Tumor specimens and corresponding precancerous normal tissues were collected from 26 colorectal cancer patients from October 2023 to January 2024, excluding patients with advanced disease, active malignancies, or those who had received radiotherapy or chemotherapy. All tissue specimens were available from patients diagnosed with colorectal cancer by histopathological examination at Wuxi People’s Hospital, and all specimens were frozen in liquid nitrogen and transferred to the specimen bank of the Clinical Research Center. 26 patients were identified by their sample name, sample type, sampling time, sex, age, height, weight, differentiation grade, TNM staging, pathology, clinical staging, immunohistochemistry, tumor size, vascular invasion, and nerve invasion, as described in the following table. Invasion, and nerve invasion, as detailed in Supplementary Table S1. This study was conducted in accordance with the ethical standards of Helsinki, and the Medical Ethics Committee of Wuxi People’s Hospital approved the study.

Sample DNA was extracted using the TIANNamp Genomic DNA Kit (Tiangen Biotech, China), and DNA quality control was performed using the following two methods: Nanodrop for DNA concentration and Agilent 4150/5400 for DNA integrity.

Using the IGT Enzyme Plus Library Prep Kit V3 (iGeneTech, China), the DNA was cut into fragments of approximately 350 bp, end repaired, “A” tailed, and annealed to the sequencing junctions, and then the products were purified by magnetic beads. Then the product was purified by magnetic beads, and the original library with moderate fragment length was selected for the next step of PCR amplification. The amplified product was purified by magnetic beads, and the small fragment library for sequencing was obtained.

After the library construction was completed, Qubit was used for quantification, followed by Agilent 4150/Qseq400 to test the insert size of the library to ensure its quality. Sequencing was performed on an Illumina Noveaseq 6000 (Kokuryo et al., 2025). The sequencing mode was PE150, which means that in the constructed DNA small fragment library, each insert fragment was sequenced at both ends, 150 bp at each end (Li X. et al., 2021).

Data filtering primarily eliminates the following three situations: the length after truncation of the splice sequence is less than 50 bpm; the number of bases at one end of a single piece of data is more than 5% of the total base ratio of that piece of data; and the number of inferior bases at one end of a single piece of data is more than 50% of the total bases of that piece of data.

The filtered data were compared to the human reference genome using the Burrows-Wheeler alignment tool (BWA) (Li and Durbin, 2009). The initial results in BAM format were obtained after BMA processing, and the final comparison results in BAM format were obtained by further using Picard software for processing, such as labeling repetitive sequences. These aligned results were used to perform statistics on sequencing depth and coverage.

GATK-Mutect2 was used to detect somatic single nucleotide mutations, insertion and deletion mutations (Pei et al., 2021), and these sequences were compared and corrected with paracancerous samples to filter out normal tissue information. The tested somatic mutation data were annotated using ANNOVAR software. Detection of germline single nucleotide mutations and insertion and deletion mutations was performed using GATK.

Single nucleotide mutations based on somatic cells were analyzed from multiple perspectives, and the mutation spectrum and mutation characteristics of the derived tumor samples were plotted using R language software. Fisher’s exact test method was used to screen for high-frequency mutated genes in the somatic cell mutation data (Mateo et al., 2020). The distribution of copy number mutations in the samples as well as on the chromosomes was analyzed using GISTIC software along with somatic single nucleotide mutation analysis.

This genetic screen is based on the screening of paraneoplastic normal tissue germline variation data to derive susceptibility genes in paraneoplastic normal tissues that may lead to the transformation of carcinoma into cancer. The screening methods include (1) removing variants with sequencing depth less than 10X; (2) removing variant loci with a high frequency of Yubo 0.0014 in the Thousands, ExAc, and esp6500 databases, directly removing loci in the dbSNP database, and retaining loci in the COSMIC database; (3) deleting variant loci in intergenic, non-coding, and intronic regions; (4) filtering out synonymous mutations as well as phase duplication genes; and (5) removing harmless mutations according to the results of the ljb23-sift, ljb23-pp2hvar, ljb23-pp2hdiv, and ljb23-mt databases.

Gene expression data were available from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression Project (GTEx) databases. This included RNA-seq data of SH3BP1 from 33 types of cancer, as well as normal colorectal tissue. The data was normalized using the log2 (TPM+1) method and then employed to subsequent analysis. The results of the analyses were visualized using public R package.

