CRC cell lines HCT116, LoVo, RKO, SW480 and SW620 were obtained from American Type Culture Collection. The 293T cell line was obtained from the Shanghai Branch, Cell Bank, Chinese Academy of Sciences and maintained in our laboratory. All cell lines were cultured in Dulbecco's Modified Eagle Medium (HyClone™; Cytiva) with 10% fetal bovine serum (cat. no. 10099141; Gibco; Thermo Fisher Scientific, Inc.), 100 U/ml penicillin and 100 µg/ml streptomycin (cat. no. C0222; Beyotime Biotechnology) in a humidified atmosphere of 5% CO₂ at 37°C. Primary CRC tumors (from 91 male patients and 78 female patients) were collected from Shanghai East Hospital (Shanghai, China) between March 2024 and February 2026, with ages ranging from 36 to 82 years and a median age of 62 years. The present study was approved by the Ethics Committee of Shanghai East Hospital (approval no. 2024-220). Cisplatin and the NF-κB inhibitor JSH-23 were purchased from Selleck Chemicals. K201 was purchased from TargetMol Chemicals Inc.

ANXA3 expression in colon adenocarcinoma (COAD) and rectum adenocarcinoma (READ) was analyzed using Gene Expression Profiling Interactive Analysis (GEPIA), an online tool that integrates tumor tissue data from The Cancer Genome Atlas (TCGA) and normal tissue data from the Genotype-Tissue Expression (GTEx) project. The COAD cohort comprised 275 tumor samples and 349 normal samples, while the READ cohort included 92 tumor samples and 318 normal samples (http://gepia2.cancer-pku.cn). Pearson correlation analyses among ANXA3 expression and excision repair cross-complementing 1 (ERCC1), GSTP1 or ATP7B in CRC were also performed using the GEPIA database. The online public tumor database UALCAN, which integrates protein expression data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC), was used to analyze the protein expression of ANXA3 in colon cancer (https://ualcan.path.uab.edu/).

To evaluate the mRNA levels, reverse transcription-quantitative PCR (RT-qPCR) was performed. Briefly, total RNA was extracted from cultured cells using RNAiso Plus (Takara Bio, Inc.), and cDNA was reverse-transcribed with the PrimeScript™ RT reagent Kit (Takara Bio, Inc.) according to the manufacturer's protocol. Subsequently, qPCR was carried out using SYBR Green qPCR Master Mix (Clontech; Takara Bio, Inc.) with Roche LightCycler^® 480II real-time PCR system (Roche Diagnostics). GAPDH was used as the reference gene, and relative expression levels were quantified using the 2^−ΔΔCt method as described previously (28). Thermal cycling conditions included initial denaturation at 95°C for 30 sec, followed by 40 cycles of 95°C for 5 sec and 60°C for 30 sec. Primer sequences used for qPCR are shown in Table SI.

Tissue arrays containing 169 pairs of CRC tissues and corresponding paracancerous tissues were prepared for IHC, and the patients' information has been described in our previous study (29). Tissues were formalin-fixed at room temperature for 24 h, paraffin-embedded and cut into 4-µm sections. Sections were permeabilized with 0.1% Triton X-100, blocked with 5% BSA at room temperature for 1 h, and incubated with an anti-ANXA3 antibody (cat. no. ab127924; Abcam; 1:250) at 4°C overnight. HRP-conjugated goat anti-rabbit IgG (cat. no. ab205718; Abcam; 1:500) was applied for 1 h at room temperature, followed by DAB chromogen detection. Staining was visualized using a light microscope. The staining of ANXA3 was scored using a semi-quantitative scale, as follows: 0, negative; 1, weak; 2, moderate; 3, strong. The scoring was performed in a blinded manner. For survival analysis, the ANXA3 score was categorized as follows: Scores of 0–1 indicated low ANXA3 expression, while scores of 2–3 indicated high ANXA3 expression.

