Hypoxia-induced lncRNA STEAP3-AS1 activates Wnt/β-catenin signaling to promote colorectal cancer progression by preventing m⁶A-mediated degradation of STEAP3 mRNA

Antibodies used are as follows: anti-STEAP3 (cat# sc-376327, 1:1000 dilution), anti-Axin (cat# sc-293190, 1:500 dilution), anti-p-GSK3β (Ser 9) (cat# sc-373800, 1:1000 dilution) and anti-GSK3β (cat# sc-9166, 1:1000 dilution) were purchased from Santa Cruz Biotechnology; anti-ZO-1 (cat# 8193, 1:1000 dilution), anti-E-cadherin (cat# 3195, 1:1000 dilution), anti-Vimentin (cat# 5741, 1:1000 dilution), anti-Claudin-1 (cat# 13995, 1:1000 dilution), anti-GAPDH (cat# 5174, 1:2000 dilution), anti-β-catenin (cat# 8480, 1:1000 dilution), anti-Histone H3 (cat# 4499, 1:1000 dilution), anti-Snail (cat# 3879, 1:1000 dilution), anti-Slug (cat# 9585, 1:1000 dilution) and anti-HIF-1α (cat# 36169, 1:1000 dilution) were purchased from Cell Signaling Technology; anti-m⁶A (cat# ab208577, 1:1000 dilution), anti-YTHDF1 (cat# ab252346, 1:1000 dilution), anti-YTHDF2 (cat# ab246514, 1:1000 dilution), anti-METTL3 (cat# ab195352, 1:1000 dilution), anti-METTL14 (cat# ab220030, 1:1000 dilution) were purchased from Abcam.

Regents used in this study: Dimethyloxalylglycine (DMOG, cat# S7483) and actinomycin D (Act D, cat# S8964) were obtained from Selleck; DMSO (cat# D2650), Crystal Violet (cat# C0775), MTT (cat# M2128), FeSO₄ (cat# 215422) and CoCl₂ (cat# 15862) were obtained from Millipore Sigma. DMOG, actinomycin D and CoCl₂ were dissolved in DMSO; Crystal Violet, MTT and FeSO₄ were dissolved in ddH₂O.

CRC cells (HCT116, RKO, DLD-1, LoVo, SW480, SW620, HT29), the nonmalignant human colon epithelial cell line NCM460 and HEK293T were obtained from ATCC and maintained in Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 10% fetal bovine serum (BI), 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen).

The generation of patient-derived organoid was performed as previously described [34]. Briefly, CRC tissue from patients was minced into small fragments (1–3 mm³) pieces and digested by collagenase II and TrypLE Express Enzyme at 37 °C. The organoid was then embedded within Matrigel and cultured in 48-well plates supplemented with human complete feeding medium in a humidified incubator with 5% CO₂ at 37 °C.

Lentiviral vectors or target plasmid were co-transfected into HEK293T cells with packaging vectors. Lentivirus particles were harvested at 24 and 48 h after transfection to infect CRC cells. shRNA plasmids were synthesized, annealed and cloned into pLKO.1 vector. Stable overexpressed gene were generated by using the pCDH expression vectors.

The sequences of shRNA and overexpression primers were as follows: STEAP3-AS1 shRNA#1, F 5′-CCGGGCACCTTTAAACTGTCCTACACTCGAGTGTAGGACAGTTTAA AGGTGCTTTTTG-3′, R 5′-AATTCAAAAAGCACCTTTAAACTGTCCTACACTCGAG TGTAGGACAGTTTAAAGGTGC-3′; STEAP3-AS1 shRNA#2, F 5′-CCGGG CTGTTCCGTGGAGCCATTATCTCGAGATAATGGCTCCACGGAACAGCTTTTTG-3′, R 5′-AATTCAAAAAGCTGTTCCGTGGAGCCATTATCTCGAGATAATGGCTCCACGGAA CAGC-3′; STEAP3-AS1 overexpression primer, F 5′-GAATTCAGACCC AAACCCCAGAGTCAT-3′, R 5′-GGATCCAGAGATGGGACCTCCCTGTGT-3′.