Duplicate and missing RNA-seq data from the GTEx and TCGA databases were exempted. The expression levels of SH3BP1 in various cancer types were tested and visualized in R Studio (version 4.2.3). Statistical comparisons were carried out using the “stats” (version 4.2.1) and “car” (version 3.1.0) packages. We chose analysis of variance (ANOVA) as the statistical method for assessing differential expression of SH3BP1. The analysis primarily examined differences in SH3BP1 expression between various cancer and normal tissues.

We assessed the expression levels of SH3BP1 mRNA in colon (COAD) and rectal (READ) adenocarcinomas during gene expression comparisons between cancerous and normal tissues. Box plots were subsequently generated. Meanwhile, the expression levels of cancer susceptibility genes obtained from sequencing were compared in COAD and READ, and heat maps were created.

Two human colorectal cancer cell lines (HCT116, RKO) and one normal human intestinal epithelial cell line (NCM460) were employed in this experiment. All cells were available from the Clinical Research Center of Wuxi Medical Center, Nanjing Medical University. All cells were grown in DMEM medium (Gibco, United States) containing fetal bovine serum (Biological Industries, Israel) and double antibody (NCM Biotech, China). They were cultured at 37°C in a constant temperature incubator containing 5% carbon dioxide.

RIPA lysis buffer (CWBIO, China) was mixed with EDTA, protease inhibitor, and phosphatase inhibitor (Beyotime, China) at a ratio of 1:100 (1000µlRIPA, 10µlEDTA, 10 µL protease inhibitor, 10 µL phosphatase inhibitor) to lose the cells and allowed to stand on ice for 20 min. The cells were centrifuged at 12,000 rpm for 10 min, and the supernatant was combined with loading buffer (CWBIO, China) at a ratio of 4:1. The mixture was boiled for 5 min, removed, and permitted to stand at room temperature before sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The protein was then transferred to a polypropylene fluoride membrane (PVDF, Millipore, United States) by membrane transfer, followed by rapid sealing with sealing solution (Beyotime, China) for 30 min, and then placed in the box with diluted primary antibodies SH3BP1 (1:5000, Proteintech, China) and GAPDH (1:2000, Affinity Biosciences, United States) overnight at 4°C in a refrigerator shaker. The next day, the antibody was retrieved, washed three times with TBS (Tris-HCl buffered saline) and Tween (TBST, CWBIO, China), diluted with secondary antibody (1:2000, Proteintech, China), and shaken for 1 h, and then washed three times with TBST. The 1:1 developer (ThermoFisher, United States) was exposed. Data quantification was conducted using the Fiji software (ImageJ, United States). GAPDH served as a control in this study.

Tissue was trimmed, fixed in 10% paraformaldehyde for 24 h, dehydrated, cleared, embedded in wax, embedded, and cut into 4-μm thick sections. The sections were baked, delayed, hydrated, antigen retrieved, and blocked with endogenous peroxidase. The sections were subsequently incubated with 4% rabbit serum for 15 min at room temperature to reduce specific staining. The sections were subsequently incubated with primary antibody (1:500, Proteintech, China) for 2 h. This was accompanied by incubation with secondary antibody at room temperature for 30 min. Next, DAB was added dropwise for color development, monitored by staining with hematoxylin for 30 s, differentiation with 0.1% hydrochloric acid, rinsing with tap water for 3 min, blueing with ammonia, and further rinsing with tap water for 3 min. Finally, the slides were dehydrated, made transparent, and sealed. The slides were celebrated under a fluorescence microscope (ZEISS, Germany).

Cellular SH3BP1 expression was silenced by small interfering RNA (RIBOBIO, China). The sequences of shRNAs are the following.

siSH3BP1-1:5′-CCAGCAACATCGCCATAGT-3';

Opti-MEM medium (Gibco, United States) 125 µL was used to add to tubes containing 10 µL siRNA and lipo3000 (ThermoFisher, United States), respectively; then siRNA dilution was added to lipo3000 dilution, mixed well, and allowed to stand for 15 min. Transfection reagent was prepared for transfection.

