The PROMO database (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF8.3) was used to predict the upstream transcription factors of HIF1A-AS2, and the promoter sequence of HIF1A-AS2 used is presented in Supplementary Table SI . The binding sites of the transcription factors c-Jun and HIF1A-AS2 were predicted via the JASPAR database (https://jaspar.genereg.net/analysis). miRDB (http://mirdb.org/) and TargetScan (https://www.targetscan.org/vert_71/) were used to predict the target miRNAs of HIF1A-AS2 and the target genes of miR-34b-5p and potential binding sites. All the predicted data are presented in Supplementary Table SII - Supplementary Table SIV .

Cervical cancer tissues and paired adjacent normal tissues (n = 20) were collected from The Affiliated Hexian Memorial Hospital of Southern Medical University. The median age was 50 years and the range was 36–66 years. The clinical characteristics of CC patients are presented in Supplementary Table SV . All resected tissue samples were frozen immediately in liquid nitrogen after collection. All enrolled patients provided informed consent before inclusion.

All the cell lines used were obtained from iCell Bioscience Inc. (Shanghai, China), including one human immortalized epidermal cell line (HaCat), one human embryonic kidney cell line (293T), and three cervical cancer cell lines (HeLa, SiHa, CaSki). All the cell lines passed the STR authentication report. HeLa, SiHa, HaCat and 293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA). CaSki cells were cultured in RPMI 1640 with 10% fetal bovine serum. All the cells were cultured at 37°C in a humidified incubator (Eppendorf, German) with 5% CO_2. When the cell density reached 80%, the cells were harvested for subsequent experiments.

Adenoviruses for the overexpression of HIF1A-AS2 (adv-HIF1A-AS2) and the control adenovirus vector (adv-NC) were constructed by Hanbio Company (Shanghai, China), and small interfering RNAs (siRNAs) for the knockdown of HIF1A-AS2 (si-HIF1A-AS2, 5’GAGTTGGAGGTGTTGAAGCAAATAT 3’), the negative control siRNA (si-NC), the miR-34b-5p mimic (ATCCGTCACAGTAATCGACTAAC), the negative control miRNA (miR-NC) and the miR-34b-5p inhibitor (UAGGCAGUGUCAUUAGCUGAUUG) were purchased from Ribo (Guangzhou, China). The sequence of si-NC and miR-NC were not provided because it involves the company’s confidential information. The pcDNA3.1 plasmid was used to overexpress the transcription factor c-Jun, and the pmirGLO, pGL3-basic, pGL3-P1, pGL3-P2-MUT and pRL-TK (internal reference) plasmids were used for the dual-luciferase reporter assay.

All the siRNAs (50 nM) and miRNA mimics (50 nM) were transfected for 6 h using 1 µL of Lipofectamine 2000 reagent (Invitrogen, USA). The recombinant adenovirus vector (adv-HIF1A-AS2, MOI = 50) and control adenovirus vector (adv-NC, MOI = 50) were used to infect cells according to the manufacturer’s protocol.

Total RNA was extracted from tissues with an RNA isolation Kit (Omega, USA), and cell total RNA was extracted with TRIzol reagent (AG, Hunan, China). Using NanoDrop 2000 (Thermo Fisher Scientific, USA), the quality and quantity of RNA were measured. The Primer Script RT Kit (AG, Hunan, China) was applied for reverse-transcription of 1µg total RNA into cDNA, the reverse transcription condition was set to 37°C for 15min, 85°C for 5s. The expression of lncRNA and mRNA was measured using pre-mix SYBR (AG, Hunan, China) by ABI ViiA7 system (Thermo Fisher Scientific, USA). The amplification conditions: 95°C predegeneration for 30s; 95°C denature for 5s; 60°C anneal for 30s; the denature and anneal processes lasts for 40 cycles.

miRNAs were extracted from cells with a miRNA isolation kit (Omega, USA), and 10 ng of RNA was reverse-transcribed into cDNA using TaqMan™ MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific, USA), Quantitative Real-time PCR was performed by Taqman™ probes (Thermo Fisher Scientific, USA). The expression level of HIF1A-AS2 was normalized to the expression levels of β-actin, and the expression of miRNA was normalized to the expression levels of U6, calculated by 2^{- ∆∆CT}. The sequences for all primers in this study were listed in Supplementary Table SVI .

