Graphical abstract graphic file with name fx1.jpg [59]Open in a new tab Highlights * • ilCAFs, characterized by the expression of IRF1 and CCL4, are responsive to RT * • IFN-γ signaling drives the polarization of ilCAFs to secrete CCL4 and CCL5 * • Activation of the IFN-γ/STING signaling pathway in ilCAFs enhances RT efficacy * • STING agonists synergize with RT, enhancing its efficacy and overcoming radioresistance __________________________________________________________________ Huang et al. identify a distinct IRF1^+ interferon-licensed CAF subset (ilCAF) in rectal cancer that reshapes the immune microenvironment and enhances response to neoadjuvant radiotherapy. Activation of the IFN-γ/STING pathway in ilCAFs boosts anti-tumor immunity, supporting combination of STING agonists with radiotherapy. Introduction Colorectal cancer (CRC) ranks as the third most lethal malignancy worldwide, with more than 30% of patients with CRC diagnosed at an advanced stage.[60]^1 Neoadjuvant radiotherapy (RT), either alone or combined with chemotherapy, plays a pivotal role in both curative and palliative settings.[61]^2^,[62]^3 It is increasingly recognized as a standard and effective approach for treating locally advanced cases.[63]^2^,[64]^3 However, only 15%–30% of patients with rectal cancer (RC) achieve a pathological complete response (pCR), while others may experience minimal or no response, transient remission, or even rapid disease progression due to radioresistance.[65]^4 Understanding the mechanisms contributing to RT failure in RC is thus of paramount clinical importance. Emerging evidence highlights the complex nature of RC, which is intricately linked to its unique tumor microenvironment (TME).[66]^5^,[67]^6 The TME in RC is marked by fibroinflammatory responses, characterized by a dynamic interplay between malignant cells and nonmalignant stromal components, including various immune cells, endothelial cells, cancer-associated fibroblasts (CAFs), and the extracellular matrix (ECM).[68]^7^,[69]^8 This complex network significantly contributes to RC’s resistance to therapy.[70]^7^,[71]^8 Moreover, analysis of receptor-ligand interactions inferred from the gene expression profiles of patients with RC suggests that CAFs serve as critical nodes in the fibroinflammatory communication network between cellular compartments and the ECM within the TME.[72]^9^,[73]^10 In addition, CAFs exhibit significant phenotypic and functional heterogeneity, contributing to divergent treatment outcomes.[74]^8^,[75]^11^,[76]^12 Recent advancements in single-cell RNA sequencing (scRNA-seq) across diverse tumor types have revealed distinct CAF subclusters, including myofibroblastic (myCAFs) and inflammatory CAFs (iCAFs).[77]^13^,[78]^14^,[79]^15 The polarization of these subclusters is influenced by their proximity to cancer cells and their responsiveness to transforming growth factor β activation or interleukin (IL)-1 signaling, resulting in either distinct matrix-producing contractile or immunomodulating secretome phenotypes.[80]^8^,[81]^16^,[82]^17 Notably, iCAFs, characterized by IL-1α secretion, play a critical role in influencing the response to chemoradiotherapy in CRC through p53-mediated senescence.[83]^8^,[84]^18^,[85]^19 Additionally, specific CAF subgroups have been identified in other cancers, such as interferon (IFN)-dependent SLC14A1^+ CAFs in bladder cancer and CD105^+ CAFs in pancreatic cancer, underscoring their regulatory roles within neighboring cell populations.[86]^20^,[87]^21 However, the plasticity and heterogeneity of CAFs, along with their interactions with the immune landscape in RC following RT, remain poorly understood. In this study, we leverage unbiased scRNA-seq to explore the complexities of CAFs in RC following RT. We identify a previously unrecognized CAF subpopulation distinguished by the expression of IFN regulatory factor 1 (IRF1) in patients with RC. Through a multidisciplinary approach, we delineate the clinical significance of these immunoregulatory CAFs and elucidate their fibroinflammatory crosstalk with endogenous immune cells. Our study aims to unravel the signals that initiate and sustain this distinct CAF subpopulation within the TME. We further explore strategies for their repolarization, aiming to enhance anti-tumor responses and improve long-term survival in both RT-sensitive and RT-resistant CRC models. Results scRNA-seq analysis identifies IFN-γ-dependent CAFs enriched in RC responsive to RT To comprehensively unravel the cellular intricacies of CAFs and intercellular interactions in patients with RC post-RT, we procured rectal biopsies or surgical specimens from 14 patients with RC (7 post-RT and 7 pre-RT) at our institution. Therapeutic response was assessed using radiological evaluations and pathological regression grades ([88]Figure S1A). Following rigorous quality control, we performed scRNA-seq on the collected samples, generating high-resolution transcriptomic profiles of 64,436 individual cells ([89]Figure 1A). Using unsupervised clustering and annotation of integrated gene expression data, we identified multiple clusters corresponding to eleven distinct cellular compartments. These included epithelial cells (EPCAM and KRT8), CD4 T cells (CD3D and CD4), CD8 T cells (CD3D, CD8A, and CD8B), natural killer (NK) cells (GNLY), B cells (MS4A1, CD79A, and CD19), myeloid cells (AIF1 and CD14), mast cells (MS4A2 and KIT), endothelial cells (PECAM1 and CDH5), plasma B cells (JCHAIN and CD79A), and CAFs/stromal cells, which were identified based on mesenchymal genes (COL1A, COL3A1, and DCN) ([90]Figures 1B, [91]S1B, and S1C). Subsequently, we conducted a more in-depth characterization of the stromal cell compartment to delineate its heterogeneity post-RT, leading to the identification of five distinct CAF clusters ([92]Figures 1C and 1D). Figure 1. [93]Figure 1 [94]Open in a new tab Single-cell RNA sequencing analysis identifies IFN-γ-dependent CAFs enriched in rectal cancer responsive to radiotherapy (A) Schematic representation of the workflow for scRNA-seq and validation experiments conducted on rectal tumors pre- and post-RT (n = 7 for each group). (B) Uniform manifold approximation and projection (UMAP) plot of all cells, representing eleven cell types. Cell clusters are colored by cell identity. (C) Reclustering of CAFs in the dataset visualized using UMAP, demonstrating five distinct CAF clusters: iCAFs (dark blue), myCAFs (orange), ilCAFs (light blue), SOX6^+ CAFs (red), and CXCL1^+ CAFs (purple). (D) Heatmap displaying differentially expressed genes across all CAF clusters. (E) GSEA depicting the top upregulated pathways and core enrichment genes in five distinct CAF clusters, with all pathways filtered by false discovery rate < 0.05. (F) Density plot of different CAF clusters pre- or post-RT. (G) Slingshot and tradeSeq trajectory analysis of RC CAF scRNA-seq data indicating predicted lineage trajectory. The trajectory path from iCAFs-ilCAFs is overlaid on the cluster-based UMAP and colored by pseudotime of this respective lineage. Trajectory analysis overlaid IRF1 expression vs. pseudotime scatterplot of iCAFs and ilCAFs along the lineage. (H) Quantitative PCR mRNA expression analysis of representative genes of ilCAFs (IRF1, CCL4, STAT1, and STING1), iCAFs (C3 and CFD), myCAFs (RGS5 and MCAM), SOX6^+ CAFs (CXCL14 and PDGFRA), and CXCL1^+ CAFs (CXCL1 and CCL11) in primary CAFs treated with RT, compared to untreated controls (n = 4 for each group). (I) Representative flow cytometric plots (top) and quantification (down) of IRF1 expression in CAFs pre- and post-RT (n = 7 for each group). (J) Representative multiplex immunofluorescence image depicting the localization of ilCAFs (COL3A, PDPN, and IRF1) and tumor cells (pan-cytokeratin) in rectal tumors pre- and post-RT (n = 5 for RT group, and n = 7 for untreated group). Scale bars, 50 μm. (K) Quantification of ilCAFs is shown in the adjacent bar graphs. (L) Kaplan-Meier survival curves of CRC patients with low (blue) and high (red) expression of ilCAFs in total CRC samples (n = 165). Student’s t tests were performed for (F), (H), (I), (J), and (K). For (L) (survival curves), the log rank test was performed. Data are presented as mean ± SEM and are representative of at least three independent experiments. A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. The myCAF cluster was enriched in mural cell- and myCAF-related genes (RGS5, MCAM, NOTCH3, ACTA2, TINAGL1, and MYH11) ([95]Figure 1D)[96]^20 and showed significant enrichment in pathways related to focal adhesion, myogenesis, Notch signaling, and vascular smooth muscle contraction ([97]Figure 1E). Consequently, we designated this stromal cell cluster as the “classic myCAFs.” The SOX6^+ CAF cluster was enriched in genes associated with the mesenchymal cell state (SOX6, WNT4, and TCF4) and matrix production (POSTN and PDGFRA) ([98]Figure 1D).[99]^11 Gene set enrichment analysis (GSEA) revealed that this stromal cluster showed enrichment in cytokine activity, mesenchymal stem cell differentiation, regulation of fibroblast growth factor receptor signaling, and Hedgehog signaling ([100]Figure 1E). Collectively, we characterized this CAF cluster as “stem-like proliferative CAFs.” The remaining three stromal cell clusters were highly enriched in genes associated with multiple inflammatory pathways, including inflammatory response, type II IFN, and IL-1 signaling ([101]Figure 1E).