The efficacy of anlotinib in CRC

Anlotinib, a novel multi-target TKI approved in China, exhibits significant anti-tumor activity against VEGFR. It has been demonstrated to inhibit the PI3K/AKT, MAPK/ERK, and RAF/MRK pathways, showing potential in treating various cancers [8]. Our preliminary in vitro findings suggest that anlotinib may exert direct cytotoxic effects on CRC cell lines SW620, HT29, SW480, and HCT116, thus assessing its efficacy. The results consistently show that anlotinib inhibits CRC cell line proliferation in a dose- dependent and time-dependent manner (Fig. S1A, B). Furthermore, the EdU incorporation and colony formation assays confirmed significant inhibition of SW620 and HCT116 proliferation by anlotinib (Fig. S1C, D). Flow cytometry analysis of cell cycle progression in SW620 and HCT116 cells indicated a decrease in S phase cells and an increase in G0/G1 phase cells under anlotinib treatment, hindering cell entry into the S phase (Fig. S1E). To explore the specific mechanism of anlotinib in colorectal cancer, mRNA sequencing was performed in the SW620 cell line. This sequencing revealed alterations in various tumor-related signaling pathways, including transcriptional misregulation in cancer, cell cycle, and the p53 signaling pathway (Fig. S1F). Further examination of anlotinib’s regulatory effect on the CRC cell cycle, using Western blot analysis of SW620 and HCT116 cells for CCNE1, CCNE2, CDK2, CCNA2, CDK1, and Histone H3, showed reduced protein levels following anlotinib treatment, especially at high concentrations (Fig. S1G). In vivo, xenograft models using CRC cell lines SW620 and HCT116 were established to validate anlotinib’s inhibitory effects. Anlotinib was administered at one-day intervals, and subcutaneous tumors were collected post-treatment. Treatments consistently led to reductions in tumor volume and weight (Fig. S2A-C), without significant differences in body weight, indicating its well tolerability (Fig. S2D). Anlotinib treatment also markedly increased survival in treated mice (Fig. S2E). The proliferation marker Ki-67 and microvascular density marker CD31 were reduced, but no difference was observed in the apoptosis marker TUNEL (Fig. S2F). In summary, Anlotinib exhibits promising anti-tumor effects on CRC cells both in vitro and in vivo.

Anlotinib-regulated circHAS2 highly expressed in CRC

Noncoding RNAs (ncRNAs) play pivotal roles in cancer regulation, yet their involvement in drug mechanisms remains elusive. To address this gap, we conducted whole transcriptome RNA sequencing in SW620 cells treated with anlotinib (Fig. 1A). Our analysis revealed 113 differentially expressed lncRNAs and 29 differentially expressed circRNAs, all exhibiting a fold-change > 2 and a P-value < 0.05 (Fig. S3A, H). Subsequently, qRT-PCR was performed to assess top 10 up-regulated and down-regulated ncRNAs in five cases of CRC tumors and matched normal tissues, confirming that hsa_circRNA_0005015 exhibited the most significant up-regulation in tumors and the most pronounced down-regulation under anlotinib treatment in SW620 and HCT116 cells (Fig. 1B and S3B). Based on data from the UCSC Genome Browser (https://genome.ucsc.edu/index.html), hsa_circRNA_0005015 originates from the back-splicing of exon 2 of the HAS2 gene located on chromosome 8, with a length of 627nt. Sanger sequencing verified the splicing site, leading us to designate hsa_circRNA_0005015 as circHAS2 (Fig. 1C). Divergent primers for circHAS2 amplified cDNA but not gDNA in PCR analysis, with amplification products analyzed in SW620 and HCT116 cells by agarose gel electrophoresis. GAPDH served as a negative control (Fig. 1D and S3C). Further investigation involving RNase R and actinomycin D treatments revealed that circHAS2 exhibited greater stability in cells compared to linear HAS2 mRNA in SW620 and HCT116 cells (Fig. 1E, F and S3D, E). Moreover, nuclear-cytoplasmic fractionation and the FISH assay demonstrated that circHAS2 predominantly localized in the cytoplasm in SW620 and HCT116 cells (Fig. 1G, H and S3F, G). These findings collectively indicate that circHAS2 is highly expressed and stable in colorectal cancer cell lines, and is regulated by anlotinib.

