A Paracrine-to-Autocrine Shunt of GREM1 Fuels Colorectal Cancer Metastasis via ACVR1C

A Paracrine-to-Autocrine Shunt of GREM1 Fuels Colorectal Cancer Metastasis via ACVR1C | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A Paracrine-to-Autocrine Shunt of GREM1 Fuels Colorectal Cancer Metastasis via ACVR1C Huaixiang Zhou, Qunlong Jin, Zhang Fu, Yanming Yang, Yunfei Gao, and 16 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7484753/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 24 Jan, 2026 Read the published version in Molecular Cancer → Version 1 posted 12 You are reading this latest preprint version Abstract Background: Tumor cells typically rely on paracrine stromal signals to guide malignant behavior, yet whether they gain signaling autonomy and thereby reduce microenvironment dependency during metastasis remains unclear. Methods: Colorectal cancer (CRC) specimens from stages I–IV were analyzed by immunohistochemistry and single-cell transcriptomics to assess GREM1 and ACVR1C expression and localization. The GREM1–ACVR1C interaction was validated by interaction proteomics, co-immunoprecipitation, immunofluorescence, and microscale thermophoresis (MST). Functional roles of the axis in metastasis were examined by transcriptomic profiling, pathway analysis, immunoblotting, RT–qPCR, scratch and transwell assays, and genetically engineered and xenograft mouse models. An inhibitory peptide targeting the GREM1–ACVR1C interface was designed and evaluated. Results: While GREM1 remains restricted to stromal cells in earlier-stage (I–III) CRC, its ectopic expression in tumor epithelium increases markedly in stage IV. Mechanistically, we identify activin A receptor type 1C (ACVR1C) as a direct, high-affinity epithelial receptor for GREM1. Their interaction, independent of canonical TGFβR and BMP signaling, activates SMAD2/3, which in turn induces the transcription of SNAI1 and GREM1 , thereby establishing a self-sustaining feedback loop that amplifies epithelial-mesenchymal transition (EMT). Disrupting this loop via stromal GREM1 deletion, epithelial ACVR1C knockdown, kinase inhibition, or a novel GREM1-blocking peptide targeting the GREM1-ACVR1C binding interface significantly impairs CRC metastasis in vivo . Clinically, epithelial GREM1 or ACVR1C expression predicts aggressive disease and poor survival. Conclusions: Our findings define a paradigm in which tumor cells hijack stromal GREM1 to establish a GREM1–ACVR1C autocrine loop that sustains EMT and metastasis, marking a shift toward signaling autonomy and revealing a targetable vulnerability in advanced CRC. Colorectal cancer GREM1–ACVR1C axis paracrine-to-autocrine shift signaling autonomy EMT Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Distant metastasis remains the leading cause of death among patients with colorectal cancer (CRC)[ 1 ]. A complex and dynamic interplay between tumor epithelial cells and various non-malignant components of the tumor microenvironment (TME) is recognized as a central driver of tumor initiation, progression, and phenotypic plasticity[ 2 ]. However, once tumor cells detach from the primary site to metastasize, they inevitably lose the continuous support of the local TME—including stromal cells and localized signaling cues. This abrupt interruption of paracrine support represents a major bottleneck that restricts most tumor cells from successfully establishing distant colonies[ 3 ]. This phenomenon raises a fundamental question about how a minority of tumor cells sustain metastasis-associated phenotypes without ongoing microenvironmental support. The capacity to sustain malignant behavior without external cues can be viewed as a form of ‘signaling autonomy,’ consistent with the classic cancer hallmark of “self-sufficiency in growth signals” described by Hanahan and Weinberg[ 4 ]. While signaling autonomy has been extensively explored in the context of proliferation and survival, it remains unclear whether metastasizing tumor cells can acquire autonomous control over the programs that sustain their migratory capacity. Within the TME, cancer-associated fibroblasts (CAFs) are among the most active stromal components, driving tumor initiation and progression through bidirectional interactions with cancer cells[ 5 ]. Among the highly heterogeneous CAF population, a GREM1-expressing subset has gained attention for promoting tumor progression via paracrine GREM1 secretion in CRC[ 6 – 8 ] and breast cancer[ 9 ]. Intriguingly, GREM1 expression is not restricted to stromal CAFs. In pancreatic[ 10 ] and prostate cancers[ 11 ], tumor epithelial cells can express and secrete GREM1 to regulate their own phenotypic plasticity, and epithelial-specific GREM1 is markedly upregulated in Hereditary Mixed Polyposis Syndrome (HMPS)[ 12 ]. Nonetheless, whether such epithelial GREM1 expression occurs in sporadic CRC and contributes to cancer cell plasticity or metastasis remains unknown. Studies, including ours, show that modulating GREM1 expression in CRC cell lines affects cell migration and epithelial–mesenchymal transition (EMT)-related phenotypes[ 13 , 14 ]. Yet, whether CRC epithelial cells can “hijack” stromal GREM1 signals to ultimately activate their own GREM1 expression for malignant progression remains an open question. Mechanistically, GREM1 is a canonical antagonist of bone morphogenetic protein (BMP) signaling[ 15 ]. Recent studies indicate that GREM1 also possesses cytokine-like functions, binding noncanonical receptors, including VEGFR2[ 16 ], FGFR1[ 11 ], and EGFR[ 17 ], suggesting that it may influence cancer cell behavior through multiple signaling routes. However, it remains unclear whether GREM1 drives signaling autonomy in CRC via these receptors or others yet unidentified. In this study, we identify a paracrine-to-autocrine shunt of GREM1 in CRC, driven by the newly recognized receptor ACVR1C[ 18 ], a TGFβ superfamily type I receptor, thereby leading to SMAD2/3 phosphorylation. This self-sustaining loop confers tumor cells with signaling autonomy and metastatic potential, uncovering a previously unrecognized mechanism of tumor evolution and a potential approach for therapeutic intervention in CRC. Results Ectopic expression of GREM1 during CRC progression In the gut, GREM1 marks a subpopulation of fibroblasts in both normal[ 19 ] and tumor tissues[ 6 ]. Using Grem1-CreER T2 ; Rosa-mTmG mice[ 19 ], we confirmed that Grem1 + cells are exclusively confined to the stromal compartment, and distributed along the intestinal isthmus and adjacent to α-SMA + myofibroblasts[ 20 ], two months post-tamoxifen (TMX) injection (Figures S1 A–C). Extending this observation to humans, we found that GREM1 was likewise absent from epithelial cells in normal intestinal tissues, but sporadically expressed in stromal cells (Figure S1 D), suggesting a conserved stromal specificity. Similarly, in human stage I-III CRC samples, GREM1 staining co-localized with VIMENTIN (VIM, a stromal cell marker, encoded by VIM )[ 20 , 21 ] and fibroblast activation protein (FAP, an activated fibroblast marker)[ 20 ], but was mutually exclusive with β-CATENIN (β-CAT, a CRC cell marker, encoded by CTNNB1 )[ 22 ], CD68 (a macrophage marker)[ 21 ], or α-SMA (a myofibroblast marker) (Figures S1 E–J). These findings confirm that GREM1 is a bona fide stromal factor and GREM1 + stromal cells are a subtype of cancer-associated fibroblasts (CAFs), potentially contributing to CRC progression[ 6 , 8 ]. To systematically investigate the distribution and clinical significance of GREM1, we first performed immunohistochemical (IHC) staining on 106 human primary CRC samples spanning all four stages. We observed a stage-dependent redistribution of GREM1⁺ cells: in early-stage tumors (stage I–II), GREM1⁺ stromal cells were predominantly restricted to the peritumoral stroma; in stage III, these cells more frequently infiltrated the tumor parenchyma; in stage IV, infiltration of GREM1⁺ stromal cells showed a decreasing trend (with no statistically significant difference compared to stage III). Notably, in stage IV tumors, strong GREM1 staining emerged in a subset of tumor epithelial regions (Figs. 1 A–C). These findings suggest a potential shift of GREM1 expression from stroma to epithelium during tumor progression. To dissect the cellular sources and stage-specific dynamics of GREM1 expression at higher resolution, we analyzed publicly available single-cell RNA sequencing (scRNA-seq) datasets covering CRC samples from stages I–IV. Consistent with IHC data, GREM1 expression was primarily detected in fibroblasts and epithelial cells (Figures S2 A–C). Stage-specific analysis revealed a marked upregulation of GREM1 in fibroblasts at stages III and IV (Figure S2 D). Strikingly, GREM1⁺ epithelial cells were detected almost exclusively in stage IV tumors (Figure S2 E), supporting the notion that ectopic GREM1 expression by tumor cells is a late event in CRC progression. This observation was further validated by colocalization of GREM1⁺ with an epithelial marker epithelial cell adhesion molecule (EPCAM)[ 21 ] in tumor cells (Fig. 1 D). Finally, survival analysis demonstrated that patients with high GREM1 expression in tumor cells had significantly shorter overall survival compared to those with low expression (Fig. 1 E). Taken together, these results reveal that while GREM1⁺ CAFs are present throughout CRC progression, GREM1⁺ tumor epithelial cells emerge predominantly in advanced CRC, indicating that GREM1 expression, initially restricted to stromal cells, is progressively co-opted by tumor epithelial cells as CRC advances. ACVR1C is a novel GREM1 receptor in CRC This spatiotemporal ectopic expression of GREM1 suggests that tumoral autocrine signaling may be initiated by preceding stromal paracrine cues. Given the potential for intercellular communication mediated by the infiltration of GREM1⁺ stromal cells into the tumor parenchyma, we hypothesized that CRC tumor cells might express GREM1 receptor(s), through which downstream signaling cascades drive tumoral GREM1 expression and promote CRC progression. To identify potential GREM1 receptors, we overexpressed HA-tagged GREM1 in the human CRC cell line HCT116 (Fig. 2 A). Mass spectrometry analysis of proteins pulled down using anti-HA beads identified activin A receptor type 1C (ACVR1C), a member of the TGFβ superfamily, as a potential GREM1 receptor (Figs. 2 B and S3A). To validate this interaction, we performed co-immunoprecipitation (co-IP) assays using HA-tagged GREM1 in HCT116 cells. Immunoblotting revealed that GREM1 interacted with ACVR1C specifically, while no such interaction was detected with other members of the TGFβ superfamily such as TGFβR1 (Fig. 2 C). Similarly, co-IP assays using Flag-tagged ACVR1C confirmed the interaction with GREM1, which was abolished by a GREM1-blocking antibody (BAb) (Fig. 2 D). Confocal microscopy revealed the co-localization of ACVR1C with GREM1 in SW480 CRC cells (Fig. 2 E). We next examined whether a direct interaction exists between GREM1 and ACVR1C. We found that Fc-tagged ACVR1C extracellular domain (ACVR1C-ECD, AA 1 − 113 ) and His-tagged full-length GREM1 were pulled down together, demonstrating a direct physical association between ACVR1C and GREM1 (Fig. 2 F). Further analysis of the binding affinity of GREM1 for ACVR1C using microscale thermophoresis (MST) revealed that ACVR1C-ECD exhibited a 12.6-fold higher affinity for GREM1 ( K d = 67.67 ± 10.35 nM) than that for ACTIVIN B ( K d = 854.1 ± 127.47 nM), a known ligand of ACVR1C[ 23 ] (Fig. 2 G). To further delineate the explicit interaction mode of GREM1 and ACVR1C, we constructed truncated GREM1 and ACVR1C-ECD. Co-IP assays showed that deletion of amino acids 100–157 (AA 100–157 ) in GREM1 or 68–113 (AA 68–113 ) in ACVR1C-ECD effectively abolished their interaction in HCT116 cells (Figs. 2 H, I). Based on these findings, we aimed to identify key residues mediating the interaction between GREM1 and ACVR1C. The HDOCK platform ( http://hdock.phys.hust.edu.cn/ ) was utilized to simulate potential docking modalities between GREM1 (PDB: 5AEJ)[ 24 ] and ACVR1C (AlphaFold prediction, https://alphafold.ebi.ac.uk/entry/Q8NER5 ) based on their structures. Residues Q101/T102/T112/N115 in GREM1 and Q72/E85/T101 in ACVR1C were predicted to be essential for binding (Fig. 2 J). To validate, we performed site-directed mutagenesis of the predicted residues to assess binding. Notably, the Q101A/T102A/T112A/N115A quadruple mutation in GREM1, or single mutations E85A or T101A in ACVR1C, significantly abrogated the GREM1-ACVR1C interaction (Figs. 2 K, L). Further, we found no detectable interaction between recombinant GREM1 and the ACVR1C-ECD double mutant (E85A/T101A) by assessing their binding affinity using MST (Figure S3 B). These data suggest that Q101/T102/T112/N115 in GREM1, and E85/T101 in ACVR1C are key residues mediating their interaction. Clinically, IHC and scRNA-seq revealed that ACVR1C is expressed in tumor cells, with markedly elevated levels in stage IV CRC (Figures S3 C–E). Notably, high ACVR1C expression correlates with poor prognosis in stage IV CRC patients, supporting ACVR1C’s tumor-promoting role and consistent with the clinical significance of GREM1 (Figure S3 F). Taken together, these results demonstrate that ACVR1C is a novel receptor of GREM1 in CRC. Secretory GREM1 induces EMT via the ACVR1C-SMAD2/3 pathway but not TGFβR/BMPR pathways To explore whether GREM1 serves as a functional ligand for ACVR1C in CRC cells, we first generated GREM1-enriched conditioned medium (GREM1-CM) and control conditioned medium (Vec-CM) using HEK293 cells (Figures S4 A–C). We then performed RNA-sequencing (RNA-seq) on HCT116 cells treated with GREM1-CM or Vec-CM. Gene Set Enrichment Analysis (GSEA) revealed that SMAD2/3 and EMT pathways were significantly enriched in CRC cells treated with GREM1-CM (Fig. 3 A). ACVR1C is one of the receptors of the TGFβ superfamily, and it transduces signals primarily through the phosphorylation of SMAD2/3 (p-SMAD2/3)[ 18 ]. Since commercial antibody for phosphorylated ACVR1C is not available, detection of p-SMAD2/3 serves as an effective proxy to reflect ACVR1C-SMAD2/3 activation. To confirm the effect of secretory GREM1 on the ACVR1C-SMAD2/3 pathway and EMT in CRC cells, we performed immunoblotting and found that GREM1-CM significantly increased p-SMAD2/3 levels in HCT116 and SW480 cells, which was effectively blocked by a GREM1 BAb (Fig. 3 B). Moreover, immunoblotting and RT–qPCR analyses also revealed that GREM1-CM induced significant downregulation of E-CADHERIN (E-CAD, encoded by CDH1 )[ 25 ] and upregulation of mesenchymal markers, including SNAIL (encoded by SNAI1 ), ZEB1, and β-CAT[ 22 , 26 ] in SW480 and HCT116 cells. GREM1 BAb effectively blocked GREM1-CM-induced EMT activation (Figs. 3 B and S4D, E). Considering that EMT serves as an effective mechanism through which tumor cells acquire stroma-like traits to promote invasion and metastasis, we tested whether blocking GREM1 could