[102]^11 Specifically, iCAF cluster prominently expressed CFD, C3, DCN, and CLU ([103]Figure 1D), aligning with the previously reported iCAF signature in pancreatic adenocarcinomas.[104]^22 GSEA further unveiled that this CAF cluster showed enrichment in inflammatory response, signaling receptor regulator activity, and a complement signature ([105]Figure 1E), supporting its designation as “classic iCAFs.” The CXCL1^+ CAF cluster exhibited a similar enrichment of the inflammatory response signature to that of classic iCAFs. However, it uniquely expressed IL-1-responsive genes, such as CXCL1, CCL11, and CCL13, along with high expression of ECM-related genes, such as FGFR2, distinguishing it from classic iCAFs ([106]Figure 1D). Furthermore, this cluster displayed enrichment in pathways involving cytokine activity, cytokine and cytokine receptor interaction, and response to chemokine and chemokine receptor binding ([107]Figure 1E). Consequently, we designated this cluster as “IL-1 signaling CAFs (CXCL1^+ CAFs).”[108]^23 The remaining CAF cluster showed enrichment in IFN-γ-related genes, particularly those involved in the stimulator of IFN genes (STING) pathway, including IRF1, CCL4, and CCL5 ([109]Figure 1D). Additionally, this CAF cluster was strongly enriched in the IL-6-JAK-STAT3, inflammatory response, IFN gamma response, and tumor necrosis factor alpha (TNF-α) signaling via nuclear factor κB (NF-κB) ([110]Figure 1E). Thus, we termed this stromal cell cluster “IFN-licensed CAFs (ilCAFs).” To investigate whether RT induces an inflammatory response in stromal cells, we assessed the changes in CAFs post-RT compared to pre-RT. Analysis of distinct CAF cluster abundances revealed a decrease in myCAFs, SOX6^+ CAFs, and CXCL1^+ CAFs, while iCAFs and ilCAFs increased in tumors following RT ([111]Figure 1F), providing further evidence of RT-mediated reprogramming of CAFs within the TME. The immunoregulatory response of fibroblasts to RT is a programmed feature of both adaptive and innate immunity within the TME. Thus, it is plausible that ilCAFs in RC arise from activation of a programmed immunoregulatory mechanism triggered by RT. Interestingly, cellular trajectory reconstruction analysis using gene counts and expression (CytoTRACE), which predicts developmental potential, revealed a gradual decline in scores from classic iCAFs to ilCAFs, with classic iCAFs showing the highest and ilCAFs the lowest scores ([112]Figure S1D).[113]^11 These patterns suggest a cellular trajectory from iCAFs toward ilCAFs ([114]Figure S1D). In light of this, we performed trajectory analysis on our scRNA-seq data. This analysis predicted that the trajectory originated in the iCAFs cluster and terminated in the ilCAFs cluster ([115]Figure 1G). We then employed trajectory-based differential expression analysis for single-cell sequencing data (tradeSeq) to identify differential gene expression changes along this trajectory’s pseudotime ([116]Figure 1G). Notably, the ilCAF trajectory was associated with the upregulation of IRF1 ([117]Figure 1G). These results indicate that RT activates IFN-related programs in iCAFs and suggest that IRF1 is a key driver in ilCAF development. To validate our scRNA-seq data, human CAFs were isolated from primary rectal tumors and subjected to RT for 48 h ex vivo. Consistently, upregulation of ilCAF-related genes (IRF1, CCL4, STAT1, and STING1) and iCAF-related genes (C3 and CFD), along with downregulation of myCAFs-related genes (RGS5 and MCAM), SOX6^+ CAF-related genes (CXCL14 and PDGFRA), and CXCL1^+CAF-related genes (CXCL1 and CCL11), were observed in human CAFs post-RT compared to naive CAFs. These changes were confirmed by qPCR and flow cytometry ([118]Figures 1H, 1I, and [119]S1E). Consequently, IRF1 and CCL4 were designated as distinctive biomarkers for ilCAFs. Multi-color immunofluorescence (IF) staining targeting pan-cytokeratin (PanCK, tumor), COL3A (pan-stroma), podoplanin (PDPN, pan-stroma), and IRF1/CCL4 (ilCAFs) was performed on untreated and RT-treated RC tumors to validate the increased presence of ilCAFs in our patient cohort ([120]Figures 1J, 1K, and [121]S1F). Cells co-expressing PDPN, COL3A, and IRF1 but lacking CK19 were identified as ilCAFs. In accordance with our scRNA-seq data, IRF1^+ and CCL4^+ CAFs were more abundant in tumors after RT than in untreated tumors ([122]Figures 1J, 1K, and [123]S1F). Collectively, these data reveal an enrichment of the ilCAF subpopulation in RC tumors post-RT. To further elucidate the clinical relevance of IRF1^+ ilCAFs, we conducted an analysis of The Cancer Genome Atlas Program (TCGA) datasets, probing correlations between IRF1^+ ilCAFs and the prognosis and treatment responses in patients with CRC. Notably, elevated IRF1 expression was associated with improved overall survival in patients with CRC ([124]Figure S1G). Leveraging the gene signature of ilCAFs derived from our scRNA-seq data, we found a correlation between ilCAF expression and the prognosis of patients with CRC ([125]Figure 1L). Collectively, these results indicate that ilCAF enrichment is linked to favorable clinical outcomes in CRC, including enhanced survival and improved response to chemoradiotherapy. To verify whether ilCAFs are present in the TME of other human tumors following RT, we analyzed two other published cohorts: one of oropharyngeal cancer (GEO: [126]GSE157517) and another of breast cancer (GEO: [127]GSE59733) ([128]Figure S1H). Consistently, we observed an upregulation of ilCAF signature scores in non-recurrent and post-RT tumors ([129]Figure S1H). Moreover, we extended our analysis to published scRNA-seq data (GEO: [130]GSE207718) from the Kras^G12D/+;Trp53^R172H/+;Pdx-1-Cre (KPC) mouse pancreatic cancer model ([131]Figure S1I). Tissues were collected both before treatment and 14 days after stereotactic body radiotherapy. Consistent with our findings in patients with RC, we identified five subgroups among KPC CAFs ([132]Figure S1I). Uniform manifold approximation and projection (UMAP) analysis revealed five CAF populations: apCAFs, myCAFs, iCAFs, ilCAFs, and MKI67^+ CAFs. Subsequent enrichment analysis showed that the ilCAF cluster was strongly enriched for IFN signature sets, including type II IFN, cytokine-mediated signaling pathways, and positive regulation of T cell activation ([133]Figure S1J). Furthermore, we observed that type II IFN- and cytokine-mediated signaling pathways were upregulated in ilCAFs relative to other CAF subtypes post-RT ([134]Figure S1K). Together, these findings suggest that ilCAFs are a conserved stromal subpopulation broadly induced across multiple solid tumor types in response to RT. RT induces enrichment of ilCAFs to promote anti-tumor responses To evaluate the functional implications of ilCAFs in RC tumors following RT, we conducted Gene Ontology (GO) pathway enrichment analysis on the transcriptomic profiles of ilCAFs in RC tumors pre-RT versus post-RT. Notably, post-RT tumors exhibited a marked enrichment of ilCAFs in signaling pathways intricately linked to immune responses, encompassing the regulation of cytokine production, innate immune response, as well as T cell differentiation and activation ([135]Figure 2A). This observation underscores the pivotal role of ilCAF enrichment in RC tumors post-RT, facilitating the priming of both innate and adaptive immune responses. Subsequently, we aimed to elucidate how IFN-responsive fibroblasts modulate anti-tumor immunity in RC tumors. To this end, we performed scRNA-seq analysis of CD45^+ immune cells from our dataset. UMAP analysis revealed various immune cell subsets, including CD4^+ T cells, CD8^+ T cells, myeloid cells, NK cells, B cells, and plasma B cells ([136]Figures S2A and S2B). Further analysis of the abundance of distinct CD45^+ clusters demonstrated an increase in B cells, NK cells, and CD8^+ T cells, while a decrease in CD4^+ T cells, plasma B cells, and myeloid cells was observed in tumors post-RT compared to pre-RT ([137]Figure S2C). These findings provide additional evidence of the RT-mediated reprogramming of anti-tumor immunity within the TME. Given the active reprogramming of RC TME immunity by RT, we sought to investigate whether ilCAFs might contribute to T cell-mediated anti-tumor immunity. Consistently, the expression of IRF1 in ilCAFs was positively correlated with CD8^+ T cell abundance ([138]Figure 2B), suggesting that ilCAFs modulate T cell responses following RT and may serve as a potential therapeutic target for enhancing anti-tumor immunity. Figure 2. [139]Figure 2 [140]Open in a new tab RT induces enrichment of ilCAFs to enhance anti-tumor responses (A) Bar graph illustrating the upregulated pathways in ilCAFs post-RT compared to pre-RT based on GO enrichment analysis (n = 7 per group), with all pathways filtered by adjusted p value (P[adj]) < 0.05. (B) Correlation analysis between IRF1 expression in ilCAFs (x axis) and the proportion of CD8^+ T cell subsets (red, y axis) in scRNA-seq data. Spearman correlation analysis was performed to determine the correlation coefficient and two-sided p value. (C) Reclustering