Fig. 1

Characteristic of circHAS2 and identification as a biomarker of CRC. A Schematic diagram and heatmap depicting whole-transcriptome RNA sequencing results from three independent replicates of SW620 cells cultured with either vehicle control or anlotinib (10 µM) for 24 h. B Relative expression of top 10 most upregulated and downregulated circRNAs in 5 CRC tumor tissues and paired normal tissues. C Schematic diagram illustrating the formation of circHAS2 from host gene and the back-splice junction site verified by Sanger sequencing. D PCR and agarose gel electrophoresis analysis in SW620 cells verified that the divergent primers for circHAS2 could be amplified from cDNA but not gDNA. E Relative expression of circHAS2 and HAS2 mRNA after treatment with RNase R in SW620 cells. F RNA abundance of circHAS2 and HAS2 mRNA after treatment with Actinomycin D in PCR and agarose gel electrophoresis analysis in SW620 cells. G Representative images of the FISH assay in SW620 cells revealed circHAS2 predominantly located in the cytoplasm with the target probe labeled with Cy3 and nuclei stained with DAPI (scale bar: 20 μm). H Nuclear and cytoplasm fractionation assays in SW620 cells showed the subcellular location of circHAS2. I The expression levels of circHAS2 in CRC tissues and paired normal tissues of microarrays were verified by FISH (scale bar: 100 μm). J The expression levels of circHAS2 in CRC tissues and paired normal tissues of microarrays were confirmed by IHC scores. K The expression of circHAS2 in different TNM stages was detected by FISH (scale bar: 400 μm). L Overall survival curves of 70 CRC patients based on circHAS2 expression levels in our research center. Statistical significance in two-group experiments was evaluated using a two-sided Student’s t-test, and survival analysis was conducted using the Log-rank method. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

CircHAS2 expression was also studied in tissue microarrays with 70 CRC patient tumor samples and matched adjacent normal tissues. The results revealed significantly higher circHAS2 levels in tumor tissues compared to adjacent non-tumor tissues (Fig. 1I, J). Subsequently, RT-PCR analysis of 70 tumor samples and their paired normal tissues revealed that circHAS2 was significantly up-regulated in tumors (Fig. S3J). This high expression was closely related to pathological staging and T-stage of patients (Fig. 1K and S3I). Detailed clinical data analysis was conducted to evaluate circHAS2’s clinical and prognostic significance in CRC. We found a significant correlation between circHAS2 expression and pathological stage, T-stage, and lymph node metastasis (Table S1). Cox regression analysis identified circHAS2 expression as an independent prognostic marker for CRC patients (Table S2). ROC curve analysis demonstrated the circHAS2 level could effectively distinguish CRC from non-tumor tissue, with an AUC of 0.759 (Fig. S3K). Kaplan-Meier survival analysis revealed that high circHAS2 expression correlates with poor prognosis in CRC patients (Fig. 1L and S3L, M).

Promotion of anlotinib-regulated circHAS2 in CRC proliferation

Our study aimed to delineate circHAS2’s role in CRC by overexpressing it in SW620 cell lines across four groups: vector, circHAS2 oe, vector with anlotinib, and circHAS2 oe with anlotinib (Fig. S4A). CRC cell proliferation, assessed through CCK-8, colony formation, EdU incorporation, and cell cycle analysis via flow cytometry, showed that high circHAS2 expression significantly accelerated proliferation, while anlotinib attenuated the proliferative promotion in both cell lines (Fig. 2A-D and S4B-F). Employing Patient-derived organoids (PDO) models, we observed that CRC36 vector organoids (IC50 = 15.68 µM) were more sensitive to anlotinib than those with circHAS2 overexpression (IC50 = 25.35 µM), aligning with cell line findings (Fig. 2E, F). In vivo, a xenograft model using SW620 cells confirmed anlotinib’s inhibitory effect on tumor growth, with circHAS2 overexpression countering this effect (Fig. 2G-I). There was no statistically significant difference in body weight among groups (Fig. S4G). Immunohistochemical analysis using anti-Ki67 and CD31 antibodies on xenograft tissues indicated increased Ki67 in the circHAS2 overexpression group, while CD31 levels were unaffected (Fig. 2J and S4H). These results support the notion that circHAS2 promotes CRC tumor growth and is modulated by anlotinib treatment.

