Formalin-fixed paraffin-embedded (FFPE) tumor tissue samples from 4 neoadjuvant treatment-sensitive and 4 treatment-resistant patients with LARC (T3-4N0 or any T, N+) who were treated at The First Affiliated Hospital of Army Medical University (Chongqing, China) between July 2018 and August 2022 were included in the present study. The cohort comprised of 7 men and 1 woman, with a median age of 64 years (range: 42–81 years). From retrospectively screened cases with pelvic MRI available both before and after nCRT, cases with clear radiological response categories according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 (https://project.eortc.org/recist/wp-content/uploads/sites/4/2015/03/RECISTGuidelines.pdf) and with adequate available FFPE material for pilot DIA-MS analysis were identified. The tumor staging was recorded according to the 8th edition of the American Joint Committee on Cancer (and Union for International Cancer Control Staging System for Colorectal Cancer) (26). Tumor response was evaluated 4–8 weeks after treatment according to the RECIST guidelines. Complete response (CR) and partial response (PR) were categorized as treatment-sensitive, whereas stable disease (SD) and PD were categorized as treatment-resistant. Response categories in the proteomic discovery set were defined primarily according to RECIST guidelines, using pelvic MRI before and after nCRT. Pathological tumor regression grade was recorded descriptively when available but was not used for the primary response grouping. Biopsy tissues were collected before neoadjuvant treatment and tumor tissues were collected during surgery after neoadjuvant therapy. The present study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Army Medical University, PLA [approval no. (A)KY2025105]. Written informed consent was obtained from all participants or their legal guardians, where applicable.

Pre-nCRT biopsy samples were obtained through colonoscopic biopsy. Fresh samples were fixed in 10% neutral buffered formalin at room temperature for 12–24 h, followed by routine pathological examination whereby ~4 µm sections were cut from FFPE blocks and stained with H&E for pathological determination of adenocarcinoma.

Post-nCRT samples were fresh, unfixed and uncut. After collection, images of gross samples were routinely captured and sectioned. To accurately assess tumor regression and account for intratumoral heterogeneity post-treatment, the entire tumor was sampled. All specimens were fixed in 10% neutral buffered formalin (Chongqing Boyi Chemical Reagent Co., Ltd.) at room temperature for 24–48 h, sectioned for routine histopathological examination and 4 µm FFPE sections were stained with H&E for analysis.

H&E staining of both pre-nCRT biopsy specimens and post-nCRT resection specimens was performed using the same staining program on an automated slide stainer (Ningbo Chiwell Biotechnology Co., Ltd.) at room temperature. Briefly, FFPE sections were deparaffinized in xylene, rehydrated through graded ethanol and water, stained with hematoxylin for 5 min, differentiated for 8 sec, blued for 15 sec and counterstained with eosin for 7 sec. All H&E-stained sections from both pre-nCRT biopsy specimens and post-nCRT resection specimens were visualized using an Olympus BX43 microscope (Olympus Corporation).

For 15 FFPE rectal cancer tissues from 8 patients (8 before and 7 post-nCRT treatment), 8 sections (10-µm thick) were cut using a microtome and transferred to 5 ml Eppendorf tubes. Before protein extraction, FFPE sections were reviewed by an experienced pathologist, but no macrodissection or laser-capture microdissection was performed. Proteins were extracted using an EasyPept FFPE micro sample preparation kit (cat. no. OSFP0006; Shanghai Omicsolution Co., Ltd.). Briefly, tissue sections were lysed using 50 µl kit-provided lysis buffer with steel balls and subjected to 55 Hz grinding for 4 min at 4°C; the samples were then incubated with agitation at 95°C and 1,000 rpm for 30 min. The sample was left to stand and after it reached room temperature, the supernatant was collected after centrifugation (4°C; 12,000 × g; 20 min) and the protein concentration was determined using the BCA method. A total of 30 µg of protein was incubated at 95°C for 2 h for reduction and alkylation and cooled to room temperature. The samples were then digested with the kit-provided protease at 37°C for 3 h, after which it was centrifuged (room temperature; 20,000 × g; 1 min) and the supernatant was collected. The peptide-containing supernatant was loaded onto a desalting column. The column was then washed twice and peptides were eluted with 100 µl kit-provided elution buffer, which was then dried using a JM50-Plus vacuum freeze dryer (Beijing JM Technology Co., Ltd.) for subsequent analysis.