CRC cells were harvest after washing twice with cold PBS. The pellet was resuspended and incubated on ice in lysis buffer for 30 min, and then the lysate was obtained by centrifugation at 12000×g for 10 min. Proteins were separated by SDS-PAGE, transferred onto PVDF membranes, blocked in 5% nonfat milk and then blotted with specific antibodies.

Total RNA was extracted from tissues and cells using Trizol (Invitrogen, USA) and reverse transcribed to cDNA by using a Reverse Transcription Kit (Takara, Dalian, China). The RNA transcripts levels were analyzed using a Bio-rad CFX96 real-time PCR system (Biorad, USA) and normalized to GAPDH. Primers used in RT-qPCR were listed in Supplementary Table S1.

Cell viability was detected by adding 5% MTT and incubation at 37 °C for 2 h at 0, 24, 48, 72 and 96 h. The absorbance of each well was measured at 570 nm. All experiments were performed in at least triplicate.

For colony formation assay, 500 cells were planted and maintained in each well of 24 well plates for 2 weeks. The medium was refreshed every 3 days. Colonies were fixed with 4% paraformaldehyde for 1 hour and then stained with 0.1% crystal violet for 30 min and washed with ddH₂O. The colony numbers of each well were counted.

Migration and invasion assays were performed using Transwell chamber system (Corning, USA). For migration assay, 5 × 10⁴ cells were seeded in the upper chamber of an insert with 0.2 ml FBS-free starvation medium, and 0.6 ml culture media with 20% FBS were added outside the chamber in the wells of the plate. For invasion assays, the upper chamber of the insert was pre-coated with Matrigel (Millipore Sigma) before plating cells. After incubation for 48 h, cells were fixed with 4% paraformaldehyde for 1 hour and then stained with 0.1% crystal violet for 30 min. After rinsing with water and airing, migrating or invading cells were imaged and counted using a Leica DM2500 microscope.

After seeding onto the glass cover slides (WHB scientific, cat# whb-24-cs) and leaving at 37 °C overnight, cells were fixed by 4% formaldehyde, and then permeabilized with 0.3% Triton X-100 and blocked with 5% BSA. After being incubated with indicated primary antibodies (1:100 dilution) at 4 °C overnight, slides were incubated with Alexa Fluor 488/594-conjugated secondary antibodies (1:200 dilution) for 1 hour at room temperature, followed by DAPI staining of the nuclei (Solarbio, cat# C0060, 1:4000 dilution) at room temperature for 10 minutes. Finally, images were captured using confocal laser scanning microscopy (Carl Zeiss Microimaging) in Pub-lab of West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University.

Immunohistochemistry was performed following a standard protocol. Briefly, slides were deparaffinized with xylene and ethanol, and the endogenous peroxidase was blocked by 3% H₂O₂ for 10 minutes. After being incubated in retrieval buffer and boiled for 3.5 minutes, slides were washed with PBS for 3 times and blocked with 5% normal serum. Then the slides were incubated with primary antibody at 4 °C overnight followed by 60 minutes-treatment of MaxVision HRP solution (MXB Biotechnology, cat# 5020). After being stained with DAB Peroxidase Substrate (MXB Biotechnology, cat# 0031), the antigen levels were detected using EnVision Detection System (Agilent Technologies, K5007).

After being collected and washed 2 times with pre-cooled PBS, cells were lysed using 1 mL complete protease and phosphatase inhibitor added IP lysis buffer (100 mM NaCl, 20 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 0.1% NP-40) and incubated on ice for 30 min. After centrifugation at 12000 rpm at 4 °C for 10 min, 80 μL of the supernatant was transferred and mix with 5× lording buffer in a new tube as the control. The remaining supernatant was transferred to a new tube and incubated with 1 μg indicated antibody overnight at 4 °C. After being washed 3 times with IP lysis buffer, 30 μL of protein A/G agarose beads (GE Healthcare, cat# 17-0963-03) were added into the mixture, and rotated for another 2 h at 4 °C. The beads were washed using washing buffer (150 mM NaCl, 0.5 mM EDTA, 20 mM Tris-HCl, pH 7.4, 0.5% NP-40) for 3 times, and proteins were separated by SDS-PAGE loading buffer with 10 min incubation at 100 °C, followed by immunoblotting analysis.