HeLa cells were crosslinked with 1% formaldehyde for 10 min at room temperature and quenched with 0.125 M glycine for 5 min. The cells were subjected to multiple washes with PBS and subsequently lysed for 10 min in lysis buffer composed of 50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 0.1% sodium deoxycholate, and 1% Triton X-100 supplemented with a protease inhibitor cocktail. After centrifugation, the pellet was lysed in lysis buffer and subjected to sonication. The sonication conditions were as follows: power 20%, ultrasonic shock for 2 s, gap for 3 s, and ice bath cooling for 5–15 min. The chromosome fragments are in the range of about 0.1–0.5 kb. The disposed chromatin was subjected to immunoprecipitation with a normal IgG antibody (ab172730, Abcam, USA) as a negative control (NC) and an antibody (9165S, CST, USA) against c-JUN as a positive control and then incubated at 4°C overnight. The mixture was captured by protein-G magnetic beads. After immunoprecipitation, the protein–DNA crosslinks were reversed, the DNA was purified, and the precipitated DNA was subjected to qPCR. The six pairs of primers used are listed in Supplementary Table SVI .

SiHa and HeLa cells were plated in 6-well plates and then transfected with si-HIF1A-AS2 or si-NC to knockdown the expression of HIF1A-AS2, and the concentrations of si-HIF1A-AS2 and si-NC were both 50 nM. Adv-HIF1A-AS2 and adv-NC were used to overexpress HIF1A-AS2, and viruses were used at a multiplicity of infection (MOI) of 50, when the cell density reached 60%. After 48 h, the transfected cells were harvested and then seeded into 96-well plates at a concentration of 2000 cells/well. Then, cell viability was assessed at different time points (0, 24, 48, and 72 h) by adding 10 µL of CCK-8 reagent (Dojindo, Japan) to each well. After incubation at 37°C for 2 h, cell viability was determined by measuring the OD value at 450 nm via a microplate reader (Biotek Synergy H1, USA).

SiHa and HeLa cells were plated in 6-well plates and transfected with si-HIF1A-AS2, si-NC, or adv-HIF1A-AS2, adv-NC. After 48 h, the transfected cells were collected and resuspended in serum-free DMEM. Then, 4 × 10⁴ cells transfected with adv-HIF1A-AS2 or adv-NC, 8 × 10⁴ cells transfected with si-HIF1A-AS2 or si-NCwere seeded into the upper chambers of a prepacked Matrigel 24-well Transwell plate (Corning, USA), and 700 µL of medium containing 10% fetal bovine serum (10% DMEM) was added to the lower chambers. After incubation for 48 h, the cells in the upper chamber were removed with a cotton swab, and the cells on the lower membrane were fixed with 4% paraformaldehyde and stained with 1% crystal violet solution (Beyotime, Beijing, China). The cells in three randomly selected fields were counted and analyzed.

After incubation for 48 h, the transfected SiHa and HeLa cells were harvested. Then, the cells were washed twice with precooled PBS. The cells were then stained with an Annexin V-PE, 7-AAD Kit (BD, USA) for 15 min at room temperature in the dark. Flow cytometry was used to analyze cell apoptosis via a NovoCyte (Agilent, USA). Apoptotic cells (annexin V-PE-positive and 7-AAD-negative for early apoptosis and both annexin V-PE- and 7-AAD-positive for late apoptosis) were identified.

The P1 and P2 sequences of the HIF1A-AS2 promoter were subsequently cloned and inserted into the luciferase reporter vector pGL3-basic for the promoter activity assay. The pGL3-P1, pGL3-P2, and pGL3-P2-MUT plasmids were transfected into 293T cells, and the luciferase activities were subsequently measured 48 h later. In addition, pGL3-P1, pGL3-P2,was co-transfected with pcDNA3.1-c-Jun or pGL3-basic, and a dual-luciferase reporter assay kit (Promega, USA) was used to measure luciferase activity after 48 h. Moreover, the HIF1A-AS2-WT, HIF1A-AS2-MUT, Radixin-WT and Radixin-MUT sequences were cloned and inserted into the luciferase reporter vector pmirGLO. HeLa cells were plated in 24-well plates, and then HIF1A-AS2-WT/MUT and Radixin-WT/MUT were co-transfected with miR-34b-5p mimics or miR-NC into HeLa cells with Lipo2000 reagent (Invitrogen, USA). Luciferase activity was subsequently measured 48 h later via a dual-luciferase reporter assay kit (Promega, USA). The experiments were conducted in triplicate and repeated three times.

