Patients and tissues

A total of 52 pathological specimens were obtained from CRC patients between January 2017 and January 2021 at the Department of Pathology, Nanfang Hospital, Southern Medical University. All cases used in the current study were collected from pre-treatment colonoscopy specimens of patients with nCRT and without any antitumor therapies, including surgery, targeted therapy, biological therapy, and immunotherapy. The 8th edition of the American Joint Council on Cancer (AJCC) tumor regression grading system (TRG) was used for the pathological analysis of tumor specimens after nCRT to evaluate the therapeutic effect [20]. TRG 0 represents complete response (absence of residual cancer cells), TRG 1 represents moderate response (only small clusters or single cancer cells remain), TRG 2 represents minimal response (residual cancer is present with predominant fibrosis), and TRG 3 represents poor response (almost no fibrosis and a large amount of residual cancer). We identified patients with TRG 0 and TRG 1 as radiotherapy responders and those with TRG 2 and TRG 3 as radiotherapy-resistant patients. All specimens were collected and analyzed after obtaining written informed consent from the patients. This study was approved by the Ethics Committee of Nanfang Hospital of Southern Medical University (Guangzhou, China) (Approval No. NFEC-2023-026).

Cell culture

The cell lines used in this study, including the human CRC cell lines SW480, CaCo2, HCT15, HCT116, SW620, HT29, LoVo, RKO, and mouse CRC cell line MC38, were purchased from the American Type Culture Collection (ATCC) and were stored in the Department of Pathology of Southern Medical University. SW480 and SW620 cells were cultured in Leiboviz’s L-15 medium (Gibco) supplemented with 10% fetal bovine serum (FBS) (Gibco). CaCo2, LoVo, HT29, and MC38 cells were cultured in DMEM (Gibco) supplemented with 10% FBS. HCT15 and RKO cells were cultured in RPMI 1640 medium (Gibco) supplemented with 10% fetal bovine serum (FBS). HCT116 cells were cultured in McCoy’s 5A medium (Gibco) supplemented with 10% fetal bovine serum (FBS). All cells were cultured at 37 °C in a 5% CO2 atmosphere.

Ionizing radiation and construction of IR-resistant SW480 and CMT93 cell lines

The cells or mouse tumors were irradiated with defined doses of ionizing radiation using 6 MV X-rays from linear accelerators (Varian 2300EX, Varian, 58 Palo Alto, CA, USA). To construct IR-resistant SW480 cell lines, the SW480 cells were exposed to radiation doses of 4 Gy with a dose rate of 5 Gy/min once daily at 24 h intervals, and the culture was continued after the end of the irradiation; the cycle lasted 8 times. The irradiation can reach 40 Gy, which is the total planned dose of clinical radiotherapy. The cells that survived all the IR cycles were named IR-SW480 and IR-CMT93. The surviving cells were evaluated for radiosensitivity using clonogenic survival assays.

Plasmids and construction of stable cell lines

PREX2 was constructed by cloning the PCR-amplified full-length human PREX2 cDNA into pEZ-Lv201. The pLKD-Puro vector was used to clone short hairpin RNA (shRNA) targeting PREX2. The shRNA sequences targeting PREX2 were obtained from the shRNA sequence prediction website portal (Additional file 1: Table S1). pEZ-Lv201 and pLKD-Puro were purchased from Addgene (Cambridge, MA). The PREX2 overexpressed or knocked down lentivirus was generated by co-transfecting 293T cells with two packing vectors, psPAX and pMD2G, and Lipofectamine 2000 (Invitrogen), and then collecting the supernatants of 293T culture medium after 48 h and filtering through 0.45-mm filters (Millipore, Temecula, CA, USA), and they were concentrated. The expression of PREX2 was detected using real-time PCR and Western blotting.

Enzyme-linked immunosorbent assay (ELISA)

The amount of HMGB1 protein in the supernatant was determined using human HMGB1-specific ELISA kits (Cusabio, #CSB-E08223h) according to the manufacturer’s instructions. The experiment was performed three times with three replicates in each experiment.

Extracellular ATP assessment

After treatment, the cell culture medium was collected, and the supernatant was processed by centrifugation. ATP levels were determined by using the reagent kit (Beyotime, #S0027) and luciferin-based assay for ATP concentration. The experiment was performed three times with three replicates in each experiment.

RNA extraction and real-time PCR

RNA isolation was performed as described previously [21]. Total RNA was extracted from cell lines using TRIzol (Invitrogen, USA), quantified by measuring the absorbance at 260 nm, and then reverse transcribed into cDNAs according to the manufacturer’s instructions.

Real-time quantitative reverse transcription PCR (RT-qPCR) was performed at least three times in triplicate using SYBR Green mix (Vazyme, Q711-02) and the ABI PRISM 7500 Sequence Detection System (Applied Biosystems, USA). Target gene expression was normalized to the geometric mean of the housekeeping gene GAPDH and was calculated using the 2−ΔΔCT method. Primer Express was used to design the real-time PCR primers, and the primer sequences for amplification are shown in Additional file 1: Table S1. The experiment was performed three times with three replicates in each experiment.