Western blot analysis was carried out as described previously (30). Briefly, the total protein from cells or exosomes was extracted using RIPA lysis buffer (Beyotime Biotechnology) supplemented with protease and phosphatase inhibitors. The total protein concentration was detected using the BCA Protein Assay Kit (cat. no. P0009; Beyotime Biotechnology), and 30 µg protein per lane was separated by 10% SDS-PAGE. Subsequently, the proteins were transferred onto polyvinylidene fluoride membranes (MilliporeSigma), and membranes were blocked with 5% (w/v) non-fat milk at room temperature for 2 h. After blocking, the membranes were incubated with the indicated primary antibodies overnight at 4°C. The anti-ANXA3 antibody (cat. no. ab127924; 1:1,000), anti-CD63 antibody (cat. no. ab68418; 1:500) and anti-CD81 antibody (cat. no. ab79559; 1:500) were obtained from Abcam. The anti-Flag-tag antibody (cat. no. M185-3; 1:2,000) was purchased from Medical & Biological Laboratories Co., Ltd. The anti-GAPDH antibody (cat. no. 97166; 1:1,000), anti-phosphorylated (p)-NF-κB p65 (Ser536) antibody (cat. no. 3033; 1:1,000) and anti-NF-κB p65 antibody (cat. no. 8242; 1:1,000) were purchased from Cell Signaling Technology, Inc. The anti-GSTP1 antibody (cat. no. F0781; 1:500) was obtained from Selleck Chemicals. After washing with TBS-Tween (0.1% v/v), the membranes were incubated with appropriate HRP-conjugated secondary antibodies (cat. no. A0208 or A0216; Beyotime Biotechnology; 1:500) for 1 h at room temperature. Finally, the immunoreactive bands were visualized using an ECL substrate (Beyotime Biotechnology) and captured with a chemiluminescence imaging system (Tanon Science and Technology Co., Ltd.). Semi-quantitative analysis of western blot band intensities was performed using ImageJ software (version 2.9.0, National Institutes of Health).

Cell viability was detected using the Cell Counting Kit-8 (CCK-8) assay (cat. no. C0042; Beyotime Biotechnology) as described previously (31,32). Briefly, CRC cells (5,000 cells/well) were seeded in 96-well plates and allowed to adhere overnight. For the cisplatin sensitivity assay, various CRC cell lines, lentivirus-infected cells, or siRNA-transfected cells were incubated with increasing concentrations of cisplatin (0, 10, 20, 40, 80 and 160 µM) for 24 h at 37°C. For combination treatment, the cells were pretreated with 10 µM K201 or vehicle (0.1% DMSO in cell culture medium) for 1 h at 37°C, followed by co-exposure to increasing concentrations of cisplatin (0, 10, 20, 40, 80 and 160 µM) for 24 h at 37°C in the continued presence of K201 or vehicle. Subsequently, 10 µl CCK-8 reagent was added to each well and the cells were cultured for a further 1 h at 37°C. After incubation with CCK-8 reagent, OD values were measured. Wells were assigned randomly and the investigator performing the OD measurement was blinded to the treatment groups. Experiments were repeated independently three times.

The human full-length ANXA3 cDNA sequence was amplified by PCR and cloned into the lentivirus vector PLVX (Addgene, Inc.) with a Flag tag (PLVX-Flag-ANXA3). An empty PLVX vector (PLVX-NC) was used as the negative control. The ANXA3-overexpressing plasmids and packaging plasmids were co-transfected into 293T cells to produce the lentivirus and the viral particles were prepared as described previously (33). In detail, lentiviral particles were produced by using a second-generation lentiviral packaging system in the present study. When 293T cells reached 70–80% confluence, they were co-transfected with the lentiviral expression plasmid, psPAX2 and pMD2.G at a mass ratio of 4:3:1 by using Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instruction. The total amount of plasmid DNA used for each 10-cm dish was 20 µg. The culture supernatant containing lentiviral particles was collected at 48 and 72 h after transfection. LoVo and RKO cells were infected with the indicated lentiviral particles at a multiplicity of infection of 20 and 30, respectively, in the presence of 5 µg/ml polybrene (cat. no. H9268; Sigma-Aldrich; Merck KGaA). Following 48 h infection, stable cells were selected using 2 µg/ml puromycin (cat. no. P9620; Sigma-Aldrich; Merck KGaA) for 7 days and maintained in medium containing 1 µg/ml puromycin.