For each sample, 200 ng total peptide was mixed with a kit containing 11 standard peptides (cat. no. Ki-3002-2; Biognosys AG) at a 1:10 ratio, then separated using a nanoElute LC system (Bruker Daltonics GmbH). Peptides were first trapped on a thermo trap cartridge using mobile phase A (0.1% formic acid in water) and then separated on a 15 cm reversed-phase C18 PepSep^® column (Bruker Daltonics GmbH; particle size: 1.9 µm; inner diameter: 75 µm) maintained at 50°C. Gradient elution was performed at a flow rate of 0.3 µl/min: Mobile phase B proportion (0.1% formic acid in acetonitrile) increased from 2 to 24% in the first 30 min, then to 38% in the next 5 min and to 80% in 7 min (total duration: 42 min) and finally maintained at 80% for 3 min.

Ultra-high-performance LC was coupled to a timsTOF Pro2 mass spectrometer (Bruker Daltonics GmbH) equipped with a nano-electrospray ion source operated in positive ion mode. The nitrogen gas temperature was set to 180°C, the nebuliser pressure was 2.90 psi and the dry gas flow rate was 3.0 l/min. Data were acquired in Parallel Accumulation-Serial Fragmentation mode. The electrospray ion source voltage was set to 1,400 V, precursor scanning range was 100–1,700 m/z, ion mobility range (1/K0) was 0.6–1.6 Vs/cm² and both trapped ion mobility spectrometry ramp time and accumulation time were set to 100 ms, with a duty cycle close to 100%. The secondary mass spectrometry scanning range was 400–1,200 m/z and the collision energy increased linearly with ion mobility (20 eV at 0.85 Vs/cm² and 59 eV at 1.3 Vs/cm²). The isolation window was set to 25 Th, with a total of 32 windows.

Raw MS files were processed in library-free mode using Spectronaut^® software (version 18.2.230802.50606; Biognosys AG) with its built-in Pulsar search engine. MS spectra lists were searched against their species-level UniProt FASTA database (uniprot_Homo sapiens_9606_reviewed_2024_05_fasta, http://www.uniprot.org/proteomes/UP000005640), with carbamidomethylation of cysteine as a fixed modification and oxidation of methionine and acetylation of protein N-termini as variable modifications. Trypsin was used as the protease. A maximum of two missed cleavages was allowed. The false discovery rate at both peptide and protein levels was set to 0.01. The maximum allowable initial precursor mass deviation for peptide identification was 20 ppm and the maximum allowable fragment mass deviation was 20 ppm. The raw data screening criterion was that each protein contained at least 1 unique peptide. Since samples were collected both before and after treatment, protein abundance may have been downregulated; thus, missing values were replaced with half of the detected minimum value.

Bioinformatic analyses were performed using R software (version 3.6.3; R Foundation for Statistical Computing) (27). Principal component analysis (PCA) was performed using the R package ‘ropls’ (version 1.16.0) (28) and visualized using ‘ggplot2’ (version 3.3.2) (29). Proteins were considered differentially abundant proteins (DAPs) if they met a nominal P<0.05 by Student's t-test and had a fold change ≥1.2 or ≤0.83. DAPs were visualized using bar plots generated with ‘ggplot2’. Hierarchical clustering heatmaps were generated using ‘pheatmap’ (version 1.0.12) (30). Subcellular localization analysis was performed using ‘loctree3’ (version 1.0.1) (31) and visualized using ‘ggplot2’. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed using the KEGG database (http://rest.kegg.jp) (32) and the results were visualized using ‘ggplot2’. The public transcriptomic dataset GSE209746 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE209746) (15) and public proteomic dataset PXD060201 (https://www.ebi.ac.uk/pride/archive/projects/PXD060201) (25) were downloaded and analyzed for cross-dataset prioritization of candidate proteins associated with nCRT response.

