Acquisition of microarray data

In this study, one microarray dataset was downloaded from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) database. Five microarray datasets were acquired from Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo) database, containing GSE41328 [20], GSE44076 [21], GSE44861 [22], GSE136735 [23], and GSE17536 [24]. In our study, data from all databases was accessed on December 1, 2023. The cohorts included: TCGA (26 normal colorectal specimens and 270 CRC specimens), GSE41328 (10 CRC specimens and 10 adjacent normal colorectal specimens), GSE136735 (6 CRC specimens and 6 adjacent normal colorectal specimens), GSE44076 (98 CRC specimens and 98 adjacent normal colorectal specimens), GSE44861 (55 CRC specimens and 55 adjacent normal colorectal specimens), and GSE17536 (177 CRC specimens). The microarray data from GSE41328 and GSE17536 were based on the GPL570 platform (HG-U133_Plus_2) Affymetrix Human Genome U133 Plus 2.0 Array. The microarray data from GSE44076 was based on the GPL13667 platform (HG-U219) Affymetrix Human Genome U219 Array. The microarray data from GSE44861 was based on the GPL3921 platform (HT_HG-U133A) Affymetrix HT Human Genome U133A Array. The microarray data from GSE136735 were based on the GPL16699 platform Agilent-039494 SurePrint G3 Human GE v2 8 × 60 K Microarray 039381 (Feature Number version). In addition, TCGA and GSE17536 cohorts included 270 and 177 CRC patients’ clinical data respectively. We further merged these cohorts into a single dataset called the Entire cohort.

Identification of SULF1

Firstly, we used the “RMA” and “Affy” packages in R studio to identify differentially expressed genes (DEGs) between CRC and normal colorectal specimens by analyzing microarrays from GSE41328, GSE44076, GSE44861 and GSE136735 based on the screening criteria |logFC|>1 and adjusted p < 0.05. A Venn plot was generated to screen out the overlapping DEGs among datasets GSE41328, GSE44076, GSE44861 and GSE136735. Subsequently, we constructed a protein–protein interaction (PPI) network of these overlapping DEGs using Search Tool for the Retrieval of Interacting Genes (String, http://string-db.org; Version:11.0) online database [25]. The CytoHubba plug-in in Cytoscape software was used to identify the top 10 hub genes of the PPI network. SULF1 was one of the top 10 hub genes. Lastly, we explored the expression levels, overall survival (OS) and disease-free survival (DFS) of 10 hub genes in CRC patients according to gene expression profiling interactive analysis (GEPIA, http://gepia.cancer-pku.cn/) online database. Notably, SULF1 exhibited statistically significant associations with both OS and DFS. Hence, we chose SULF1 for our subsequent study.

GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of genes and genomes) enrichment analyses

We used cBioPortal (https://www.cbioportal.org/) [26] online database to identify 361 co-expression genes of SULF1 (|Spearman| > 0.9; Supplementary Table 6). Based on these co-expression genes, “ggplot2”, “enrichplot”, “org.Hs.eg.db”, and “clusterProfiler” packages were used to perform GO and KEGG enrichment analyses.

Patient specimens and cell culture

All CRC tissues and adjacent cancer tissues were collected from the Second Affiliated Hospital of Nanchang University (Nanchang, China) between September 2018 and September 2023. All of these patients did not receive any radiotherapy or chemotherapy before surgery. The collected tissues were then stored in 4% paraformaldehyde or at -80℃.

All cell lines were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). HCT116 was cultured in RPMI-1640 medium supplemented with penicillin G (100 mg/mL), streptomycin (100 mg/mL), and 10% fetal bovine serum (FBS; Gibco; USA). NCM460, SW620, SW480, HT29, and DLD1 were cultured in DMEM medium with penicillin G (100 mg/mL), streptomycin (100 mg/mL), and 10% fetal bovine serum (FBS; Gibco; USA). All cells were incubated in a 37 °C incubator with 5% CO2.

Cell transfection

Lentiviruses for SULF1 silencing (sh-SULF1#1, sh-SULF1#2, and sh-SULF1#3) and the silencing control (NC) were purchased from HANBIO (Shanghai, China). ARSH plasmid (ARSH) and ARSH siRNA (si-ARSH) were purchased from Genechem (Shanghai, China). Lentiviral transduction was performed according to the manufacturer’s instructions. For plasmid and siRNA transfection, we used Lipofectamine 3000 Transfection Reagent (Invitrogen, Waltham, Massachusetts, USA).