Six to eight weeks old male BALB/c nu/nu mice were used. For orthotopic implantation, 1 × 10⁷ PBS suspended cells was injected subcutaneously. Tumors were collected and sliced into 3 × 3 mm pieces for orthotopic implantation once their diameter reached 1 cm. After being anesthetized and laparotomized, CRC tissues were positioned in the wound and tied down using a suture. The intestines were then placed back followed by closing the peritoneum after sterilization. For the spleen injection model, 5 × 10⁵ cells were injected into the spleen through an incision on the left side of abdomen. Mice were sacrificed at 6–8 weeks after implantation or injection to examine the lung and liver metastases. H&E staining was performed after tissues were fixed in 4% formaldehyde.

Tg (flk1:eGFP) zebrafishes were used to establish zebrafish xenograft model of human CRC. After being anesthetized with 0.04 mg/mL tricaine (Millipore Sigma), zebrafishes received a microinjection of 200 mCherry stably expressed CRC cells. The tumor cells-bearing zebrafishes were randomly divided into two groups after examination of mCherry fluorescent signal in the next day. Zebrafishes were maintained under normal oxygen or hypoxic conditions for a period of 3 days. Finally, the mCherry fluorescent signal was read to examine the distribution and metastasis of CRC cancer cells using a stereo microscope.

In situ hybridization assays were performed to evaluate the STEAP3-AS1 levels in the CRC xenograft model. Sections were deparaffinized with xylene and ethanol, and the endogenous peroxidase was blocked by 3% H₂O₂ for 10 minutes at room temperature. After being incubated with 3% citric acid and freshly diluted pepsin for about 60 s at 37 °C, slides were washed 3 times with PBS and fixed with 1% formaldehyde with addition of 0.1% DEPC for 10 min at room temperature. The sections were then pre-hybridized at 40 °C for 2 hours in a hybrid box with 20 mL 20% glycerinum placed in the bottom. Twenty microlitre hybridization liquid was then added and left at 40 °C overnight. After being washed successively with 2 × SSC, 0.5 × SSC, 0.2 × SSC (15 minutes for each), the sections were blocked with blocking reagent for 30 min at 37 °C. Next, sections were incubated with biotin-digoxigenin for 1 hour at 37 °C followed by Strept Avidin-Biotin Complex (SABC) for 20 min at 37 °C. After being incubated with biotin peroxidase, sections were subjected to DAB. This was followed by hematoxylin redye, dehydration using graded ethanol and vitrification with dimethylbenzene. Sections were analysed using an EnVision Detection System (Agilent Technologies, K5007).

After being fixed with 4% formaldehyde, cells were permeabilized with 0.3% Triton X-100 and blocked with 5% BSA. Cells were then pre-hybridized at 37 °C for 30 min followed by incubation with lncRNA FISH Probe Mix for hybridization at 37 °C overnight. After being washed three times, DAPI was added to stain the nucleus. Images were captured at 555 nm using confocal laser scanning microscopy (Carl Zeiss Microimaging).

Cells were harvested for nuclear isolation before incubating with m⁶A antibody for 4 h at 4 °C in 1× immunoprecipitation buffer supplemented with RNase inhibitors. Prewashed protein A/G magnetic beads (30 μL) were added and incubated overnight at 4 °C. After washing 3 times and incubating with proteinase K digestion buffer, RNA was finally extracted using phenol-chloroform and analyzed by qPCR.

Briefly, the in vitro biotin-labelled RNAs were transcribed with 10 × Biotin RNA labeling mix (Roche, cat# 1165597910) and T7 enzyme mix (New England Biolabs, cat# M0251S), and heated at 65 °C for 5 min. Samples were cooled to room temperature to form the proper secondary structure in the presence of 10 mM HEPES, 10 mM MgCl₂ and 0.1 M NaCl. The RNAs were then incubated with Streptavidin Magnetic Beads (Beyotime Biotechnology, cat# P2151) for 15–30 minutes at room temperature with agitation. Protein lysates were then mixed with the RNA-beads complex for 30–60 minutes at 4 °C with agitation or rotation. The pulldown complexes were then washed and boiled at 95–100 °C for 5–10 minutes, followed by immunoblotting.