inhibit the invasive and migratory capacity of CRC cells. Indeed, in vitro scratch and transwell assays showed that GREM1-CM significantly enhanced migration and invasion abilities of HCT116 and SW480 cells. Remarkably, these effects were abolished by GREM1 BAb treatment (Figures S4 F–I). These findings suggest that secretory GREM1 activates the ACVR1C-SMAD2/3 pathway and promotes EMT and subsequent cellular behavior, such as migration and invasion of CRC cells. The TGFβ superfamily signals through BMPRs, ACVRs, and TGFβRs. Although both ACVRs and TGFβRs converge on the SMAD2/3 axis[ 18 ], and GREM1-CM robustly activated TGFβ superfamily signaling in HEK293 cells (Figure S4 J), our co-IP analysis showed no direct interaction between GREM1 and TGFβR1 (Fig. 2 C). To determine whether GREM1 activates SMAD2/3 via TGFβR1 or ACVR1C, we treated SW480 and HCT116 cells with increasing concentrations of recombinant human GREM1 (rhGREM1). This led to a dose-dependent increase in p-SMAD2/3 (Fig. 3 C), suggesting activation of a SMAD2/3-coupled receptor. To exclude the possibility that GREM1 indirectly stimulates TGFβ signaling, we examined whether rhGREM1 alters the expression of TGFβ or TGFβR1 in CRC cells. No changes were observed across all doses tested (Fig. 3 C), indicating that GREM1 does not upregulate endogenous TGFβ signaling components. Together, these data suggest that GREM1-induced SMAD2/3 activation is unlikely to be mediated by TGFβR1 and instead proceeds via the ACVR1C pathway (Fig. 3 D). Considering that GREM1 is a canonical antagonist of BMP and that BMP receptors (BMPRs) exert their function through the phosphorylation of SMAD1/5/9 (p-SMAD1/5/9), we sought to investigate whether GREM1 regulates EMT via BMPR superfamily pathways. As expected, rhGREM1 suppressed the p-SMAD1/5/9 (Fig. 3 C), in keeping with the canonical role of GREM1 as a BMP inhibitor. However, rescue of BMP signaling using the specific agonist sb4, which exclusively increases p-SMAD1/5/9 levels without affecting p-SMAD2/3[ 27 ], failed to reverse GREM1-induced EMT marker changes (Fig. 3 E). This definitive exclusion of BMPR-SMAD1/5/9 involvement establishes that GREM1 promotes EMT independently of its classical role as a BMP inhibitor. Subsequently, to examine whether GREM1 promotes EMT through activation of the ACVR1C-SMAD2/3 pathway, we either stably knocked down ACVR1C (shACVR1C) or inhibited SMAD2/3 phosphorylation using SB505124[ 28 ]. Our immunoblotting and RT–qPCR analysis revealed that shACVR1C or SB505124 significantly blocked GREM1-CM-induced changes in EMT marker expression ( i.e. E-CAD, ZEB1, β-CAT and SNAIL), indicating that ACVR1C-SMAD2/3 activation is required for GREM1-driven EMT in SW480 and HCT116 CRC cells (Fig. 3 F, g and S5A, B and S6A, B). In addition, GREM1-CM-induced invasion and migration of these CRC cells was abolished by shACVR1C or SB505124 (Figures S5 C–G and S6C–F). Collectively, these findings demonstrate that GREM1 induces EMT, as well as subsequent migration and invasion, by activating the ACVR1C-SMAD2/3 pathway. p-SMAD2/3 form a complex with SMAD4 that translocates into the nucleus and acts as a transcriptional regulator[ 29 ]. SNAI1 has been identified as a transcriptional target of the SMAD2/3/4 complex in several cancers, including CRC[ 30 – 32 ]. To define the direct binding sites involved in this regulation in CRC cells, we queried the JASPAR database ( http://jaspar.genereg.net/ ) and identified three candidate SMAD2/3/4 binding sites (–967, − 787, and − 186) within the SNAI1 promoter (Figure S6G). Chromatin immunoprecipitation (ChIP) followed by qPCR confirmed specific binding of SMAD2/3/4 to the − 787 site. Importantly, inhibition of ACVR1C with SB505124 significantly decreased this binding (Figure S6H). Overall, our data reveal that secretory GREM1 is a specific functional ligand that activates the ACVR1C–SMAD2/3–SNAIL signaling axis, thereby promoting EMT, invasion and migration of CRC cells. Exogenous GREM1 induction of endogenous GREM1 transcription reinforces EMT in CRC cells via the ACVR1C-SMAD2/3 pathway Our data above show that there is a marked increase in the proportion of GREM1 + epithelial tumor cells in stage IV CRC (Figs. 1 C and S2E). To determine whether exogenous GREM1 can regulate endogenous GREM1 expression, we treated SW480 cells with increasing concentrations of rhGREM1. RT-qPCR and immunoblotting analyses revealed a dose-dependent upregulation of tumor GREM1 expression in response to rhGREM1 stimulation (Figs. 4 A–C). However, the mechanism by which exogenous GREM1 triggers endogenous GREM1 transcription in advanced CRC cells remains unclear. Notably, we also observed that ACVR1C expression was markedly upregulated in stage IV CRC (Figures S3 D, E), coinciding with the emergence of epithelial GREM1 expression, whereas stromal GREM1 was already abundantly present in both stage III and IV tumors (Figs. 1 B, C and S2D, E). This spatiotemporal concordance suggests that epithelial GREM1 induction may depend on elevated ACVR1C expression. To investigate the role of the ACVR1C-SMAD2/3 pathway in the exogenous GREM1-mediated tumor GREM1 transcription, we either overexpressed ACVR1C or inhibited the pathway using SB505124 in SW480 cells following rhGREM1 treatment. Interestingly, ACVR1C overexpression further enhanced endogenous GREM1 expression, whereas SB505124 treatment completely reversed the effect of rhGREM1 (Fig. 4 D). To investigate whether SMAD2/3/4 act as transcription factors for GREM1 , we utilized the JASPAR database and identified five candidate SMAD2/3/4 binding sites (-733, -612, -446, -316, and − 3) in the GREM1 promoter region (Fig. 4 E). To validate the predicted results, we performed a ChIP–qPCR analysis, which revealed that SMAD2/3/4 bound the GREM1 promoter at the − 733 and − 612 sites. Notably, SB505124 treatment significantly reduced this binding (Fig. 4 F). These data demonstrate that exogenous GREM1 efficiently induces endogenous GREM1 transcription in CRC cells via the ACVR1C-SMAD2/3 signaling pathway. Having established that exogenous GREM1 promotes EMT in CRC cells, we sought to determine whether endogenous GREM1 exerts a similar function. In normal epithelial cells, β-CAT is localized at the cellular membrane with the adhesion molecule E-CAD[ 33 ]. However, during tumor progression and the onset of EMT, E-CAD is gradually lost, and a portion of β-CAT translocates to the cytoplasm and nucleus[ 34 , 35 ]. To delineate the bona fide correlation between epithelial GREM1 expression and EMT hallmarks in CRC, we performed immunofluorescence (IF) staining on stage IV CRC clinical samples, which contained both GREM1 + and GREM1 − CRC cells. Remarkably, GREM1 + CRC cells exhibited a significant loss of E-CAD, along with increased β-CAT expression and its translocation to the cytoplasm and nucleus, compared to adjacent GREM1 − counterparts within the same tumor (Figs. 4 G–J). We previously found that forced GREM1 expression (pLV-GREM1) in CRC cells enhanced their EMT and metastatic traits[ 13 ]. Given the typically low expression of GREM1 in CRC cell lines compared to normal and tumoral fibroblasts (Fig. 4 K), we sought to model the therapeutic blockade of endogenous GREM1 in advanced tumors. To this end, we used CRISPR/Cas9 techniques to knock out GREM1 in SW480 and HCT116 cells that stably expressed pLV-GREM1. RT–qPCR analysis revealed that pLV-GREM1 resulted in a significant change in EMT markers, including a marked decrease in CDH1 and an increase in SNAI1, VIM , and ZEB1 , while GREM1 knockout significantly restored the expression of these markers, indicating that endogenous GREM1 promotes EMT within CRC cells (Figs. 4 L, M). Collectively, these data demonstrate that exogenous GREM1 induces endogenous GREM1 transcription in CRC cells through the ACVR1C–SMAD2/3 pathway, establishing a self-propelling loop that promotes EMT. Given our observation of a reduction in GREM1⁺ stromal cells in stage IV CRC compared to stage III (Fig. 1 B), we speculate that this self-sustaining GREM1 feedback loop enables CRC cells to maintain EMT independently of stromal inputs, reducing their reliance on stromal GREM1. This autonomous signaling cascade comprises multiple nodes that may be amenable to therapeutic intervention aimed at halting CRC metastasis. Depletion of Grem1 + stromal cells inhibits EMT and metastasis of CRC in vivo To validate the GREM1-ACVR1C-induced autocrine GREM1 feedback loop, we applied genetic and pharmacological strategies (Figure S6I). First, we evaluated the impact of stromal paracrine GREM1 on CRC in vivo . To block exogenous GREM1, we crossed Grem1-CreER T2 ; Rosa-LSL-DTA mice with APC Min/+ mice[ 36 ], a well-established model for CRC proliferation and EMT that develops intestinal tumors by 10 weeks[ 34 , 37 , 38 ], generating AGD mice for TMX-induced depletion of GREM1 + stromal cells (Fig. 5 A). Intriguingly, ablation of Grem1 + stromal cells for 6 weeks postnatally did not significantly alter the number or size of intestinal tumors in APC Min/+ mice (Figures S7A, B). Subsequently, we delved deeper into whether loss of paracrine GREM1 could restrain the malignant potential of intestinal tumor cells. As expected, IF staining in the GREM1 + stromal cell infiltration zone of the APC Min/+ intestines revealed a marked loss of E-cad in tumor cells (Fig. 5 B, upper left panels), accompanied by increased β-cat expression and its translocation to the cytoplasm and nucleus (Fig. 5 B, upper right panels). In contrast, upon depletion of GREM1 + stromal cells in AGD mice, E-cad expression was significantly elevated (Fig. 5 C), with intense and continuous localization along the tumor cell membrane (Fig. 5 B, bottom left). In parallel, β-cat staining was reduced and restricted to the cell membrane (Fig. 5 D, E), co-localizing with E-cad, and rarely observed in the cytoplasm or nucleus (Fig. 5 B, bottom right panels). These results suggest a critical role for exogenous Grem1 in orchestrating malignant cell behaviors. Next, to evaluate the impact of paracrine Grem1 on CRC metastasis, we injected luciferase-labeled murine rectal cancer cells (MC38-luc) into Grem1 + cell-depleted (GD) or control mice (Figures S7C, D). Cells were administered via the tail vein to induce lung metastasis, or into the spleen or cecum wall to induce liver metastasis[ 39 ] (Figures S7E–G). Strikingly, we observed a significant reduction in lung and liver metastases of CRC cells in GD mice compared with Grem1-CreER T2 or Rosa-LSL-DTA controls (Figs. 5 F–I and S7H–J), suggesting that the stromal factor Grem1 is vital for CRC metastasis. Inhibition of ACVR1C-SMAD2/3 pathway inhibits EMT and metastasis of CRC in vivo To determine the role of ACVR1C and its downstream SMAD2/3 pathway in GREM1-mediated EMT in vivo , we first pre-treated HCT116 cells carrying a luciferase reporter (HCT116-luc), stably expressing either shACVR1C or scramble shRNA, with GREM1-CM, followed by subcutaneous transplantation into nude mice. Notably, ACVR1C knockdown significantly inhibited subcutaneous tumor growth (Figs. 6 A–C), accompanied by a significant increase in epithelial gene expression (e.g. CDH1 ) and a marked decrease in mesenchymal gene expression (e.g. SNAI1, CTNNB1 , and ZEB1 ), as shown by RT–qPCR analysis (Fig. 6 D). These findings were further corroborated by IF staining of EMT markers, including E-CAD and SNAIL (Figs. 6 E–G). Meanwhile, we also harnessed SB505124 as a pharmacological alternative to knock down ACVR1C. Consistent with the genetic approach, SB505124 treatment yielded similar effects on local tumor progression, further substantiating the contextual role of the ACVR1C-SMAD2/3 pathway (Figures S8A–G). Importantly, to investigate the impact of the ACVR1C-SMAD2/3 axis on metastasis, we injected GREM1-CM-pre-treated HCT116-luc cells, stably expressing either shACVR1C or scramble shRNA, into the tail vein of nude mice. Remarkably, lung metastasis was profoundly suppressed in the ACVR1C knockdown group compared with controls (Figs. 6 H, I). In parallel, we inoculated GREM1-CM-pre-treated HCT116-luc cells into the cecum wall of NOG (NOD/Shi-scid/IL-2Rγ) mice, a new generation of severely immunodeficient mice[ 40 ]. We found that SB505124 treatment resulted in a significant reduction in liver metastasis (Figs. 5 I and S8 H, I). Collectively, these findings establish the ACVR1C-SMAD2/3 axis as a critical effector of stroma-derived GREM1, and reveal that its inhibition provides an effective strategy to counteract GREM1-induced EMT and metastasis in CRC in vivo . Epithelial GREM1 enhances EMT and metastasis of CRC in vivo Further, to examine the contribution of tumor-autocrine GREM1 to EMT and metastasis of CRC in vivo , we first inoculated HCT116 cells carrying pLV-GREM1 into nude mice, which led to significantly enhanced subcutaneous tumor growth (Figures S9A–C). Subsequent RT–qPCR analysis demonstrated that autocrine GREM1 upregulated SNAI1, VIM , and ZEB1 while downregulating CDH1 (Figure S9D). These findings were confirmed by IF staining for EMT markers, including E-CAD and SNAIL (Figures S9E–G), supporting that autocrine GREM1 enhances the EMT process in CRC cells. Next, to confirm the role of tumoral GREM1 in enhancing CRC metastasis, we conducted metastasis assays by injecting HCT116-luc cells expressing either pLV-GREM1 or the control vector via the tail vein or into the cecum wall. We found that overexpression of tumoral GREM1 resulted in a significant increase in lung and liver metastasis compared with controls (Figs. 7 A, B and S9H, I). In summary, these findings demonstrate that tumoral GREM1 promotes both EMT and metastasis of CRC cells in vivo . Targeting the GREM1-ACVR1C interaction interface to inhibit metastasis of CRC in vivo In clinical research, the molecular complexity and shared signaling of CRC limit the effective targeted therapy. While the TGFβ superfamily is a promising therapeutic axis[ 41 ], its intertwined and overarching roles in development and physiology make selective targeting difficult without systemic toxicity[ 42 , 43 ]. Targeting protein–protein interaction (PPI) interfaces is thought to further enhance specificity and reduce off-target effects[ 44 ]. Based on the principle of GREM1-ACVR1C interface disruption, we designed a peptide inhibitor derived from amino acid residues 84–102 (AA84-102) of ACVR1C (hereafter referred to as the ACVR1C peptide) (Fig. 7 C). We further examined the binding affinity between GREM1 and the ACVR1C peptide using MST. We found that GREM1 exhibited an affinity for the ACVR1C peptide comparable to that for the ACVR1C-ECD ( K d = 92.30 ± 7.51 nM) (Fig. 7 D). Pull-down assays showed that the introduction of ACVR1C peptide significantly blocked the GREM1-ACVR1C binding (Fig. 7 E), suggesting that the peptide disrupts the GREM1-ACVR1C interaction by potently and competitively binding to GREM1. Further, we employed a spleen-to-liver metastasis model to evaluate the functional effect of the ACVR1C peptide in CRC