For IHC experiments, a tissue microarray (TMA) containing 92 CRC and 87 matched adjacent non-tumor tissues was acquired from Shanghai Outdo Biotech Co., Ltd. (cat. no. HColA180Su24). IHC staining was conducted using an IHC SP Kit (cat. no. SP-9001; Beijing Zhongshan Jinqiao Biotechnology Co., Ltd.) following the manufacturer's instructions. The primary antibody used was anti-branched-chain keto acid dehydrogenase E1 subunit α (BCKDHA; cat. no. 30028-1-AP; 1:6,000; Proteintech Group, Inc.) overnight at 4°C. Staining results were assessed independently by two pathologists. Staining intensity was classified into four levels: Negative, score 0; weak, score 1; medium, score 2; and strong, score 3. The percentage of positively stained cells ranged from 0–100%. The final IHC score was calculated by multiplying the percentage of positive cells by the staining intensity. High and low BCKDHA expression were defined using an IHC score cut-off of 2.375; scores ≤2.375 were classified as low expression, whereas scores >2.375 were classified as high expression.

The human CRC SW480 cell line was obtained from American Type Culture Collection (cat. no. CCL-228), grown in medium supplemented with 10% FBS (Sangon Biotech Co., Ltd.) and maintained in a humidified 5% CO₂ incubator at 37°C, with cells passaged using standard cell culture techniques.

To construct the BCKDHA-overexpressing cell lines, the BCKDHA coding sequence was synthesized by Beijing Tsingke Biotech Co., Ltd., and then inserted into the XhoI-NotI sites of the pLV10ltr-PGK-Hyg-CMV vector (Beijing Tsingke Biotech Co., Ltd.). The short hairpin (sh-)RNAs were integrated into the pLKO.1-mCherry-puro vector (Beijing Tsingke Biotech Co., Ltd.). The scrambled RNA sequence was 3′-CCGGGGTTCTCCGAACGTGTCACGTCTCGAGACGTGACACGTTCGGAGAACCTTTTTT-5′ and the shBCKDHA sequence was 3′-CCGGGGGTATGGCATCATGTCAATCCTCGAGGATTGACATGATGCCATACCCTTTTTT-5′. For lentiviral packaging, 5 µg transfer plasmid was co-transfected with 3.5 µg psPAX2 and 1.5 µg pMD2.G plasmids into 293T cells in 6-cm dishes using polyethylenimine (Yeasen Biotechnology Co., Ltd.). The plasmid-PEI complexes were formed at room temperature for 15 min and then added to 293T cells. After incubation at 37°C for 48 h, the supernatant was harvested and used to infect SW480 cells. SW480 cells were screened using 200 µg/ml hygromycin for 7 days or 3 µg/ml puromycin for 3 days. The construct efficiency was determined using a western blotting.

For the colony formation assay and western blotting analyses after irradiation, cells were seeded into 6-well plates in 2 ml DMEM supplemented with 10% FBS (both Sangon Biotech Co., Ltd.). Irradiation was performed using a Varian 600CD machine (Varian Medical Systems, Inc.) with 6 MV X-rays at a dose rate of 4 Gy/min. The source-to-surface distance was set to 100 cm with a field size of 37.0×39.4 cm. For dose buildup, the plates were covered with 1 cm solid water and placed on a 4 cm-thick solid water slab beneath. Irradiation was delivered at a gantry angle of 0°.

SW480 cells were seeded at 1,000 or 2,000 cells per well in 6-well plates. The following day, cells were exposed to 0, 1, 2, 3, 4 or 5 Gy radiation at a dose rate of 4 Gy/min at room temperature, corresponding to irradiation durations of 0, 15, 30, 45, 60 and 75 sec, respectively. Cells were then cultured in a humidified 5% CO₂ incubator at 37°C for 10–14 days to allow colony formation. Colonies were fixed with 4% formaldehyde at room temperature for 20 min, stained with 0.05% crystal violet, and analyzed using ImageJ (version 1.54 g; National Institutes of Health). Colonies containing >50 cells were considered surviving colonies. The survival fraction was calculated as follows: Planting efficiency (PE) of untreated (UT) cells (PE_UT)=(number of colonies/number of seeded cells) ×100; survival fraction=number of colonies/(number of seeded cells × PE_UT/100).