Fig. 2

Promotion of Anlotinib-regulated circHAS2 in CRC proliferation in vitro and in vivo. A The viabilities of SW620/vector, SW620/circHAS2 oe with the addition of vehicle control or anlotinib were detected by CCK-8 assays in SW620 cells. B Quantitative results and representative images of cell proliferation were conducted by colony formation assay with indicated treatment in SW620 cells. C Quantitative results and representative images of cell proliferation were evaluated by EdU incorporation with indicated treatment in SW620 cells (scale bar: 50 μm). D The percentage of cells in the G1, S, and G2 phases of the whole cell population was determined by flow cytometry with indicated treatment in SW620 cells. E Representative images of organoids in the vector- and circHAS2 overexpression-treated groups treated with the indicated concentrations of anlotinib (scale bar: 10 μm). F Dose–effect curves of organoids in the vector- and circHAS2 overexpression-treated groups treated with the indicated concentrations of anlotinib. G Images of subcutaneous tumors formed by SW620/vector, SW620/circHAS2 oe cells (n = 5). The mice were administered either a vehicle control or anlotinib (3 mg/kg, orally, every two days), respectively. H Tumor weights were analyzed at the endpoint. I Tumor volumes were measured every two days. J IHC scores of IHC staining (Ki-67 and CD31) assay of tumors of indicated treatment. Cell viability was determined by CCK-8 assay, employing two-way analysis of variance. Statistical differences in clone formation, EdU experiments, tumor weights, tumor volumes and IHC scores were assessed using two-sided Student’s t-test. Survival analysis was performed using the Log-rank method. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

CircHAS2 as a miR-1244 sponge to regulate CCNE2 in CRC

Most circRNAs in the cytoplasm act as competitive endogenous RNAs (ceRNAs) by sponging for miRNAs [29]. Based on previous research, we hypothesized that circHAS2, predominantly found in the cytoplasm, might exert its effects by competitively inhibiting miRNAs. We used miRNA sequencing in SW620 cells to screen for differentially expressed miRNAs in anlotinib-treated and control groups (Fig. 3A). AGO2, essential for circRNA-miRNA interaction, was found enriched with circHAS2, suggesting circRNA-miRNA binding through AGO2 by RIP assays in SW620 cells (Fig. 3B). Cross-analysis of the sequencing results and the online prediction database CircBank predicted miR-1244 and miR-29a-5p as downstream targets (Fig. S5A). Subsequent experiments confirmed miR-1244 was identified as a circRNA-regulated candidate gene through verification following the knockdown or overexpression of circHAS2 in SW620 cells (Fig. 3C). AGO2 RIP assays then revealed a higher relative enrichment of circHAS2 and AGO2 antibody binding in miR-1244 mimic-transfected cells compared to the miR-1244 NC group (Fig. 3D, E). The co-localization of circHAS2 and miR-1244 was observed in both cell lines (Fig. 3F and S5B). In our 70-patient cohort, miR-1244 was significantly down-regulated in CRC tumors, negatively correlating with circHAS2 and indicating poor prognosis (Fig. S5D-F). Dual-luciferase reporter assays using circHAS2 WT and MUT plasmids in SW620 and HCT116 cells demonstrated that circHAS2 directly bound and regulated miR-1244 (Fig. 3G and S5C). Moreover, miR-1244 was up-regulated under anlotinib treatment in the SW620 cell line (Fig. 3H).

Fig. 3

CircHAS2 as a miR-1244 sponge to modulate CCNE2 expression in CRC. A Heatmap showing differentially expressed miRNAs in SW620 cells treated by either vehicle control or anlotinib (10 µM) for 24 h. B The relative expression of circHAS2 in SW620 cells was analyzed by RIP assay with AGO2 and IgG antibodies. C Relative miRNAs expressions were verified after indicated treatment in SW620 cells. D The RIP assay was used to detect the relative expression of circHAS2 in SW620 cells with indicated treatment. E The RIP assay was used to detect the expression of AGO2 protein in SW620 cells with indicated treatment. F Representative images of the FISH assay revealed the colocalization of circHAS2 and miR-1244 in the cytoplasm of SW620 cells with the target probe labeled with Cy3 and nuclei stained with DAPI (scale bar: 20 μm). G Luciferase activity of circHAS2 in SW620 cells was verified with co-transfected Luc circHAS2 WT or MUT and miR-1244 mimics or inhibitor. H Relative miRNA expressions were confirmed after anlotinib treatment (L: 5 µM, H: 10 µM) in SW620 for 24 h. I Prediction of potential downstream target genes for miR-1244 using RNA sequencing, TargetScan, miRDB, and miRWalk. J Relative CCNE2 expressions were verified after indicated treatment in SW620 cells. K Overall survival curves of 70 CRC patients based on CCNE2 expression levels in our center. L The correlation between circHAS2 and CCNE2 was determined by Pearson correlation analysis from our center. M Luciferase activity of CCNE2 in SW620 cells was verified with co-transfected Luc CCNE2 WT or MUT and miR-1244 mimics or inhibitor. N Representative Western blotting images in SW620 cells with indicated treatments for CCNE2, CDK2, p53, and p21 proteins. GAPDH was used as an internal control. Statistical significance in two-group experiments was evaluated using a two-sided Student’s t-test. One-way analysis of variance was employed for multiple-group comparisons. Survival analysis was performed using the Log-rank method. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