metastasis. Notably, administration of the ACVR1C peptide effectively suppressed the increase in liver metastasis induced by tumor-specific GREM1 overexpression (Figs. 7 F, G), suggesting that our ACVR1C peptide significantly attenuated CRC progression by blocking the metastasis-promoting effects of tumor-autonomous autocrine GREM1-ACVR1C signaling, laying the foundation for the development of a novel class of peptide-based targeted therapies in CRC. Discussion Whether tumor cells can internalize stromal paracrine signals and convert them into autocrine loops during metastasis has remained unclear. Here, we show that in advanced colorectal cancer, GREM1 undergoes a stromal-to-epithelial shunt and binds ACVR1C to activate SMAD2/3, inducing both SNAI1 and GREM1 expression. This establishes a self-amplifying autocrine circuit that drives EMT, conferring signaling autonomy and promoting metastasis (Fig. 7 H). A functional peptide we designed for targeting the GREM1–ACVR1C interface effectively disrupts this loop and demonstrates therapeutic potential. During embryonic development and tissue repair, intercellular communication is precisely regulated by paracrine and autocrine signaling pathways[ 45 , 46 ]. Similarly, tumor cells exploit these dual signaling modes to enhance their survival and invasive capacity[ 47 ]. Moreover, during dissemination to distant sites, metastatic tumor cells have been observed to bring along components of the primary tumor stroma, including CAFs, reflecting their persistent reliance on microenvironmental support[ 48 , 49 ]. Sporn et al. first proposed that a potential mechanism of malignant transformation is the autocrine production of growth factors to which the cell itself can respond, a process that may originate from the reactivation of autocrine strategies employed during early embryonic development, enabling cells to survive even in the absence of external support[ 50 , 51 ]. Weinberg et al. further suggested that paracrine signals from the stroma might trigger the emergence of signaling autonomy in tumor cells. While neuron-secreted NLGN3 has been reported to upregulate NLGN3 expression in glioma cells, it has not been demonstrated whether this process evolves into a self-sustaining autocrine loop [ 52 ]. Thus, whether tumor cells can transition from paracrine induction to a self-sustaining autocrine circuit, remains unresolved. In this study, we show that during CRC metastasis, GREM1 expression undergoes a stromal-to-epithelial shunt, establishing an autocrine signaling mode initiated by paracrine cues. This transition suggests that metastatic cells acquire signaling autonomy by reactivating developmental autocrine programs, enabling them to survive and disseminate in distant organs with reduced microenvironmental support. Such a shift from “dependence on soil” to “self-construction of soil” reflects the remarkable adaptability of tumor cells. As a canonical BMP antagonist in the TGF-β superfamily, GREM1 plays a pivotal regulatory role across diverse physiological and pathological contexts. During embryogenesis, GREM1 modulates BMP gradients to orchestrate the development of organs such as the kidney[ 53 ], skeleton, and gut[ 19 ]. In the post-developmental setting, GREM1-mediated BMP signaling also plays an important role in benign conditions such as tissue repair[ 54 ], and fibrosis[ 55 ]. In the context of cancer, GREM1 displays marked tissue-specific functionality. In hereditary mixed polyposis syndrome (HMPS), duplication of an upstream enhancer leads to aberrant overexpression of GREM1 in epithelial cells, disrupting the BMP-driven stemness gradient and initiating polyp formation[ 12 , 56 ]. GREM1 can also enhance stemness and drive cancer progression by suppressing BMP signaling[ 6 ]. In contrast, in pancreatic ductal adenocarcinoma, GREM1 expression constrains EMT and promotes differentiation toward a less invasive epithelial phenotype[ 10 ]. While traditionally categorized as a BMP antagonist, portraying GREM1 solely as an inhibitor overlooks its diverse, sometimes paradoxical roles in various biological settings[ 57 ]. Accumulating evidence suggests that GREM1 possesses non-canonical, BMP-independent activities, engaging multiple receptors and activating diverse downstream signaling pathways. Previous studies have implicated EGFR[ 58 ] and VEGFR2[ 16 ] as putative GREM1 targets, though their binding affinities have not been determined. Zhu et al. reported that GREM1 binds to FGFR1 with a dissociation constant of K d = 10.6 nM[ 11 ]. In this study, we further identified ACVR1C[ 59 ] as a novel binding receptor for GREM1, with a dissociation constant of K d = 67.67 ± 10.35 nM. Notably, this binding affinity is markedly higher than that of the canonical ACVR1C ligand Activin B ( K d = 854.1 ± 127.47 nM), representing a 12.6-fold increase. These results suggest that GREM1 engages ACVR1C with superior affinity, indicating its potential to act as a dominant ligand within this signaling pathway. Unlike the mechanism described by Lan et al. in pancreatic cancer[ 10 ], where GREM1 inhibits EMT via classical BMP antagonism, our findings demonstrate that GREM1 binds ACVR1C and activates the SMAD2/3 axis independently of BMP signaling. This interaction initiates a positive autocrine feedback loop that promotes EMT and metastasis in CRC. These contrasting roles highlight the context-dependent plasticity of GREM1, which engage distinct signaling programs across tissue types to drive divergent outcomes. Although GREM1 is not traditionally classified as a cytokine, emerging evidence suggests it acts as an adipokine in adipose tissue, modulating metabolic homeostasis[ 60 ]. Our study extends this concept by showing that GREM1 exhibits key cytokine characteristics, including active secretion, receptor engagement, downstream signaling activation, and autocrine feedback amplification. Reframing GREM1 as a cytokine beyond BMP antagonism may provide new insights into its pleiotropic roles in cancer, metabolic disorders, and developmental or regenerative processes. The advent of cancer genomics and precision medicine has enabled targeted therapies for advanced CRC, with agents against VEGF, EGFR, BRAF V600E, and HER2 showing efficacy in selected patient subsets. However, issues like limited bioavailability, resistance, and narrow indications restrict their broader use[ 1 ]. Unlike mutation-specific targets, GREM1 is a widely expressed cytokine with sustained, context-dependent activity, making it a promising and potentially broadly applicable therapeutic target in CRC. Therapeutic targeting of GREM1 to date has focused largely on full neutralization strategies using monoclonal antibodies. Fully humanized monoclonal antibodies such as Ginisortamab and TST003 have advanced into phase I/II clinical trials[ 61 , 62 ]. Although these antibodies have shown efficacy in preclinical models, the indispensable roles of GREM1 in physiological processes such as intestinal homeostasis and bone marrow hematopoiesis raise the risk of adverse effects from systemic inhibition, posing a major challenge to their clinical translation[ 54 , 63 ]. In recent years, PPI have emerged as attractive therapeutic targets, with small peptides showing particular promise due to their high binding affinity and specificity[ 64 ]. In this study, we exploited the structural interface of the GREM1–ACVR1C interaction as a therapeutic entry point. Through structure-guided molecular design, we developed a high-affinity peptide capable of specifically disrupting this interaction, thereby markedly suppressing metastatic potential. Unlike conventional antibodies, our approach selectively targets the pathogenic GREM1–ACVR1C axis, highlighting the translational promise of rational PPI-targeted cancer therapies. In summary, our findings reveal that the GREM1–ACVR1C axis acts as a key mediator of the stromal-to-epithelial shunt and the establishment of autonomous GREM1 signaling in CRC. By uncovering this pathway, we not only deepen our understanding of tumor self-sufficiency mechanisms, but also identify a functionally precise intervention point that opens new avenues for disrupting metastatic competence through targeted dismantling of self-reinforcing oncogenic circuits. Declarations Acknowledgements We thank Ruohan Li and Changxue Li for assisting with IHC data scoring; Guihua Wang for providing the APC Min/+ mouse line; Pengcheng Bu for assisting with the construction of the cecum to liver metastasis mouse model; Lei Zhou for drawing the GREM1-ACVR1C protein docking; and Xingqiao Xie from the Sample Preparation and Analysis Core Facility of Shenzhen Medical Academy of Research and Translation (SMART) for technical support of MST. Author contributions H.Z. and N.L. conceived and designed the study and wrote the manuscript. H.Z. performed most of the experiments and analyses. Q.J. and Y.J. helped with RT-qPCR analysis and clinical tissue IF experiments. Z.F. and Y.Y. helped with subcutaneous tumor IF experiments. Y.Y. assisted with the design and drawing of schematics and mechanistic diagrams. Y.G. assisted with histopathological assessment. N.W. and B.Z. performed ACVR1C-ECD and ACVR1C-ECD-doble-mutant proteins expression and purification. J.L. helped with single-cell RNA-seq data analysis. Z.Z. helped with the establishment of the cecum to liver metastasis mouse model. L.G., Y.Z. (Affiliation 1), Y.H., Y.Z. (Affiliation 7), and J.Z. provided intellectual feedback. Note: Y.Z. (Affiliation 1) and Y.Z. (Affiliation 7) are distinct individuals. S.L. and L.F. assisted with data interpretation and analysis and text proofreading. X.W. helped with CHIP experiments. T.W. helped with RNA-seq data analysis. X.S., T.W., Y.J. and N.L. supervised the study. Funding This work was supported by grants from the National Natural Science Foundation of China (81874176, 82072766, 32471261), Guangdong Provincial Key Laboratory of Digestive Cancer Research (2021B1212040006), the Sanming Project of Medicine in Shenzhen (SZSM202111005), Funding of Shenzhen Clinical Research Center for Gastroenterology (Gastrointestinal Surgery) (LCYSSQ20220823091203008), Guangdong Medical Science and Technology Research Foundation (A2022068). Ethics approval and consent to participate The use of human CRC tissues was approved by the Ethics Committee of Sun Yat-sen University (Approval No. KY-2023-071-01). Informed consent was obtained from all participants, and the study adhered to the principles of the 1975 Declaration of Helsinki. All animal experiments were conducted in accordance with institutional guidelines and approved by the Animal Ethics Committee of Sun Yat-sen University (Approval No. SYSU-IACUC-2024-001631). Consent for publication Consent was obtained from each patient. Competing interests Authors declare that they have no competing interests. References Ciardiello F, Ciardiello D, Martini G, Napolitano S, Tabernero J, Cervantes A: Clinical management of metastatic colorectal cancer in the era of precision medicine. CA Cancer J Clin 2022, 72: 372-401. de Visser KE, Joyce JA: The evolving tumor microenvironment: From cancer initiation to metastatic outgrowth. Cancer Cell 2023, 41: 374-403. Lambert AW, Zhang Y, Weinberg RA: Cell-intrinsic and microenvironmental determinants of metastatic colonization. Nat Cell Biol 2024, 26: 687-697. Hanahan D, Weinberg RA: The hallmarks of cancer. Cell 2000, 100: 57-70. Cords L, de Souza N, Bodenmiller B: Classifying cancer-associated fibroblasts-The good, the bad, and the target. Cancer Cell 2024, 42: 1480-1485. Kobayashi H, Gieniec KA, Wright JA, Wang T, Asai N, Mizutani Y, Lida T, Ando R, Suzuki N, Lannagan TRM, et al: The Balance of Stromal BMP Signaling Mediated by GREM1 and ISLR Drives Colorectal Carcinogenesis. Gastroenterology 2021, 160: 1224-1239 e1230. Li T, Fei Y, Sun X, Zhu J, Fang X, Meng F, Wang D, Zhang X, Liu C, Du Q: Over-expressed Phospholamban in Cancer-associated Fibroblasts Is a Driver for CMS4 Subtype Colorectal Cancer Formation. Cell Mol Gastroenterol Hepatol 2025, 19: 101524. Jiang Y, Fu Z, Chen Y, Jin Q, Yang Y, Lin Z, Li C, Gao Y, Dong Z, He Y, et al: Mapping and tracing Grem1(+) stromal cells in an Apc (Min/+) mouse utilizing cryopreserved intestinal sections prepared via modified Swiss-roll technique. iScience 2024, 27: 111173. Ren J, Smid M, Iaria J, Salvatori DCF, van Dam H, Zhu HJ, Martens JWM, Ten Dijke P: Cancer-associated fibroblast-derived Gremlin 1 promotes breast cancer progression. Breast Cancer Res 2019, 21: 109. Lan L, Evan T, Li H, Hussain A, Ruiz EJ, Zaw Thin M, Ferreira RMM, Ps H, Riising EM, Zen Y, et al: GREM1 is required to maintain cellular heterogeneity in pancreatic cancer. Nature 2022, 607: 163-168. Cheng C, Wang J, Xu P, Zhang K, Xin Z, Zhao H, Ji Z, Zhang M, Wang D, He Y, et al: Gremlin1 is a therapeutically targetable FGFR1 ligand that regulates lineage plasticity and castration resistance in prostate cancer. Nat Cancer 2022, 3: 565-580. Jaeger E, Leedham S, Lewis A, Segditsas S, Becker M, Cuadrado PR, Davis H, Kaur K, Heinimann K, Howarth K, et al: Hereditary mixed polyposis syndrome is caused by a 40-kb upstream duplication that leads to increased and ectopic expression of the BMP antagonist GREM1. Nat Genet 2012, 44: 699-703. Li R, Zhou H, Li M, Mai Q, Fu Z, Jiang Y, Li C, Gao Y, Fan Y, Wu K, et al: Gremlin-1 Promotes Colorectal Cancer Cell Metastasis by Activating ATF6 and Inhibiting ATF4 Pathways. Cells 2022, 11 . Liu Y, Li Y, Hou R, Shu Z: Knockdown GREM1 suppresses cell growth, angiogenesis, and epithelial-mesenchymal transition in colon cancer. J Cell Biochem 2019, 120: 5583-5596. Brazil DP, Church RH, Surae S, Godson C, Martin F: BMP signalling: agony and antagony in the family. Trends Cell Biol 2015, 25: 249-264. Mitola S, Ravelli C, Moroni E, Salvi V, Leali D, Ballmer-Hofer K, Zammataro L, Presta M: Gremlin is a novel agonist of the major proangiogenic receptor VEGFR2. Blood 2010, 116: 3677-3680. Park SA, Sung NJ, Choi BJ, Kim W, Kim SH, Surh YJ: Gremlin-1 augments the oestrogen-related receptor alpha signalling through EGFR activation: implications for the progression of breast cancer. Br J Cancer 2020, 123: 988-999. Derynck R, Zhang YE: Smad-dependent and Smad-independent pathways in TGF-beta family signalling. Nature 2003, 425: 577-584. Worthley DL, Churchill M, Compton JT, Tailor Y, Rao M, Si Y, Levin D, Schwartz MG, Uygur A, Hayakawa Y, et al: Gremlin 1 identifies a skeletal stem cell with bone, cartilage, and reticular stromal potential. Cell 2015, 160: 269-284. Kalluri R, Zeisberg M: Fibroblasts in cancer. Nat Rev Cancer 2006, 6: 392-401. Pastushenko I, Brisebarre A, Sifrim A, Fioramonti M, Revenco T, Boumahdi S, Van Keymeulen A, Brown D, Moers V, Lemaire S, et al: Identification of the tumour transition states occurring during EMT. Nature 2018, 556: 463-468. Zhao H, Ming T, Tang S, Ren S, Yang H, Liu M, Tao Q, Xu H: Wnt signaling in colorectal cancer: pathogenic role and therapeutic target. Mol Cancer 2022, 21: 144. Tsuchida K, Nakatani M, Yamakawa