The binding between HIF1A-AS2 and miR-34b-5p, and between radixin and miR-34b-5p was explored by using an RNA-Binding Protein Immunoprecipitation Kit (EMD Millipore, USA) according to the manufacturer’s instructions. Briefly, the cells were lysed, and the resulting lysis solutions were subsequently incubated with an antibody against Ago2 (CST, 2897S, USA) or an isotype control IgG. The RNA-protein complexes were immunoprecipitated with protein A agarose beads, and the RNA-protein complexes were subsequently washed 5 times. After washing, the RNA/bead complexes were incubated with proteinase K at 55°C for 30 min to digest the proteins and purify the immunoprecipitated RNA. The final concentration of RNA was 115 ng/µL. The RNA was reverse transcribed to cDNA and analyzed by RT-qPCR. The relevant primers used are detailed in Supplementary Table SVI .

The cells were lysed with RIPA buffer (Thermo Fisher Scientific, USA) supplemented with protease inhibitors (CWBIO, Beijing, China). Of protein, 20µg was separated on 10% SDS-PAGE (Beyotime, Beijing, China) and then transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, USA). The molecular weight of radixin is about 80 kDa, and in our study, the target protein band was between 70–100 kDa. The membranes were subsequently incubated overnight at 4°C with suitable dilutions of primary antibodies against the following antigens: radixin (1:2000; ab52495, Abcam, USA) and GAPDH (1:3000, ab9485, Abcam, USA). The membranes were subsequently washed with TBST three times for 10 min each in a shaker. After washing, the membranes were incubated with goat anti-rabbit secondary antibodies (1:5000, 7074P2, CST, USA) for 120 min at room temperature. The membranes were subsequently rewashed three times with TBST buffer. The protein bands were detected via enhanced chemiluminescence and imaged via a gel imager system (UE, UK). Finally, densitometric analysis of the protein bands was performed via ImageJ.

SPSS Statistics 26.0 software (IBM, SPSS, Chicago, IL, USA) was used for analyzing the data. The measurement data were expressed as mean ± standard deviation. When data were normally distributed and had equal variance, t test was used for analyzing comparisons between two groups, while comparisons among multiple groups, one-way analysis of variance (ANOVA) with Tukey’s post hoc test. Paired samples were compared using a Wilcoxon matched pairs signed rank sum test. P<0.05 indicated statistically significant difference, while P<0.01 indicated extremely significant difference.

To determine the expression of HIF1A-AS2 in CC tissues, we collected 20 cervical cancer patient tissues from 2020–2022. All biopsies of squamous type are shown in Supplementary Table SV 18 of 20 patients were diagnosed with squamous cell carcinomas and just 2 patients were diagnosed with adenocarcinomas. QRT-PCR was used to detect the expression of HIF1A-AS2 and radixin in 20 cervical cancer patient tissues and their matched normal tissue samples. The results revealed that the expression of HIF1A-AS2 was significantly greater in cervical cancer tissues (61.864 ± 100.309) than in normal tissues (1.054 ± 1.210) (P<0.01; Figure 1A ). The expression of radixin (176.472 ± 42.6) was also greater than that in normal tissues (12.7 ± 19.76) (P<0.05; Figure 1B ). To further explore the expression of HIF1A-AS2 in CC cells, the expression of HIF1A-AS2 and radixin was detected in CC cells via qRT-PCR, and the results revealed that HIF1A-AS2 was upregulated in three CC cell lines (HeLa (2.02 ± 0.19), Siha (5.56 ± 0.72) and Caski (2.63 ± 0.62)) compared with HaCat cells (1.0 ± 0.1) (P<0.05; Figure 1C ). As for radixin, the results revealed that radixin was upregulated in three CC cell lines (HeLa (4.67 ± 1.15), Siha (3.85 ± 0.53) and Caski (1.84 ± 0.22)) compared with HaCat cells (1.0 ± 0.09) (P<0.05; Figure 1D ).