Chromatin immunoprecipitation assay was performed using a ChIP kit (Millipore Corp.) following the manufacturer’s protocol. Firstly, after being cross-linked with 1% formaldehyde, DLD-1 or SW480 cells (1 × 10⁷) were sonicated at 30% maximum power for 8 min (5 s pulse after every 10 s). Supernatants were transferred into a new tube for immunoprecipitation with 1 μg of specific antibodies or IgG antibody after centrifugation at 15000×g for 10 min. The target protein and their binding DNA complex was sedimented using prewashed agarose beads (GE Healthcare, cat# 17–0963-03). After elution and purification, DNA was analyzed by RT-qPCR. Primers used in ChIP-qPCR are listed in Supplementary Table S2.

Gene expression data and the corresponding clinical information were obtained from TCGA repository using the *GDCquery* function of the TCGAbiolinks R package as well as a recent updated clinical data resource [35]. LncRNAs upregulated in CRC were analyzed using DESeq2 R package (log2FoldChange > 0.5 & padj< 0.05). The hypoxia signature score was calculated based on a gene set as previously described [36] using ‘ssgsea’ method of the GSVA R package, and correlations with each lncRNAs were analyzed by Pearson Correlation (coefficient > 0.15 & padj < 0.05). For the survival and kaplan-meier analysis, patients were stratified into two groups (high & low expression) using surv_cutpoint and surv_categorize function of the survminer R package and the progression-free survival was analyzed.

Total RNA was extracted using TRIzol (Invitrogen, Carlsbad, CA, USA). RNA-sequencing analysis was performed at Novogene (Tianjin, China). TCGA data was downloaded and extracted using TCGAbiolinks v.2.22. The rlog transformation and differential expression analysis were performed using DESeq2 v.1.34. The differentially expressed genes (Padj ≤0.05 and log2 (fold change) ≥ 0.5) were subjected to enriched biological pathways analysis using DAVID bioinformatic Resources (2021 Update). Enriched pathways were visualized using clusterProfiler v.4.22 and ComplexHeatmap v.2.10.

All data were from at least three independent experiments and presented as the mean ± SD. The P value was calculated using GraphPad (version 9). P < 0.05 was considered statistically significant. Comparisons between two groups and repeated measurements over a period of time were performed by two-tailed Student t test, one-way analysis of variance (ANOVA) or two-way ANOVA. Correlation between two independent groups were performed using Pearson’s Chi-square test. Kaplan-Meier method were used to generate survival curves.

To screen for essential lncRNAs that relate to CRC progression under hypoxia, we designed the screening workflow shown in Fig. 1A. Briefly, we analyzed the expression of antisense lncRNAs in the CRC cohort from TCGA database and found 212 candidate lncRNAs that were highly expressed in tumor tissues. The prognostic significance of these lncRNAs were then analyzed, revealing a total of 98 antisense lncRNAs that correlated with poor survival. Finally, we identified 19 hypoxia-related antisense lncRNAs as determined by a correlation analysis with a hypoxic signature as previously described [36] (Fig. 1A). Among these candidates, lncRNA STEAP3-AS1 was highly correlated with HIF-1α (R = 0.46) and hypoxia signature genes in CRC tissues from TCGA (Fig. 1B and Fig. S1A). Importantly, lncRNA STEAP3-AS1 was highly expressed in clinical CRC tissues and positively correlated with poor prognosis of CRC patients (Fig. 1C-D), suggesting STEAP3-AS1 as a potential diagnostic or prognostic biomarker for CRC.

Next, we ascertained the increased expression of STEAP3-AS1 in response to hypoxic stress. As shown in Fig. 1E-G, STEAP3-AS1 expression was elevated in hypoxic CRC cells following treatment with 1% oxygen, DMOG, or CoCl₂ in vitro. Consistently, results from qPCR and ISH analysis on DLD-1 xenografts revealed that the levels of STEAP3-AS1 in the inner regions were significantly higher than those in marginal regions (Fig. S1B-G). To further verify the involvement of STEAP3-AS1 in hypoxia promoted CRC progression, a zebrafish orthotopic CRC model was built using SW480 mCherry cells. The zebrafishes were maintained in normal or 8% oxygen condition [37]. As shown in Fig. 1H-I, hypoxia promoted the dissemination of CRC cells and upregulated the expression of STEAP3-AS1, as well as other downstream hypoxic targets, indicating that STEAP3-AS1 upregulation may be essential for hypoxia-mediated tumor metastasis. Collectively, these data suggest that STEAP3-AS1 is upregulated under hypoxic condition and may participate in hypoxia-promoted tumor progression both in in vitro and in vivo CRC models.