To explore the underlying functions and mechanisms of miR-1244 in CRC, we used TargetScan, miRDB, miRWalk databases, and RNA sequencing cross-analysis to identify ten potential target genes, including NOVA1, NCR3LG1, and DGKH (Fig. 3I). CCNE2 emerged as a co-regulated target, negatively correlating with miR-1244 and positively with circHAS2 (Fig. 3J and S5G, I). Furthermore, using our specimen database center, we identified that CCNE2 was up-regulated in tumors and associated with poor prognoses (Fig. 3K and S5H). CCNE2 expression correlated negatively with miR-1244 expression and positively with circHAS2 expression (Fig. 3L and S5J). Notably, CCNE2 is regulated by multiple genes and pathways and is involved in the CRC cell cycle and proliferation [30]. Based on these findings, we proposed that CCNE2 is a miR-1244 functional target gene. Dual-luciferase reporter assays in SW620 and HCT116 cells revealed miR-1244 binding to the 3’-UTR of CCNE2 (Fig. 3M and S5K). Therefore, we propose that circHAS2 appears to regulate CCNE2 through miR-1244 inhibition, with circHAS2 overexpression activating the CCNE2/CDK2 axis. Nevertheless, transfecting miR-1244 mimics partially reversed these effects, without influencing p53 and p21 levels (Fig. 3N and S5L). Additional potential mechanisms may also contribute to regulating CCNE2 expression.

Identification of circHAS2 binding to USP10

CircRNAs, known for their roles as miRNA sponges and protein interaction platforms, can form functional circRNP complexes and modulate mRNA expression [31]. Using an RNA pull-down assay and silver staining in SW620 cells, we investigated whether circHAS2 fulfilled its functional role by interacting with RNA-binding proteins (Fig. S6A). We generated MS2-CP-MS2-circRNA complexes in SW620 cells through co-transfection with circHAS2-MS2 and MS2-CP vectors (Fig. S6B). Pull-down assays confirmed the specificity of these complexes, showing significant circHAS2 enrichment (Fig. 4A). Subsequent mass spectrometry analysis and silver staining of the differential protein expression profiles between circHAS2-MS2 + MS2-CP and circHAS2 + MS2-CP pull-down experiments identified 110 kDa ubiquitin specific peptidase 10 (USP10) as a likely circHAS2 associate, suggesting its presence in high quantities in circHAS2 protein complexes (Fig. 4B and Table S3). Direct interaction between USP10 and circHAS2 was validated by RNA pull-down and RIP assays of SW620 cells, with a noted increase in USP10 enrichment correlating with circHAS2 expression levels (Fig. 4C, D). FISH analysis demonstrated circHAS2 and USP10 colocalization in the cytoplasm in both cell lines (Fig. 4E and S6C). CircHAS2 and USP10 co-localized at the tissue level in tissue microarrays containing 70 CRC tumor samples and matched adjacent normal tissues (Fig. 4F). The binding affinity between USP10 and circHAS2 was analyzed using the catRAPID database, revealing strongest interactions at specific regions of circHAS2 and USP10. We identified the strongest binding between circHAS2 at positions 100–300 nt and 500–627 nt, and USP10 at positions 1–100 aa (Interaction Propensity = 62, Discriminative Power = 97%) (Fig. 4G, I). Focused on these findings, we investigated the interaction regions between circHAS2 and USP10 using various truncated circHAS2 forms in HEK-293T cells. RIP analysis indicated interactions predominantly between circHAS2-2 and circHAS2-4 with USP10 (Fig. 4H). However, altering circHAS2 expression did not impact USP10 levels at mRNA and protein stages (Fig. S6D-E), and using miR-1244 mimics and inhibitors showed no regulation of USP10 mRNA expression levels (Fig. S6F). Additionally, modulating USP10 expression did not alter circHAS2 and miR-1244 levels, indicating a complex interaction mechanism warranting further study (Fig. S6G-H).