Before the cells were passaged into the six-well plate, the scratch film (Beyotime, China) was placed in the plate, and UV light irradiation was applied for 20 min to start the passage and spreading of the plate, and the length was up to 90%. On the back of the six-well plate, two horizontal lines were drawn along the ruler as a marking line to observe the cells; the original culture medium was discarded, the film was gently torn off, and the scratch film intersected with the marking line perpendicularly, and the intersection point was used as a fixed detection point. Immediately after serum-free DMEM medium (Gibco, United States) was added, the cell photographs were observed with a microscope (Nikon, Japan) as a control and then placed in an incubator, and the growth of cells near the scratches was recorded at different times: 0 h, 24 h, and 48 h.

RKO and HCT116 colorectal cancer cells, which were transformed with either si-NC or si-SH3BP1, were inoculated into 96-well plates at a density of 2,000 cells per well. Three groups were established: experimental, control, and blank, with six replicates in each group. At 0, 1, 2, 3, 4and 5days, 10 µL of CCK-8 reagent was in addition to each well. The wells were not subject to light and incubated for 2 hours in an incubator. Measure the optical density (OD) at 450 nm using a spectrophotometer. Generate a growth curve.

All data are expressed as mean ± SEM (±SD). Analytical plotting for this postgraduate letter section was performed by R4.2.3. Descriptive statistics were utilized to summarize sample data characteristics as well as mutation frequencies, while Fisher’s exact test was applied. All raw p values for the above tests were adapted by a very conservative Bonferroni correction. Images were analyzed and processed in the experimental part using Fiji software (ImageJ, United States) and GraphPad Prism 10.1.2 (GraphPad Software, United States). All p-values <0.05 were examined statistically significant.

We performed whole-genome sequencing of colorectal tumors and their paracancerous normal tissues, and collected samples of tumors and their paracancerous normal tissues from 26 patients diagnosed with colorectal cancer; all samples were collected with the approval of the institutional ethical review board. Supplementary Tables S1 and Table 1 show the relevant clinical characteristics of the group. At the time of diagnosis, 42.31% of the patients were females and another 57.69% were male. Overall, 65.38% of patients in this cohort had adenocarcinoma, 7.69% and 3.85% had mucinous and tubular adenocarcinoma, respectively, and 19.23% had specific types of adenocarcinoma, including adenocarcinoma with mucinous adenocarcinoma and tubular-papillary adenocarcinoma. Vascular invasion and nerve invasion were found in 34.62% and 38.46% of patients, respectively. The percentages of patients classified as having clinical stage I, II, III, and IV disease were 7.69%, 30.77%, 61.54%, and 0%, respectively. In addition, 76.92% of the tumors had grade 2 differentiation, 19.23% had grade 2–3 as well as 3.85% had grade 3.

Multidimensional analysis of somatic single nucleotide mutations (point mutations), including mutation spectrum and mutation signature. From these results, we can clearly understand the characteristics of cancer mutations at the point mutation level, i.e., at the base level.

Copy number variation (CNV) is an important part of genomic structural variation (SV), which can be categorized into two types: deletions and duplications. In the distribution map of copy number mutations in tumor samples, most of the samples showed copy number gains on chromosomes 7, 13, and 20, and mainly copy number losses on chromosomes 4, 6, 17, and 18 (Figure 4A). Meanwhile, in the high-frequency copy number mutation distribution map, we used the GISTIC software to score the high-frequency copy number mutation region; the higher the score, the higher the frequency of copy number mutation in this region, and we found that the copy number increase was mainly shown on chromosomes 1, 2, 3, 6, 7, 13, and 20, and the copy number decrease was mainly shown on chromosomes 4, 6, 17, 18, and 21 (Figure 4B). This indicates that the genomes of tumor samples show different alterations of gene fragments on different chromosomes, which is more favorable evidence that tumors are caused by a series of mutations or aberrations accumulated at the level of somatic genome, and that genomic alterations are crucial in the process of tumorigenesis and development.

