_0009035/miR-889-3p/HOXB7 axis in CC progression and radiosensitivity. In CC, hsa_circ_0009035 was overexpressed and the increased level of hsa_circ_0009035 reduced the miR-889-3p level. Then, the downregulation of miR-889-3p increased HOXB7 expression. Finally, HOXB7 overexpression promoted cell proliferation, migration, invasion, and radiation resistance and suppressed cell apoptosis, thereby enhancing CC progression and radiation resistance. progression and radioresistance in vitro. Previous work also discovered the potential involvement of miR-889-3p in cell survival of radiation in nervous system cancer (27).
HOXB7, a member of the HOX gene family, has been shown as a strong oncogene in various human cancers, such as hepatocellular carcinoma, breast cancer, and osteosarcoma (28)(29)(30). Here, we first identified the role of hsa_circ_0009035 as a posttranscriptional regulator of HOXB7 expression by targeting miR-889-3p. Furthermore, for the first time, we showed that HOXB7 was a functional target of miR-889-3p in modulating CC progression and radioresistance in vitro. Similarly, How et al. reported that miR-196b hindered CC development by targeting HOXB7 (16). Previous studies also reported that several other miRNAs, such as miR-377 and miR-384, impacted human carcinogenesis by directly targeting HOXB7 (31,32). The circRNA/miRNA/mRNA regulatory networks are intricate, and a circRNA can act as an inhibitor of many miRNAs (33). There may be alternative miRNA/ mRNA axes that remain to be uncovered in the modulation of hsa_circ_0009035 in CC. Future studies will build on the findings by identifying precisely how the novel network modulates CC progression and radioresistance in vitro and in vivo.
In summary, our findings identify hsa_circ_0009035 as an important modulator in CC progression and radioresistance at least in part by targeting the miR-889-3p/HOXB7 axis. To our knowledge, this is the first report of hsa_circ_0009035 in human carcinogenesis, pointing to its significance as a potential therapeutic target for CC treatment.

MATERIALS AND METHODS
Human specimens and cells. Specimen tissues, including cancer tissues and adjacent healthy cervical tissues, were obtained from 82 consecutive patients with CC (average age, 46.2 6 8.1; 36 patients with primary CC and 46 with recurrent CC after radiation treatment) who underwent surgical resection at The First Affiliated Hospital of University of South China. All specimens were stored at 280°C until quantitative real-time PCR (qRT-PCR) analysis for the expression levels of hsa_circ_0009035, miR-889-3p, and HOXB7. The human study project was approved by the Ethics Committee of The First Affiliated Hospital of University of South China, and all patients gave written informed consent.
Human ectocervical Ect1/E6E7 cells and two CC cells (HeLa and Siha) were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). HeLa and Siha cells were cultivated in accordance with the standard protocols provided by ATCC. Ect1/E6E7 cells were propagated in keratinocyte serum-free medium (Gibco, Lucerne, Switzerland) as reported elsewhere (34).
RNA extraction and qRT-PCR. Total RNA containing small RNA was prepared from cells and tissues with the RNAiso Plus (TaKaRa, Beijing, China) as per the accompanying instructions. For hsa_circ_0009035 and mRNAs qRT-PCR, reverse transcription (RT) was done using the PrimeScript RT kit (TaKaRa). For miRNAs qRT-PCR, cDNA was generated with a TaqMan microRNA RT kit (Applied Biosystems, Courtaboeuf, France) based on the protocols of the manufacturers. qRT-PCR analysis with designed primers (Table 1) was carried out in triplicate using a SYBR green kit (TaKaRa) on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Human b-actin or U6 was used for normalization. Fold changes in gene expression levels were calculated by the 2 2DDCT method (35).
RNase R assay. RNase R treatment was carried out by adding 3 U of RNase R (Geneseed) to 1 mg of total RNA and incubating for 20 min at 37°C. Dual-luciferase reporter assay. The fragments of the HOXB7 39 UTR encompassing either an intact or a mutated miR-889-3p binding sequence were synthesized by BGI and individually cloned into the pMIR-REPORT vector (Promega, Leiden, The Netherlands). HeLa and Siha cells at ;60% confluence in 24well plates were cotransfected with 200 ng of each reporter construct, 20 ng of pRL-TK control vector (Promega), and 50 nM miR-889-3p mimic or negative mimic control. Each sample was assayed in triplicate using the dual-luciferase assay system as per the manufacturing instructions (Promega).
RIP and RNA pulldown assays. HeLa and Siha cells were homogenized in xTractor buffer as per the manufacturer's protocol. In RIP assays, cell lysates were incubated with protein A/G beads (Invitrogen) conjugated with Argonaute2 antibody (AGO2; MA5-23515; Invitrogen) or isotype IgG control (10400C; Invitrogen) overnight at 4°C. In RNA pulldown assays, cell lysates were incubated with a biotin-labeled miR-899-3p mimic (Bio-miR-889-3p; Ribobio), a biotinylated antisense oligomer (CTTTAGTCT) spanning the junction of hsa_circ_0009035 (Bio-anti-hsa_circ_0009035), or a negative control (Bio-NC; Ribobio) for 3 h at 4°C before addition of the streptavidin beads (Thermo Fisher Scientific) for 3 h. In both assays, total RNA was prepared from the beads for the assessment of hsa_circ_0009035, HOXB7, and miRNA levels.
Animal studies. All animal procedures and experimental protocols were approved by the Animal Care and Use Committee of The First Affiliated Hospital of University of South China. Approximately 5 Â 10 6 sh-NC-infected, sh-hsa_circ_0009035-transduced, lenti-NC-transduced, or lenti-hsa_circ_0009035infected HeLa cells were subcutaneously injected into the left flanks of female BALB/c nude mice aged 6 weeks (6 per group; Vital River Laboratory, Beijing, China). Tumor length and width were gauged every 5 days, and tumor volume was determined with the formula length Â width 2 Â 0.5. All mice were euthanized after 25 days, and the xenograft tumors were collected for weight and gene expression analysis.
Statistical analysis. Unless otherwise indicated, means and standard deviations are representative of the average of at least three independent experiments. Statistical significance (a P value less than 0.05 was considered significant) was determined by a two-sided Student's t test, Mann-Whitney U test, and analysis of variance (ANOVA) with Dunnett's multiple-comparison test. For survival data, a log-rank test was used. Correlations among hsa_circ_0009035, miR-889-3p, and HOXB7 expression levels in CC tissues were evaluated using the Spearman rank correlation test.