The siRNA against ANXA3 (siANXA3), the siRNA against GSTP1 (siGSTP1) or the negative control siRNA (siNC) were transfected into CRC cells at a final concentration of 10 nM using Lipofectamine^® RNAiMAX (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Transfections were carried out at 37°C in a humidified incubator with 5% CO₂, and the transfection medium was replaced with complete culture medium after 6 h. These cells were prepared for subsequent experiments 24 or 48 h post-transfection. The sequences were as follows: siANXA3 sense strand, 5′-GGAAUAUCAAGCAGCAUAUdTdT-3′; antisense strand, 5′-AUAUGCUGCUUGAUAUUCCdTdT-3′; siGSTP1 sense strand, 5′-CAUCAAUGGCAACGGGAAAdTdT-3′; antisense strand, 5′-UUUCCCGUUGCCAUUGAUGdTdT-3′; and siNC sense strand, 5′-UUCUCCGAACGUGUCACGUdTdT-3′; antisense strand, 5′-ACGUGACACGUUCGGAGAAdTdT-3′ (all synthesized by Shanghai GenePharma Co., Ltd.).

For pharmacological inhibition of NF-κB signaling, LoVo cells were seeded in 6-well plates at 5×10⁵ cells/well and allowed to adhere overnight at 37°C with 5% CO₂. And then, cells were incubated with 5 µM NF-κB inhibitor JSH-23 (Selleck Chemicals) or vehicle control (DMSO) for 48 h at 37°C with 5% CO₂, followed by RNA and protein extraction for further analysis.

Exosomes were isolated as described previously (34). Briefly, LoVo cells infected with PLVX-NC or PLVX-Flag-ANXA3-derived lentiviruses were incubated for 72 h at 37°C. Cell culture supernatants were then collected and subjected to sequential centrifugation: 300 × g for 10 min at 4°C to remove cells, 2,000 × g for 20 min at 4°C to remove debris and 10,000 × g for 30 min at 4°C to remove large vesicles. Exosomes were then isolated from the resulting supernatant using the Hieff^® Quick Exosome Isolation Kit Plus (cat. no. 41205ES10; Shanghai Yeasen Biotechnology Co., Ltd.) according to the manufacturer's instructions. The total protein concentration of the isolated exosomes was detected using BCA Protein Assay Kit according to the manufacturer's protocol. The expression levels of exosomal biomarkers CD63 and CD81 were then detected by western blot analysis as aforementioned. For exosome treatment, LoVo cells were incubated with 50 µg/ml OE-ANXA3-exosomes or NC-exosomes at 37°C with 5% CO₂ for 24 h.

The GSTP1 luciferase reporter gene construct (−336 to +74) was generated as reported previously (35). For the luciferase assay, LoVo cells (5×10⁴ cells/well) were seeded in 24-well plates and cultured overnight. The cells were then co-transfected with 400 ng GSTP1 luciferase reporter plasmid (constructed by inserting GSTP1 promoter into PGL4 vector, and 40 ng pRL-TK Renilla internal control vector (Promega Corporation) using Lipofectamine 3,000 according to the manufacturer's instructions. After 24 h, the cells were infected with the PLVX-Flag-ANXA3-derived lentiviruses for 48 h or incubated with the ANXA3-overexpressing exosomes for 24 h, followed by luciferase assay using the Dual-Luciferase Reporter Assay System (Promega Corporation) according to the manufacturer's instructions. Firefly luciferase activity was normalized to the corresponding Renilla luciferase activity for each well. All transfections and treatments were performed in three independent replicates.