of CD8 T cells in the dataset visualized by UMAPs, demonstrating six distinct clusters: naive T (light green), effector T (blue), T[EMRA] (light purple), T[RM] (dark purple), T[EX] (red) and MAIT (orange). (D) UMAP nucleus densities of different CD8 T cell clusters pre- or post-RT. (E) Effector memory and exhaustion scores of effector T cells and T[EMRA] cells pre- and post-RT. (F) ELISA for CCL4 and CCL5 content in the supernatant of primary CAFs pre-RT and post-RT (n = 3 per group). (G) Quantification of CD8^+ T cells and GZMB^+ CD8^+ T cells after CAFs were exposed to CCR5i combined with RT (n = 5 per group). (H) Differential interaction strength between post-RT and pre-RT among ilCAFs and antigen-presenting cell subsets in RC scRNA-seq data. (I) GO enrichment analysis showing top upregulated pathways in cDC1s compared to other DCs pre-RT and post-RT in rectal cancers, with all pathways filtered by P[adj] < 0.05. (J and K) Representative multiplex immunohistochemistry staining images (left) and quantification (right) for COL3A1, IRF1 (J), or CCL4 (K) in tumors from day 12 RT-treated and untreated MC38-bearing mice (n = 5 per group). Scale bars, 20 μm. (L–O) Evaluation of the impact of RT on the growth of established MC38 colorectal tumors co-inoculated with Irf1^−/− or WT fibroblasts in syngeneic mice. (L) Experimental design for the treatment of MC38 colorectal tumor-bearing C57BL/6 mice. (M) Average growth curve of colorectal cancer treated with vehicle and RT (n = 5 per group). (N) Quantification of CD4^+ T cells, CD8^+ T cells, and cDCs among CD45^+ cells in tumors at day 10 post-RT treatment (n = 5 per group). (O) Modified Kaplan-Meier curves for each treatment cohort in the mouse model (n = 5 mice per group). (P) Evaluation of the role of the IFN-γ pathway in RT-mediated tumor control in established MC38 tumor cells co-inoculated with fibroblasts in syngeneic mice. Average tumor growth curves of colorectal tumors treated with vehicle or anti-IFN-γ, with or without RT (n = 5 per group). (Q) Evaluation of the impact of RT-induced CCL4-ilCAFs on tumor growth in established MC38 tumor cells co-inoculated with si-Ccl4 or WT fibroblasts in syngeneic mice. Average tumor growth curve of colorectal tumors treated with vehicle and RT (n = 5 per group). (R) Evaluation of the impact of CCR5i combined with RT on tumor growth in established MC38 tumor cells co-inoculated with fibroblasts in syngeneic mice. Average tumor growth curve of colorectal cancer treated with vehicle and CCR5i ± RT (n = 5 per group). One-way or two-way ANOVA was performed for (G), (M), (N), (P), (Q), and (R). For (O), the log rank test was performed. Data are presented as mean ± SEM and are representative of at least three independent experiments. A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. To further investigate the correlation between ilCAFs and T cell responses, we analyzed the lymphocyte population dynamics in the scRNA-seq dataset of RC tumors ([141]Figures 2C–2E). Employing UMAP and clustering approaches, we identified six CD8^+ T cell subsets, including anti-tumor effector CD8^+ T cells, terminally differentiated effector memory (T[EMRA]) or effector T cells, naive T cells, tissue-resident memory T cells (T[RM]), exhausted T (T[EX]) cells, and mucosal-associated invariant T cells (MAITs) ([142]Figures 2C and [143]S2D). Within the CD8^+ T cell compartment, notable changes were observed in cytotoxic CD8^+ T cells (comprising effector CD8^+ T cells and T[EMRA]) post-RT ([144]Figure 2D). Notably, effector CD8^+ T cells and T[EMRA] exhibited enhanced effector memory scores and reduced exhaustion scores post-RT compared with pre-RT conditions ([145]Figure 2E). At the gene level, CD8^+ T cells exhibited elevated expression of inflammatory mediators (CCR5 and CXCR3) and cytotoxic effectors (IFN gamma [IFN-γ] and granzyme H) ([146]Figure S2D). Furthermore, in post-RT tumors, ilCAFs exhibited enrichment in multiple signaling pathways associated with both adaptive and innate immune responses, which were strongly correlated with the upregulation of IRF1, CCL4, and CCL5 ([147]Figure 1E). This observation suggests that ilCAFs may contribute to fostering anti-tumor immunity by secreting CCL4 and CCL5. Consequently, we hypothesized that ilCAFs could directly modulate T cells via the CCL4/CCL5-CCR5 axis. To validate this hypothesis, we conducted in vitro transwell assays with primary CAFs, either untreated or treated with RT, which were seeded in the bottom chamber, and T cells isolated from mouse spleens were added to the upper chamber. Consistent with our hypothesis, ELISA measurements revealed significantly elevated levels of CCL4 and CCL5 in the supernatant of the bottom chamber in the post-RT condition, leading to enhanced transmigration of CD8^+ T cells, particularly Granzyme B^+ (GZMB^+) T cells, into the bottom chamber ([148]Figures 2F and 2G). Subsequently, to disrupt the interaction between ilCAFs and T cells through the CCL4/CCL5-CCR5 axis post-RT, we employed a CCR5 inhibitor (CCR5i) ([149]Figure 2G). Remarkably, CCR5i treatment of RT-exposed fibroblasts reduced T cell migration and effector function in vitro ([150]Figure 2G). These findings collectively indicate that ilCAFs can directly induce the infiltration and effector phenotype differentiation of CD8^+ T cells. Additionally, the anti-tumor T cell response in CRC mediated by RT has been attributed, in part, to conventional dendritic cells (cDCs).[151]^24 Consequently, we aimed to elucidate whether cDC phenotypes are modulated by ilCAFs. To address this question, we analyzed our RC scRNA-seq dataset. We re-clustered myeloid cells and identified distinct clusters representing macrophages, monocytes, cDC1s, cDC2s, mature dendritic cells (DCs) (LAMP3^+ DCs), and plasmacytoid DCs ([152]Figures S2E and S2F). Further analysis using CellChat revealed increased interaction strength between ilCAFs and cDC1 subsets post-RT within our dataset ([153]Figure 2H). Pathway enrichment analysis indicated that cDC1 subsets in post-RT tumors were enriched for gene sets associated with T cell proliferation and migration, response to type II IFN, antigen processing and presentation of exogenous antigens, as well as dendritic cell antigen processing and presentation ([154]Figure 2I). To further investigate alterations within the cDC compartment, we assessed in vitro DC migration and found that RT enhanced DC migration ([155]Figure S2G), suggesting that ilCAFs may contribute to immune modulation within tumors post-RT. To mechanistically validate this interaction, we disrupted the crosstalk between ilCAFs and DCs by pharmacologically inhibiting CCR5 or genetically silencing CCL4 expression in ilCAFs. These interventions demonstrated that cDC1s migration toward ilCAFs post-RT was mediated by the CCL4/CCL5-CCR5 signaling axis ([156]Figure S2H). Moreover, integrative analysis of CRC data from TCGA and four independent datasets (two CRC, one oropharyngeal cancer, and one breast cancer dataset) revealed a strong positive correlation between the ilCAF gene signature and infiltration of CD8^+ T cells, CD4^+ T cells, and cDC1 subsets across various human solid tumors and murine models ([157]Figures S2I and S2J). These findings underscore the critical role of ilCAFs in orchestrating anti-tumor immune responses following RT. To further delineate the epithelial landscape of the tumor, we re-clustered all malignant and normal epithelial cells and identified seven major subclusters ([158]Figure S3A). All malignant epithelial subclusters exhibited prominent copy-number variation (CNV) signals, in contrast to non-epithelial populations such as endothelial and mast cells, which showed minimal CNV alterations ([159]Figures S3B and S3C). Notably, subcluster 0 was characterized by elevated expression of mesenchymal markers such as ERBB3 and CD44, which are suggestive of enhanced tumor cell proliferation, differentiation, and adhesion potential. Similarly, subcluster 5 exhibited high expression of MKI67 and ASCL2, genes associated with tumor aggressiveness and proliferative capacity ([160]Figure S3D). Moreover, both subclusters 0 and 5 displayed higher stemness signature scores compared to other epithelial subpopulations ([161]Figure S3E). Following RT, we observed a marked reduction in the proportion of cells in subcluster 0 and 5, indicating their radiosensitivity ([162]Figure S3F). To further elucidate intercellular communication, we conducted CellChat analysis, which revealed extensive intercellular interactions between the stem-like epithelial subclusters (0 and 5) and endothelial cells ([163]Figures S3G and S3H). These findings suggest a potential role for epithelial-endothelial crosstalk in modulating tumor response to RT and may provide mechanistic insights into radiation resistance. Next, to elucidate the mechanisms underlying the crosstalk between ilCAFs and immune cells in RC tumors after RT, we established a preclinical CRC model designed for in vivo monitoring of radiation effects. In this model, mouse MC38 CRC cells were subcutaneously co-inoculated with mouse fibroblasts at a 1:1 ratio, mimicking the desmoplastic TME characteristic of CRC. Upon achieving a tumor volume of 200–300 mm^3, syngeneic tumor-bearing mice were subjected to a single 20 Gy dose of irradiation, mirroring clinically relevant irradiation regimens ([164]Figure S4A). Tumors were harvested 10 days after RT for multiparametric flow cytometric analysis. Using previously established CAF identification criteria, myCAFs (Ly6C^−MHC^−), iCAFs (Ly6C^+MHC^−), and apCAFs (MHC^+Ly6C^−) were identified within platelet-derived growth factor receptor α (PDGFRα)^+PDPN^+ cell populations lacking expression of CD45, epithelial cell adhesion molecule (EpCAM), and CD31 ([165]Figures S4B and S4C).