N, Hashimoto O, Hasegawa Y, Sugino H: Activin isoforms signal through type I receptor serine/threonine kinase ALK7. Mol Cell Endocrinol 2004, 220: 59-65. Kisonaite M, Wang X, Hyvonen M: Structure of Gremlin-1 and analysis of its interaction with BMP-2. Biochem J 2016, 473: 1593-1604. Dongre A, Weinberg RA: New insights into the mechanisms of epithelial-mesenchymal transition and implications for cancer. Nat Rev Mol Cell Biol 2019, 20: 69-84. Stemmler MP, Eccles RL, Brabletz S, Brabletz T: Non-redundant functions of EMT transcription factors. Nat Cell Biol 2019, 21: 102-112. Bradford STJ, Ranghini EJ, Grimley E, Lee PH, Dressler GR: High-throughput screens for agonists of bone morphogenetic protein (BMP) signaling identify potent benzoxazole compounds. J Biol Chem 2019, 294: 3125-3136. DaCosta Byfield S, Major C, Laping NJ, Roberts AB: SB-505124 is a selective inhibitor of transforming growth factor-beta type I receptors ALK4, ALK5, and ALK7. Mol Pharmacol 2004, 65: 744-752. Shi Y, Massagué J: Mechanisms of TGF-beta signaling from cell membrane to the nucleus. Cell 2003, 113: 685-700. Asnaghi L, White DT, Key N, Choi J, Mahale A, Alkatan H, Edward DP, Elkhamary SM, Al-Mesfer S, Maktabi A, et al: ACVR1C/SMAD2 signaling promotes invasion and growth in retinoblastoma. Oncogene 2019, 38: 2056-2075. Yang S, Zhang H, Yang H, Zhang J, Wang J, Luo T, Jiang Y, Hua H: SEPHS1 promotes SMAD2/3/4 expression and hepatocellular carcinoma cells invasion. Exp Hematol Oncol 2021, 10: 17. Cai J, Xia L, Li J, Ni S, Song H, Wu X: Tumor-Associated Macrophages Derived TGF-beta ‒Induced Epithelial to Mesenchymal Transition in Colorectal Cancer Cells through Smad2,3-4/Snail Signaling Pathway. Cancer Res Treat 2019, 51: 252-266. Clevers H: Wnt/beta-catenin signaling in development and disease. Cell 2006, 127: 469-480. Sánchez-Tilló E, de Barrios O, Siles L, Cuatrecasas M, Castells A, Postigo A: β-catenin/TCF4 complex induces the epithelial-to-mesenchymal transition (EMT)-activator ZEB1 to regulate tumor invasiveness. Proc Natl Acad Sci U S A 2011, 108: 19204-19209. Inomata M, Ochiai A, Akimoto S, Kitano S, Hirohashi S: Alteration of beta-catenin expression in colonic epithelial cells of familial adenomatous polyposis patients. Cancer Res 1996, 56: 2213-2217. Moser AR, Pitot HC, Dove WF: A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science 1990, 247: 322-324. Ren J, Sui H, Fang F, Li Q, Li B: The application of Apc(Min/+) mouse model in colorectal tumor researches. J Cancer Res Clin Oncol 2019, 145: 1111-1122. Babaei-Jadidi R, Li N, Saadeddin A, Spencer-Dene B, Jandke A, Muhammad B, Ibrahim EE, Muraleedharan R, Abuzinadah M, Davis H, et al: FBXW7 influences murine intestinal homeostasis and cancer, targeting Notch, Jun, and DEK for degradation. J Exp Med 2011, 208: 295-312. Zhang L, Zhu Z, Yan H, Wang W, Wu Z, Zhang F, Zhang Q, Shi G, Du J, Cai H, et al: Creatine promotes cancer metastasis through activation of Smad2/3. Cell Metab 2021, 33: 1111-1123.e1114. Ito M, Hiramatsu H, Kobayashi K, Suzue K, Kawahata M, Hioki K, Ueyama Y, Koyanagi Y, Sugamura K, Tsuji K, et al: NOD/SCID/gamma(c)(null) mouse: an excellent recipient mouse model for engraftment of human cells. Blood 2002, 100: 3175-3182. Jung B, Staudacher JJ, Beauchamp D: Transforming Growth Factor β Superfamily Signaling in Development of Colorectal Cancer. Gastroenterology 2017, 152: 36-52. Liu S, Ren J, Ten Dijke P: Targeting TGF β signal transduction for cancer therapy. Signal Transduct Target Ther 2021, 6: 8. Peng D, Fu M, Wang M, Wei Y, Wei X: Targeting TGF- β signal transduction for fibrosis and cancer therapy. Mol Cancer 2022, 21: 104. Modell AE, Blosser SL, Arora PS: Systematic Targeting of Protein-Protein Interactions. Trends Pharmacol Sci 2016, 37: 702-713. Wang Y, Bikle DD, Chang W: Autocrine and Paracrine Actions of IGF-I Signaling in Skeletal Development. Bone Res 2013, 1: 249-259. Liu S, Hur YH, Cai X, Cong Q, Yang Y, Xu C, Bilate AM, Gonzales KAU, Parigi SM, Cowley CJ, et al: A tissue injury sensing and repair pathway distinct from host pathogen defense. Cell 2023, 186: 2127-2143 e2122. Basu R, Boguszewski CL, Kopchick JJ: Growth Hormone Action as a Target in Cancer: Significance, Mechanisms, and Possible Therapies. Endocr Rev 2025, 46: 224-280. Ao Z, Shah SH, Machlin LM, Parajuli R, Miller PC, Rawal S, Williams AJ, Cote RJ, Lippman ME, Datar RH, El-Ashry D: Identification of Cancer-Associated Fibroblasts in Circulating Blood from Patients with Metastatic Breast Cancer. Cancer Res 2015, 75: 4681-4687. Hurtado P, Martinez-Pena I, Pineiro R: Dangerous Liaisons: Circulating Tumor Cells (CTCs) and Cancer-Associated Fibroblasts (CAFs). Cancers (Basel) 2020, 12 . Sporn MB, Todaro GJ: Autocrine secretion and malignant transformation of cells. N Engl J Med 1980, 303: 878-880. Wydooghe E, Vandaele L, Heras S, De Sutter P, Deforce D, Peelman L, De Schauwer C, Van Soom A: Autocrine embryotropins revisited: how do embryos communicate with each other in vitro when cultured in groups? Biol Rev Camb Philos Soc 2017, 92: 505-520. Venkatesh HS, Johung TB, Caretti V, Noll A, Tang Y, Nagaraja S, Gibson EM, Mount CW, Polepalli J, Mitra SS, et al: Neuronal Activity Promotes Glioma Growth through Neuroligin-3 Secretion. Cell 2015, 161: 803-816. Michos O, Panman L, Vintersten K, Beier K, Zeller R, Zuniga A: Gremlin-mediated BMP antagonism induces the epithelial-mesenchymal feedback signaling controlling metanephric kidney and limb organogenesis. Development 2004, 131: 3401-3410. Koppens MAJ, Davis H, Valbuena GN, Mulholland EJ, Nasreddin N, Colombe M, Antanaviciute A, Biswas S, Friedrich M, Lee L, et al: Bone Morphogenetic Protein Pathway Antagonism by Grem1 Regulates Epithelial Cell Fate in Intestinal Regeneration. Gastroenterology 2021, 161: 239-254.e239. Baboota RK, Rawshani A, Bonnet L, Li X, Yang H, Mardinoglu A, Tchkonia T, Kirkland JL, Hoffmann A, Dietrich A, et al: BMP4 and Gremlin 1 regulate hepatic cell senescence during clinical progression of NAFLD/NASH. Nat Metab 2022, 4: 1007-1021. Mulholland EJ, Belnoue-Davis HL, Valbuena GN, Gunduz N, Ligeza A, Lin M, Biswas S, Gil Vasquez E, Omwenga S, Nasreddin N, et al: Epithelial GREMLIN1 disrupts intestinal epithelial-mesenchymal crosstalk to induce a wnt-dependent ectopic stem cell niche through stromal remodelling. Nat Commun 2025, 16: 5167. Gao Z, Houthuijzen JM, Ten Dijke P, Brazil DP: GREM1 signaling in cancer: tumor promotor and suppressor? J Cell Commun Signal 2023, 17: 1517-1526. Park SA, Sung NJ, Choi BJ, Kim W, Kim SH, Surh YJ: Gremlin-1 augments the oestrogen-related receptor α signalling through EGFR activation: implications for the progression of breast cancer. Br J Cancer 2020, 123: 988-999. Tsuchida K, Sawchenko PE, Nishikawa S, Vale WW: Molecular cloning of a novel type I receptor serine/threonine kinase for the TGF beta superfamily from rat brain. Mol Cell Neurosci 1996, 7: 467-478. Hedjazifar S, Khatib Shahidi R, Hammarstedt A, Bonnet L, Church C, Boucher J, Blüher M, Smith U: The Novel Adipokine Gremlin 1 Antagonizes Insulin Action and Is Increased in Type 2 Diabetes and NAFLD/NASH. Diabetes 2020, 69: 331-341. Sarker D, Blagden S, Cook N, Evans TR, Plummer R, Angelakas A, Bashir S, Badri SE, Mavroeidis L, Morton C, et al: Abstract CT094: Phase I dose-escalation study evaluating the safety an d tolerability of ginisortamab (UCB6114), a first-in-class anti-gremli n-1 monoclonal antibody (mAb), as monotherapy in advanced solid tumors. Cancer Research , 83: CT094-CT094. Rivera IR, Shen L, Kummar S, Barve M, Germa C, Qi C, Chen L, Milata J, Yao J, Shen L, Zhu X: Abstract 5977: A first-in-human, open-label, multi-center phase 1 stud y of TST003, a GREM 1 inhibitor, in subjects with locally advanced or metastatic solid tumors. Cancer Research , 84: 5977-5977. Rowan SC, Jahns H, Mthunzi L, Piouceau L, Cornwell J, Doody R, Frohlich S, Callanan JJ, McLoughlin P: Gremlin 1 depletion in vivo causes severe enteropathy and bone marrow failure. J Pathol 2020, 251: 117-122. Lu H, Zhou Q, He J, Jiang Z, Peng C, Tong R, Shi J: Recent advances in the development of protein-protein interactions modulators: mechanisms and clinical trials. Signal Transduct Target Ther 2020, 5: 213. Gibson DG, Young L, Chuang RY, Venter JC, Hutchison CA, 3rd, Smith HO: Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat Methods 2009, 6: 343-345. Flanagan SP: 'Nude', a new hairless gene with pleiotropic effects in the mouse. Genet Res 1966, 8: 295-309. Muzumdar MD, Tasic B, Miyamichi K, Li L, Luo L: A global double-fluorescent Cre reporter mouse. Genesis 2007, 45: 593-605. Voehringer D, Liang HE, Locksley RM: Homeostasis and effector function of lymphopenia-induced "memory-like" T cells in constitutively T cell-depleted mice. J Immunol 2008, 180: 4742-4753. Yan Y, Zhang D, Zhou P, Li B, Huang SY: HDOCK: a web server for protein-protein and protein-DNA/RNA docking based on a hybrid strategy. Nucleic Acids Res 2017, 45: W365-w373. Methods Cell lines Human colon cancer cell lines (HCT116, SW480, DLD1, LOVO, LS174T, and HT29), mouse colon cancer cell line (MC38) and human embryonic kidney cell lines (HEK293 and its derivative 293T) were purchased from ATCC and tested negative for mycoplasma contamination. Colon cancer cells, HEK293 cells and 293T cells were cultured in RPMI1640 (HyClone, SH30809.01) and DMEM (SH30022.01), respectively, supplemented with 10% fetal bovine serum (Biological Industries, 04-001-1ACS) and 1% Penicillin/Streptomycin (Biological Industries, 03-031-1B) at 37°C, 5% CO 2 . Patients and tissue samples A total of 106 archived human colorectal cancer specimens were obtained from the colorectal cancer database and tissue bank of the First Affiliated Hospital of Sun Yat-sen University (SYSU). These tissues were collected from patients who underwent radical resection for colorectal cancer between 2008 and 2015, and followed up until December 2017. All patients had provided written informed consent, and the use of these samples was approved by the Institutional Review Board of the First Affiliated Hospital, SYSU. In addition, a commercial human CRC tissue microarray containing primary tumor tissues from patients with stage I–III CRC (n = 93) was purchased from Shanghai Outdo Biotech Company (Shanghai, China), with ethics approval documented by the company. Conditioned medium (CM) To investigate the effect of secreted GREM1 on CRC cells, an in vitro GREM1 secretion system was established using HEK293 cells. Stably GREM1-expressing cells (pcDNA3.1-GREM1) and empty-vector transfected cells (pcDNA3.1) were cultured in DMEM/F12 medium with 10% FBS until 40% confluence. After complete removal of the normal culture medium, HEK293-GREM1 and HEK293-Vec cells were continuously cultured in DMEM/F12 medium without FBS for 5 days before medium collection. GREM1 or Vector conditioned medium (GREM1-CM/Vec-CM) was then centrifuged at 1000 rpm for 30 min and the supernatant was collected for further study. Isolation of fibroblasts from normal and tumoral human intestinal tissues Fibroblasts were isolated from normal and tumoral human intestinal tissues. Briefly, a cell dissociation buffer was prepared using DMEM supplemented with 10% FBS, 1% Penicillin/Streptomycin, collagenase type D (1 mg/mL, Roche, 11088866001), and DNase I (20 µg/mL, Roche, 10104159001). Tumor tissues were washed twice with DMEM or phosphate-buffered saline (PBS), transferred to a 100-mm culture dish containing 15 mL of cell dissociation buffer, and minced into fragments (< 1 mm³) using sterile razor blades. The tissue fragments were enzymatically digested at 37 °C for 30 min. Following digestion, the cell suspension was filtered through a 70-µm cell strainer to obtain a single-cell suspension. The cell suspension was centrifuged at 1000 rpm for 5 min, and the pellet was resuspended in DMEM. This step was repeated, and the final pellet was resuspended in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37 °C in a humidified incubator with 5% CO₂. Lentiviral plasmid construction, lentivirus production and infection Human GREM1 CDS (NM_013372.7) with a HA-mCherry tag or human ACVR1C CDS (NM_145259.3) with a Flag-GFP tag was cloned into a pCDH-CMV-MCS-EF1-puro vector. Truncated or point mutations of GREM1 or ACVR1C were cloned from entire GREM1- or ACVR1C-expressing plasmids by PCR. Human GREM1 CDS (NM_013372.7) was cloned into the pLV-EF1a-IRES-Puro lentiviral vector. CRISPR-mediated gene knockout: The sequences targeting GREM1 were GREM1 KO27 (gRNA1: 5′– GCAAATACCTGAAGCGAGAC –3′) and GREM1 KO28 (g RNA2: 5′– AAGCAGACCATCCACGAGGA –3′). The Cas9 lentivirus and gRNA1/2 lentivirus were purchased from GenePharma. shRNA-mediated silencing: The human ACVR1C shRNA target sequences are listed as follows: shACVR1C#1 (5′– CGGAGGAATTGTTGAGGAGTA –3′); shACVR1C#2 (5′– GCAACACCTCAACTCATCTTT –3′). All inserts and vectors were purified from agarose gel using the FastPure ® Gel DNA Extraction Mini Kit (Vazyme, DC301-01) and assembled with Gibson Assembly Master Mix[65] (NEB, E2611) according to the manufacturers’ protocols. All plasmids were verified by Sanger sequencing. HEK 293T cells were seeded at a density expected to reach 70-80% confluence at the time of transfection. To produce lentivirus, plasmids mentioned above together with packaging plasmid ( psPAX2 ) and envelope plasmid ( pMD2.G ) were mixed in a 3.9:2.1:1 ratio and transfected into the cells using polyethylenimine (PEI). After 48-72 hours, supernatant containing lentivirus was collected, filtered, and either used immediately or stored at -80℃ for later applications. HCT116 and SW480 tumor cells were infected with lentiviral particles in the presence of 5 μg/mL polybrene. To establish stable cell lines, these infected cells were selected with 1.25 μg/mL puromycin for 2 weeks. Animal experiments The immunocompromised nude[66] and NOG[40] female mice (6 weeks old) were purchased from Guangzhou Vital River Laboratory Animal Technology Co., Ltd. Grem1-CreER T2 (stock no. 027039)[19], Rosa-mTmG (stock no. 007576)[67], Rosa-LSL-DTA (stock no. 007900) mice[68] were obtained from the Jackson Laboratory. APC Min/+ mice were obtained from the Gempharmatech Co., Ltd (stock no. 002020)[36]. Grem1-CreER T2 mice were crossed with Rosa-mTmG mice or Rosa-LSL-DTA mice to generate GR or GD mice, respectively. The Cre recombinase activity was induced by the ER antagonist tamoxifen (TMX), allowing Grem1 + cells to express GFP in GD mice. GD mice were further crossed with APC Min/+ mice to generate AGD mice. AGD and control mice were administered with 100mg/kg tamoxifen (TMX) through oral gavage at 4-week-old time, when tumor initiates. In AGD and its control mice, activation of Cre lead to the expression of DTA (diphtheria toxin A chain), which removed the population of Grem1 + cells from the APC Min/+ mice. All animals were maintained at the Animal Experiment Center of Sun-Yat-Sen University, and all procedures were approved by the Animal Care and Use Committee of Sun-Yat-Sen University. Mice were randomized at the beginning of each experiment. For tail vein-to-lung metastasis model, 5×10 5 MC38-luc cells were resuspended in 100 μl of PBS and injected into the tail veins of GD mice or Grem1-CreER T2 /Rosa-LSL-DTA mice (n = 5 mice in each group); 1×10 6 HCT116-luc cells, transduced with lentivirus carrying a control shRNA or two ACVR1C shRNAs or carrying a pLV or pLV-GREM1, were injected into the tail veins of nude mice (n = 5 mice in each group). For spleen-to-liver metastasis model, 5×10 5 MC38-luc cells were resuspended in 50 μl of PBS and injected into the spleen of GD mice or Grem1-CreER T2 /Rosa-LSL-DTA mice (n = 5 mice in each group). 