Patients who underwent surgical procedures at The University of Osaka Hospital between October 2023 and May 2024 and were diagnosed with CRC according to the guidelines of the Japanese Society for Cancer of the Colon and Rectum (21) were considered for inclusion. Exclusion criteria included inflammatory bowel disease and familial adenomatous polyposis. No exclusions were made based on age or disease stage. All cases that underwent surgery during the specified period and were evaluated by DR classification were included in the study. A total of 59 CRC tissue samples were included using opportunistic sampling based on the availability of cases meeting the inclusion criteria; the sample size was not determined by formal power calculation. The study cohort consisted of 26 males and 33 females, with a median age of 73 years (range: 40–88 years). All patients received a definitive diagnosis of CRC. Clinicopathological classification was determined using the criteria of the Japanese Society for Cancer of the Colon and Rectum. After surgery, patients with lymph node metastasis generally received adjuvant chemotherapy. Collected samples were immersed in 10% buffered formalin and fixed overnight at 4°C, then dehydrated through a graded ethanol series and embedded in paraffin.

The current study was approved by the ethics committee of the Graduate School of Medicine, University of Osaka (approval nos. 15144 and 19020; Suita, Japan). Written informed consent was obtained from all participants prior to inclusion in the study. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki.

Between October 2023 and May 2024, 59 CRC tissue samples were collected during surgeries performed at the Department of Gastroenterological Surgery, University of Osaka. Samples were formalin-fixed at 4°C overnight, processed through graded ethanol and embedded in paraffin. A pathologist blinded to the clinical outcomes classified 16 cases as having immature DR and 43 as mature based on established histological criteria.

CAFs and normal fibroblasts (NFs) were isolated from tumor and adjacent normal tissues, respectively, immediately after CRC resection. To prevent cross-contamination, separate surgical blades were used for each tissue type. Tissue fragments underwent enzymatic digestion with collagenase, and the resulting cell pellet was suspended in Dulbecco's Modified Eagle Medium (DMEM; Sigma-Aldrich; Merck KGaA; RRID:SCR_001905) with antibiotics. Cells were seeded in 6-well plates with DMEM supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.; RRID:SCR_008452) and cultured at 37°C with 5% CO₂.

CAFs and NFs were cultured in DMEM supplemented with 10% FBS under normoxic (37°C, 5% CO₂) and hypoxic (37°C, 1% O₂) conditions. The human CRC cell lines HT29 (RRID: CVCL_0320), HCT116 (RRID: CVCL_0291) and RKO (RRID: CVCL_0504) were obtained from the American Type Culture Collection and authenticated using short tandem repeat profiling within 6 months of experimental use. Cell lines were routinely tested for mycoplasma contamination using polymerase chain reaction and were maintained in DMEM supplemented with 10% FBS under the same conditions as fibroblasts.

Cells were cultured in 96-well plates for 24 h, fixed with 4% paraformaldehyde at room temperature (RT) for 15 min and treated with Triton X-100 and 5% bovine serum albumin (MilliporeSigma) at RT for 10 min. Subsequently, cells were washed twice with phosphate-buffered saline (PBS), and primary antibodies including anti-α-smooth muscle actin (α-SMA; 1:400; cat. no. 19245; Cell Signaling Technology, Inc.; RRID: AB_2734735), anti-hypoxia-inducible factor-1α (HIF-1α; 1:500; cat. no. ab51608; Abcam; RRID: AB_880418) and anti-epithelial cell adhesion molecule (EpCAM; 1:500; cat. no. 2929; Cell Signaling Technology, Inc.; RRID: AB_2098834) were applied. Cells were incubated at RT for 1 h, washed twice with PBS and incubated with Alexa Fluor 647 anti-rabbit IgG secondary antibody (1:500; cat. no. 4414; Cell Signaling Technology, Inc.; RRID: AB_10694544) for 30 min. After washing with PBS, cells were counterstained with 4′,6-diamidino-2-phenylindole (1 µg/ml) and visualized using a confocal microscope.