The efficiency of siRNA knockdown was detected by qRT-PCR (P<0.01; Figure 2A ), and 50 nM siRNA had the highest knockdown efficiency (si-HIF1A-AS2 0.346 ± 0.017, si-NC 1.000 ± 0.117) and was used for the following experiments. The results of the CCK-8 assay revealed that, compared with the control, the knockdown of HIF1A-AS2 inhibited the proliferation of both HeLa and SiHa cells (P<0.01; Figure 2B ). Knockdown of HIF1A-AS2 significantly inhibited the invasion ability of both HeLa and SiHa cells (P<0.01; Figure 2C ). To explore the cell apoptosis rate, flow cytometry was used, and the results demonstrated that knockdown of HIF1A-AS2 markedly increased the apoptosis rate of both HeLa and SiHa cells (P<0.01; Figure 2D ).

To elucidate the functions of HIF1A-AS2, the full-length cDNA of HIF1A-AS2 was cloned and inserted into an adenovirus encoding the green fluorescent protein (GFP) gene. The overexpression efficiency was determined using both fluorescence microscopy and qRT-PCR (P<0.05; Figure 3A ). According to the results, an MOI = 50 was used for the following experiments.

The results of the CCK-8 assay revealed that overexpression of HIF1A-AS2 (adv-HIF1A-AS2) significantly promoted the proliferation of both HeLa and SiHa cells (P<0.01; Figure 3B ). The Transwell assay results indicated that the invasion capacity of the HeLa and SiHa cells was significantly greater than that of the control cells (P<0.01; Figure 3C ). Furthermore, the flow cytometry results revealed that, compared with the negative control (adv-NC), adv-HIF1A-AS2 impaired the apoptosis rate of both HeLa and SiHa cells (P<0.01; Figure 3D ).

The PROMO and JASPAR databases were used to predict the potential transcription factors of HIF1A-AS2 and binding sites, and all the predicted transcription factors are shown in Supplementary Table SII . These results indicate that c-Jun may be a crucial transcription factor. In addition, Liu C et al. reported that the expression of c-Jun was elevated in cervical intraepithelial neoplasia (19). Hence, c-Jun is considered a transcription factor that can bind to the promoter of HIF1A-AS2. According to the prediction data, there are two suggested binding sites between c-Jun and the promoter of HIF1A-AS2. To validate which binding sites had promoter activity, dual-luciferase assays were performed. We first constructed two dual-luciferase reporter vectors that contain two sequences of the HIF1A-AS2 promoter, including two binding sites, P1 and P2 ( Figure 4A ). The results revealed that luciferase activity increased when the pGL3-P2 vector was transfected, whereas pGL3-P1 had the same effect as the empty vector (P<0.01; Figure 4B ). We subsequently constructed a pGL3-P2-MUT vector, and the results revealed that the luciferase activity did not significantly change compared with that of the empty vector (P>0.05; Figure 4B ). The above results indicate that the P2 sequence of the HIF1A-AS2 promoter has promoter activity.

To further explore whether HIF1A-AS2 was activated by c-Jun, pGL3-P2 and pcDNA3.1-c-Jun were cotransfected into 293T cells, and the results revealed that luciferase activity increased when pcDNA3.1-c-Jun and pGL3-P2 were cotransfected (P<0.01; Figure 4C ). Moreover, ChIP assays confirmed that c-Jun could bind to the P2 sequence of the HIF1A-AS2 promoter (P<0.01; Figure 4D ). Taken together, these results demonstrated that the expression of HIF1A-AS2 may be regulated by the transcription factor c-Jun in cervical cancer cells.

The TargetScan and miRDB databases were used for the prediction of the target miRNAs of HIF1A-AS2 and the binding sites, and the prediction results are shown in Supplementary Table SIII . Research has reported that the expression of miR-34b-5p is reduced in cervical cancer and promotes the progression of cervical cancer (22). Therefore, miR-34b-5p was chosen, as shown in Figure 5A . miR-34b-5p has a binding site for HIF1A-AS2 ( Figure 5A ). To validate the interaction relationship, we constructed two sequences, one containing the binding site (HIF1A-AS2-WT) and the other containing the mutant binding site (HIF1A-AS2-Mut) ( Figure 5A ). These two sequences were cloned downstream of the luciferase reporter gene. The effects of miR-34b-5p on the HIF1A-AS2-WT or HIF1A-AS2-Mut luciferase reporter system were subsequently determined. The results demonstrated that cotransfection with wild-type (HIF1A-AS2-WT) plasmids and miR-34b-5p mimics significantly decreased the luciferase activity of HIF1A-AS2-WT but did not affect that of the mutant type (HIF1A-AS2-MUT) (P<0.01; Figure 5B ).