Fig. 4

Interaction and co-localization of circHAS2 with USP10. A The enrichment of circHAS2-MS2 complex formation was detected with MS2-CP-Flag in SW620 cells. B The Silver staining of circHAS2-associated proteins by the biotin-labeled probe. C The RNA pull-down confirmed the interaction between circHAS2 and USP10. The relative expression of circHAS2 was determined using the RIP assay with AGO2 and IgG antibodies in SW620 cells. D The RIP assays were conducted under the specified treatment conditions using anti-USP10 and anti-IgG antibodies in SW620 cells. E Representative images of the FISH assay in SW620 cells revealed the colocalization of circHAS2 and USP10 in the cytoplasm with the target probe labeled with Cy3 and nuclei stained with DAPI (scale bar: 20 μm). F The colocalization of circHAS2 and USP10 in CRC tissues and paired normal tissues of microarrays were verified by FISH (scale bar: 400 μm; scale bar: 100 μm). G The interaction profile between circHAS2 fragments and USP10 was predicted by the catRAPID database. H The interaction between circHAS2 full-length and truncations with USP10 was validated using RIP assays in HEK-293T cells. I The heat map illustrated the interaction between circHAS2 and the N-terminal region (1-100 amino acids) of USP10 by the catRAPID database. Statistical significance in two-group experiments was assessed using a two-sided Student’s t-test, and multiple-group comparisons were conducted through one-way analysis of variance. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

Disruption of USP10-p53 interaction by circHAS2

To investigate the effect of circHAS2 on the USP10, we constructed Flag-tagged full-length and truncated proteins of USP10’s functional domains (Fig. 5A). RIP assays and RNA pull-down assays in SW620 cells identified the N-terminal region (1–100 aa) of USP10 as critical for circHAS2 binding (Fig. 5B-C), confirming their direct interaction. These findings supported the direct binding between circHAS2 and the deubiquitinating enzyme USP10. Previous studies have reported that USP10 can bind to p53 and stabilize its expression in HCT116 cells [32]. We performed Co-IP assays to confirm the presence of this interaction, which revealed that USP10 and p53 could immunoprecipitate each other (Fig. 5D). This led us to hypothesize that circHAS2 might disrupt USP10-p53 interactions. RIP assays in SW620 cells with circHAS2 overexpression supported this, showing increased binding between overexpressed circHAS2 and USP10 diminished p53 enrichment (Fig. 5E). Furthermore, ubiquitination-related experiments in SW620 and HCT116 demonstrated that circHAS2 overexpression increased p53 protein ubiquitination, resulting in reduced expression, which could be reversed by USP10 (Fig. 5F, G and S6I, J). P53 is readily visible in the cell nucleus, where it is primarily active, and USP10 modulates the nuclear export mechanism of p53, causing it to be ubiquitinated in the cytoplasm [33]. We observed that circHAS2 increased the p53 nuclear export (Fig. 5H), confirmed by cell fractionation experiments of SW620 cells, suggested that circHAS2 induces p53 ubiquitination and nuclear export, counteracted by USP10 co-expression or LMB treatment (Fig. 5I). In addition, circHAS2 overexpression increased p53 ubiquitination in the cytoplasm but not in the nucleus (Fig. 5J). RT-qPCR results indicated that circHAS2 overexpression that altered USP10 without affecting p53 mRNA levels, but downregulated p21 mRNA (Fig. 5K). CircHAS2 reduced p53 and p21 protein levels in CRC cells, partially reversed by USP10 in both cell lines (Fig. 5L and S6K). Overall, our findings in Figs. 5 and 6 underscored circHAS2’s pivotal role in disrupting p53-USP10 binding and influencing p53 nuclear localization.