[166]^13 Although apCAFs showed negligible change, a marked decrease in myCAFs and an increase in iCAFs were observed in post-RT tumors ([167]Figure S4C). Multiplexed IF staining further demonstrated that RT significantly upregulated Ly6C expression in PDPN^+ stromal cells compared to untreated tumors ([168]Figure S4D). To delve deeper into the changes in CAF subsets in the mouse model, bulk RNA sequencing (RNA-seq) was performed on untreated and post-RT CRC tumors. Signature scores for mouse ilCAFs and iCAFs were significantly elevated in post-RT tumors compared to untreated ones ([169]Figure S4E). Consistent with our human scRNA-seq data, post-RT tumors exhibited enrichment in signaling pathways related to both adaptive and innate immune responses. Moreover, RT-treated mouse tumors demonstrated enhanced cell killing, TNF-α mediated by NF-κB, and immune responses against tumor cells ([170]Figure S4F). To validate these findings at the protein level, multiplex IF staining of mouse tumor tissues confirmed increased expression of IRF1 and CCL4 colocalized in COL3A^+ cells after RT ([171]Figures 2J and 2K). Collectively, our data indicate that this mouse model phenotypically recapitulates the enrichment of ilCAFs observed in human RC tumors post-RT. Subsequently, we aimed to elucidate the mechanisms governing the crosstalk between ilCAFs and immune cells in CRC tumors post-RT. Immune profiling using flow cytometry revealed a significant increase in CD8^+ T cells, CD4^+ T cells, and CD103^+ conventional dendritic cells (cDC1s) in both tumors and tumor-draining lymph nodes after RT (20 Gy × 1) ([172]Figures S5A–S5E). This indicates that, beyond enriching ilCAFs, RT also primes local and systemic anti-tumor responses by recruiting T cells and cDC1s. To further investigate the role of ilCAFs in modulating immune interactions and contributing to RT efficacy, we treated MC38 tumors co-inoculated with either wild-type (WT) or Irf1 knockout (Irf1^−/−) fibroblasts with RT ([173]Figures 2L–2N). Compared to controls, Irf1 deficiency significantly promoted tumor progression and diminished the anti-tumor effects of RT ([174]Figures 2M–2O and [175]S6A). Flow cytometric analysis revealed that Irf1 deficiency reduced the infiltration and accumulation of CD4^+ and CD8^+ T cells, as well as cDC1s, within tumors, accompanied by a decrease in GZMB^+ CD8^+ T cells ([176]Figures 2N and [177]S6B). Consistently, there was a trend toward increased CD8^+ T cells, CD4^+ T cells, and CD103^+ cDC1s in tumor-draining lymph nodes following RT ([178]Figure S6C). In parallel, multiplex IF staining confirmed a significant reduction in ilCAF abundance at the protein level in post-RT tumors derived from Irf1^−/− mice ([179]Figure S6D), underscoring a pivotal role for IRF1 in maintaining the ilCAF phenotype in response to RT-induced stress. To further investigate the role of IFN-γ signaling in ilCAF activation, we administered neutralizing anti-IFN-γ antibodies to inhibit this pathway in vivo. Strikingly, blockade of IFN-γ signaling significantly compromised the therapeutic efficacy of RT, as evidenced by diminished tumor control ([180]Figures 2P and [181]S6E), thereby underscoring the critical role of IFN-γ-dependent activation of ilCAFs in orchestrating effective anti-tumor immune responses following RT. In line with these findings, multiplex IF staining showed a marked reduction in protein-level ilCAF abundance in post-RT tumors treated with anti-IFN-γ antibodies ([182]Figure S6F), further confirming the regulatory role of IFN-γ signaling in sustaining the ilCAF phenotype. To further elucidate the mechanisms by which ilCAFs modulate T cell and DC activity, we employed an in vivo model in which MC38 colorectal tumors were co-inoculated with si-Ccl4 fibroblasts and subjected to RT, with CCR5i administration every other day ([183]Figures 2Q and 2R). Notably, Ccl4 knockdown significantly impaired the therapeutic efficacy of RT ([184]Figure 2Q). Similarly, pharmacological inhibition of CCR5 attenuated the anti-tumor effects of RT ([185]Figures 2R, [186]S7A, and S7B), and this was associated with reduced infiltration of CD8^+ T cells, CD4^+ T cells, and CD103^+ cDC1s in tumor tissues, as well as a trend toward decreased immune cell infiltration in tumor-draining lymph nodes ([187]Figures S7C and S7D). To further validate these findings, we treated Irf1^−/− fibroblasts with a CCR5i in vitro. As expected, Irf1 deficiency led to reduced secretion of CCL4 and CCL5 in the fibroblast-conditioned medium ([188]Figure S7E). In line with this, the inhibition of CCR5 diminished the migration of both T cells and cDC1s in vitro ([189]Figures S7F and S7G). Collectively, these results demonstrate that RT-induced ilCAF enrichment promotes the recruitment of anti-tumor effector cells, specifically T cells and cDC1s, via the CCL4/CCL5-CCR5 axis, thereby enhancing immune infiltration and RT efficacy in CRC. IRF1^+ ilCAFs promote RT-induced senescence through the IFN-γ/STAT1 pathway Next, we endeavored to elucidate the molecular mechanisms underlying the polarization of ilCAFs in CRC tumors following RT. Given the strong association between ilCAFs and IFN-γ/TNF-α-responsive pathways revealed by our scRNA-seq analysis, we hypothesized that type II IFN signaling is elevated in CRC tumors post-RT, thereby promoting the polarization and function of ilCAFs. To validate this, we conducted gain- and loss-of function assays using human and mouse primary CAFs in vitro. Flow cytometry showed that RT significantly increased the expression of IRF1 and CCL4 in CAFs, which was further enhanced by IFN-γ stimulation, a canonical type II IFN ([190]Figures 3A and [191]S8A). Inhibition of signal transducer and activator of transcription 1 (STAT1), a key downstream transcription factor in type II IFN signaling,[192]^25 using a STAT1 inhibitor (fludarabine), led to a significant decrease in IRF1 and CCL4 expression in CAFs following RT ([193]Figures 3A and [194]S8A). Thus, our flow data suggest a link between activation of IFN-γ/STAT1 pathway and ilCAF polarization following RT. Immunocytochemistry and western blot (WB) analyses further confirmed that IFN-γ stimulation increased the expression of CCL4 and IRF1 in RT-treated CAFs ([195]Figures 3B and 3C). Importantly, IFN-γ also upregulated the expression of cyclic guanosine monophosphate-AMP synthase (cGAS) and the phosphorylation of STING (p-STING) and phosphorylated STAT1 (p-STAT1) ([196]Figure 3C). Notably, the inhibition of STAT1 also suppressed STING expression, thereby abrogating RT-induced activation of IFN/STING pathway genes in CAFs ([197]Figures 3B and 3C). Given that RT induces DNA damage and activates the cGAS-STING pathways, we further observed that IFN-γ promoted γH2AX expression (a marker of DNA double-stranded breaks) in CAFs, suggesting that RT may promote ilCAF polarization via this cGAS-STING (DNA-sensing) pathway ([198]Figure S8B). Collectively, our findings reveal that the cGAS-STING pathway, potentially engaged downstream of type II IFN signaling, contributes to the RT-induced polarization of ilCAFs. Figure 3. [199]Figure 3 [200]Open in a new tab ilCAFs mediate radiation-induced senescence via the IFN-γ/STAT1 pathway (A) Representative histogram plots (left) and quantification (right) showing mean fluorescence intensity (MFI) of IRF1 or CCL4 in CAFs isolated from treatment-naive patients with rectal cancer, analyzed by flow cytometry. Comparisons include vehicle control, RT, IFN-γ, IFN-γ + RT, STAT1 inhibitor (STAT1i), and RT + STAT1i, all evaluated at 48 h post-RT (n = 3 per group). (B) Representative immunofluorescent staining images (left) and quantification (right) depicting the expression of α-SMA, IRF1, CCL4, or STAT1 in primary CAFs treated with vehicle, IFN-γ, IFN-γ ± RT, STAT1i, or STAT1i ± RT at specified time points (n = 5 per group). Scale bars, 10 μm. (C) Western blot images showing the expression levels of cGAS, p-STAT1/STAT1, p-STING/STING, IRF1, and CCL4 in primary CAFs treated with IFN-γ (20 ng/mL) or STAT1i (5 μM) combined with RT for 48 h. (D) GSEA depicting the upregulated STING pathway signature in ilCAFs compared to other CAFs from rectal cancers. (E) Heatmap depicts a list of differentially expressed genes sourced from inflammatory modulation, type II IFN, and TNF-α between vehicle- and RT-treated groups. (F) Representative flow cytometric dot plots displaying the MFI of IRF1 or CCL4 in CAFs isolated from treatment-naive rectal cancer patients (n = 3 per group), comparing vehicle control, RT, STING agonist SR717, and SR717 + RT at 48 h post-RT treatment. (G) Western blot images depicting the expression levels of cGAS, p-STAT1/STAT1, p-STING/STING, and IRF1 in primary CAFs treated with SR717 (3 