The interaction of lncRNAs with miRNAs causes RNA-induced silencing complex (RISC) formation, and Ago2 is a critical protein of the RISC [12]. Therefore, we performed a RIP assay to further verify the binding interaction of anti-Ago2. qRT-PCR was used to determine the expression of HIF1A-AS2 and miR-34b-5p in the Ago2 immunoprecipitation complex. The results demonstrated that HIF1A-AS2 and miR-34b-5p were significantly enriched in Ago2-containing beads in both HeLa and Siha cells compared with those in the IgG group (P<0.01; Figure 5C ).

To determine the regulatory relationship, we performed qRT-PCR to determine the expression of HIF1A-AS2 and miR-34b-5p in HeLa and Siha cells. The upregulation of HIF1A-AS2 led to a reduction in miR-34b-5p levels, whereas the silencing of HIF1A-AS2 resulted in an increase in miR-34b-5p expression, with statistically significant differences (P<0.01; Figure 5D ). Taken together, these data revealed that HIF1A-AS2 could bind with miR-34b-5p and that the expression of miR-34b-5p was negatively regulated by HIF1A-AS2.

miRNAs exert their functions by binding to their targets. TargetScan and miRDB were used to predict the targets of miR-34b-5p, and all the targets of miR-34b-5p are presented in Supplementary Table SIV . Specifically, increased expression of the radixin protein can promote the progression of cervical cancer (23). Therefore, radixin was considered a potential target, as shown in Figure 6A . To verify the interaction between radixin and miR-34b-5p, a dual-luciferase assay was performed as previously described. We constructed two sequences, one containing the binding site (Radixin-WT) and the other containing the mutant binding site (Radixin-Mut) ( Figure 6A ). These two sequences were cloned downstream of the luciferase reporter gene. The effects of miR-34b-5p on the Radixin-WT or Radixin-Mut luciferase reporter system were subsequently determined. The results revealed that the luciferase activity decreased when the cells were cotransfected with wild-type (Radixin-WT) and miR-34b-5p mimics but had no significant effect on the mutant type (Radixin-MUT) (P<0.01; Figure 6B ). The results of the luciferase activity assay indicated that miR-34b-5p might bind to radixin mRNA.

Moreover, to confirm that miR-34b-5p could indeed bind with radixin mRNA in CC cells, a RIP assay was performed using anti-Ago2. The results revealed that the expression of both miR-34b-5p and radixin mRNAs was increased in the anti-Ago2 immunoprecipitation complex (P<0.001; Figure 6C ). Subsequently, qRT-PCR was used to explore how miR-34b-5p regulates the expression of radixin mRNA. We transfected miR-34b-5p mimics to overexpress miR-34b-5p and miR-34b-5p inhibitors to knockdown miR-34b-5p in both HeLa and Siha cells, the inhibiting effect of miR-34b-5p was shown in Supplementary Figure S1 . The results revealed that the overexpression of miR-34b-5p decreased the expression of radixin mRNA, whereas the knockdown of miR-34b-5p increased the expression of radixin mRNA (P<0.05; Figure 6D ). In addition, we performed Western blot experiments to further verify the regulatory relationship. Compared with the control, transfection with the miR-34b-5p mimics decreased the amount of the radixin protein, whereas the miR-34b-5p inhibitors increased the amount of the radixin protein ( Figure 6E , Supplementary Figure S2 - Supplementary Figure S5 ). The above results confirmed that miR-34b-5p could negatively regulate the expression of radixin at both the mRNA and protein levels. In addition, western blot results revealed that the expression of the radixin protein decreased when cells were transfected with si-HIF1A-AS2 to knockdown HIF1A-AS2, whereas the expression of the radixin protein increased when they were cotransfected with si-HIF1A-AS2 and miR-34b-5p inhibitors ( Figure 6F ). These findings indicated that HIF1A-AS2 could bind with miR-34b-5p to positively regulate radixin.