To determine whether STEAP3-AS1 is a direct target of HIF-1α, we explored the characteristics of the genomic locus of STEAP3-AS1. Seven putative hypoxia response elements (HREs) were located near STEAP3-AS1 locus (Fig. S1H). Next, we performed chromatin immunoprecipitation (ChIP) assays to unbiasedly demonstrate the binding of HIF-1α with the predicted HREs in STEAP3-AS1. These results suggested that HIF-1α was significantly enriched at HRE 1, 5, and 7 (Fig. 1J). In addition, analysis from coding potential calculator (CPC) and coding potential assessment tool (CPAT) databases both revealed that STEAP3-AS1 had a low potential for protein encoding (Fig. S1I). Then, the subcellular distribution of STEAP3-AS1 was determined using fluorescence in situ hybridization (FISH) analysis. The results suggested that lncRNA STEAP3-AS1 was mainly located in the nucleus (Fig. 1K). The nuclear localization of STEAP3-AS1 was further validated by the nuclear/cytoplasmic RNA fractionation assay (Fig. 1L and Fig. S1J). Collectively, these findings reveal that lncRNA STEAP3-AS1 is a direct HIF-1α target gene and mainly located in the nucleus in human CRC.

To study the biological role of STEAP3-AS1 in CRC, we firstly determined the basal level of STEAP3-AS1 expression in several CRC cell lines (Fig. 2A). Subsequent viability and colony formation assay suggested that stably knocked down STEAP3-AS1 expression could clearly attenuate proliferative capability in CRC cells (Fig. 2B-F and Fig. S2A-B). In contrast, overexpression of STEAP3-AS1 significantly accelerated cell proliferation rate (Fig. S2C-E). This effect of STEAP3-AS1 on promoting proliferation was further evidenced by EdU staining assay (Fig. 2G and Fig. S2F). To further ascertain the function of STEAP3-AS1 in CRC proliferation, a subcutaneous xenograft model was established. Consistent with the in vitro results, STEAP3-AS1 knock down significantly decreased tumor weight and tumor volume compared with those in the control group (Fig. 2H-J). Moreover, the protein level of Ki67, a proliferative maker was detected using immunohistochemistry (IHC) assays. The results confirmed that STEAP3-AS1 knocked down caused reduction in Ki67 expression (Fig. 2K). In addition, we also used a patient-derived organoid (PDO) model to further verify the function of STEAP3-AS1. As shown in Fig. 2L, infection with lenti-shSTEAP3-AS1 or treatment of antisense oligo targeting STEAP3-AS1 significantly suppressed the growth of PDOs, indicating the important role of STEAP3-AS1 in promoting CRC progression. Taken together, these results indicate that lncRNA STEAP3-AS1 promotes CRC cell proliferation both in vitro and in vivo.

We then investigated the role of STEAP3-AS1 in the mobility of CRC cells. To this end, we firstly constructed tail vein injection models using DLD-1 cells. Following sacrificed, their lungs were isolated for further investigation of the metastatic nudes. Hematoxylin and eosin (H&E) staining results suggested that lncRNA STEAP3-AS1 deficiency remarkably attenuated the number of lung metastatic nodes (Fig. 3A and Fig. S3A). Furthermore, two metastatic liver colonization models were established by inoculating DLD-1 cells into the spleens of BALB/c nude mice or orthotopically implanting DLD-1-derived CRC xenograft into the cecum of BALB/c nude mice (Fig. S3B). H&E staining of liver isolated from these two models also revealed that knockdown of lncRNA STEAP3-AS1 decreased the number of liver metastatic lesions compared with those in the control group (Fig. 3B). In vitro experiments including wound healing and transwell assays were performed to demonstrate the role of lncRNA STEAP3-AS1 on cell mobility of CRC cells. Consistently, knockdown of lncRNA STEAP3-AS1 significantly suppressed CRC cell migration and invasion (Fig. 3C-G), while overexpression of it increased the migratory and invasive abilities of CRC cells (Fig. S3C-D). Next, the expression of several epithelial-to-mesenchymal transition (EMT) markers was tested, as it is well documented that EMT is one of the hallmarks of elevated cell mobility. Results from western blotting and qPCR analysis revealed that STEAP3-AS1 knockdown up-regulated the epithelial markers E-cadherin, claudin-1 and Zona occludin-1 but suppressed the mesenchymal markers Snail, Slug and vimentin in DLD-1 and SW480 cells (Fig. 3H-J). However, overexpression of STEAP3-AS1 could up-regulate the mesenchymal markers but suppress the epithelial markers (Fig. S3E). This was further validated by immunofluorescence (IF) assay, in which the fluorescence intensity of ECAD was distinctly enhanced with STEAP3-AS1 knockdown (Fig. S3F). Taken together, these data show that lncRNA STEAP3-AS1 facilitates tumor metastasis in CRC cells both in vitro and in vivo.