5×10 5 HCT116-luc cells carrying a pLV or pLV-GREM1 were resuspended in 50 μl of PBS and intrasplenically injected in NOG mice (n = 5 mice in each group). ACVR1C peptide was administered via tail vein injection at a dose of 10 mg/kg once every other day. For cecum-to-liver metastasis model, 1×10 6 MC38-luc cells were resuspended in 50 μl of PBS and injected into the cecum of GD mice or Grem1-CreER T2 /Rosa-LSL-DTA mice (n = 6 mice in each group); 5×10 6 HCT116-luc cells transduced with or without lentivirus carrying a pLV or pLV-GREM1 were injected into the cecum of NOG mice (n = 5 mice in each group). SB505124 was administered via intraperitoneal injection at a dose of 10 mg/kg once every other day. The metastases were examined every 5 days post injection using an IVIS Lumina Imaging System. Mice were euthanized between 2-6 weeks after injection. For subcutaneous transplantation, 1×10 6 HCT116 cells, either unmodified or transduced with lentivirus carrying a control shRNA, two ACVR1C shRNAs, pLV, or pLV-GREM1, were subcutaneously injected into mice (n = 6-8 mice in each group). Mice were euthanized 4 weeks after injection. The tumor tissues were collected for further evaluation. Immunohistochemical (IHC) staining Immunohistochemical staining of GREM1 (1:50, Biorbyt, orb10741), ACVR1C (1:50, Thermo, PA587475) and Ki67 (1:100, Servicebio, GB111499) was performed on primary tumors tissues. After dewaxing, hydration, and antigen retrieval, the rest of the experimental procedures were performed according to the instructions of the SP Immunohistochemistry Kit (ZSBIO, PV9000). Finally, after DAB staining, hematoxylin re-staining, and neutral resin sealing, the sections were observed under a microscope. Images were taken with a Slide Scanning Imaging System (SQS-1000, sqray). Quantification of positive staining was performed using Fiji (ImageJ). Immunofluorescence (IF) staining Tissue was fixed in 4% paraformaldehyde (Thermo Scientific, I28800) for 24 h at 4 °C, washed with PBS, embedded in paraffin, and sectioned at 5 μm thickness. Antigen retrieval was performed using target retrieval solution, pH 9.0 in a pressure cooker for 15–20 min. Non-specific binding was then blocked with 10% normal donkey serum (Abcam, ab7475) and 0.3% Triton X-100 in PBS for 30 min at room temperature. Cells for IF were fixed with 4% paraformaldehyde for 20 min at room temperature, washed with PBS, and permeabilized with 0.2% Triton X-100 in PBS for 20 min. Cells were then blocked in PBS with 5% BSA for 30 min at room temperature. Subsequently, the samples were incubated with goat anti-GREM1 (3 µg/mL, R&D, AF956), mouse anti-β-Catenin (1:100, BD, 610154), rabbit anti-β-Catenin (1:100, Absea, RC-6352), rabbit anti-Vimentin (1:100, CST, 5741), rabbit anti-CD68 (1:100, CST, 26042), rabbit anti-FAP (1:50, Proteintech, 15384-1-AP), rabbit anti-α-SMA (1:100, Abcam, ab5694), rabbit anti-E-Cadherin (1:200, CST, 3195), rabbit anti-ACVR1C (1:50, Thermo, PA587475), rabbit anti-Snail (1:200, Abcam, ab224731) overnight at 4 °C. The tissues were incubated with Alexa-Fluor-conjugated secondary antibodies (Invitrogen) in PBS with 1 % normal donkey serum for 1 h at room temperature. DAPI was then used for counterstaining the nuclei, and images were obtained by a laser scanning confocal microscope (LSM880, Zeiss). Analysis of scRNA-seq data Single-cell RNA sequencing (scRNA-seq) data from the Gene Expression Omnibus (GEO) database colorectal cancer datasets (GSE200997 and GSE221575) were processed using the R 'Seurat' package (v4.4). Initial quality control involved rigorous filtering of low-quality cells: Cells expressing fewer than 200 genes or more than 10,000 genes were excluded, and cells with mitochondrial gene content exceeding 25% were discarded to remove potential apoptotic cells or debris. After quality control, a total of 34,675 high-quality cells were retained for downstream analysis. Gene expression matrices were normalized using the "LogNormalize" method implemented in the NormalizeData function, which scales feature counts per cell by total expression and multiplies by a scale factor (10,000), followed by natural log transformation. To identify biologically relevant features, the FindVariableFeatures function was employed to select the top 2,000 highly variable genes (HVGs) exhibiting the highest cell-to-cell variation. Dimensionality reduction was performed using principal component analysis (PCA) on scaled expression data of the identified HVGs. To address technical batch effects between samples and datasets, we applied multiple Canonical Correlation Analysis (CCA) as implemented in Seurat's integration workflow. Cell clustering was performed using a graph-based approach: The FindNeighbors function constructed a shared nearest neighbor (SNN) graph based on the first 30 principal components, followed by the FindClusters function using the Louvain algorithm at a resolution of 0.8 to identify distinct cell subpopulations. Finally, non-linear dimensionality reduction was achieved through t-distributed Stochastic Neighbor Embedding (t-SNE) using the same principal components. Immunoprecipitation (IP) HCT116 cells were transfected with the indicated plasmids and lysed in NP40 lysis buffer (Beyotime, P0013F) supplemented with protease inhibitor cocktail (Thermo, 78446). Lysates were incubated with the indicated Anti-Flag nanobody magarose beads (Ktsm-life, KTSM1338), Anti-HA nanobody magarose beads (Ktsm-life, KTSM1335) or Anti-GFP nanobody magarose beads (Ktsm-life, KTSM1334) overnight at 4 °C. The protein complex was washed four times with the NP40 lysis buffer, eluted with 1×loading buffer (Beyotime, P0015) by boiling for 5 min, followed by mass spectrometry and immunoblotting with the indicated antibodies. Mass spectrometry (MS) analysis Proteins were separated by 10% SDS-PAGE and visualized using Coomassie Brilliant Blue staining before mass spectrometry analysis. The stained gel bands were excised (~1–2 mm), washed with MilliQ water, and destained using 25 mM NH₄HCO₃ and 50% acetonitrile (ACN) at 37°C. The gel pieces were dehydrated with ACN, reduced with 10 mM dithiothreitol (DTT) in 25 mM NH₄HCO₃ at 37°C for 1 h, and alkylated with 30 mM iodoacetamide (IAA) in 25 mM NH₄HCO₃ in the dark for 45 min. After sequential washing with MilliQ water and 50% ACN, the gel pieces were dehydrated with ACN and digested overnight at 37°C with trypsin (20 ng/μL) in 25 mM NH₄HCO₃. Peptides were extracted using 60% ACN followed by pure ACN, pooled, lyophilized, resuspended in 0.1% formic acid (FA), and purified using ZipTip C18 before analysis. Mass spectrometry was performed using a Thermo Fisher Orbitrap HF-X coupled with an Easy-nLC 1200 system and a C18 column, employing a 90-min gradient of 5–35% ACN in 0.1% FA at a flow rate of 300 nL/min. MS1 scans were acquired at a resolution of 60,000 with an AGC target of 3 × 10⁶, a maximum injection time of 20 ms, and a scan range of m/z 350–1800. MS2 scans were performed at a resolution of 15,000 with an AGC target of 2 × 10⁵, a maximum injection time of 100 ms, TopN of 20, and a normalized collision energy (NCE) of 32. Raw MS data were analyzed using Proteome Discoverer 2.4, with protein identification performed against the SwissProt human database using trypsin specificity (allowing one missed cleavage site), cysteine alkylation with MMTS, a precursor mass tolerance of 10 ppm, a fragment mass tolerance of 0.02 Da, and a false discovery rate (FDR) threshold of <1%. Immunoblotting (IB) Protein was extracted from the cells with RIPA buffer (Beyotime, P0018) or NP40 lysis buffer (Beyotime, P0013F) and separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Primary antibodies against GREM1 (1:1,000, SinoBiological, 50016-R117), ACVR1C (1:1,000, Thermo, PA587475), Flag-tag (1:1,000, CST, 14793), HA-tag (1:1,000, CST, 3724), E-Cadherin (1:1,000, CST, 3195), β-Catenin (1:1,000, CST, 8480), ZEB1 (1:1,000, CST, 3396), Snail (1:1,000, CST, 3879), SMAD2/3 (1:1,000, CST, 8685), p-SMAD2/3 (1:1,000, CST, 8828), SMAD1 (1:1,000, CST, 6944), p-SMAD1/5/9 (1:1,000, CST, 13820), TGFβ (1:1,000, CST, 3709), TGFβR1 (0.3µg/mL, R&D, AF3025), β-actin (1:5,000, Beyotime, AF0003) and GAPDH (1:5,000, Beyotime, AF0006) were used in this study. Peroxidase-conjugated secondary antibody (1:10,000, Cell Signaling Technology, 7074, 7076) was used and signal was visualized using an enhanced chemiluminescence assay (ECL, Thermo), according to the manufacturer’s protocol. Band intensity was quantified using Fiji (ImageJ) by grayscale analysis. Recombinant protein production and purification Expi293F cells were transfected with a pcDNA3.4-ACVR1C-ECD-Fc and pcDNA3.4-ACVR1C-ECD-Fc-double mutant (E85A/T101A) expression vector to produce the target protein, which was subsequently purified using a Protein G column. Briefly, the coding sequence (CDS) of the human ACVR1C extracellular domain ( ACVR1C-ECD , NM_145259.3, residues 1-339) fused to an Fc tag was cloned into the pcDNA3.4 vector. Expi293F cells were transfected with this construct, the supernatant was harvested 5 days post transfection. The supernatant was first centrifuged at 1000 rpm for 20 minutes to remove cell debris, and the supernatant was further centrifuged at 8000 rpm for 30 minutes, followed by filtering with a 0.45 μm PES filter. The protein in the supernatant was then purified using a Protein G column equilibrated with binding buffer (0.15 M NaCl, 20 mM Na₂HPO₄, pH 7.0). The target protein was eluted with 0.1 M glycine (pH 2.5) and immediately neutralized with 1 M Tris-HCl (pH 8.5). Subsequently, the protein buffer was exchanged into a 20 mM Tris-HCl (pH 7.5) system. To further purify the sample, it was centrifuged at 12000 rpm and 4°C for 10 minutes to remove impurities and precipitates. The clarified sample was then loaded onto an ion exchange column (HiTrap™ Capto™ Q ImpRes) equilibrated with binding buffer (20 mM Tris-HCl, pH 7.5). The ACVR1C-ECD-Fc protein, having an opposite charge to the resin, was bound to the column. Finally, the target protein was eluted with a linear gradient (0-100%) of elution buffer (20 mM Tris-HCl, 1 M NaCl, pH 7.5) over 6 column volumes. Protein pull-down assay Protein pull-down assay was performed using purified recombinant human His-tagged GREM1 protein and recombinant human ACVR1C-ECD and Fc chimera protein. Protein was enriched by Pierce Protein A magnetic beads (MCE, HY-K0202) or Ni Sepharose 6 Fast Flow (GE, 17531801) following the manufacturer’s instructions. Pulled-down proteins were detected by Coomassie Brilliant Blue staining. MicroScale thermophoresis (MST) MST was carried out on a Monolith NT.115 instrument (NanoTemper Technologies GmbH). To evaluate ACVR1C-ECD or ACVR1C peptide or ACVR1C-ECD-double mutant (E85A/T101A) binding to GREM1-His or ACTIVINB-His, an increasing concentration of purified ACVR1C-ECD-Fc protein (0–27.5 μM) or ACVR1C peptide (0–2.3 μM) or ACVR1C-ECD-double mutant (0–27.5 μM) was incubated with 50 nM RED-labeled (NanoTemper Technologies GmbH) GREM1-His protein (R&D, 5190-GR) or ACTIVINB-His protein (SinoBiological, 10814-H08H). Experiments were carried out in a PBS buffer pH 7.4 using premium capillaries. Protein-protein interaction docking study GREM1 (PDB: 5AEJ) was selected as the ligand and ACVR1C (PDB: AF-Q8NER5-F1) as the receptor for protein-protein docking. The HDOCK web service was used for docking with default parameters (http://hdock.phys.hust.edu.cn/). Key amino acid residues in the binding pocket between GREM1 and ACVR1C were further identified based on the docking module[69]. RNA-seq and gene set enrichment analysis (GSEA) Total RNA was extracted using Trizol reagent (Invitrogen, 15596026) and quantified with a NanoDrop spectrophotometer (Thermo Fisher Scientific). RNA integrity was assessed using an Agilent 2100 Bioanalyzer. mRNA was enriched using oligo(dT) magnetic beads, fragmented, and reverse-transcribed into cDNA. After adapter ligation and PCR amplification, libraries were sequenced on an Illumina platform, generating 150-bp paired-end reads. Raw reads were trimmed and aligned to the human reference genome (GRCh38) using STAR. Differential gene expression analysis was conducted using Limma, with significance thresholds set at |log2FoldChange| >1.5 and adjusted P-value <0.05. Gene set enrichment analysis (GSEA) was performed using the GSEA software (Broad Institute) with the MSigDB gene sets to identify enriched biological pathways, employing 1,000 permutations and FDR <0.25 as the cutoff for significance. RT–qPCR Total RNA was extracted using Trizol reagent (Invitrogen, 15596026). According to the instruction, cDNA was generated using the PrimeScript RT reagent Kit with gDNA Eraser (Accurate Biology, AG11706). The SYBR Green Premix Pro Taq HS qPCR Kit (Accurate Biology, AG11701) was then used to quantify mRNA expression according to the manufacturer’s instruction. All results were calculated using the 2 -ΔΔct method. Primers used in the study are listed in Supplementary Table 1. ChIP SW480 cells were starved in DMEM with 1% FCS overnight before treatment with vehicle, 10 μM SB505124 for 24 hours. Cells were fixed in 1 % paraformaldehyde for 10 min at RT for DNA-protein cross-linking, followed by quenching with glycine. Cross-linking chromatin was prepared using the SimpleChIP® Enzymatic Chromatin IP Kit (CST, 9002) according to the manufacturer’s instructions. For immunoprecipitation, 10 μg chromatin was incubated with 10 μL anti-histone H3 rabbit IgG (CST, 14269, positive control), 2 μL normal Rabbit IgG (CST, 2729) or 5 μL anti-SMAD2/3 rabbit IgG (CST, 8685) at 4 °C overnight. 