Western blot

Western blotting was performed as previously described [22]. The antibodies used are listed in Additional file 2: Table S2.

Immunohistochemistry

Immunohistochemical (IHC) staining was performed as previously described using specific antibodies [22]. The antibodies used are listed in Additional file 2: Table S2. Two observers independently reviewed and scored independently by two observers. Staining intensity was graded according to the following criteria: 0 (no staining), 1 (weak staining or light yellow), 2 (moderate staining or yellow-brown), and 3 (strong staining or brown). The proportion of tumor cells was scored as follows: 0 (no positive tumor cells), 1 (< 25% positive tumor cells), 2 (26–50% positive tumor cells), 3 (55–75% positive tumor cells), and 4 (> 75% positive tumor cells). The staining index was calculated as the staining intensity score × proportion of positive tumor cells. Optimal cut-off values were identified: a staining index ≥ 6 was used to define tumors with high PREX2 expression, and an index < 6 was used to define tumors with low PREX2 expression.

Immunofluorescence

The cells were grown on coverslips after irradiation with 4 Gy. Paraformaldehyde (4%) was used to fix the cells for approximately 10 min, followed by incubation with 0.1% Triton X-100 for approximately 30 min. All samples were placed in a 1% BSA solution (dilution ratio; see instructions) and closed on a horizontal shaker for 30 min. The primary antibody solution and incubated overnight at 4 °C. The corresponding fluorescent dye-conjugated secondary antibody was added and incubated in the dark for 1 h, and 10 μL DAPI was added for 15 min. Finally, images were collected using fluorescence microscopy or confocal microscopy. The antibodies used are listed in Additional file 2: Table S2. The experiment was performed three times with three replicates in each experiment.

Radiation clonogenic assay

Cells were seeded on six-well plates at a density of 4 × 102, 8 × 102, 1 × 103, 5 × 103, and 8 × 103 cells per well and exposed to IR at doses of 0 Gy, 2 Gy, 4 Gy, 6 Gy, and 8 Gy respectively. After incubation for 14 d at 37 °C, the plated cells were washed with PBS, fixed with 4% paraformaldehyde for 30 min, and stained with crystal violet for 20 min. Colonies (≥ 50 cells/colony) were counted under a dissecting microscope. The surviving fraction (SF) curve was revised via a multi-target single-hit model with the following formula “SF = 1 − (1 − e−D/D0) N”. The experiment was performed three times with three replicates in each experiment.

Comet assay

For the comet assay, single-cell suspensions were prepared and collected for the experiments. After gel preparation, the gels were poured into a pre-chilled lysis buffer, split at 4 °C for 2 h. The slides were removed, washed, and placed in anhydrous ethanol for 5 min after electrophoresis. After adding PI staining solution to each slide and using 515–560 nm excitation light for fluorescence microscopy, DNA images of PI samples were red. Finally, nuclear DNA and migrated DNA were observed using Open Comet software to measure nuclear DNA diameter and DNA migration length. The experiment was performed three times with three replicates in each experiment.

Flow cytometry

For apoptosis analysis, the cells were irradiated and collected at 0 h and 48 h cells to prepare single-cell suspensions. The cells were then operated according to the APC Annexin V/PI Apoptosis Detection Kit (BioLegend #640932). For cell cycle analysis, the harvested cells were incubated with DNA staining and permeabilization solution (protected from light, room temperature, 30 min) and then analyzed by flow cytometry. To determine the expression of calreticulin, the cells were stained with the respective primary antibodies, followed by Alexa Fluor 488 anti-mouse IgG. The experiment was performed three times with three replicates in each experiment.

To investigate immune cell infiltration, we first isolated tumor-infiltrating lymphocytes (TILs) by Percoll. The TILs were incubated with antibodies conjugated with fluorescent biotin specific for CD45, CD4, and CD8A for 30 min on ice. After incubation, the cells were resuspended and analyzed using flow cytometry. The antibodies used are listed in Additional file 2: Table S2.