Since most antisense lncRNAs exert their biological function by regulating their neighboring genes, we examined whether lncRNA STEAP3-AS1 has an effect on STEAP3 expression. Bioinformatic analysis revealed that the mRNA level of STEAP3 was positively correlated with STEAP3-AS1 expression both in CRC tissues from the TCGA datasets and CRC cell lines from the cancer cell line encyclopedia (CCLE) (Fig. 4A-B). Consistently, the mRNA (Fig. 4C) and protein (Fig. 4D) levels of STEAP3 were both decreased in STEAP3-AS1-knockdown DLD-1 and SW480 cells. Given that STEAP3-AS1 is transcriptionally induced by HIF-1α under hypoxia, we wondered whether the expression of STEAP3 is responsive to hypoxic stimuli. The data showed that the protein (Fig. 4E, S4A-B) and mRNA (Fig. 4F, S4A-B) levels of STEAP3 were significantly upregulated in CRC cells under hypoxia induced by 1% oxygen, DMOG, or CoCl₂. Furthermore, the expression of STEAP3 was obviously increased in SW480 mCherry cells under hypoxia (as indicated by the upregulation of HIF-1α and HIF-1α target genes) (Fig. 4G), as well as in the inner region of tumors from DLD-1 xenografts (Fig. S4C-D). In conclusion, these results suggest that STEAP3 is a hypoxia-induced gene positively regulated by STEAP3-AS1.

To further verify the regulatory role of STEAP3 in STEAP3-AS1-induced CRC proliferation, cell growth rate (Fig. 4H and Fig. S4E) and colony formation ability (Fig. 4I-K and Fig. S4F) were analyzed in STEAP3-AS1- and STEAP3-manipulated CRC cells. As shown in Fig. 4H-K and Fig. S4E-F, overexpression of STEAP3 could partly reverse the decreased cell proliferation induced by STEAP3-AS1 knockdown in CRC cells and STEAP3 knockdown could suppress cell proliferation promoted by STEAP3-AS1 overexpression. Consistently, replenishment of STEAP3 could also rescue the attenuated migration and invasion in STEAP3-AS1-knockdown CRC cells, as determined by wound healing (Fig. 4L and Fig. S4G-H) and cell migration/invasion assays (Fig. 4N and Fig. S4I). In contrast, knockdown of STEAP3 in STEAP3-AS1-overexpressing CRC cells displayed the opposite phenotype (Fig. 4M and Fig. S4J-K). Taken together, these results demonstrate that STEAP3 is responsible for STEAP3-AS1-mediated CRC progression.

To further dissect the mechanism underlying the regulatory effect of STEAP3-AS1 on STEAP3 mRNA, we predicted the potential binding proteins of STEAP3-AS1 using the AnnoLnc2 database (Fig. 5A). Among the putative binding targets, a reader for RNA N⁶-methyladenosine (m⁶A) modification, YTHDF2, was of particular interest for further study, as m⁶A modification can regulate translation, degradation and other RNA processes. The results of methylated RNA immunoprecipitation (MeRIP) confirmed m⁶A modification of both STEAP3-AS1 and STEAP3 mRNA (Fig. 5B). In addition, RNA pulldown and RIP analysis showed that STEAP3-AS1 could directly bind to YTHDF2 (Fig. 5C-D). To map the binding fragment of STEAP3-AS1 responsible for YTHDF2 interaction, we generated truncated mutants of STEAP3-AS1 (1–758 bp (F1), 759–1988 bp (F2) and 1989–3218 bp (F3)) (Fig. 5E). Further RNA pulldown assay revealed that the 3′-fragment (1989–3218 bp, F3) of STEAP3-AS1 was responsible for its interaction with YTHDF2 (Fig. 5E), which was consistent with the predicted sequence from the AnnoLnc2 database (Fig. S5A).

To determine the related regulators for STEAP3 m⁶A modification, we knocked down m⁶A writers (METTL3, METTL14) and m⁶A readers (YTHDF1, YTHDF2) in CRC cells. As shown in Fig. S5B and Fig. S5C, the protein level and mRNA level of STEAP3 were decreased by METTL14 and YTHDF2, indicating that METTL14-mediated m⁶A modification of STEAP3 might promote its degradation via YTHDF2. Moreover, MeRIP analysis of STEAP3 in METTL14-knockdown CRC cells confirmed that METTL14 is the m⁶A writer for STEAP3 (Fig. S5D). The putative m⁶A modification sites are in exon 3 and exon 2, respectively, for STEAP3 and STEAP3-AS1, as predicted by SRAMP database (http://​www.​cuilab.​cn/​sramp) (Fig. S5E).

As STEAP3 mRNA underwent m⁶A modification (Fig. 5B) but exhibited no obvious interaction with YTHDF2, we speculated that STEAP3-AS1 may bind to YTHDF2 thus disrupting STEAP3-YTHDF2 interaction. To verify this hypothesis, RNA pulldown and RIP assay were performed in control or STEAP3-AS1-knockdown CRC cells. The data revealed that knockdown of STEAP3-AS1 significantly increased STEAP3 binding to YTHDF2 (Fig. 5F-G). Furthermore, STEAP3-AS1-knockdown resulted in accelerated STEAP3 degradation as examined by relative mRNA levels after actinomycin D (ACD) treatment (Fig. 5H), while overexpression of STEAP3-AS1 showed the opposite effect (Fig. 5I). Collectively, these results indicate that STEAP3-AS1 binds to YTHDF2 to prevent m⁶A-mediated degradation of STEAP3 mRNA.

To identify the specific signaling pathway responsible for STEAP3-AS1-mediated CRC progression, RNA sequencing and the following pathways enrichment analyses were performed using control and STEAP3-AS1-knockdown DLD-1 cells. As shown in Fig. 6A, enriched biological pathways analysis revealed Wnt signaling as one of the most significantly changed pathways in STEAP3-AS1-knockdown cells, with the expression levels of 48 related genes significantly altered (Fig. 6B).

To validate the effect of STEAP3-AS1, activation of the Wnt signaling pathway was firstly examined by nuclear translocation of β-catenin in CRC cells with STEAP3-AS1 manipulation (the GSK3β inhibitor CHIR-99021 was used as a positive control). As shown in Fig. 6C-E, STEAP3-AS1 knockdown resulted in a decreased level of nuclear β-catenin, while STEAP3-AS1 overexpression increased the level of nuclear β-catenin. Consistently, immunofluorescence analysis demonstrated that β-catenin was primarily located in cytoplasm in STEAP3-AS1-knockdown CRC cells (Fig. S6A). In addition, the interaction of the β-catenin/Axin/GSK3β complex, which is essential for Wnt signaling inhibition, was impaired in HCT116 cells following STEAP3-AS1 overexpression (Fig. S6B-C). Furthermore, results of the TOP/FOP flash assay, which reflects the transcriptional activity of β-catenin-TCF, showed that STEAP3-AS1 knockdown significantly weakened the transcriptional activity of β-catenin-TCF in CRC cells (Fig. 6F), whereas enhanced expression of STEAP3-AS1 increased the activity (Fig. 6G). Also, relative mRNA levels of several Wnt downstream genes (including Wnt1, Wnt3a, Wnt5b, LGR5, AXIN2, and CCND1) were decreased, while the Wnt inhibitor protein DKK1 was enhanced in STEAP3-AS1-knockdown CRC cells (Fig. 6H). The results in STEAP3-AS1-overexpressing CRC cells were exactly opposite (Fig. S6D). Altogether, these results illustrate that STEAP3-AS1 activates Wnt/β-catenin signaling to promote CRC progression. More importantly, STEAP3-AS1 knockdown-triggered cytoplasmic retention of β-catenin was rescued by STEAP3 reintroduction (Fig. S6A), suggesting that STEAP3 plays an important part in STEAP3-AS1/Wnt/β-catenin axis.