μM). (H) Bubble plots illustrating upregulated pathways in fibroblasts co-cultured with tumor cell-derived media (TCMs) after RT + SR717 treatment compared to RT alone, based on GO biological processes. Pathways were filtered using adjusted p values (P[adj] < 0.05). (I) Quantification (right) of flow cytometry analysis of CD8^+ T cells, GZMB^+ CD8^+ T cells, CD11c^+ DCs, and CD103^+ CD11c^+ cDC1s in T cells or DCs co-cultured with mouse fibroblasts from the indicated combination treatment groups (n = 3 per group). For (A), (B), (F), and (I), one-way ANOVA was performed. Data are presented as mean ± SEM and are representative of at least three independent experiments. A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. We then reanalyzed human scRNA-seq data to investigate the correlation between STING signaling and the enrichment of ilCAFs. Intriguingly, we observed an upregulation of STING signaling[201]^26 in ilCAFs post-RT ([202]Figure 3D), consistent with our WB data presented earlier in the in vitro culture system ([203]Figure 3C). Further exploration of the TCGA dataset revealed that the expression of the STING gene TMEM173 correlated with genes associated with ilCAFs (IRF1 and CCL4) in rectal adenocarcinoma (READ) and colon adenocarcinoma (COAD) tumors ([204]Figure S8C). Similarly, our mouse tumor bulk RNA-seq data revealed a notable correlation between Sting1 expression and that of IFN signaling genes, including Irf1, Stat1, Ccl4, and Ccl5 ([205]Figures 3E and [206]S8D). To further validate the regulatory role of STING signaling in ilCAF activation, we collected conditioned media from co-cultures of Sting1-deficient fibroblasts and MC38 CRC cells ([207]Figures S8E and S8F). WB and ELISA analyses confirmed that the disruption of the STING pathway significantly impaired the expression of key ilCAF markers, indicating that STING signaling plays a pivotal role in driving the inflammatory activation state of CAFs ([208]Figures S8E and S8F). Collectively, these findings underscore the critical role of the STING-IFN axis in regulating ilCAF function and suggest that targeting STING signaling could be a promising strategy to modulate the tumor stroma and enhance the efficacy of RT. To directly assess whether the activation of STING can induce the polarization of ilCAFs, we applied the STING agonist SR717 to human primary CAFs that were pretreated with RT. Flow cytometric analysis revealed that, compared to vehicle controls, SR717 or RT alone, the combination of RT followed by SR717 further upregulated the expression of IRF1 and CCL4, confirming that the STING agonist can enhance the polarization of ilCAFs ([209]Figure 3F). WB analysis further demonstrated that RT followed by SR717 increased the expression of proteins involved in the IFN-γ/STING pathway, including p-STAT1, p-STING, IRF1, and cGAS, in CAFs ([210]Figure 3G). Moreover, pathway analysis revealed a strong inflammatory response in fibroblasts exposed to the combination of SR717 and RT, with notable enrichment of cytokine-mediated signaling and cytokine-binding pathways, as well as leukocyte migration, which mirrors the immune response observed in the TME ([211]Figure 3H). Furthermore, in our transwell migration assay, we observed that RT- and SR717-treated CAFs significantly promoted the transmigration of CD8^+ T cells, GZMB^+ CD8^+ T cells, DCs, and cDC1s ([212]Figure 3I). This provides additional evidence that the activation of the cGAS-STING pathway further promotes the polarization and formation of ilCAFs upon RT, enhancing their interactions with T cells and DCs. However, it remains unclear whether STING is necessary for the activation of ilCAFs. To address this, we isolated primary fibroblasts from Tmem173 knockout (Tmem173^−/−) mice and their parental WT counterparts and treated them with RT or a vehicle control ([213]Figure 4A). Compared to the vehicle control, we observed a marked increase of IRF1 and CCL4 in the fibroblasts subjected to RT ([214]Figures 4A and 4B). Importantly, this RT-induced effect was significantly diminished in the fibroblasts derived from Tmem173^−/− mice and was not rescued by SR717 treatment ([215]Figures 4A and 4B). In a murine model where MC38 CRC cells were co-inoculated with either WT or Tmem173^−/− fibroblasts and treated with RT (with the vehicle as the control), our in vivo data mirrored the in vitro findings ([216]Figures 4C–4G). Notably, Tmem173 knockout attenuated the anti-tumor efficacy of RT, and this effect persisted despite SR717 treatment ([217]Figure 4D). Multiplexed IF staining further revealed that Tmem173 deficiency reduced the accumulation of ilCAFs and decreased the infiltration of CD4^+ T cells, CD8^+ T cells, and cDC1 cells into the TME ([218]Figures 4H and 4I). Together, these findings identify STING as a pivotal regulator of ilCAF polarization in response to RT. Figure 4. [219]Figure 4 [220]Open in a new tab STING knockout in CAFs modulates the stromal landscape and enhances T cell infiltration (A and B) Histogram plots (left) and quantification of MFI of IRF1 (A) or CCL4 (B) in primary fibroblasts isolated from treatment-naive wild-type (WT) and Tmem173^−/− (STING knockout) mice, analyzed by flow cytometry. Comparisons include vehicle, RT, STING agonist (SR717), and combination therapy (SR717 + RT) in Tmem173^−/− fibroblasts and vehicle and RT in WT fibroblasts post-RT (vehicle group in A: n = 4; all other groups: n = 3 per group). (C–G) Impact of RT on the growth of established MC38 colorectal tumors co-inoculated with Tmem173^−/− or WT fibroblasts in syngeneic C57BL/6 mice. (C) Schematic of the experimental design. (D) Average tumor growth curve. (E) Tumor weight at endpoint. (F) Tumor-draining lymph node sizes. (G) Modified Kaplan-Meier survival curves for each treatment group (n = 5 per group). (H) Percentage of CD4^+ T cells, CD8^+ T cells, and cDC1s among CD45^+ tumor-infiltrating cells at day 14 post-RT treatment (n = 5 per group). (I) Representative multiplex immunofluorescence images (left) and quantification (right) of IRF1^+ CAFs (green: COL1A, red: IRF1) and CCL4^+ CAFs (green: COL1A, red: CCL4). Scale bars, 100 μm (n = 5 per group). For (A), (B), (E), (F), (H), and (I), one-way ANOVA was performed. For (D), a two-way ANOVA was conducted. For (G), the log rank test was performed. Data are presented as mean ± SEM. Results are representative of at least three independent experiments. A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. STING agonist enriched ilCAFs and enhanced endogenous anti-tumor immunity to synergize with RT efficacy Although RT remains a primary treatment modality for patients with RC, the average complete response rate is approximately 20%. Thus, improving RT efficacy could substantially benefit patients with RC. Given the strong activation and enhanced function of ilCAFs by STING agonist SR717 in vitro, we hypothesized that SR717 may potentiate RT efficacy via IFN/STING pathway activation. We treated syngeneic C57BL/6 mice bearing MC38 and fibroblasts co-injection tumors with vehicle, SR717, RT, or a combination of RT and SR717 ([221]Figure 5A). While SR717 alone showed minimal inhibition of tumor growth, RT exhibited robust inhibition, indicating the sensitivity of this mouse syngeneic co-injection tumor model to RT rather than STING agonist. Importantly, the combination of RT with SR717 led to enhanced tumor regression and prolonged survival in tumor-bearing mice ([222]Figures 5B–5D and [223]S9A). Next, we collected post-treatment tumors and performed bulk RNA-seq and multiplex IF to investigate potential reprogramming of CAFs. Consistent with earlier findings, RT significantly upregulated the expression of ilCAFs and iCAFs, with even greater induction observed in tumors receiving RT followed by SR717 ([224]Figure 5E). Additionally, the enrichment of ilCAFs and iCAFs in tumors treated with RT or RT combined with SR717 was associated with the upregulation of Sting1 expression ([225]Figure 5E). Multiplex IF further confirmed that RT followed by SR717 significantly increased STING and downstream p-STAT1 expression in CAFs compared to other treatment groups ([226]Figures 5F and 5G). Collectively, these results indicate that STING agonist enhances RT efficacy through IFN-γ pathway activation and ilCAF polarization. Figure 5. [227]Figure 5 [228]Open in a new tab STING agonist augments the presence of ilCAFs and enhances endogenous anti-tumor immunity to synergize with radiotherapy for enhanced efficacy (A) Experimental design illustrating the treatment regimen administered to C57BL/6 mice bearing MC38 colorectal tumors. (B) Average tumor growth curve in syngeneic mice co-inoculated with fibroblasts and treated with vehicle, SR717 (STING agonist), RT, or the combination SR717 with RT. (C) Tumor weights at endpoint. (D) Modified Kaplan-Meier survival curves representing outcomes for each treatment group (n = 5 per group). (E) Dot plots illustrating expression signatures of different CAF clusters in the mouse colorectal tumor model, determined by bulk RNA-seq. Violin plots depict cDC1s signature expression in MC38 tumors following the indicated treatments (n = 3 per group). (F and G) Representative multiplex IF images (F) and quantification (G) of STING^+ CAFs (green: STING, red: COL1A) and p-STAT1^+ CAFs (green: p-STAT1, red: COL1A; n = 5 per group). Scale bars, 100 μm. (H) Heatmap depicting the expression levels of multiple signaling pathways in tumors treated with vehicle control, SR717, RT, and RT + SR717 (n = 3 per group). (I and J) Representative multiplex IF images (left) and quantification (right) of endpoint tumors stained for colocalized CD103 (green) and DC (red) (I) and CD4 (red) and CD8 (green) (J) (n = 5 per group). Nuclei were stained with DAPI (blue). Scale bar, 100 μm. (K) Proportion of CD4^+ T cells, CD8^+ T cells, and cDC1s among CD45^+ cells in tumors and tumor-draining lymph nodes at day 12 post-RT in vehicle control, SR717, RT, and RT + SR717 groups (n = 5 mice/group). For (C), (E), (G), (I), (J), and (K), one-way ANOVA was performed. For (B), a two-way ANOVA was conducted. For (D), the log rank test was utilized. Data are presented as mean ± SEM. Results are representative of at least three independent experiments. A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. In addition to ilCAF polarization, gene set variation analysis (GSVA) of bulk RNA-seq from mouse tumors revealed enrichment of multiple immune-related signaling pathways in response to RT and SR717 treatment. These pathways included TNF-α, IFN-α, and IFN-γ signaling, along with increased ilCAF-associated gene expression ([229]Figure 5H). Furthermore, tumors treated with RT followed by SR717 exhibited increased activation of CCR5 chemokine receptor binding, chemokine activity, and DC cytokine production ([230]Figure 5H). These effects potentially contributed to the upregulation of CCL4 expression in ilCAFs and the recruitment of T cells and cDC1s into tumors ([231]Figure 5E). We next performed multiplex IF and multiparametric flow cytometry at the study endpoint to characterize changes in the immune landscape. Notably, we observed increased infiltration of cDC1s, CD4^+, and CD8^+ T cells in tumors and tumor-draining lymph nodes of mice treated with RT followed by SR717, compared to other groups ([232]Figures 5I–5K). Moreover, peripheral blood analysis revealed increased levels of NK cells, as well as CD4^+ and CD8^+ T cells, in mice treated with the RT and SR717 combination ([233]Figure S9B), suggesting robust priming of systemic immune responses. We next employed flow cytometry to phenotypically assess the functional status of tumor-infiltrating CD8^+ T cells. RT combined with STING agonist markedly increased the expression of activation markers (PD-1 and CD69) and expanded GZMB^+ CD8^+ and TCF1^+ CD8^+ T cell populations relative to other treatment groups ([234]Figures S9C and S9D). These findings indicate that CD8^+ T cells adopt an activated effector phenotype in response to RT combined with STING agonist therapy, thereby contributing to enhanced anti-tumor immunity. Taken together, our findings provide direct evidence that RT combined with an STING agonist reprograms CAFs into ilCAFs, reshapes the tumor immune microenvironment to support T cell priming, and drives durable anti-tumor responses. To assess the safety of SR717 treatment followed by RT, we systematically collected blood and various organs to evaluate potential side effects or toxicity ([235]Figure 5A). As previously reported,[236]^27 we observed an increase in granulocyte levels in blood of mice after RT or combination therapy ([237]Figure S10A). Although changes were noted in alanine aminotransferase levels, no significant impairment of liver or kidney function was observed in mice after combination therapy or RT alone ([238]Figure S10B). Moreover, no evidence of organ damage was detected in any of the mice ([239]Figure S10C). Collectively, these results indicate that RT combined with STING agonists induces only minimal systemic toxicity in tumor-bearing mice. STING agonist augments the anti-tumor efficacy of RT in spontaneous CRC mouse models Motivated by the potent anti-tumor efficacy observed with the combination of RT and STING agonist in a syngeneic subcutaneous mouse model, we sought to assess the translational relevance of this finding using a more clinically pertinent genetic engineering mouse model. We employed the Apc^min/+ mouse model, a well-established model for multiple intestinal neoplasia, mimicking human familial adenomatous polyposis and colorectal tumors.[240]^6 Characterized by a mutation in the Apc gene, a pivotal tumor-suppressor gene in the Wnt signaling pathway, these mice develop tumors in the small intestine.[241]^28 At 15 weeks of age, when tumors had already developed in the small intestine, Apc^min/+ mice were treated with RT followed by the STING agonist ([242]Figure 6A). Macroscopic analysis at the endpoint revealed that SR717 alone exerted minimal impact on tumor burden in Apc^min/+ mice, while RT significantly suppressed tumor growth ([243]Figure 6B), consistent with the subcutaneous model findings ([244]Figures 5B–5D). Notably, a profound reduction in both the overall number and size of lesions was observed in mice treated with RT combined with STING agonists ([245]Figures 6B and 6C). Similarly, RT followed by STING activation markedly reduced tumor number and tumor load in the small intestine of Apc^min/+ mice ([246]Figures 6B and 6C), providing additional evidence of its potent anti-tumor activity. Figure 6. [247]Figure 6 [248]Open in a new tab STING agonist enhances the therapeutic efficacy of radiotherapy in spontaneous colorectal cancer mouse models (A) Experimental timeline illustrating the treatment regimen administered to Apc^min/+ mice bearing spontaneous colorectal tumors. (B) Hematoxylin and eosin (H&E) staining of small intestinal sections from 17-week-old mice. Scale bars indicate 1 mm (top) and 100 μm (bottom). (C) Quantification of tumor number and tumor burden in the small intestine and total intestinal tract (n = 4 per group), comparing treatments with vehicle control, radiotherapy (RT), STING agonist (SR717), and combination therapy (RT + SR717). (D) Schematic timeline of the experimental design evaluating the impact of combination therapy on dextran sodium sulfate (DSS)-induced tumorigenesis in C57BL/6J wild-type mice injected intraperitoneally with azoxymethane (AOM, 10 mg/kg). (E) Representative images of colorectal tumors treated with RT versus RT + SR717 (n = 5 per group). (F) H&E staining of colorectal sections from DSS-induced tumorigenesis. Scale bars: 1 mm (top) and 100 μm (bottom). (G) Quantification of tumor burden and tumor number in the colorectal and total intestinal tract (n = 5 per group), comparing RT with RT + SR717 treatment groups. (H and I) Representative multiplex IF images (H) and quantification (I) showing the localization of STING^+ CAFs and p-STAT1^+ CAFs (green: STING, red: p-STAT1, orange: COL1A, blue: DAPI) (n = 5 per group). Scale bars indicate 100 μm. Data are presented as mean ± SEM. Statistical analyses included one-way ANOVA for (C) and Student’s t tests for (G) and (I). A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Furthermore, we employed the azoxymethane (AOM)/dextran sodium sulfate (DSS) model of colitis-associated cancer (CAC), which recapitulates key aspects of human CAC, including distally located tumors and invasive adenocarcinomas.[249]^29 This model was utilized to validate the synergistic effect of the STING agonist with RT in inhibiting the development and progression of CRC ([250]Figure 6D). Encouragingly, long-term follow-up revealed that AOM-DSS-induced spontaneous CRC tumors exhibited significantly reduced tumor burden, in both numbers and sizes, when treated with RT followed by SR717 than with RT alone ([251]Figures 6E–6G). Moreover, our multiplex IF analysis validated that RT followed by SR717 significantly upregulated STING and downstream p-STAT1 levels in the CAFs compared to RT ([252]Figures 6H and 6I). Furthermore, we examined the activation status of ilCAFs following treatment with the STING agonist in combination with RT. Notably, the combination therapy induced significant upregulation of IRF1 and CCL4 expression in ilCAFs compared to RT alone ([253]Figures S11A and S11B), indicating enhanced IFN pathway activity in the stromal compartment. These findings collectively underscore the potent anti-tumor efficacy elicited by the combination of RT and a STING agonist across two spontaneous CRC models, highlighting the strong translational potential of this therapeutic strategy for clinical application in solid tumors. To further evaluate the generalizability of this combination strategy across solid tumor types, we established a subcutaneous Lewis lung carcinoma (LLC) mouse model ([254]Figures S11C and S11D). As anticipated, the combination of RT and the STING agonist resulted in a pronounced anti-tumor effect in the LLC model. Together, these results support the broad therapeutic potential of RT + STING combination treatment across multiple solid tumor settings, paving the way for future translational and clinical investigations. STING agonist reversed the radioresistant effect of colorectal tumors In the context of neoadjuvant chemoradiotherapy for patients with RC, the emergence of acquired resistance poses a formidable challenge, often culminating in treatment failure. Our investigative endeavor sought to ascertain the potential of a STING