Immunohistochemistry was performed as previously described (10,11). Primary antibodies used included anti-HIF-1α (1:200), anti-periostin (1:400; cat. no. 20302; Cell Signaling Technology, Inc.; RRID: AB_2798819), anti-parathyroid hormone-related protein (PTHrP; 1:200; cat. no. 10817-1-AP; Proteintech Group, Inc.; RRID: AB_2174535) and anti-vitamin D receptor (VDR; 1:200, cat. no. BS-2987R; Bioss Antibodies; RRID: AB_11058910). VECTASTAIN^® Elite^® ABC Rabbit IgG Kit (Vector Laboratories, Inc.; RRID: AB_2336817) was used according to manufacturer's instructions. Primary antibodies were incubated overnight at 4°C. Staining intensity was evaluated by two pathologists independently and scored as +2 (equivalent to positive control), +1 (weaker than positive control), or 0 (unstained) for cytoplasmic and peritumoral stroma. A score of +2 was considered positive.

For indirect co-culture experiments, supernatants from CAFs cultured under hypoxic and normoxic conditions were collected. HT29 cells were then cultured in a medium containing a 1:1 ratio of DMEM supplemented with FBS and the respective CAF-conditioned supernatant.

NFs and CAFs were cultured in DMEM supplemented with PBS and PTHrP (1 µg/ml) at 37°C and 5% CO₂. For vitamin D experiments, CAFs were cultured in DMEM supplemented with 1α,25-dihydroxyvitamin D3 (10 µg/ml) dissolved in dimethyl sulfoxide (DMSO; final DMSO concentration ≤0.5%) at 37°C and 5% CO₂.

Cells were seeded in 96-well plates at a density of 1.0×10³ cells/well. Cell proliferation was assessed at 24, 48 and 72 h after treatment using 10 µl of Cell Counting Kit-8 (cat. no. CK04; Dojindo Molecular Technologies, Inc.) at 37°C for 2 h according to the manufacturer's protocol. The association between absorbance measurements and manual cell counting was verified before the experiment. Absorbance was measured at 450 nm.

The wound healing assay was used to assess cell migration. Cells were plated in 6-well dishes at 2.0×105 cells/well and cultured to 60–80% confluency. A 200-µl pipette tip was used to create a linear scratch in the cell monolayer. Culture medium was switched to DMEM containing 1% FBS to inhibit cell proliferation. Images were captured using a BZ-X710 microscope (Keyence Corporation) at 0, 24 and 48 h post-scratch and analyzed using ImageJ (version 1.54f; National Institutes of Health) (RRID:SCR_003070). Cell migration was quantified by measuring the mean area between wound edges at three randomly selected locations by an investigator blinded to the experimental conditions.

HT29 cells co-cultured with supernatant from normoxia- or hypoxia-cultured CAFs were seeded in 96-well plates at 1.0×10⁴ cells/well for 24 h. Cells were then exposed to varying concentrations of oxaliplatin. Cell viability was evaluated using Cell Counting Kit-8 according to the manufacturer's protocol.

Sample preparation and library construction were performed using the TruSeq stranded mRNA sample preparation kit (Illumina, Inc.) following the manufacturer's protocols. HCT116, HT29 and RKO cell lines were cultured under normoxic (37°C, 5% CO₂) and hypoxic (37°C, 1% O₂) conditions. Sequencing was performed on a DNBSEQ-G400 sequencer (MGI Tech Co., Ltd.) in 100-base single-read mode. Adapter sequences were removed using Trimmomatic (version 0.38; RRID:SCR_011848). Cleaned reads were aligned to the hg19 human reference genome using TopHat2 (version 2.1.1; RRID:SCR_013035). Fragments per kilobase of exons per million mapped fragments were calculated using Cufflinks (version 2.2.1; RRID:SCR_014597). Gene Set Variation Analysis was performed using R (version 4.3.2; RRID:SCR_001905). RNA quality and integrity were assessed using the Agilent 2100 Bioanalyzer system. RNA integrity was evaluated by electrophoresis to measure the degree of degradation. Sequencing was performed using the NovaSeq 6000 S1 Reagent Kit v1.5 (200 cycles; cat. no. 20028318; Illumina, Inc.). Final library concentrations were quantified using the KAPA Library Quantification Kit (cat. no. KK4824/D; Roche Diagnostics) by real-time PCR. Libraries were normalized to 2 nM before loading onto the sequencer.

Supernatants from CAFs cultured under hypoxic and normoxic conditions were collected. Total protein was extracted using radioimmunoprecipitation assay buffer containing protease and phosphatase inhibitors (Thermo Fisher Scientific, Inc.). Proteins were processed using nano liquid chromatography-tandem mass spectrometry configured with an Ultimate 3000 Nano LC system, Q-Exactive (Thermo Fisher Scientific, Inc.). Raw data were analyzed using Scaffold 5 (Proteome Software Inc.; RRID:SCR_014345). Ionization mode was positive (nano-ESI with capillary voltage of 1.8 kV). Flow rate was 300 nl/min.

CAFs cultivated under both hypoxic and normoxic conditions were prepared for single-cell RNA sequencing. Libraries were prepared according to the Chromium Next GEM Single Cell 3′Reagent Kits (version 3.1; 10X Genomics, Inc.) protocols. Sequencing was performed on an Illumina HiSeq X platform. The Cell Ranger pipeline (version 6.1.1; RRID:SCR_017344) was used to generate the data matrix. Raw sequencing reads were mapped to the human reference genome (GRCh 38) using the STAR aligner (RRID:SCR_015899). Data visualization and analysis were performed using Loupe Browser (version 6.0.1; 10X Genomics, Inc.).

For Gene Ontology (GO) analysis, the top 100 differentially expressed genes (DEGs) ranked by P-value were analyzed using the Metascape platform (https://metascape.org; RRID: SCR_016620). Volcano plots and correlation analyses (Pearson's correlation coefficient) were performed using RStudio. Copy Number Variation (CNV) data were obtained from The Cancer Genome Atlas (TCGA) dataset through the cBioPortal platform (RRID:SCR_014555). Gene information and chromosomal locations were obtained from the National Center for Biotechnology Information (NCBI) Gene database (RRID:SCR_002472).

Statistical analyses were performed using GraphPad Prism (Dotmatics; RRID: SCR_002798). Data are presented as the mean ± standard deviation from at least three independent experiments. Comparisons between two groups were performed using two-tailed unpaired Student's t-test. For multiple comparisons, one-way analysis of variance followed by Tukey's post-hoc test was used. Correlation analyses were performed using Pearson's correlation coefficient and Spearman's rank correlation coefficient. P<0.05 was considered to indicate a statistically significant difference.

All RNA and single-cell RNA sequencing data generated in the present study have been deposited in the Gene Expression Omnibus. CNV data from TCGA were accessed through the cBioPortal platform (https://www.cbioportal.org/). Gene information was obtained from the NCBI database (https://www.ncbi.nlm.nih.gov/).

Both CAFs and NFs exhibited a spindle-shaped morphology, but NFs appeared slightly more flattened (Fig. 1A). Immunofluorescence staining revealed that CAFs were positive for α-SMA, while NFs were negative. EpCAM, an epithelial marker, was not detected in either cell type, confirming the successful establishment of CAFs and NFs culture systems (Fig. 1B).

Both CAFs and NFs exhibited a significantly enhanced proliferative capacity under hypoxic conditions compared with normoxic conditions (P<0.05; Fig. 1C).

HT29 cells cultured with the supernatant from CAFs cultured under both normoxic and hypoxic conditions showed no significant difference in proliferative capacity (Fig. 2A). However, HT29 cells treated with supernatants from hypoxia-conditioned CAFs exhibited a significantly enhanced migratory ability (P<0.05; Fig. 2B) and increased resistance to oxaliplatin (cat. no. S1224; Selleck Chemicals) and 5-Fluorouracil (5-FU; cat. no. 16220-14; Nacalai Tesque, Inc.) (P<0.05; Fig. 2C), a key chemotherapeutic drug for CRC.

Shotgun proteomic analysis identified 59 proteins in the supernatant of CAFs. Furthermore, a t-test comparing protein levels in the supernatant of CAFs cultured under hypoxic and normoxic conditions identified six proteins with P<0.01 and the average value fold change greater than two (Table I). Among these, DBP exhibited the most significant increase.

In a single-cell analysis of CAFs cultured under hypoxic and normoxic conditions, the cells were classified into 15 clusters (Fig. 3A). Clusters exhibiting upregulation of CAF markers were identified. Clusters 2, 5, 6 and 7, characterized by high expression of α-SMA, periostin and HIF-1α, were selected for further analysis (Fig. 3B and C). In addition, examination of our dataset for the expression of CAF marker genes reported by Elyada et al (22) revealed that several markers such as FAP and MYL9 were expressed at high levels in clusters 2, 5, 6 and 7 (Fig. S1). These clusters were then re-clustered, and DEGs between hypoxic and normoxic CAFs were identified within each cluster (Fig. 3D). The proportions of hypoxic and normoxic cells in each cluster can be confirmed in Table SI.

GO and Kyoto Encyclopedia of Genes and Genomes pathway analyses revealed that genes upregulated in hypoxic CAFs were markedly enriched in pathways related to ossification, skeletal development and the phosphoinositide 3-kinase-AKT signaling pathway (PI3K-AKT) signaling pathway (Fig. 3E and F).