Virtual screening was performed using molecular docking as described previously (36). ANXA3 structure (UniProt ID: P12429; http://www.uniprot.org/uniprotkb/P12429/entry) was obtained from AlphaFold database (version 2.3.2; http://alphafold.ebi.ac.uk/entry/P12429) and the scoring of the molecular docking was completed using the Schrödinger Suite (HTVS/SP/XP modes, Schrödinger LLC), followed by Sybyl clustering conducted in Sybyl-X 2.0 (Certara L.P.). Finally, the candidate compound K201 was screened out based on the scores and PubMed literature search (https://pubmed.ncbi.nlm.nih.gov). The docking score of K201 was −5.378 kcal/mol and Molecular Mechanics/Generalized Born Surface Area of K201 was −54.204 kcal/mol.

In vitro experiments were independently repeated at least three times unless otherwise indicated. The software GraphPad Prism 5 (Dotmatics) was used to generate figures and perform statistical analysis. Continuous data are presented as the mean ± SD, and were compared between two independent groups using unpaired Student's t-test. When data were paired or matched, paired Student's t-test was used. Categorical data are presented as median with interquartile range (IQR); the Wilcoxon signed-rank test was applied for non-continuous data (such as IHC scores). One-way or two-way ANOVA was used to compare three or more groups, followed by Tukey's post hoc test for multiple comparisons. The Kaplan-Meier method and log-rank test were used for survival analysis. For IHC score distributions (scores 0, 1, 2 and 3), Bowker's test was used to compare CRC and NT groups, and Fisher's exact test was used for comparisons across clinical stages I–IV. P<0.05 was considered to indicate a statistically significant difference.

In our previous study, ANXA3 was identified as being upregulated in both primary and liver metastatic CRC (37). To further validate this finding, the publicly available cancer data from the GEPIA database were analyzed, which integrates TCGA normal and GTEx datasets, and the results revealed significantly upregulated mRNA expression levels of ANXA3 in COAD and READ tissues (Fig. 1A). Concurrently, data from the UALCAN database (matched with CPTAC samples) revealed that the protein levels of ANXA3 were elevated in the primary tumor tissues of colon cancer (Fig. 1B). To validate these findings, a tissue array was prepared and subsequently evaluated by IHC staining. These results confirmed that ANXA3 was significantly upregulated in CRC tumor tissues (Fig. 1C and 1D; Table SII). Furthermore, Kaplan-Meier survival analysis revealed that high expression of ANXA3 was associated with poor prognosis in patients with CRC (Fig. 1E). Taken together, these preliminary data indicated that ANXA3 was upregulated in CRC tumors and elevated ANXA3 was associated with a poor prognosis for patients with CRC.

First, the expression levels of ANXA3 in five CRC cell lines were detected, including HCT116, LoVo, RKO, SW480 and SW620. As shown in Fig. 2A, ANXA3 was markedly downregulated in LoVo and RKO cell lines compared with in the other cell lines. Subsequently, these five CRC cell lines were incubated with cisplatin and CCK-8 assays were performed to assess drug sensitivity. The results showed that CRC cell lines with high ANXA3 expression exhibited reduced sensitivity to cisplatin compared with those with low ANXA3 (Fig. 2B). To further validate this relationship, ANXA3 was overexpressed by lentiviral plasmid transduction in LoVo and RKO cells, which exhibited low endogenous ANXA3 (Fig. 2C), and the results showed that overexpression of ANXA3 significantly reduced the sensitivity of cisplatin in CRC cells (Fig. 2D and E). By contrast, when ANXA3 was silenced in HCT116 and SW480 cells with high endogenous ANXA3 (Fig. 2F and G), these CRC cells became more sensitive to cisplatin (Fig. 2H and I). Taken together, the aforementioned results suggested that ANXA3 mediates cisplatin sensitivity in CRC cells.