agonist in overcoming acquired radioresistance and impeding the growth of CRC tumors. To this end, we devised a mouse model that mirrored resistance to RT through an in vivo selection approach akin to that employed in patient-derived xenograft models. Concretely, syngeneic mice were implanted with high-dose-radiated MC38 colorectal cells. Non-responsive tumors were harvested and subsequently transplanted into naive recipients, iterating this process for five generations until all subsequent generations displayed insensitivity to radiation ([255]Figure S12A). Verification of RT resistance was accomplished by subcutaneously implanting tumors from the third generation into naive mice; as anticipated, tumors from the sixth generation were unresponsive to RT compared to those from the first generation ([256]Figure S12A). This procedure was reiterated for three additional generations to establish stable radioresistance ([257]Figure S12A). Subsequently, tumors from the sixth generation were treated with RT, SR717, and the combination of RT and SR717 ([258]Figure 7A). Intriguingly, similar to the vehicle and RT alone, SR717 alone did not impact tumor growth. In contrast, SR717, when combined with RT, markedly reduced tumor growth and formation ([259]Figures 7B and 7C), accompanied by a notable decrease in draining lymph node sizes in tumors treated with RT followed by SR717 ([260]Figure 7D). Our preliminary data suggest that the STING agonist re-sensitizes RT-resistant CRC tumors to radiation. Figure 7. [261]Figure 7 [262]Open in a new tab STING agonist reverses radioresistance in colorectal tumors (A) Treatment protocol: C57BL/6 mice bearing radioresistant MC38 tumors received 20 Gy of RT, followed by treatment with a STING agonist. (B–D) Average tumor growth curves (B), tumor weight at the endpoint (C), and tumor-draining lymph node sizes (D) were evaluated across different treatment cohorts in a syngeneic transplant mouse model (n = 5 per group). (E) Heatmap illustrating the transcriptional profiles of STING and type II IFN signaling pathway-associated genes in radioresistant tumors treated with vehicle control, SR717, RT, or RT + SR717, as revealed by bulk RNA-seq (n = 3 per group). (F) Bubble plots illustrating the upregulated pathways in tumors treated with RT + SR717 relative to RT alone, based on known biological functions in GO databases. Pathways were filtered using adjusted p values (P[adj] < 0.05). (G) Representative multiplex IF image (left) and quantification (right) demonstrating the localization of STING^+ CAFs and p-STAT1^+ CAFs within the tumor microenvironment (green: STING, red: p-STAT1, orange: COL1A, blue: DAPI). Scale bar, 100 μm (n = 5 per group). (H) Quantification of CD4^+ T cells, CD8^+ T cells, and cDC1s among CD45^+ cells in tumors and tumor-draining lymph nodes from vehicle control, SR717, RT, and RT + SR717 treatment groups at day 14 post-RT (n = 5 per group). Data are presented as mean ± SEM. Statistical analyses were performed using one-way ANOVA for (C), (D), (G), and (H) and two-way ANOVA for (B). A p value less than 0.05 indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. To comprehensively elucidate the mechanisms underlying RT resistance and the restoration of responsiveness through the combination of RT and STING agonist, we conducted bulk RNA-seq analysis on the collected tumors. Strikingly, tumors treated with RT followed by SR717 exhibited significant upregulation of type II IFN signaling gene sets (Sting1, Irf1, Stat1, and Ccl4) and STING-mediated immune response pathways ([263]Figures S12B–S12D). In contrast, RT alone exhibited a modest response in STING pathways ([264]Figures S12B–S12D). Notably, the combination therapy markedly upregulated ilCAF signature genes compared to other CAF subtypes ([265]Figures 7E and [266]S12B). Importantly, tumors treated with RT followed by the STING agonist demonstrated enrichment in multiple pathways associated with adaptive and innate immune responses, as well as the activation of immune cells, compared to RT alone ([267]Figure 7F). Furthermore, higher inflammatory responses and ECM disassembly were observed in tumors treated with the combination of RT and SR717 ([268]Figure 7F). Together, our bulk RNA-seq analysis indicated that the STING agonist renders RT-resistant tumors sensitive to irradiation by reshaping the TME, including the enrichment of ilCAFs and recruitment of anti-tumor immune cells. To validate our findings from bulk RNA-seq analysis, we conducted multiplex IF and multiparametric flow cytometry to analyze STING/STAT1 activation in tumors. Consistently, we found that SR717 increased the numbers of STING^+ and p-STAT1^+ cells among total COL1A^+ fibroblasts in irradiated tumors ([269]Figure 7G). Quantification of ilCAFs further revealed their substantial enrichment in the combination group ([270]Figures S13A and S13B). Moreover, flow cytometric analysis revealed that combination therapy resulted in the highest infiltration of cDC1s, CD8^+ T cells, and CD4^+ T cells in tumors, draining lymph nodes, and peripheral blood among all groups ([271]Figures 7H and [272]S12D). Collectively, these data substantiate the conclusion that the STING agonist reverses RT-resistant tumors, rendering them sensitive to RT by activating STING/STAT1 signaling, enriching ilCAFs, and recruiting anti-tumor immune cells. To further evaluate the applicability of this strategy to other radioresistant tumor types, we used the MCA205 fibrosarcoma model ([273]Figure S13C). Consistent with our previous findings, the combination of RT and SR717 also markedly induced an anti-tumor response in the MCA205 model ([274]Figures S13D and S13E), further reinforcing the therapeutic potential of this strategy across diverse solid tumor. Discussion Beyond its primary effect of inducing cancer cell death, RT has significantly expanded our understanding by revealing its ability to induce rapid and consistent remodeling of the TME.[275]^30^,[276]^31^,[277]^32 Despite these advancements, the impact of RT on CAFs, particularly the potential effects and mechanisms through which CAFs influence tumor radioresistance, remains poorly understood. In this study, we employed scRNA-seq and bulk RNA-seq in human and murine RC tumors, respectively. Our analysis identified a subset of ilCAFs driven by IFN-γ signaling. These unique ilCAFs were enriched in tumors that responded to RT. Notably, ilCAFs played a crucial role in enhancing local and systemic adaptive anti-tumor immunity by secreting chemokines CCL4 and CCL5, leading to the recruitment of CD4^+, CD8^+ T cells, and cDC1s into the tumors. Furthermore, our investigation revealed that cGAS-STING activation was a major contributor to type II IFN signaling. This activation induced the polarization and enrichment of the ilCAF subpopulation. Importantly, patients with a high proportion of ilCAFs exhibited a better prognosis and better responses to RT across various solid tumors, including CRC. In conclusion, our data contribute to a deeper understanding of the dynamic nature of CAFs in response to RT. We propose that STING activation can synergize with the anti-tumor efficacy of RT, partially through the polarization and enrichment of ilCAFs. Recent studies such as that by Chatila et al. have provided a comprehensive molecular characterization of RC, identifying key genomic alterations and immune profiles associated with response to neoadjuvant therapy.[278]^33 Notably, mutations in KRAS and variations in immune infiltration markers have been implicated in differential treatment outcomes.[279]^33 While these findings offer valuable insights into the genetic and transcriptomic landscape of RC, they predominantly emphasize tumor-intrinsic features and immune components in the TME, with limited exploration of the functional contributions of CAFs in modulating therapeutic response. In contrast, our study centers on the heterogeneity of CAFs and their functional implications in the response to RT in RC. Specifically, we identify a distinct subset of iCAFs expressing IRF1 as critical modulators of RT efficacy through immune regulation. Prior investigations have suggested that certain CAF subsets, including fibroblast activation protein (FAP)^+ CAFs, influence treatment responses via mechanisms such as immune activation or the induction of epithelial-mesenchymal transition.[280]^34 Moreover, iCAFs have been implicated in promoting radioresistance in RC through IL-1 signaling and IL-6-mediated pathways.