STEAP3 is a well-documented metalloreductase involved in reducing Fe³⁺ to Fe²⁺, thus playing an essential role in cancer progression [38‐40]. Given the role of Fe²⁺ on phosphorylating and inactivating GSK3β [41‐44], we therefore speculated that STEAP3-AS1 activated Wnt/β-catenin signaling might be related to STEAP3-mediated Fe²⁺ generation. We firstly set about investigating the effect of STEAP3-AS1 on Fe²⁺ levels. Results suggested that STEAP3-AS1 knockdown distinctly decreased the cellular Fe²⁺ level in DLD-1 and SW480 cells, while STEAP3-AS1 overexpression elevated cellular Fe²⁺ level in HCT116 cells (Fig. 7A and S7A). Furthermore, the TOP/FOP flash assays suggested that STEAP3-AS1 knockdown impaired the transcriptional activity of Wnt/β-catenin signaling, while supplementation of exogenous Fe²⁺ could restore the decreased transcriptional activity of Wnt/β-catenin signaling in DLD-1 and SW480 cells (Fig. 7B and S7B). As nuclear localization of β-catenin is the prerequisite for exerting transcriptional function, we investigated the role of Fe²⁺ on the nuclear localization of β-catenin. Nuclear/cytoplasmic fractionation and subsequent WB analysis, together with IF assay revealed that STEAP3-AS1 knockdown hindered the nuclear localization of β-catenin, while supplementation of exogenous Fe²⁺ could restore its nuclear localization (Fig. 7C-D). As activity of β-catenin was reported to be regulated by GSK3β [44], we conducted experiments to determine whether GSK3β was involved in the effect of Fe²⁺ on β-catenin activity. Co-IP and subsequent WB analysis suggested that STEAP3-AS1 knockdown decreased Ser 9 phosphorylation of GSK3β (p-GSK3β) with little influence on the total GSK3β protein levels, and STEAP3-AS1 deficiency significantly enhanced the interaction between β-catenin and GSK3β or Axin (Fig. 7E-F). As expected, addition of Fe²⁺ could diminish the interaction between β-catenin and GSK3β or Axin (Fig. 7E-F). To further ascertain the function of GSK3β in STEAP3-AS1-mediated CRC progression, we performed siRNA-mediated knockdown of GSK3β expression under the manipulation of STEAP3-AS1. As shown in Fig. 7G-I and Fig. S7C-D, knockdown of GSK3β could partly reverse the decreased cell proliferation and rescue the migratory and invasive ability of STEAP3-AS1-knockdown CRC cells, further confirming the regulatory role of GSK3β in STEAP3-AS1-mediated CRC progression. In summary, our data reveal that the activation of Wnt/β-catenin signaling relies on STEAP3-AS1/STEAP3/Fe²⁺ axis mediated GSK3β inactivation.

Next, we employed multiple assays to ascertain the role of Fe²⁺ in STEAP3-AS1 mediated CRC progression. Cell viability and colony formation assay suggested that supplementation of exogenous Fe²⁺ could partly counteract the corresponding decrease in proliferation rate induced by STEAP3-AS1 knockdown (Fig. 7J and S7E-F), indicating that Fe²⁺ participated in STEAP3-AS1 promoted CRC proliferation. Moreover, wound healing and transwell assays also demonstrated that Fe²⁺ addition rescued the migration and invasion in STEAP3-AS1 knockdown DLD-1 and SW480 cells (Fig. 7K and S7G-H). Additionally, the epithelial phenotype caused by STEAP3-AS1 knockdown could be partially reversed by Fe²⁺ treatment, as evidenced by the reduction in epithelial markers and augmentation of mesenchymal markers (Fig. S7I). Collectively, our results suggest that Fe²⁺ participates in STEAP3-AS1/STEAP3/Wnt/β-catenin axis mediated CRC proliferation and metastasis.