Quantitative real-time PCR (qRT-PCR)

Total tissue and cell RNA were extracted using the Trizol method, which were reverse transcribed into cDNA (TAKARA, RR047A). The cDNA was used for real-time quantitative PCR (TAKARA, RR420A). The 2− ρρCt method was used for data analysis.

The primer sequences used are presented in Supplementary Table 9.

Cell proliferation assay

EdU, CCK8 and colony formation assays were performed to testify proliferation capacity of HCT116 and SW480 cells. For the EdU assay, cells were seeded in 96-well plates with 2 × 104 cells per well. After the cells were incubated at 37 ℃ for 8 h, EDU incubation, fixation, and staining were performed according to the instructions of the YF®594 Click-iT EDU staining kit (UE, Shanghai, China). Photographs were taken under a fluorescence microscope finally. For CCK8 assay, cells (5000 cells/well) were inoculated in a 96-well plate. The medium containing 10% CCK-8 (Biosharp, Beijing, China) was added to each well at the specified time (6 h, 24 h, 48 h, 96 h). After incubation at 37℃ for 2 h, the absorbance of each well was detected at 450 nm, and the data were analyzed to evaluate the cell proliferation ability. For colony formation assay, cells (1000 cells/well) were seeded into 6-well plates. After incubation for two weeks, the cells were fixed with 4% paraformaldehyde, stained with crystal violet, and photographed to calculate the number of spherical cells to evaluate the proliferation ability of the cells.

Transwell assay

We performed Transwell invasion and migration assays to explore cell invasion and migration ability respectively. For the invasion assay, each chamber was pre-coated with Matrigel diluted 1:8 in medium. For both invasion and migration assays, 3 × 104 HCT116 or SW480 cells were seeded in 200 µL serum-free medium into each upper chamber while 600 µl complete medium with 20% fetal bovine serum were filled with each lower chamber. After 48–72 h of incubation, HCT116 or SW480 cells that invaded the lower chamber were fixed with 4% paraformaldehyde and stained with crystal violet.

Wound healing assay

Wound healing assay was performed to explore the migration capacity of HCT116 and SW480 cells. 6 × 104 HCT116 or SW480 cells per well were seeded into six-well plate.

We performed the wound healing assay using a 200µL sterile pipette when cell monolayers were adherent. The complete medium was then replaced with serum-free medium to minimize cell proliferation and promote migration. Finally, wound closure was monitored by photographing the wells at 0 and 24 h.

Cell apoptosis and cell cycle

For cell apoptosis, the cells (104 cells /mL) were seeded in a 6-well plate. After cell growth reached 60-70%, the cells were collected, washed with pre-cooled phosphate-buffered saline (PBS) without Ca2+ and Mg2+, and then centrifuged. Annexin V-FITC and PI (UE, Shanghai, China) were then added and incubated for 15 min in the dark. Finally, 400µL buffer was added for re-suspension and then detected by flow cytometry. For cell cycle, cells were collected, centrifuged, washed with pre-cooled PBS, fixed with pre-cooled 70% ethanol, and placed in a -20℃ refrigerator overnight. On the next day, the fixed cells were washed with pre-cooled PBS, and PI (UE, Shanghai, China) were added after washing. Cell cycle detection was performed by flow cytometry.