Next, the public tumor database GEPIA was analyzed and it was found that ANXA3 expression in CRC was positively correlated with the expression of several important genes associated with chemoresistance, including ERCC1 (r=0.24; P<0.05), GSTP1 (r=0.45; P<0.05) and ATP7B (r=0.27; P<0.05) (Fig. 3A). The expression levels of these genes were also confirmed using RT-qPCR and GSTP1 was revealed to be the most prominently upregulated following ANXA3 overexpression in CRC cells (Fig. 3B and C). Western blot analysis further revealed that ANXA3 overexpression markedly increased GSTP1 protein levels in CRC cells (Fig. 3D). Combined with the mRNA levels, it was hypothesized that ANXA3 may regulate the transcriptional expression of GSTP1. To assess this, a luciferase reporter system driven by the GSTP1 promoter was used, and the luciferase assay showed that ANXA3 significantly increased the activity of the GSTP1 promoter (Fig. 3E). By contrast, silencing ANXA3 decreased GSTP1 expression (Fig. 3F). The aforementioned results indicate that ANXA3 promoted the transcriptional expression of GSTP1 in CRC cells. Given that NF-κB has been reported to be an important transcription factor mediating the expression of GSTP1 (38), whether ANXA3 exerted its function by activating NF-κB was investigated. As shown in Fig. 3G, overexpression of ANXA3 upregulated the expression levels of p-NF-κB p65. Moreover, the regulatory effect of ANXA3 on GSTP1 expression was also blocked by the NF-κB inhibitor JSH-23 (Fig. 3H and I), which indicated that ANXA3 regulated GSTP1 expression in CRC cells in an NF-κB-dependent manner. In addition, as demonstrated in Fig. S1, knockdown of GSTP1 significantly suppressed ANXA3-mediated cisplatin resistance, which further suggested that GSTP1 may serve as a key downstream molecule through which ANXA3 exerts its effect on drug sensitivity.

For the present study, OE-ANXA3-exosomes were also generated. As shown in Fig. 4A and B, ANXA3-overexpressing CRC cells were successfully established, and exosomes isolated from the culture supernatants were subsequently characterized by western blot analysis, which confirmed the expression of exosomal biomarkers CD63 and CD81. Subsequently, CRC cells were incubated with OE-ANXA3-exosomes or NC-exosomes, and the results indicated that OE-ANXA3-exosomes significantly upregulated the transcriptional expression of GSTP1 (Fig. 4C and D). Additionally, OE-ANXA3-exosomes significantly increased the luciferase activity of the GSTP1 promoter (Fig. 4E). Subsequently, CRC cells were incubated with OE-ANXA3-exosomes or cisplatin, and the results indicated that OE-ANXA3-exosomes significantly decreased the drug sensitivity of cisplatin in CRC cells (Fig. 4F). Moreover, the cisplatin-triggered downregulation of GSTP1 was partially reversed by the treatment with OE-ANXA3-exosomes (Fig. 4G and H). Based on these findings, it is speculated that ANXA3 may be involved in the process of exosome-mediated cisplatin resistance in CRC cells through regulating GSTP1 expression.

Finally, molecular docking-based virtual screening identified K201 as a potential compound interacting with ANXA3. As shown in Fig. 5A-C, K201 formed π-π interactions with His8 (HIE8) and Phe206, as well as a hydrogen bond and a salt bridge with Asp283 of ANXA3. Subsequently, CRC cells harboring different expression levels of ANXA3 were incubated with K201, and the results showed that K201 exhibited stronger cytotoxicity in HCT116 cells with high ANXA3 expression than in RKO and LoVo cells with low ANXA3 expression (Fig. S2). CRC cells with high ANXA3 expression (including HCT116 and SW480) were then incubated with cisplatin alone or in combination with K201, and K201 was shown to significantly enhance the cytotoxicity of cisplatin in CRC cells (Fig. 5D and E). Critically, the enhanced cytotoxicity observed in the combination group (K201 + cisplatin) was significantly greater than that of K201 alone, arguing against a simple additive toxic effect and supporting a specific sensitizing mechanism (Fig. 5F). Moreover, ectopic overexpression of ANXA3 partially reversed the K201-mediated cisplatin sensitization in CRC cells (Fig. 5F), which suggested that the level of ectopically expressed ANXA3, while sufficient to significantly attenuate K201 activity, may not fully saturate all available targets in the presence of the drug. These results collectively suggested that targeting ANXA3 activity with K201 may enhance cisplatin efficacy in CRC cells.