2% chromatin prior to immunoprecipitation was used as input. Chromatin-protein-antibody complex was captured by protein G magnetic beads, and chromatin was released by reversal of cross-links and purified using the SimpleChIP® Enzymatic Chromatin IP Kit (CST, 9002) according to the manufacturer’s instructions. DNA was quantified by qPCR with primers targeting predicted SMAD2/3/4 binding regions on GREM1 or SNAI1 promoters. DNA levels were normalised to the input, and the fold-change of enrichment was calculated over the control. ChIP–qPCR primers are listed in Supplementary Table 2. Scratch assay Cells were seeded into 6-well plates after centrifugation and digestion with 0.05% trypsin. When the cell density reached 90%, three vertical lines were scratched in each well with a 10 μL pipette tip and the floating cells were gently washed away with 1×PBS. Complete medium was added, and images of the scratch area were taken at 0 h. Three different fields of view were selected for each well. After photography, the medium was replaced with serum-free medium. Wound healing was documented at the same location after 24 h or 48 h of incubation. Transwell invasion assay Cells (1×10 5 ) were seeded in serum-free medium in the Matrigel-coated (Corning, 354480) transwell chambers (24-well insert, 8-μm pore size; BD Biosciences) for invasion experiments. The lower chamber was filled with RPMI1640 or DMEM containing 20% FBS. The migration of HCT116 and SW480 cells was measured in three random visual fields and quantified by microscopy after 48 h of incubation, followed by staining with DAPI or crystal violet. The invasive capacity of the cells was assessed using ImageJ software for quantification. Statistical analysis All the statistical analyses were performed using GraphPad Prism 9, and error bars indicate s.e.m. Student’s t-test assuming equal variance and one-way analysis of variance for independent variance were used. Growth curves were generated using ANOVA for repeated measurement. P <0.05 was considered significant. The number of independent experiments, the number of events and information about the statistical details and methods are indicated in the relevant figure legends. Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformation44.docx ThemassspectrometrydatasupportFig.2A2BandFig.S3A.xlsx RNAseqgenefpkmTheRNAseqdatasupportFig.S4J.xlsx RNAseqgenefpkmTheRNAseqdatasupportFig.3A..xlsx rawimagesoftheimmunoblottingexperiments.docx Cite Share Download PDF Status: Published Journal Publication published 24 Jan, 2026 Read the published version in Molecular Cancer → Version 1 posted Editorial decision: Revision requested 30 Sep, 2025 Reviews received at journal 26 Sep, 2025 Reviews received at journal 26 Sep, 2025 Reviewers agreed at journal 22 Sep, 2025 Reviewers agreed at journal 21 Sep, 2025 Reviewers agreed at journal 20 Sep, 2025 Reviewers agreed at journal 20 Sep, 2025 Reviewers agreed at journal 20 Sep, 2025 Reviewers invited by journal 09 Sep, 2025 Editor assigned by journal 01 Sep, 2025 Submission checks completed at journal 01 Sep, 2025 First submitted to journal 29 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7484753","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":513887232,"identity":"29cad479-724a-41f6-ac33-3ad5b32c87f4","order_by":0,"name":"Huaixiang Zhou","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Huaixiang","middleName":"","lastName":"Zhou","suffix":""},{"id":513887233,"identity":"607373ce-00c4-4239-b0f5-3fb922f56f3e","order_by":1,"name":"Qunlong Jin","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Qunlong","middleName":"","lastName":"Jin","suffix":""},{"id":513887234,"identity":"994ac45b-8d09-4964-a15e-8b9d5eaefabf","order_by":2,"name":"Zhang Fu","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Zhang","middleName":"","lastName":"Fu","suffix":""},{"id":513887235,"identity":"6f03a35b-884b-4793-93a8-201358f86aed","order_by":3,"name":"Yanming Yang","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Yanming","middleName":"","lastName":"Yang","suffix":""},{"id":513887236,"identity":"0bea8104-fe0a-4248-acb4-95bd74058776","order_by":4,"name":"Yunfei Gao","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Yunfei","middleName":"","lastName":"Gao","suffix":""},{"id":513887238,"identity":"13067148-1cc9-464a-af7a-4cd74aac9713","order_by":5,"name":"Niu Wang","email":"","orcid":"","institution":"Shenzhen Campus of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Niu","middleName":"","lastName":"Wang","suffix":""},{"id":513887239,"identity":"fd7e6710-2030-4601-a2c2-aa6c0cd03f62","order_by":6,"name":"Bo Zhao","email":"","orcid":"","institution":"Shenzhen Campus of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Bo","middleName":"","lastName":"Zhao","suffix":""},{"id":513887240,"identity":"5af71371-ae0c-4524-a0c7-9f5a1a944846","order_by":7,"name":"Long Gui","email":"","orcid":"","institution":"Shenzhen Medical Academy of Research and Translation","correspondingAuthor":false,"prefix":"","firstName":"Long","middleName":"","lastName":"Gui","suffix":""},{"id":513887241,"identity":"5b6c6669-407b-4147-ad4b-fe9b99c99e12","order_by":8,"name":"Jiang Li","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Jiang","middleName":"","lastName":"Li","suffix":""},{"id":513887242,"identity":"8293cbde-8f93-4280-abdc-bb020cd442e0","order_by":9,"name":"Zijing Zhu","email":"","orcid":"","institution":"Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Zijing","middleName":"","lastName":"Zhu","suffix":""},{"id":513887243,"identity":"30ec5648-cfd1-44cd-b0bf-e991a169312a","order_by":10,"name":"Ying Zhang","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Zhang","suffix":""},{"id":513887244,"identity":"b791a20d-fba9-4bcc-a1f3-fe78a54553b0","order_by":11,"name":"Yulong He","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Yulong","middleName":"","lastName":"He","suffix":""},{"id":513887245,"identity":"10f39908-9a64-4efc-89bf-d90121c343ae","order_by":12,"name":"Ying Zhang","email":"","orcid":"","institution":"The Francis Crick Institute","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Zhang","suffix":""},{"id":513887246,"identity":"07eee19f-e5a2-4d06-b691-c5077720f4be","order_by":13,"name":"Shouqing Luo","email":"","orcid":"","institution":"University of Plymouth","correspondingAuthor":false,"prefix":"","firstName":"Shouqing","middleName":"","lastName":"Luo","suffix":""},{"id":513887247,"identity":"eea2394a-3d88-4c4c-a1ae-cf77f438527c","order_by":14,"name":"Li Fu","email":"","orcid":"","institution":"Shenzhen University Medical School, Shenzhen University","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Fu","suffix":""},{"id":513887248,"identity":"3c4a77aa-6385-42ee-820c-1b4bcf4dbe2f","order_by":15,"name":"Xudong Wu","email":"","orcid":"","institution":"The Province and Ministry Co-sponsored Collaborative Innovation Center for Medical Epigenetics, Tianjin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xudong","middleName":"","lastName":"Wu","suffix":""},{"id":513887249,"identity":"165cd25f-c5bc-4969-906c-e63cb054c79a","order_by":16,"name":"Junjing Zhang","email":"","orcid":"","institution":"Hohhot First Hospital","correspondingAuthor":false,"prefix":"","firstName":"Junjing","middleName":"","lastName":"Zhang","suffix":""},{"id":513887250,"identity":"3d31b00a-be15-4718-89d4-27b247a30512","order_by":17,"name":"Xuetong Shen","email":"","orcid":"","institution":"Shenzhen Bay Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Xuetong","middleName":"","lastName":"Shen","suffix":""},{"id":513887251,"identity":"e177cef3-af16-4429-8882-919c8adb4a3c","order_by":18,"name":"Tao Wang","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Tao","middleName":"","lastName":"Wang","suffix":""},{"id":513887253,"identity":"14c21bb6-6eaa-4260-979d-67c287484490","order_by":19,"name":"Youheng Jiang","email":"","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Youheng","middleName":"","lastName":"Jiang","suffix":""},{"id":513887255,"identity":"28145f85-9025-4efe-b6d5-0b438cd95e4a","order_by":20,"name":"Ningning Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA70lEQVRIiWNgGAWjYBACPmYg8aHChoGBmbkBJmiAVwsbUAvjjDNpQC2MxGoBYmbetsNAimgt7LzHHvC2nY/mb2dsYPzZVpfYwN68TYKh5g4eh/GlG0icu5074zBjA8i6xAaeY2USDMee4dHCYyZhUHY7twGkhbHtQGKDRI6ZBGPDYfxaEtjO5c4/DHOY/BsitBxoO5C7AaiFgbeNGWgLD2Etkg1nknM3ArUc5jl32LiNJ63YIuEYbi38/GfMpP9U2OXOO3/44MMfZXWy/eyHN974UINbCwo4wMgGiSmGBOI0gMAf4pWOglEwCkbByAEA679P2I/QWXAAAAAASUVORK5CYII=","orcid":"","institution":"The Seventh Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":true,"prefix":"","firstName":"Ningning","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-08-29 04:23:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7484753/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7484753/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12943-025-02554-w","type":"published","date":"2026-01-24T15:57:27+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":91536175,"identity":"32d55d01-02cc-41d3-b2ac-b98c84262df3","added_by":"auto","created_at":"2025-09-17 13:04:39","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2280641,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEctopic expression of GREM1 protein in CRC. A\u003c/strong\u003e, Representative images of GREM1 immunohistochemical (IHC) staining (Upper) on human primary CRC samples and illustration of GREM1 expression pattern (Lower). The red dashed line separates tumor from adjacent normal area. N: normal areas; T: tumoral areas; S: stromal areas; Red arrow: GREM1\u003csup\u003e+\u003c/sup\u003e CRC cells. \u003cstrong\u003eB, C\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eQuantification of GREM1⁺ stromal cell infiltration (B) and GREM1 expression in cancer cells (C) across stage I–IV primary CRC tissues from Sun Yat-sen University (SYSU). \u003cstrong\u003eD\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative immunofluorescence (IF) images of GREM1 and EPCAM staining in stage IV human primary CRC tumors. A white dashed line separates epithelial and mesenchymal cells. White arrow: GREM1\u003csup\u003e+\u003c/sup\u003e CRC cells. \u003cstrong\u003eE\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eKaplan–Meier survival curves of 106 CRC patients from SYSU stratified by GREM1 levels in cancer cells. Patients were divided into high and low groups based on the cohort median. For B and C, data are mean ± s.e.m. \u003cem\u003eP\u003c/em\u003e values were calculated using one-way ANOVA with Bonferroni multiple-comparison test (B, C). Significance was determined using a two-sided log-rank test. HR, hazard ratio (E). Scale bars, 20 μm.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/0fc7584dba39a78c39e47abf.jpeg"},{"id":91537154,"identity":"3712e224-3ebb-447d-b4c8-d74b9da2f918","added_by":"auto","created_at":"2025-09-17 13:12:39","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1427722,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eACVR1C is a novel GREM1 receptor in CRC. A\u003c/strong\u003e, Proteins extracted from the HCT116 cells transfected with HA-mCherry-tagged-GREM1 were incubated with magnetic beads conjugated with an anti-HA antibody. Bound proteins were eluted and visualized by Coomassie Brilliant Blue staining. A protein band of ~110 kDa was submitted for mass spectrometry (MS). \u003cstrong\u003eB\u003c/strong\u003e, The full amino-acid sequence of human ACVR1C. The sequences in blue are the tryptic peptides identified by MS. \u003cstrong\u003eC\u003c/strong\u003e, GREM1 coimmunoprecipitates with ACVR1C in HCT116 cells transfected with HA-mCherry-tagged GREM1 expressing plasmids. The bound proteins were immunoprecipitated with an anti-HA antibody and blotted by anti-HA, anti-ACVR1C, and anti-TGFβR1 antibodies. \u003cstrong\u003eD\u003c/strong\u003e, ACVR1C coimmunoprecipitates with GREM1 in HCT116 cells transfected with Flag-tagged ACVR1C expressing plasmids. The bound proteins were immunoprecipitated with an anti-Flag antibody and blotted by an anti-GREM1 antibody. \u003cstrong\u003eE\u003c/strong\u003e, Confocal microscopy images of GREM1 and ACVR1C in SW480 cells. Scale bars, 1 μm. \u003cstrong\u003eF\u003c/strong\u003e, Interaction of purified GREM1 and ACVR1C-ECD protein demonstrated by pull-down experiments. \u003cstrong\u003eG\u003c/strong\u003e, Increasing concentrations of recombinant ACVR1C-ECD-Fc protein (0-27.5 μM) were incubated with Red-labeled 50 nM recombinant GREM1-his or ACTIVIN B-his. MST was used to evaluate ACVR1C-ECD-Fc binding to GREM1-his or ACTIVIN B-his (n = 3 independent experiments), data are presented as mean ± s.e.m. \u003cstrong\u003eH,\u003c/strong\u003e \u003cstrong\u003eI\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eDiagrams of truncated ACVR1C (H, Left) and truncated GREM1 (I, Left), with corresponding coimmunoprecipitation results (Right) comparing truncation mutants with binding partners. \u003cstrong\u003eJ\u003c/strong\u003e, Molecular docking of GREM1 and ACVR1C-ECD simulated by HDOCK (http://hdock.phys.hust.edu.cn/). Protein structures were generated from PDB (\u003ca href=\"http://www.rcsb.org/\"\u003ehttp://www.rcsb.org/\u003c/a\u003e, GREM1: 5AEJ) and AlphaFold (\u003ca href=\"http://alphafold.ebi.ac.uk/\"\u003ehttp://alphafold.ebi.ac.uk/\u003c/a\u003e, ACVR1C: Q8NER5). Docking module highlighting key amino acid residues in the binding pocket between GREM1 and ACVR1C. \u003cstrong\u003eK\u003c/strong\u003e, Schematic of ACVR1C mutations (point mutations highlighted in red, Upper). Coimmunoprecipitation of ACVR1C and GREM1 is impaired by the AA\u003csub\u003e85 \u003c/sub\u003emutant (M2) or the AA\u003csub\u003e101\u003c/sub\u003e mutant (M3) of ACVR1C (Lower). \u003cstrong\u003eL\u003c/strong\u003e, Schematic of GREM1 mutations (point mutations highlighted in red, Upper). Coimmunoprecipitation of ACVR1C and GREM1 is impaired by the AA\u003csub\u003e100-102/112/115 \u003c/sub\u003emutant (M3) of GREM1 (Lower).\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/f619248f833eeb29b7e4e3c1.jpeg"},{"id":91536176,"identity":"81557e79-f7f3-4a39-aa48-ab9de79d62c0","added_by":"auto","created_at":"2025-09-17 13:04:39","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1258976,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSecretory GREM1 binding of ACVR1C induced EMT via SMAD2/3 pathway. A\u003c/strong\u003e, RNA sequencing was performed on HCT116 cells treated with GREM1 conditioned medium (GREM1-CM) or control conditioned medium (Vec-CM), followed by GSEAof the C2 gene sets. NES, normalized enrichment score, all P-values equal to 0.\u003cstrong\u003eB\u003c/strong\u003e, Activation levels of the ACVR1C pathway markers p-SMAD2/3 and SMAD2/3 and expression of epithelial marker E-CAD, mesenchymal markers SNAIL, ZEB1 and β-CAT were compared by immunoblotting analysis in SW480 and HCT116 cells treated with Vec-CM, GREM1-CM, GREM1-CM + GREM1 BAb or GREM1-CM + IgG. \u003cstrong\u003eC\u003c/strong\u003e, Immunoblotting for the ACVR1C pathway markers p-SMAD2/3 and SMAD2/3 and the BMPR pathway markers p-SMAD1/5/9 and SMAD1 and the TGFβR pathway markers TGFβR1 and TGFβ from the lysates of HCT116 or SW480 cells with a concentration gradient of rhGREM1 treatment. \u003cstrong\u003eD\u003c/strong\u003e, Schematic representation of the TGFβ signaling pathway, highlighting GREM1-induced activation of ACVR1C and inhibition of BMPR signaling. \u003cstrong\u003eE\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eImmunoblotting for the BMPR pathway markers p-SMAD1/5/9 and SMAD1 and expression of epithelial marker E-CAD, mesenchymal markers SNAIL and ZEB1 from the lysates of HCT116 or SW480 cells treated with GREM1-CM and different concentrations of BMP agonist, sb4. \u003cstrong\u003eF\u003c/strong\u003e, SW480 and HCT116 cells were transfected with control shRNA (Scramble shRNA) or with one of two ACVR1C-targeting shRNAs (denoted shACVR1C#1 and shACVR1C#2) and separately treated with Vec-CM or GREM1-CM; epithelial marker E-CAD and mesenchymal markers SNAIL and ZEB1 were analyzed by immunoblotting. \u003cstrong\u003eG\u003c/strong\u003e, Expression of epithelial markers E-CAD and mesenchymal markers SNAIL, ZEB1 and β-CAT were compared by immunoblotting analysis in SW480 and HCT116 cells treated with Vec-CM, GREM1-CM or GREM1-CM + SB505124 (1 μM). SB505124 inhibits the ACVR1C signaling pathway by impairing SMAD2/3 phosphorylation. CM, culture medium. For p-SMAD2/3, SMAD2/3 was used as a control. For other proteins, β-actin was used as a loading control.\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/b124817b9b8be75731d0afec.jpeg"},{"id":91536180,"identity":"dbc6dd81-72af-4d68-a15a-fe6a9794ffca","added_by":"auto","created_at":"2025-09-17 13:04:40","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1824369,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSecreted GREM1 upregulates \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eGREM1\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e transcription via ACVR1C-SMAD2/3 pathway to reinforce EMT in CRC. A,\u003c/strong\u003e \u003cstrong\u003eB\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRT–qPCR analysis of\u003cem\u003e GREM1 \u003c/em\u003emRNA levels in SW480 (A) and HCT116 (B) cells treated with increasing concentrations of rhGREM1. n = 3 biological replicates. \u003cstrong\u003eC\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eImmunoblotting of GREM1 protein levels in SW480 and HCT116 cells treated with the concentration gradient of rhGREM1 used in (A) and (B).\u003cstrong\u003e D\u003c/strong\u003e, RT–qPCR analysis of \u003cem\u003eGREM1\u003c/em\u003e expression in SW480 cells. n = 3 biological replicates. \u003cstrong\u003eE\u003c/strong\u003e, SMAD2/3/4 binding motifs in the \u003cem\u003eGREM1\u003c/em\u003e promoter, predicted by the JASPAR database. Red boxes denote predicted binding sites upstream of the transcription start site (TSS). \u003cstrong\u003eF\u003c/strong\u003e, ChIP analysis of p-SMAD2/3 binding sites on the \u003cem\u003eGREM1\u003c/em\u003e promoter in SW480 cells, treated as indicated. IgG control or SMAD2/3 antibodies were used for ChIP and DNA was quantified by qPCR. DNA levels for each condition were normalized to the input, and the fold-change was calculated over the vehicle control (n = 3 independent experiments). \u003cem\u003ens\u003c/em\u003e: not significant. \u003cstrong\u003eG\u003c/strong\u003e, Representative IF images for staining of GREM1, E-CAD and β-CAT in stage Ⅳ human primary CRC tumors from SYSU. The white square-dashed box in the left image indicates the location of the corresponding region in the right image. The white irregular-dashed box marks the GREM1⁺ CRC cells, while the brown irregular-dashed box marks the GREM1\u003csup\u003e-\u003c/sup\u003e CRC cells. Scale bars, 20 μm. \u003cstrong\u003eH, I, \u003c/strong\u003eQuantification of the fluorescence intensity of E-CAD (H) and β-CAT (I) between GREM1\u003csup\u003e+\u003c/sup\u003e and GREM1\u003csup\u003e-\u003c/sup\u003e CRC subpopulations shown in (G) (n = 3 patients). \u003cstrong\u003eJ\u003c/strong\u003e, Quantification of the percentage of cells exhibiting cytoplasmic and nuclear translocation of β-CAT in the conditions shown in (G) (n = 3 patients). \u003cstrong\u003eK\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRT–qPCR analysis of \u003cem\u003eGREM1\u003c/em\u003e mRNA expression levels in CRC cell lines (DLD1, LOVO, HCT116, LS174T, SW480, and HT29), primary fibroblasts (normal fibroblasts and tumoral fibroblasts) derived from the gut of human CRC patients. \u003cstrong\u003eL, M\u003c/strong\u003e, RT–qPCR analysis of \u003cem\u003eGREM1\u003c/em\u003e, epithelial marker \u003cem\u003eCDH1 \u003c/em\u003eand mesenchymal markers \u003cem\u003eZEB1\u003c/em\u003e and \u003cem\u003eVIM\u003c/em\u003e in HCT116 cells infected with control (pLV) or GREM1-overexpressing (pLV-GREM1) lentiviruses or GREM1-overexpressing cells transduced with one of two GREM1-targeting sgRNAs (denoted GREM1KO26 and GREM1KO27), n = 3 independent experiments. For A, B, D, F and H–M, data are presented as mean ± s.e.m. \u003cem\u003eP\u003c/em\u003e values were calculated using one-way ANOVA with Bonferroni multiple-comparison test (A, B, D, F, L, M) and two-tailed Student’s t-test (H–J).\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/13a31a79f4218fedf281b523.jpeg"},{"id":91536177,"identity":"848650df-5e91-43bf-8c77-ada092b99d4f","added_by":"auto","created_at":"2025-09-17 13:04:39","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1363731,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDepletion of Grem1\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003e+\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003e stromal cells inhibits EMT and metastasis \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo. \u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003eA\u003c/strong\u003e, Diagram of the AGD Mouse Model (\u003cem\u003eAPC\u003c/em\u003e\u003csup\u003e\u003cem\u003eMin/+\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Grem1-CreER\u003c/em\u003e\u003csup\u003e\u003cem\u003eT2\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Rosa-LSL-DTA\u003c/em\u003e) and the experimental approach. Grem1⁺ cells are selectively deleted through tamoxifen (TMX)-induced DTA expression via Cre-lox recombination. Mice received TMX (100 mg/kg) by oral gavage for five consecutive days starting at 4 weeks of age, followed by weekly TMX administration until analysis at 11 weeks. The black triangle indicates the loxP sites. \u003cstrong\u003eB\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative IF images of staining for Grem1 and E-cad and β-cat in AGD or control mice (\u003cem\u003eAPC\u003c/em\u003e\u003csup\u003e\u003cem\u003eMin/+\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Grem1-CreER\u003c/em\u003e\u003csup\u003e\u003cem\u003eT2\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e \u003c/em\u003eor\u003cem\u003e APC\u003c/em\u003e\u003csup\u003e\u003cem\u003eMin/+\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Rosa-LSL-DTA\u003c/em\u003e). The white irregular dashed box indicates the infiltrated region of Grem1\u003csup\u003e+\u003c/sup\u003e stromal cells. Scale bars, 10 μm. \u003cstrong\u003eC, D\u003c/strong\u003e, Quantification of the fluorescence intensity of E-cad (C) and β-cat (D) between AGD and control mice shown in (B) (n = 5 mice per group). \u003cstrong\u003eE\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eQuantification of the percentage of cells exhibiting cytoplasmic and nuclear translocation of β-cat under the conditions shown in (B) (n = 5 mice per group).\u003cstrong\u003e F\u003c/strong\u003e, Diagram of the GD mouse CRC model and the experimental approach. GD and control (\u003cem\u003eGrem1-CreER\u003c/em\u003e\u003csup\u003e\u003cem\u003eT2\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e \u003c/em\u003eor\u003cem\u003e Rosa-LSL-DTA\u003c/em\u003e) mice received TMX (100 mg/kg) by oral gavage for five consecutive days starting at 6 weeks of age, followed by weekly doses. At designated time points, MC38-luciferase (MC38-luc) cells were injected intravenously, intrasplenically, or into the cecal wall. Lung and liver metastases were imaged and quantified every five days using the IVIS Lumina Imaging System. The black triangle indicates the loxP sites. \u003cstrong\u003eG\u003c/strong\u003e, Representative images (Left) and quantification (Right) of GD model mice and control mice intravenously injected with MC38-luc cells. Lung metastases were imaged and quantified by IVIS Lumina Imaging System (n = 5 mice per group). \u003cstrong\u003eH\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative images (Left) and quantification (Right) of GD model mice and control mice intrasplenically injected with MC38-luc cells. Liver metastases were imaged and quantified by IVIS Lumina Imaging System (n = 5 mice per group).\u003cstrong\u003e I\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative images (Left) and quantification (Right) of the colorectum, and liver from GD model mice and control mice injected into the cecum wall with MC38-luc cells. liver metastases were imaged and quantified by IVIS Lumina Imaging System (n = 5 mice per group). met., metastases. For C–E and G–I, data are presented as mean ± s.e.m. \u003cem\u003eP\u003c/em\u003e values were calculated using two-tailed Student’s t-test.\u003c/p\u003e","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/77fbc5e761b037fb037017d3.jpeg"},{"id":91537540,"identity":"b5c86903-5880-4013-ab99-35d40b3902ce","added_by":"auto","created_at":"2025-09-17 13:20:40","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1251859,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKnockdown of ACVR1C abolishes GREM1-CM-induced EMT and metastasis in CRC \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e.\u003c/strong\u003e \u003cstrong\u003eA-C\u003c/strong\u003e, Images (A), tumor weight (B), and tumor volume (C) of subcutaneous tumors from three groups of nude mice (n = 8 per group). Mice were subcutaneously injected with HCT116 cells transduced with either Scramble shRNA or shACVR1C#1 and shACVR1C#2, all pretreated with GREM1-CM. \u003cstrong\u003eD\u003c/strong\u003e, RT–qPCR analysis of mRNA levels of epithelial marker \u003cem\u003eCDH1\u003c/em\u003eand mesenchymal markers \u003cem\u003eCTNNB1\u003c/em\u003e, \u003cem\u003eSNAI1\u003c/em\u003e and \u003cem\u003eZEB1\u003c/em\u003e in subcutaneous tumors. n = 3 independent experiments.\u003cstrong\u003e E-G\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative IF images (E) and quantification (F, G) of E-CAD (green) and SNAIL (red) expression in subcutaneous tumors. Knockdown of ACVR1C reduced SNAIL expression and increased E-CAD expression. Fluorescence intensity was quantified in 3-5 mice per group.\u003cstrong\u003e H\u003c/strong\u003e, \u003cstrong\u003eI\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepresentative IVIS images (H) and quantification (I) of lung metastases.\u003cstrong\u003e \u003c/strong\u003eMice were injected intravenously with HCT116‐luciferase (HCT116‐luc) cells, which were transduced with Scramble shRNA or shACVR1C and treated with GREM1-CM. Luminescence signals from metastatic lesions were monitored over time using the IVIS Lumina Imaging System. Fluorescence signals in the harvested lungs were visualized at day 21 post-tumor injection (upper right panel). For B–D, F, G and I, data are presented as mean ± s.e.m. \u003cem\u003eP\u003c/em\u003e values were calculated using one-way ANOVA with Bonferroni multiple-comparison test.\u003c/p\u003e","description":"","filename":"floatimage11.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/475c2626267bfd1f84e49186.jpeg"},{"id":91536205,"identity":"3f371697-3c60-4b4d-9d32-cc57364251f0","added_by":"auto","created_at":"2025-09-17 13:04:40","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":1297338,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAutocrine GREM1 and GREM1-ACVR1C binding promote EMT and metastasis of CRC \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ein vivo\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e. A\u003c/strong\u003e, \u003cstrong\u003eB\u003c/strong\u003e, Representative IVIS images (A) and quantification (B) of liver metastases. NOG mice's cecum wall was injected with HCT116‐luc cells (pLV/ pLV-GREM1). Liver metastases were imaged and quantified by IVIS Lumina Imaging System (n = 5 mice per group). \u003cstrong\u003eC\u003c/strong\u003e, Docking module of GREM1 and ACVR1C-ECD, highlighting key amino acid sequence between GREM1 and ACVR1C (ACVR1C peptide). \u003cstrong\u003eD\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eIncreasing concentrations of recombinant ACVR1C peptide (0–2.3 μM) were incubated with red-labeled 50 nM recombinant GREM1-his. MST was used to evaluate ACVR1C peptide binding to GREM1-his (n = 3 independent experiments). \u003cstrong\u003eE\u003c/strong\u003e, The pull-down assay confirmed that ACVR1C peptide blocked the interaction between GREM1 and ACVR1C-ECD protein. The black arrow indicates the GREM1-his band. \u003cstrong\u003eF\u003c/strong\u003e, \u003cstrong\u003eG\u003c/strong\u003e, Representative IVIS images (F) and quantification (G) of liver metastases. NOG mice's spleen was injected with HCT116‐luc cells (pLV/ pLV-GREM1). Treatment was started 48 h after cell injection. ACVR1C peptide was given once every other day (10 mg/kg i.v.). i.v.: intravenous injection. Liver metastases were imaged and quantified by IVIS Lumina Imaging System (n = 5 mice per group). \u003cstrong\u003eH\u003c/strong\u003e, A model for paracrine-driven autocrine of GREM1 boosts metastatic potential of CRC cells. For B, E, and G, data are presented as mean ± s.e.m. \u003cem\u003eP\u003c/em\u003e values were calculated using two-tailed Student’s t-test (B) or one-way ANOVA with Bonferroni multiple-comparison test (G).\u003c/p\u003e","description":"","filename":"floatimage13.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/8196a57854d02b3cc9097933.jpeg"},{"id":101152444,"identity":"399a06c3-edb6-43b0-a8a0-59ded2f6d15c","added_by":"auto","created_at":"2026-01-26 16:11:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":14944230,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/43e53a47-7f17-4ce0-8bfb-0913c272ad87.pdf"},{"id":91537168,"identity":"64bfe9bf-1fba-4432-b41c-43926aaf0efa","added_by":"auto","created_at":"2025-09-17 13:12:40","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":12855624,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation44.docx","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/b4ba1ba18317980a0c0c617a.docx"},{"id":91537532,"identity":"8f3e6858-9058-4658-92b3-ea7a83af397e","added_by":"auto","created_at":"2025-09-17 13:20:40","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":235968,"visible":true,"origin":"","legend":"","description":"","filename":"ThemassspectrometrydatasupportFig.2A2BandFig.S3A.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/5c68226059e9cc8f588ded30.xlsx"},{"id":91538576,"identity":"88d32c26-87a4-4230-a69a-a02b9d2c692a","added_by":"auto","created_at":"2025-09-17 13:28:40","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":3596483,"visible":true,"origin":"","legend":"","description":"","filename":"RNAseqgenefpkmTheRNAseqdatasupportFig.S4J.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/709f9e3722b6fe8bcbab5957.xlsx"},{"id":91537164,"identity":"e6f4764c-d484-49e0-a814-747de81bec84","added_by":"auto","created_at":"2025-09-17 13:12:40","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":7390398,"visible":true,"origin":"","legend":"","description":"","filename":"RNAseqgenefpkmTheRNAseqdatasupportFig.3A..xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/7e4fc130be428a4b3a3138d4.xlsx"},{"id":91536186,"identity":"0247212d-7a5f-4f29-80fd-130b45601c46","added_by":"auto","created_at":"2025-09-17 13:04:40","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":1107805,"visible":true,"origin":"","legend":"","description":"","filename":"rawimagesoftheimmunoblottingexperiments.docx","url":"https://assets-eu.researchsquare.com/files/rs-7484753/v1/1f80e03ee14611b0be4fddd8.