Western blotting assay

Total protein lysates were extracted from HCT116 and SW480 cells using radioimmunoprecipitation assay buffer (Beyotime Institute of Biotechnology, Shanghai, China). The lysates were separated by SDS-PAGE and blotted onto PVDF membranes (Bio-Rad). Then, we blocked the protein-free sites on PVDF membranes with 5% skim milk. Subsequently, the PVDF membranes were incubated with antibodies against SULF1 (1:1000, Affinity, DF13592), ARSH (1:1000, Affinity, DF9228), FAK antibody (1:1000, ZENBIO, 381,143), p-FAK antibody (1:1000, ZENBIO, R24276), PI3K antibody (1:1000, ZENBIO, 251,221), p-PI3K antibody (1:1000, ZENBIO, 310,164), AKT antibody (1:1000, ZENBIO, 342,529), p-AKT antibody (1:1000, ZENBIO, 310,021), mTOR antibody (1:5000, Proteintech, 66888-1-Ig), p-mTOR antibody (1:2000, Proteintech, 67778-1-Ig), BAX antibody (1:2000, Proteintech, 50599-2-Ig), cleaved caspase-3 antibody (1:500, Abcam, ab2302), BCL2 antibody (1:5000, Proteintech, 68103-1-Ig), CDK4 (1:1000, Affinity, DF6102), CDK6(1:1000, Affinity, DF6448), CyclinD1(1:1000, Affinity, AF0931), E-cadherin(1:1000, Affinity, AF0131), N-cadherin(1:1000, Affinity, AF5239), Snail(1:1000, Affinity, AF6032), and GAPDH (1:50000, Proteintech, 60004-1-Ig) overnight at 4℃. Then, the PVDF membranes were probed with secondary HRP Linked Secondary Antibodies (Sangon Biotech, Shanghai.). Finally, we used imaging system (DenvilleScientific Inc., Holliston, MA, USA) to visualize immunoblots. Details of all primary antibodies are provided in Supplementary Table 10.

Immunoprecipitation (IP)

Cells were lysed on ice for 30 min with pre-cooled radio immunoprecipitation assay (RIPA) buffer containing phenylmethanesulfonyl fluoride (PMSF) (100:1 ratio). The lysate was then centrifuged at 10,000 x g for 10 min. The supernatant was added with 2 µg primary antibody and incubated in a shaker for 1 h, then 40 µL protein A/G PLUS-Agarose (Santa Cruz, sc-2003) was added and incubated in a shaker at 4℃ for 12 h. The precipitation was collected after centrifugation at 2500 rpm for 5 min. The pellet was washed four times with pre-cooled 10% RIPA buffer. Finally, the precipitate was dissolved in 40 µL of 2× electrophoresis sample buffer, boiled for 10 min, and then western blotting was performed.

Subcutaneous xenograft assay

Ten 4-week-old BALB/c female nude mice, weighing 18–21 g, were purchased from Keris Biotechnology Company, Nanjing, and raised in a specific-pathogen free (SPF) environment.

Ten female BALB/c nude mice were randomly divided into 2 groups: Non-targeting control (NC) group and sh-SULF1#1 group. Cell suspensions (1 × 107) of HCT116 cells stably transfected with either NC or sh-SULF1#1 were added to 200 µL of PBS. To ensure high biological activity during the experiment, these cells were in a logarithmic growth phase. The skin on the right flank of the nude mice was disinfected with alcohol, and 100 µL of each cell suspension was then injected into the area using a 1 mL syringe. The width and length of the tumor were measured at 5-day intervals, and the tumor was calculated by the formula (V): V = 0.52× length × width2. After 40 days of inoculation, the nude mice were euthanized by means of carbon dioxide (CO2) inhalation. The euthanasia chamber was utilized to expose the nude mice to CO2 at a flow rate equivalent to 20% of the replacement volume per minute. Once complete immobility, respiratory arrest, and pupil dilation were observed in the nude mice, the administration of CO2 was terminated. Following a 3-minute observation period to confirm death, the nude mice could be removed from the euthanasia chamber. Subsequently, tumor tissue was extracted for photographic analysis. All animal assays were approved by the Laboratory Animal Science Center of Nanchang University.

Immunohistochemistry (IHC)

The tumor tissue was embedded in paraffin blocks and cut into Sect. 4 μm thick. Tissue sections were incubated with primary and then enzyme-labeled secondary antibodies. Positive staining was visualized using 3,3’-diaminobenzidine (DAB) as the chromogen. IHC staining results from patient or mouse tissues were then quantified using the positive staining cell count method. Protein expression was categorized into a four-tier score (0, 1, 2, or 3) based on the percentage of positively stained cells: 0–5%, 6–25%, 26–50%, and 51–100%. Scores of 0 and 1 were classified as low expression, while 2 and 3 were classified as high expression.

Statistical analysis

GraphPad Prism (version: 8.0.1) and R studio (version: 4.0.3) were used for data analyses Student’s t-test was employed to compare the means between two groups. One-way ANOVA test was employed to assess the means among multiple groups. Kaplan-Meier survival curves were generated and compared using the Log-rank test. All experiments were independently repeated three times. Statistical significance was considered at p < 0.05.