Significantly mutated genes (SMGs) include somatic single nucleotide mutations and insertion/deletion mutations. They are genes with a mutation frequency significantly higher than the background mutation rate (BMR). We used Fisher’s exact test to screen for SMGs. The genes with more mutations are NUDT4B, HRNR, OBSCN, ATAD3B, and NBPF20. In combination with Fisher’s test, a P value of <0.05 is examined statistically significant. Therefore, we finally selected the ATAD3B gene (Table 6). As a gene with a high mutation frequency in tumor samples, it can serve as a target for gene therapy of CRC for further research and verification.

Cancer-predisposing genes (CPGs) are genes that may encode genetic diseases or acquired susceptibility to disease under environmental stimuli. Mutations that occur in an individual’s germ cells do not necessarily lead directly to cancer, but they can significantly increase an individual’s risk of developing cancer. We screened primarily for germ line mutations derived from sequencing 26 cases of normal tissue adjacent to cancer to screen for possible cancer predisposition genes. We screened for six candidate cancer susceptibility genes, where hom_ref indicates no mutation, het indicates a heterozygous mutation, and hom_alt indicates a homozygous mutation. The probability of these six genes to be mutated in normal tissue adjacent to cancer is CPA6 (3.85%), ZNF888 (46.15%), SH3BP1 (76.92%), ANKRD16 (30.77%), ATN1 (11.54%), and C4orf54 (80.77%). Take into account this, we roughly concluded that the more ideal cancer susceptibility genes are SH3BP1 and C4orf54 (Table 7). Since the probability of mutations in these two genes in normal tissues adjacent to cancer is relatively high, these two genes may be potential genetic targets for colorectal cancer and can be further verified.

A pan-cancer expression analysis revealed that SH3BP1 is highly expressed in ten cancer types: Cervical Squamous Cell Carcinoma and Intracervical Adenocarcinoma (CESC), Colon Adenocarcinoma (COAD), Renal Clear Cell Carcinoma (KIRC), Lung Squamous Cell Carcinoma (LUSC), Ovarian plasma cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), rectal adenocarcinoma (READ), gastric adenocarcinoma (STAD), testicular germ cell tumor (TGCT), and uterine sarcoma (UCS). Two other cancers, adrenocortical carcinoma (ACC) and prostate adenocarcinoma (PRAD), showed reduced expression. Two other cancers, adrenocortical carcinoma (ACC) and prostate adenocarcinoma (PRAD), showed decreased expression (Figure 5A). Comparing the expression of SH3BP1, CPA6, ZNF888, and NKRD16 in COAD and READ tumor and normal tissues revealed that SH3BP1 expression increased in COAD and READ and was more prevalent in tumor tissues than in the other susceptibility genes obtained by whole-genome sequencing (Figures 5B,C). These findings suggest that SH3BP1 is upregulated in various tumor tissues, including COAD and READ.

Built on the genetic information obtained from susceptibility screening, we selected SH3BP1 as the target gene for experimental verification. To investigate whether SH3BP1 is differentially expressed in cancer and adjacent normal tissues, we performed protein blotting (WB) and immunohistochemistry (IHC) experiments. Our immunohistochemical analysis of pathological tissues from two patient groups revealed that SH3BP1 expression was consistently higher in cancer tissues than in normal tissues (Figures 6A,C). Conversely, SH3BP1 expression was negligible in normal human intestinal epithelial cells (NCM460) but significantly increased in RKO and HCT116 colorectal cancer cells (Figure 6B). These results imply that SH3BP1 is differentially expressed in tumors and normal tissues adjacent to cancer. To further explore SH3BP1’s potential role in tumors, we downregulated the SH3BP1 gene in HCT116 and RKO colorectal cancer cells by transfecting them with si-NC and si-SH3BP1, respectively. As shown in Figure 7A, the proliferative capacity of the colorectal cancer cells (HCT116 and RKO) was significantly reduced in the SH3BP1 knockdown group (si-SH3BP1-1, si-SH3BP1-2) compared to the si-NC group, as demonstrated by a cell viability assay (CCK8). Additionally, the migratory effect of SH3BP1 on colorectal cancer cells was evaluated using a scratch wound healing assay. This assay showed that the migratory ability of colorectal cancer cells in the si-SH3BP1 group decreased, while the migratory ability of colorectal cancer cells in the si-NC group did not differ significantly from that of the CON group (Figures 7B,C). Thus, SH3BP1 may promote tumor proliferation and migration in colorectal cancer and has the potential to be a new gene therapy target.