Western blotting analysis was performed as previously described (33). In brief, cells were lysed in RIPA buffer (cat. no. C510006; Sangon Biotech Co., Ltd.) containing a protease inhibitor (Roche Diagnostics GmbH) and incubated on a rocker at 4°C for 15 min. The protein concentration of the lysates was measured using a BCA protein kit (Sangon Biotech Co., Ltd.), normalized to equal amounts of protein (20 µg per lane), resolved on a 10–12% SDS-gel, transferred to PVDF membranes and probed at 4°C overnight with one of the following primary antibodies: BCKDHA (cat. no. A21588; 1:1,000), phosphorylated (p)-ataxia telangiectasia mutated (p-ATM; Ser1981; cat. no. AP1030; 1:500), ATM (cat. no. A19650; 1:500), p-checkpoint kinase 2 (p-Chk2; Thr68; cat. no. AP0590; 1:1,000), Chk2 (cat. no. A19543; 1:500), DNA repair protein RAD51 homolog 1 (RAD51; cat. no. A26856; 1:1,000), H2A histone family member X (γ-H2AX; cat. no. AP1555; 1:1,000), H2AX (cat. no. A11361; 1:1,000) or β-actin (cat. no. AC026, 1:1,000). All primary antibodies were purchased from ABclonal Biotech Co., Ltd. After probing with HRP-conjugated goat anti-rabbit IgG (cat. no. D110058; 1:2,500) or HRP-conjugated goat anti-mouse IgG (cat. no. D110087; 1:2,500) secondary antibodies (both Sangon Biotech Co., Ltd.) at room temperature for 1 h, signals were detected using an ECL reagent (cat. no. C520045; Sangon Biotech Co., Ltd.).

A χ² test was used to assess the association between BCKDHA expression and clinicopathological parameters. Kaplan-Meier analysis and the log-rank test were performed to assess the association between BCKDHA expression and overall survival (OS) in patients with CRC. Multivariate survival analysis was performed using Cox proportional hazards regression analysis. The survival fraction in the colony formation assay was analyzed for significance using two-way ANOVA. P<0.05 was considered to indicate a statistically significant difference.

To investigate proteins potentially associated with nCRT response in LARC, four nCRT-sensitive and four nCRT-resistant FFPE tumor tissue pairs collected before and after treatment were analyzed using DIA-MS (Fig. 1A). The clinical characteristics of the 8 patients are summarized in Table I. Representative MRI images of one nCRT-sensitive case and one nCRT-resistant case before and after nCRT are shown in Fig. 1B. The patients received intensity-modulated radiation therapy (Fig. 1C) combined with concurrent capecitabine chemotherapy. After nCRT, all patients underwent surgery except one patient with PD, yielding 15 FFPE tumor specimens for proteomic analysis. Representative H&E staining is shown in Fig. 1D.

DIA-MS identified 39,052 peptides and 6,006 proteins. Samples were labeled by treatment status and response: Before_sensitive (BS), before_resistant (BR), post_sensitive (PS) and post_resistant (PR). PCA revealed distinct separation among the four groups (Fig. 2A). Bar plots of DAPs are shown in Fig. 2B. To characterize response-associated DAPs, comparisons between BS with BR and PS with PR were performed. Under this denominator-first notation, a higher abundance in the second group corresponded to the resistant group in these two comparisons. Hierarchical clustering using proteins meeting the predefined thresholds separated sensitive and resistant samples both before treatment and after nCRT (Fig. 2C and D). The complete list of identified proteins and the DAP lists for the BS vs. BR and PS vs. PR comparisons are provided in Table SI, Table SII, Table SIII. Given this was a small proteomic discovery set, these analyses were intended for candidate prioritization rather than definitive biomarker selection.