RNA-seq and data analysis

Total RNA was extracted using the TRIzol reagent kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. RNA quality was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and verified by RNase-free agarose gel electrophoresis. After total RNA was extracted, eukaryotic mRNA was enriched with Oligo (dT) beads, fragmented into short fragments using fragmentation buffer, and reverse transcribed into cDNA using NEBNext Ultra RNA Library Prep Kit for Illumina (NEB #7530, New England Biolabs, Ipswich, MA, USA). The purified double-stranded cDNA fragments were end-repaired and a base was added and ligated to Illumina sequencing adapters. The ligation reaction was purified using AMPure XP Beads (1.0X). The ligation products were size-selected by agarose gel electrophoresis, PCR-amplified, and sequenced using an Illumina NovaSeq6000 by Gene Denovo Biotechnology Co. (Guangzhou, China). Quality control of raw reads was performed using fastp (version 0.18.0) to obtain clean reads. An index of the reference genome was constructed, and paired-end clean reads were aligned to the reference genome using HISAT2.2.4 with the parameter “-rna-strandness RF” and default settings. The aligned reads for each sample were assembled using StringTie v1.3.1 in a reference-based approach. Fragment per kilobase of transcript per million mapped reads values were calculated for each transcription region to quantify expression abundance and variations, utilizing the RSEM software. Differentially expressed genes (DEGs) within SW480 and IR-SW480 groups, or between IR-SW480 cells and PREX2 knockdown IR-SW480 cells, were identified with an adjusted p-value of < 0.05 using the R package “limma.” The “ggplot2” package facilitated the creation of a volcano plot illustrating the DEGs, while the top 50 DEGs’ expressions were visualized in a heatmap generated by the “pheatmap” package. To further explore the associated signaling pathways of PREX2, the Gene Set Enrichment Analyses (GSEA) software was employed to identify differentially enriched pathways between IR-SW480 and PREX2 knockdown IR-SW480 groups. Additionally, Gene Ontology (GO) function enrichment analysis was conducted based on DEGs between IR-SW480 cells and PREX2 knockdown IR-SW480 cells.

Animal model

Animal studies received approval from the Committee on the Ethics of Animal Experiments of Southern Medical University (Approval No. L2019023). Female C57/B6J mice (5–6 weeks) and nude mice (4–6 weeks old) were procured from the Guangdong Animal Center. For the xenograft subcutaneous tumor model, 2 × 106 cells/100 μL of tumor cells were injected subcutaneously into the rump dorsum of nude mice. CMT93 (1 × 106 cells/100 μL) was subcutaneously injected into the flanks of C57/B6J mice. Upon reaching a tumor volume of 100 mm3, they were irradiated with 8 Gy three times in total, with an irradiation interval of 1 day. The mice were anesthetized with pentobarbital of 50 mg/kg (intraperitoneally). After euthanasia by cervical dislocation within 8 weeks, tumor tissues underwent formaldehyde fixation, paraffin-embedding, and 3-mm sections were cut for H&E and IHC staining.

For the subcutaneous injection of MC38 and IR-CMT93 cells (1 × 106 cells/100 μL) into C57/B6J mice. Mice were randomly assigned to groups 2 weeks before receiving IR treatment (8 Gy ×3 fractions) at 2-day intervals. Small-molecule inhibitors of PREX2 (PREX-in1, 0.8 mg/kg; Specs, The Netherlands) were intraperitoneally injected 3 h before IR treatment. Tumor volume (V) was monitored every 2 days by measuring the short axis (W) and the long axis (L) of the xenograft tumor and calculated with the following formula: V = (L × W2)/2. At the end of the experiment, all mice were sacrificed by cervical dislocation, tumor tissues were dissected, photographed, and subjected to flow cytometry for tumor-infiltrating lymphocytes or formaldehyde fixation for IHC analyses.

Bioinformatics analysis

To investigate the correlation between PREX2 and radiosensitivity, gene expression profiles and clinical information were obtained from the public Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) and the Cancer Genome Atlas (TCGA) databases (https://www.cancer.gov/ccg/research/genome-sequencing/tcga). Our study utilized two GEO cohorts (GSE145037 and GSE150082). PREX2 expression was analyzed based on the patient’s sensitivity to radiotherapy. The radiosensitivity index (RSI) was calculated according to the reported methods [23], RSI = − 0.0098009 × AR + 0.0128283 × Jun + 0.0254552 × STAT1-0.0017589 × PRKCB-0.0038171 × RELA + 0.1070213 × ABL1-0.0002509 × SUMO-0.0092431 × CDK1-0.0204469 × HDAC1-0.0441683 × IRF1. The radiosensitivity index (RSI) was calculated, and patients were divided into RSI-low and RSI-high groups according to the median RSI value in TCGA-COADREAD. Kaplan–Meier method and log-rank method were employed to assess the relationship between PREX2 expression and progression-free interval by using the R package “survival.” The algorithm determined the split point where the p-value was minimal.

To explore the correlation between PREX2 and immune infiltration in CRC, the Estimating the Proportions of Immune and Cancer Cell (EPIC) algorithm evaluated the immunity scores for each TCGA-COADREAD patient [24].

Statistical analysis

Data statistics were performed using the IBM SPSS Statistics 22 version software. The GraphPad Prism 9 statistical software was used for data processing. Correlation coefficients were determined using Spearman’s rank correlation test. Unpaired two-tailed Student’s t-test was used to compare the two groups. Two-way ANOVA analysis of variance was used to compare the two groups in the animal experiments. For comparison of more than two groups, p-values were calculated using one-way ANOVA. Survival curves were plotted using the Kaplan–Meier method and compared by the log-rank test. *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant.