The intratumoral microbiota, as an important component of the tumor microenvironment, is increasingly recognized as a key factor in regulating responses to cancer immunotherapy. Recent studies have revealed that the intratumoral microbiota is not uniformly distributed but instead exhibits significant spatial heterogeneity, with its distribution patterns influenced by factors such as tumor anatomy, local immune status, and therapeutic interventions. This spatial heterogeneity not only alters the interactions between microbes and the host immune system but may also reshape the immunogenic and immunosuppressive landscapes of tumors. The enrichment or depletion of microbiota in different tumor regions can influence immune cell infiltration patterns, metabolic pathway activities, and immune checkpoint molecule expression, thereby driving the development of resistance to immunotherapy. Moreover, certain bacterial metabolites form concentration gradients between the tumor core and margins, thereby regulating immune cell function. Therefore, understanding and manipulating the spatial distribution of intratumoral microbiota, particularly in resistant patients, holds promise for developing new strategies to overcome immunotherapy resistance. In the future, precise modulation strategies targeting microbial spatial heterogeneity, such as engineered bacterial vectors, probiotic combinations, and phage therapy, may open new avenues for immunotherapy.

Obstructions can occur during any stage of colorectal cancer and correspond with poor prognosis. Obstructive colorectal cancer (OCRC) is harder and exhibits increased tumor budding and proliferation of myofibroblasts compared with nonobstructive colorectal cancer, suggesting that the occurrence of obstruction may be related to extracellular matrix (ECM) remodeling. In this study, we found that colorectal cancer and OCRC samples differed substantially in ECM composition, specifically in collagen (newly formed and mature) and proteoglycans (including glycosaminoglycan, hyaluronic acid, and chondroitin sulfate). OCRC also exhibited considerable changes in ECM biomechanics and collagen arrangement. Interestingly, OCRC samples presented a notable increase in matrix cancer-associated fibroblasts (mCAF). The abundance of mCAFs correlated with the accumulation of palmitic acid (PA), and high concentrations of PA increased the secretion of ECM-related proteins by mCAFs. Additionally, PA did not directly affect normal fibroblasts but rather activated the NF-κB pathway in tumor cells to stimulate secretion of CSF1, TGFβ1, and CXCL8, which promoted the activation of normal fibroblasts into mCAFs and exacerbated ECM stiffening. Drug screening with a natural compound library identified vanillylacetone as a potential inhibitor of PA-induced cytokine secretion and ECM stiffening. These findings highlight intratumoral PA accumulation as a key mechanism driving ECM alterations and OCRC progression and suggest that targeting this axis may be useful for treating patients with colorectal cancer with risk of obstruction. Significance: Palmitic acid accumulation activates the NF-κB pathway in colorectal cancer cells to promote cytokine secretion that facilitates the generation of matrix cancer-associated fibroblasts, driving extracellular matrix remodeling and development of obstructions.

Figure S1: OCRC and CRC differed substantially in clinicopathology, with poorer prognoses in OCRC. (A) Comparative analysis of OS and DFS in CRC (n=1272) and OCRC (n=198) patients. (B) Clinicopathological characteristics of CRC (n=1272) and OCRC (n=198) patients. (C) Comparative analysis of OS and DFS in 150 matched pairs of CRC and OCRC patients. (D) Clinicopathological characteristics of 150 matched pairs of OCRC and CRC. Statistical significance was determined by Chi-Squared Test. *P <0.05, **P <0.01, ***P <0.001.

Figure S6: OCRC- and CRC-derived CAFs exhibited significant functional differences. (A) The function of ECM-related proteins, matrix metalloproteinases, LOX, and tissue inhibitor of matrix metalloproteinases secretion by CAFs from OCRC and CRC tissues (n = 5). (B) Diagram of co-culture and 3D culture experimental models. (C-E) Function analysis of proliferation (C), migration (D) and invasion (E) in CRC cells facilitated by CAFs conditioned medium derived from OCRC and CRC tissues (Scale bars = 50 and 100μm). (F) GSEA pathway and GO biological function enrichment analysis of the effect of mechanical force (F) stimulation on CAFs (n = 3). Statistical significance was determined by unpaired Student t test and one-way ANOVA. *P <0.05, **P <0.01, ***P <0.001, ****P <0.0001.

Figure S7: OCRC and CRC differed metabolically, with palmitic acid levels being elevated in OCRC. (A) Analysis of serum lipids indexes, glucose, arterial calcification and atherosclerosis in 150 matched pairs of OCRC and CRC patients. (B) KEGG and GO enrichment analysis based on digital RNA sequencing of five matched pairs of stage III OCRC and CRC tissues. (C) Non-targeted metabolomics profiling and differential metabolites analysis in five pairs of stage III OCRC and CRC tissue samples (P < 0.05, fold-change [FC] >1.2, variable importance projection (VIP) ≥ 1). (D) Comparative analysis of lipid indexes in five pairs of stage III OCRC and CRC patients. (E) Correlation analysis of PA levels and serum lipids indexes in five pairs of stage III OCRC and CRC tissue samples. Unpaired Student t test, Chi-Squared test, and Spearman correlation was used for statistical methods. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure S16: Vanillylacetone inhibited PA-induced CAF activation, ECM stiffness, and OCRC progression. (A) Tumor growth and mass measurements in C57BL/6 and BALB/c mice subcutaneous tumors following vanillylacetone treatment. (B-C) Analysis of lipid deposition, collagen fiber content, CAFs abundance, and ECM biomechanics properties in BALB/c tumors after vanillylacetone treatment (Scale bars = 1mm and 50μm; AFM 10μm). (D) Quantification of PA content in subcutaneous tumor from C57BL/6 and BALB/c mice after vanillylacetone treatment. Statistical significance was determined by unpaired Student t test and two-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns, no significance.