Given the critical role of crosstalk between CAFs and tumor cells in the TME, it was aimed to elucidate common hypoxic response mechanisms across the entire TME. By identifying genes whose expression was altered not only in hypoxic CAFs but also in hypoxic CRC cells, the aim was to identify key factors involved in CAF-cancer cell interactions that contribute to tumor progression and drug resistance. Among the upregulated genes in hypoxic CAFs, 59 genes were identified that were also upregulated in hypoxic CRC cell lines (Table SII). From these 59 genes, those strongly associated with vitamin D, bone metabolism and the PI3K-AKT signaling pathway were selected, factors previously identified as crucial through hypoxic CAF secretome analysis and pathway analysis, as candidates for further investigation. Based on these findings, PTHrP was identified as a key factor in hypoxic CAFs and designated as the primary candidate for further analysis.

When PTHrP was added to NFs under normoxic conditions, both α-SMA and HIF-1α expression increased (Fig. 4A). In CAFs, PTHrP treatment induced morphological changes similar to those observed in hypoxic CAF, with cells transitioning from a spindle-shaped to a more flattened morphology and forming layered structures (Fig. 4B), along with enhanced proliferation capacity (Fig. 4C). DBP, identified through shotgun analysis of the conditioned medium from hypoxic CAFs, has been reported to reduce tissue vitamin D levels (23), which are known to exert anticancer effects (24). Given that vitamin D has been reported to downregulate PTHrP expression (25), vitamin D was added to the CAF culture medium to investigate possible attenuation of their CAF-like features. Vitamin D treatment led to morphological reversion of CAFs to an NF-like appearance (Fig. 4B), although no significant suppression of proliferation was observed (Fig. 4C).

Immunohistochemical analysis of DR patterns (16 mature, nine immature) revealed significantly higher expression of HIF-1α and periostin, one of the reported CAF markers, in the immature group, indicating that these tumors were exposed to a more hypoxic microenvironment (Fig. 5A; Table II). To perform a more detailed analysis, an expanded cohort of 59 cases (43 mature, 16 immature) was examined. In this cohort, the immature group showed significantly higher PTHrP expression in both cytoplasm of tumor cells at the invasive front and the surrounding stromal areas. Since RAS mutation has been reported as one of the upstream regulators of PTHrP (25), the results of genetic testing performed on tumor tissue samples were also analyzed. Among the 59 cases, 43 had undergone testing for RAS and B-Raf proto-oncogene mutations and microsatellite instability status, with KRAS mutations found in 20 cases and neuroblastoma RAS viral oncogene homolog mutation in one case (Table SIII) and MSI testing results showed 0% (0/13) MSI-H cases in the immature DR group and 10.0% (3/30) in the mature DR group, comparable to the reported 5–10% prevalence in general CRC populations. This suggests that the current cohort is not uniquely biased toward MSS cases.

RAS mutations were found to be significantly more frequent in the immature group (Fig. 5B; Table III). Subgroup analysis based on PTHrP expression levels in both the cancer cells at the invasive front and the surrounding stroma showed a positive correlation between PTHrP expression and RAS mutation status (Tables IV and SII).

To further investigate the association between PTHrP expression and RAS mutation observed in tumor tissue, the relationship between PTHrP and KRAS was analyzed using public datasets. Both PTHrP and KRAS are located on chromosome 12. Analysis of CNVs in patients with CRC using TCGA data revealed a significant positive correlation between KRAS and PTHrP copy number (R=0.92; P<0.001; Fig. 5C). Similar trends were observed across multiple cancer types (Fig. 5D).

Furthermore, copy number correlations were analyzed for other known driver genes located on chromosome 12, CDK4 and MDM2, and a very strong positive correlation was found between them (Spearman's r=0.91; P=5.17×10⁻¹⁹⁶; Fig. S2A). Strong copy number correlations were also observed between VDR, which is located on the same chromosome, and both PTHrP (Fig. S2B) and KRAS (Fig. S2C; PTHrP, Spearman's r=0.77; P=7.13×10⁻¹⁰⁶; KRAS, Spearman's r=0.78; P=5.90×10⁻¹¹¹). These results support the possibility that the co-amplification of PTHrP and KRAS is a passenger event associated with regional chromosomal gains on chromosome 12, and also suggest potential involvement of other functional relationships, such as vitamin D signaling through VDR.