Fig. 5

CircHAS2 interaction with USP10 to enhance p53 Ubiquitination. A Schematic diagram of USP10 full-length and truncations. B The RIP assay was performed in SW620 cells using anti-Flag antibodies to assess the expression levels of circHAS2 following transfection with USP10 full-length and truncations. C The RNA pull-down assays were performed using biotin-labeled circHAS2 probes to capture interacting proteins and detect the expression of USP10 full-length and truncations by representative Western blotting images in SW620 cells. D The direct interaction between USP10 and p53 was analyzed by Co-IP assays with USP10 and p53 antibodies in SW620 cells. E The Co-IP assays were conducted under the specified treatment conditions using USP10 and p53 antibodies in SW620 cells. F After indicated treatments were treated with various durations of cycloheximide (0.1 mg/ml) in SW620 cells, and the expression of p53 was detected by Western blotting. GAPDH was employed as an internal control. G CircHAS2 modulated the levels of p53 ubiquitination through the interaction with USP10. SW620 cells were transfected with the indicated constructs and subjected to MG132 (50 mM) for 4 h before harvest to evaluate the ubiquitination levels of p53. H Representative images of the FISH assay revealed the localization of p53 after indicated treatments with the target probe labeled with Cy3 and nuclei stained with DAPI in SW620 cells (scale bar: 20 μm). I Quantification of cells with different p53 subcellular localization after indicated treatments in SW620 cells. Nuc, Nucleus only; Cyto + Nuc, both cytoplasm and nucleus. J SW620 cells were transfected with the indicated constructs and subjected to MG132 treatment to evaluate the ubiquitination levels of cytoplasmic or nuclear p53. K Relative p53 and p21 expressions were verified after indicated treatments in SW620 cells. L Representative Western blotting images with indicated treatments in SW620 cells for Flag, p53, and p21 proteins. GAPDH was used as an internal control. Statistical significance in two-group experiments was assessed using a two-sided Student’s t-test. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

Fig. 6

CircHAS2 activating bidirectional downstream pathways to promote CRC progression. A The viabilities of SW620 cells following indicated treatments were detected by CCK-8 assays. B Quantitative results and representative images of cell proliferation were conducted by colony formation assay in SW620 cells following indicated treatments. C Quantitative results and representative images of cell proliferation were evaluated by EdU incorporation in SW620 cells following indicated treatments (scale bar: 50 μm). D The percentage of cells in the G1, S, and G2 phases of the entire cell population was determined by flow cytometry following indicated treatments in SW620 cells. E Representative Western blotting images with indicated treatments for p53, p21, CDK2, and CCNE2 proteins in SW620 cells. GAPDH was used as an internal control. F IHC scores of ISH staining (miR-1244) and IHC staining (USP10, p53, p21, CCNE2, and CDK2) assay of tumors of indicated treatments. Cell viability was determined by CCK-8 assay, employing two-way analysis of variance. Statistical differences in clone formation, EdU experiments, and IHC scores were assessed using two-sided Student’s t-test. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

CircHAS2 activating bidirectional downstream pathways to enhance CRC growth

To elucidate circHAS2’s specific functions in CRC, we established six experimental groups: control, circHAS2 oe, circHAS2 oe + miR-1244 mimics, circHAS2 oe + USP10 oe, circHAS2 oe + USP10 oe + miR-1244 mimics, and circHAS2 oe + CCNE2 sh in both cell lines. CCK-8, colony formation, EdU assays, and flow cytometry analyses indicated enhanced proliferation in circHAS2-oe groups, mitigated by miR-1244 or USP10, and fully reversed by their combination or CCNE2-knockdown (Fig. 6A-D and S7A-E). Furthermore, USP10 restoration significantly increased p53 and p21 expression, while miR-1244 and USP10 affected CDK2 and CCNE2 expressions (Fig. 6E and S8A). At the tissue level, we performed ISH with miR-1244 probes and immunohistochemical staining with antibodies against USP10, p53, p21, CCNE2, and CDK2. Post-anlotinib treatment showed that miR-1244 expression levels increased, p53 and p21 expressions were up-regulated, while CCNE2 and CDK2 were down-regulated, indicating tumor-suppressive functions (Fig. 6F and S8B). These results suggest that anlotinib-regulated circHAS2 downregulation inhibits CRC cell growth through ceRNA mechanisms and p53 disruption.