To identify candidate proteins associated with nCRT resistance, proteins that were more abundant in resistant than in sensitive tumors before nCRT (BS vs. BR) and after nCRT (PS vs. PR) were examined. Before nCRT, 133 proteins were found to be more abundant in resistant tumors, primarily localized in the cytoplasm (33.08%), secretory fraction (24.81%) and nucleus (24.06%; Fig. 3A). Preliminary KEGG analysis indicated enrichment in pathways that include ‘Adherens junction’, ‘Focal adhesion’, ‘Tight junction’, ‘Regulation of actin cytoskeleton’, ‘Propanoate metabolism’ and ‘Amino sugar and nucleotide sugar metabolism’ (Fig. 3B). After nCRT, 290 proteins were found to be more abundant in resistant tumors, with nuclear localization (36.55%) being the most prominent category (Fig. 3C). Preliminary enrichment was observed in pathways that include ‘DNA replication’, ‘RNA polymerase’, ‘Ribosome’, ‘Protein export’, ‘Nucleocytoplasmic transport’ and ‘Metabolic pathways’ (Fig. 3D). Representative proteins of interest from these comparisons are listed in Tables II and III.

To further prioritize candidate proteins associated with nCRT response, the public proteomic dataset PXD060201 (25) and transcriptomic dataset GSE209746 (15) were analyzed. BCKDHA emerged as a shared candidate across all three datasets (Fig. 4A). BCKDHA is the E1 subunit of the branched-chain keto acid dehydrogenase complex, which catalyzes the catabolism of branched-chain amino acids (34). In the present proteomic discovery set, BCKDHA exhibited higher abundance in resistant tumors both before and after nCRT (Fig. 4B and C). A similar direction of change was observed in PXD060201 (Fig. 4D) and GSE209746 (Fig. 4E).

To further assess its broader clinical relevance, IHC staining was performed on a TMA containing samples from 92 patients with CRC, with 87 matched adjacent non-tumorous tissues. BCKDHA localized to the cytoplasm (Fig. 4G) and IHC scores were significantly higher in CRC tissues compared with adjacent non-tumor tissues (Fig. 4F). The χ² test (Table IV) showed that high BCKDHA expression was significantly associated with patient age. Kaplan-Meier analysis showed that high BCKDHA expression was associated with shorter OS in patients with CRC (Fig. 4H). Cox proportional hazards regression analysis (Table V) revealed that high BCKDHA expression, pathology grade and T stage were significantly associated with worse OS in univariate analysis. Multivariate Cox regression showed that high BCKDHA expression was independently associated with a shorter OS in the CRC cohort. Collectively, these data support BCKDHA as a candidate protein associated with nCRT resistance and suggest broader clinical relevance in CRC.

To examine whether BCKDHA contributed to the response to radiotherapy, SW480 cell lines with either stable BCKDHA overexpression or stable BCKDHA knockdown were established. In BCKDHA overexpressing SW480 cells, colony formation assays exhibited significantly increased survival after irradiation (Fig. 5A and B). In GSE209746, BCKDHA mRNA expression was found to be positively correlated with a number of DDR-associated genes, including ATM, ATR, BRCA1, BRCA2, RAD51, RAD52, RPA1, RPA2, RPA3, MRE11 and H2AX (Fig. 5C), supporting a potential association between BCKDHA and DDR. In BCKDHA-overexpressing SW480 cells, higher levels of ATM, Chk2, RAD51 and H2AX were observed, together with increased phosphorylation of ATM and Chk2 at 30 min to 1 h post-irradiation compared with the control cells (Fig. 5D). In addition, BCKDHA overexpression was also accompanied by earlier induction of γ-H2AX after irradiation.

By contrast, stable BCKDHA knockdown significantly reduced colony formation after irradiation (Fig. 6A and B). BCKDHA knockdown was accompanied by lower levels of ATM, Chk2, RAD51 and H2AX, attenuated radiation-induced phosphorylation of ATM and Chk2 and lower γ-H2AX levels across the examined time points post-irradiation (Fig. 6C). Collectively, these findings provide preliminary functional evidence that BCKDHA overexpression is associated with increased clonogenic survival and a more resistant phenotype after irradiation, whereas BCKDHA knockdown is associated with reduced clonogenic survival and enhanced radiosensitivity in SW480 cells.