[281]^35 Interestingly, our findings differ from these earlier reports. Through comprehensive scRNA-seq analysis of patients with RC tumors, we observed the significant enrichment of iCAFs following RT. Importantly, the post-RT accumulation of iCAFs correlated positively with treatment response and patient survival, suggesting the presence of phenotypically and functionally distinct iCAF subsets with divergent roles in response to RT. Unbiased clustering of scRNA-seq data revealed multiple CAF subpopulations, including myCAFs, SOX6^+ CAFs, and iCAFs. Further stratification of the iCAF population identified three transcriptionally distinct subsets: classic iCAFs, CXCL1^+ CAFs, and ilCAFs. Notably, ilCAFs were markedly enriched following RT and exhibited strong activation of type II IFN signaling, along with indications of enhanced anti-tumor immunity. Crucially, gene signature analysis revealed that increased abundance of ilCAFs was correlated with improved survival outcomes across various solid tumors, including CRC, breast cancer, and esophageal cancer. These correlations were substantiated through integrative analyses of TCGA datasets and publicly available patient datasets. Collectively, our findings underscore the dynamic and heterogeneous nature of iCAFs and their potential as prognostic indicators across diverse cancer types in response to radiotherapeutic interventions. In-depth signature profiling of ilCAFs in comparison to iCAFs revealed significant enrichment of immune-related pathways, which was corroborated in vivo by increased cDC1, CD8^+, and CD4^+ T cell infiltration in tumors, draining lymph nodes, and peripheral blood, following RT. Mechanistically, RT-induced ilCAFs upregulated CCL4 and CCL5, and pharmacologic inhibition of their receptor CCR5, significantly reducing cDC1 and CD8^+ T cell recruitment and thereby diminishing the therapeutic efficacy of RT in CRC models. Functional analyses further showed that CCR5-dependent cDC1s infiltration precedes and facilitates CD8^+ T cell recruitment and activation, as reflected by elevated intratumoral GZMB^+ CD8^+ T cells. These data support a model in which the RT-ilCAF-CCL4/CCL5-CCR5 axis orchestrates a coordinated immune cascade essential for effective anti-tumor responses. Our study further elucidated that the polarization of ilCAFs in tumors following RT was contingent upon the IFN-γ/STING signaling axis. Previous literature has established that the activation of the IFN/STING pathway induces alterations in the TME, impacting immune cells, tumor cells, and stromal cells.[282]^30^,[283]^36^,[284]^37^,[285]^38 In a separate investigation employing scRNA-seq in bladder cancer, a distinct subset of CAFs characterized by SLC14A1 expression (SLC14A1^+ CAFs) was identified. This particular CAF subpopulation was found to be driven by type I IFN signaling. The study provided evidence that SLC14A1^+ CAFs, under the influence of type I IFN signaling, can promote chemotherapy resistance. The underlying mechanism involved the enhancement of cancer stemness, indicating a pivotal role for this CAF subpopulation in fostering resistance to chemotherapy in the context of bladder cancer. Consistent with these findings, our scRNA-seq analysis identified a subset of CAFs marked by the expression of IFN-γ-induced genes, specifically IRF1, downstream of the STING pathway. Remarkably, these ilCAFs demonstrated the capacity to induce T cell and cDC1 priming, leading to a favorable prognosis. Furthermore, our observations indicated an augmented presence of the STING gene signature in response to RT, suggesting a potential shift toward an enhanced anti-tumor immune phenotype mediated by ilCAFs. Crucially, our study unveiled the critical role of STING in inducing the formation of IRF1^+ ilCAFs. Activation of STING in CAFs resulted in a marked upregulation of IRF1 expression, thereby promoting the polarization of ilCAFs. These findings contribute to a comprehensive understanding of the intricate interplay between ilCAFs, immune responses, and the STING pathway in the context of radiotherapeutic interventions. A limitation of our study is that the scRNA-seq samples for pre- and post-RT conditions were obtained from different patients, rather than matched pairs. Further studies utilizing paired pre- and post-treatment samples are warranted to validate these findings. Furthermore, in our murine MC38 syngeneic subcutaneous tumor models and autochthonous Apc^min/+ CRC tumors, RT induced transient tumor control but proved ineffective in priming the T cell response. This led to a lack of durable tumor control and did not confer a survival benefit. Building upon these observations, we investigated whether the administration of a STING agonist could enhance the polarization of RT-induced ilCAFs, promote robust T cell infiltration, and augment anti-tumor immune responses in CRC. Encouragingly, the combination of RT and a STING agonist in CRC tumors exhibited potent anti-tumor efficacy and immune priming. This effect was associated with the upregulation of the ilCAF signature. Notably, this combination therapy sensitized previously RT-resistant CRC tumors in an established mouse model. The sensitization was achieved through activation of the IFN/STING pathway, which elicited responses from antigen-processing cells and enhanced cytokine signaling, resulting in a heightened overall immune response. The remarkable anti-tumor effects observed in our study provide a compelling rationale for combining RT with STING agonists in future CRC treatments. Moving forward, it will be intriguing to assess patient responses to the combination of RT followed by a combination of anti-PD-1 therapy and STING agonists across various solid tumors, as explored in preclinical and on-going trials ([286]NCT05846646 and [287]NCT05846659).[288]^39 Several STING agonists, including ADU-S100 ([289]NCT03172936) and SYNB1891 ([290]NCT04167137), have advanced into clinical trials to assess their safety and efficacy across various cancer types.[291]^40^,[292]^41 Preliminary safety and tolerability data from SYNB1891 monotherapy ([293]NCT04167137) in eleven patients with refractory advanced solid tumors or lymphoma revealed no dose-limiting toxicities and no treatment discontinuations due to adverse events.[294]^42 Notably, two patients achieved stable disease. In a phase 1b trial, ADU-S100 was administered in combination with the PD-1 inhibitor spartalizumab.[295]^41 Preliminary results indicated that, in patients with triple-negative breast cancer and melanoma, the combination therapy elicited improved responses compared with ADU-S100 monotherapy.[296]^41 However, these studies were small and preliminary, and none included patients with RC. Additionally, two clinical trials evaluating the combination of STING agonists and radiation therapy ([297]NCT05846646 and [298]NCT05846659) have been temporarily paused pending further evaluation. In summary, our work underscores the pivotal involvement of IFN-STING signaling in governing the polarization of ilCAFs. The heightened abundance of ilCAFs, capable of secreting chemokines CCL4 and CCL5, enhances endogenous immune responses and ultimately contributes to improved overall survival in patients with RC. Collectively, our findings underscore the translational potential of combining STING agonists with RT, offering a strong rationale for further exploration in the context of cancer therapy. Notably, early initiation of this combination strategy may amplify initial anti-tumor responses and attenuate the emergence of therapeutic resistance. Nonetheless, this combination has not yet been tested in RC, and preliminary safety and tolerability data will be essential prerequisites before larger efficacy trials can be considered. Limitations of the study Several limitations of this study should be acknowledged. First, although we identified a distinct CAF subpopulation associated with RT, the sample size remains relatively small, and scRNA-seq data from radioresistant patients are currently unavailable. Second, while we highlighted interactions between ilCAFs and immune cells such as T cells and DCs following RT, further mechanistic studies are required to explore the crosstalk between ilCAFs and other cellular compartments, including epithelial and endothelial cells, as well as myeloid populations such as macrophages and neutrophils. Third, although we demonstrated a role for ilCAFs in recruiting DCs, further validation through conditional gene knockout models is warranted to confirm these findings. Fourth, while our results suggest that ilCAFs may reverse tumor radioresistance in vivo, the underlying molecular mechanisms remain to be fully elucidated. These limitations present important avenues for future research. In particular, exploring this treatment strategy in other tumor types may help validate and generalize its therapeutic potential. Furthermore, the potential to enhance the synergistic efficacy of STING agonists and chemotherapeutic agents merits further exploration, especially through approaches aimed at amplifying the IFN-STING signaling axis. Resource availability Lead contact Requests for resources and reagents should be directed to the lead contact, Zhen Zhang (zhen_zhang@fudan.edu.cn). Materials availability This study did not generate new unique reagents. Data and code availability * • The scRNA-seq data from human RC have been deposited in the Genome Sequence Archive (GSA) at the National Genomics Data Center (Beijing, China), and the bulk RNA-seq data generated from mouse MC38 tumors have also been deposited in the same database. Accession numbers are listed in the [299]key resources table. * • This paper does not report the original code. All custom code used for the analyses was written with existing software, as detailed in the [300]STAR Methods section, and is available upon request. * • Any additional information required to reanalyze the data reported in this paper is available from the [301]lead contact upon request. Acknowledgments