CircHAS2 as a predictive biomarker for anlotinib sensitivity

To further confirm the effect of anlotinib on patients with CRC, we treated PDO cells with anlotinib, noting circHAS2’s influence on anlotinib sensitivity in 22 CRC patient-derived samples (Fig. 7A). When we divided the PDOs into two groups based on circHAS2 expression levels, the PDOs with high circHAS2 expression exhibited significant proliferation suppression under anlotinib, correlating with lower IC50 values. Monitoring PDO cell proliferation under various drug concentrations revealed that patients with high circHAS2 expression were more sensitive to anlotinib, while those with low circHAS2 expression exhibited higher IC50 values (Fig. 7B-C and S9A-B). These results suggest suggest circHAS2 as a potential marker for guiding anlotinib treatment in CRC. Subsequently, we conducted two patient-derived tumor xenograft (PDX) models and treated mice with anlotinib (Fig. 7D). Using immunohistochemical (IHC) score and gene expression analysis of patient tumor tissues, we determined that PDX2 expressed high levels of circHAS2, whereas PDX1 expressed lower levels (Fig. 7E and S9C). Anlotinib significantly reduced tumor growth in both PDX groups, particularly in high circHAS2-expressing tumors with minor effects in the PDX1 model (Fig. 7F-H). Furthermore, anlotinib treatment significantly prolonged the survival of mice with high circHAS2 expression (Fig. 7I). Thus, circHAS2 emerges as a key oncogenic factor in CRC and a potential biomarker for anlotinib sensitivity.

Fig. 7

CircHAS2 as a predictive biomarker for response to anlotinib. A Schematic of the experimental setup for in vitro culture experiments involving PDO models. B Representative images of PDOs treated with the indicated concentrations of anlotinib via microscopy (scale bar: 10 μm). C The IC50 values of organoids treated with the indicated concentrations of anlotinib. D Schematic of the experimental setup for in vivo experiments involving PDX models. E Representative images and IHC scores of CRC PDX models were established and evaluated for circHAS2 expression by ISH staining. F Representative images of subcutaneous PDX tumors. The mice were given vehicle control and anlotinib, respectively. G The PDX tumors growth curve with measurements taken every two days. H The weights of subcutaneous PDX tumors were analyzed at the endpoint in each group. I Kaplan–Meier survival curves of the PDX models after the indicated treatments. Statistical significance in two-group experiments was evaluated using a two-sided Student’s t-test. Survival analysis was performed using the Log-rank method. Data are represented as mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance

Clinical response to anlotinib therapy based on circHAS2 expression

Our findings suggest a specific benefit of anlotinib in CRC patients with high circHAS2 levels. A clinical study (ALTER-C001) demonstrated that anlotinib combined with the XELOX regimen for first-line and maintenance therapy had a modest benefit for patients with metastatic CRC [34]. Based on our preclinical data, we initiated a clinical trial at Ruijin Hospital (ClinicalTrials.gov identifier: NCT05262335) to evaluate the efficacy and safety of anlotinib plus the XELOX regimen in patients with advanced gastrointestinal tumors. CircHAS2 expression levels were employed as a stratification criterion to explore treatment responses in advanced CRC patients (Fig. 8A). Utilizing a tissue sample library comprising 70 CRC specimens, 52.9% of CRC cases exhibited high circHAS2 levels (Fig. 8B). Non-invasive imaging and serum tumor marker evaluations were conducted before, during, and after anlotinib-based chemotherapy to assess treatment responses (Fig. 8C). Currently, 14 patients with advanced CRC who have received treatment, 7 were categorized as having high circHAS2 expression (Fig. S9D). The clinical information of patients and genetic mutation details has been compiled (Fig. S9E and Table S4). Notably, computed tomography scans revealed a substantial reduction in the primary tumor size in the group A cohort (Fig. 8D). Compared to the low circHAS2-expressing group B cohort, group A displayed higher tumor response rates, with a Disease Control Rate (DCR) of 100.0% and an Overall Response Rate (ORR) of 57.1% (Fig. 8E). An overall incidence rate of 71.4% (10/14) for all-grade treatment-emergent adverse events (TEAEs) and a rate of 28.5% for ≥ 3 grade TEAEs. The major events included a decrease in neutrophil count (14.2%), a decrease in platelet count (14.2%), hand-foot skin reaction (14.2%), and hypertension (7.1%). Notably, reductions in serum tumor markers CEA, CA125, and CA19-9 were closely associated with favorable prognoses in CRC [