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Paracrine-to-Autocrine Shunt of GREM1 Fuels Colorectal Cancer Metastasis via ACVR1C","fulltext":[{"header":"Introduction","content":"\u003cp\u003eDistant metastasis remains the leading cause of death among patients with colorectal cancer (CRC)[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. A complex and dynamic interplay between tumor epithelial cells and various non-malignant components of the tumor microenvironment (TME) is recognized as a central driver of tumor initiation, progression, and phenotypic plasticity[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. However, once tumor cells detach from the primary site to metastasize, they inevitably lose the continuous support of the local TME\u0026mdash;including stromal cells and localized signaling cues. This abrupt interruption of paracrine support represents a major bottleneck that restricts most tumor cells from successfully establishing distant colonies[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. This phenomenon raises a fundamental question about how a minority of tumor cells sustain metastasis-associated phenotypes without ongoing microenvironmental support. The capacity to sustain malignant behavior without external cues can be viewed as a form of \u0026lsquo;signaling autonomy,\u0026rsquo; consistent with the classic cancer hallmark of \u0026ldquo;self-sufficiency in growth signals\u0026rdquo; described by Hanahan and Weinberg[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. While signaling autonomy has been extensively explored in the context of proliferation and survival, it remains unclear whether metastasizing tumor cells can acquire autonomous control over the programs that sustain their migratory capacity.\u003c/p\u003e\u003cp\u003eWithin the TME, cancer-associated fibroblasts (CAFs) are among the most active stromal components, driving tumor initiation and progression through bidirectional interactions with cancer cells[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Among the highly heterogeneous CAF population, a GREM1-expressing subset has gained attention for promoting tumor progression via paracrine GREM1 secretion in CRC[\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] and breast cancer[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Intriguingly, GREM1 expression is not restricted to stromal CAFs. In pancreatic[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] and prostate cancers[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], tumor epithelial cells can express and secrete GREM1 to regulate their own phenotypic plasticity, and epithelial-specific GREM1 is markedly upregulated in Hereditary Mixed Polyposis Syndrome (HMPS)[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Nonetheless, whether such epithelial GREM1 expression occurs in sporadic CRC and contributes to cancer cell plasticity or metastasis remains unknown. Studies, including ours, show that modulating GREM1 expression in CRC cell lines affects cell migration and epithelial\u0026ndash;mesenchymal transition (EMT)-related phenotypes[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Yet, whether CRC epithelial cells can \u0026ldquo;hijack\u0026rdquo; stromal GREM1 signals to ultimately activate their own GREM1 expression for malignant progression remains an open question. Mechanistically, GREM1 is a canonical antagonist of bone morphogenetic protein (BMP) signaling[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Recent studies indicate that GREM1 also possesses cytokine-like functions, binding noncanonical receptors, including VEGFR2[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], FGFR1[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], and EGFR[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e], suggesting that it may influence cancer cell behavior through multiple signaling routes. However, it remains unclear whether GREM1 drives signaling autonomy in CRC via these receptors or others yet unidentified.\u003c/p\u003e\u003cp\u003eIn this study, we identify a paracrine-to-autocrine shunt of GREM1 in CRC, driven by the newly recognized receptor ACVR1C[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], a TGFβ superfamily type I receptor, thereby leading to SMAD2/3 phosphorylation. This self-sustaining loop confers tumor cells with signaling autonomy and metastatic potential, uncovering a previously unrecognized mechanism of tumor evolution and a potential approach for therapeutic intervention in CRC.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eEctopic expression of GREM1 during CRC progression\u003c/h2\u003e\u003cp\u003eIn the gut, GREM1 marks a subpopulation of fibroblasts in both normal[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] and tumor tissues[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Using \u003cem\u003eGrem1-CreER\u003c/em\u003e\u003csup\u003e\u003cem\u003eT2\u003c/em\u003e\u003c/sup\u003e;\u003cem\u003eRosa-mTmG\u003c/em\u003e mice[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], we confirmed that Grem1\u003csup\u003e+\u003c/sup\u003e cells are exclusively confined to the stromal compartment, and distributed along the intestinal isthmus and adjacent to α-SMA\u003csup\u003e+\u003c/sup\u003e myofibroblasts[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], two months post-tamoxifen (TMX) injection (Figures \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA\u0026ndash;C). Extending this observation to humans, we found that GREM1 was likewise absent from epithelial cells in normal intestinal tissues, but sporadically expressed in stromal cells (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eD), suggesting a conserved stromal specificity. Similarly, in human stage I-III CRC samples, GREM1 staining co-localized with VIMENTIN (VIM, a stromal cell marker, encoded by \u003cem\u003eVIM\u003c/em\u003e)[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] and fibroblast activation protein (FAP, an activated fibroblast marker)[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], but was mutually exclusive with β-CATENIN (β-CAT, a CRC cell marker, encoded by \u003cem\u003eCTNNB1\u003c/em\u003e)[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], CD68 (a macrophage marker)[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], or α-SMA (a myofibroblast marker) (Figures \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eE\u0026ndash;J). These findings confirm that GREM1 is a bona fide stromal factor and GREM1\u003csup\u003e+\u003c/sup\u003e stromal cells are a subtype of cancer-associated fibroblasts (CAFs), potentially contributing to CRC progression[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eTo systematically investigate the distribution and clinical significance of GREM1, we first performed immunohistochemical (IHC) staining on 106 human primary CRC samples spanning all four stages. We observed a stage-dependent redistribution of GREM1⁺ cells: in early-stage tumors (stage I\u0026ndash;II), GREM1⁺ stromal cells were predominantly restricted to the peritumoral stroma; in stage III, these cells more frequently infiltrated the tumor parenchyma; in stage IV, infiltration of GREM1⁺ stromal cells showed a decreasing trend (with no statistically significant difference compared to stage III). Notably, in stage IV tumors, strong GREM1 staining emerged in a subset of tumor epithelial regions (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA\u0026ndash;C). These findings suggest a potential shift of GREM1 expression from stroma to epithelium during tumor progression.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo dissect the cellular sources and stage-specific dynamics of GREM1 expression at higher resolution, we analyzed publicly available single-cell RNA sequencing (scRNA-seq) datasets covering CRC samples from stages I\u0026ndash;IV. Consistent with IHC data, GREM1 expression was primarily detected in fibroblasts and epithelial cells (Figures \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eA\u0026ndash;C). Stage-specific analysis revealed a marked upregulation of GREM1 in fibroblasts at stages III and IV (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eD). Strikingly, GREM1⁺ epithelial cells were detected almost exclusively in stage IV tumors (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eE), supporting the notion that ectopic GREM1 expression by tumor cells is a late event in CRC progression. This observation was further validated by colocalization of GREM1⁺ with an epithelial marker epithelial cell adhesion molecule (EPCAM)[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] in tumor cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). Finally, survival analysis demonstrated that patients with high GREM1 expression in tumor cells had significantly shorter overall survival compared to those with low expression (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE). Taken together, these results reveal that while GREM1⁺ CAFs are present throughout CRC progression, GREM1⁺ tumor epithelial cells emerge predominantly in advanced CRC, indicating that GREM1 expression, initially restricted to stromal cells, is progressively co-opted by tumor epithelial cells as CRC advances.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eACVR1C is a novel GREM1 receptor in CRC\u003c/h3\u003e\n\u003cp\u003eThis spatiotemporal ectopic expression of GREM1 suggests that tumoral autocrine signaling may be initiated by preceding stromal paracrine cues. Given the potential for intercellular communication mediated by the infiltration of GREM1⁺ stromal cells into the tumor parenchyma, we hypothesized that CRC tumor cells might express GREM1 receptor(s), through which downstream signaling cascades drive tumoral GREM1 expression and promote CRC progression. To identify potential GREM1 receptors, we overexpressed HA-tagged GREM1 in the human CRC cell line HCT116 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Mass spectrometry analysis of proteins pulled down using anti-HA beads identified activin A receptor type 1C (ACVR1C), a member of the TGFβ superfamily, as a potential GREM1 receptor (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB and S3A).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo validate this interaction, we performed co-immunoprecipitation (co-IP) assays using HA-tagged GREM1 in HCT116 cells. Immunoblotting revealed that GREM1 interacted with ACVR1C specifically, while no such interaction was detected with other members of the TGFβ superfamily such as TGFβR1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Similarly, co-IP assays using Flag-tagged ACVR1C confirmed the interaction with GREM1, which was abolished by a GREM1-blocking antibody (BAb) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). Confocal microscopy revealed the co-localization of ACVR1C with GREM1 in SW480 CRC cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). We next examined whether a direct interaction exists between GREM1 and ACVR1C. We found that Fc-tagged ACVR1C extracellular domain (ACVR1C-ECD, AA\u003csub\u003e1\u0026thinsp;\u0026minus;\u0026thinsp;113\u003c/sub\u003e) and His-tagged full-length GREM1 were pulled down together, demonstrating a direct physical association between ACVR1C and GREM1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF). Further analysis of the binding affinity of GREM1 for ACVR1C using microscale thermophoresis (MST) revealed that ACVR1C-ECD exhibited a 12.6-fold higher affinity for GREM1 (\u003cem\u003eK\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e = 67.67\u0026thinsp;\u0026plusmn;\u0026thinsp;10.35 nM) than that for ACTIVIN B (\u003cem\u003eK\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e = 854.1\u0026thinsp;\u0026plusmn;\u0026thinsp;127.47 nM), a known ligand of ACVR1C[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eG).\u003c/p\u003e\u003cp\u003eTo further delineate the explicit interaction mode of GREM1 and ACVR1C, we constructed truncated GREM1 and ACVR1C-ECD. Co-IP assays showed that deletion of amino acids 100\u0026ndash;157 (AA\u003csub\u003e100\u0026ndash;157\u003c/sub\u003e) in GREM1 or 68\u0026ndash;113 (AA\u003csub\u003e68\u0026ndash;113\u003c/sub\u003e) in ACVR1C-ECD effectively abolished their interaction in HCT116 cells (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eH, I). Based on these findings, we aimed to identify key residues mediating the interaction between GREM1 and ACVR1C. The HDOCK platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://hdock.phys.hust.edu.cn/\u003c/span\u003e\u003cspan address=\"http://hdock.phys.hust.edu.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was utilized to simulate potential docking modalities between GREM1 (PDB: 5AEJ)[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] and ACVR1C (AlphaFold prediction, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://alphafold.ebi.ac.uk/entry/Q8NER5\u003c/span\u003e\u003cspan address=\"https://alphafold.ebi.ac.uk/entry/Q8NER5\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) based on their structures. Residues Q101/T102/T112/N115 in GREM1 and Q72/E85/T101 in ACVR1C were predicted to be essential for binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eJ). To validate, we performed site-directed mutagenesis of the predicted residues to assess binding. Notably, the Q101A/T102A/T112A/N115A quadruple mutation in GREM1, or single mutations E85A or T101A in ACVR1C, significantly abrogated the GREM1-ACVR1C interaction (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eK, L). Further, we found no detectable interaction between recombinant GREM1 and the ACVR1C-ECD double mutant (E85A/T101A) by assessing their binding affinity using MST (Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eB). These data suggest that Q101/T102/T112/N115 in GREM1, and E85/T101 in ACVR1C are key residues mediating their interaction. Clinically, IHC and scRNA-seq revealed that ACVR1C is expressed in tumor cells, with markedly elevated levels in stage IV CRC (Figures \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eC\u0026ndash;E). Notably, high ACVR1C expression correlates with poor prognosis in stage IV CRC patients, supporting ACVR1C\u0026rsquo;s tumor-promoting role and consistent with the clinical significance of GREM1 (Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eF). Taken together, these results demonstrate that ACVR1C is a novel receptor of GREM1 in CRC.\u003c/p\u003e\n\u003ch3\u003eSecretory GREM1 induces EMT via the ACVR1C-SMAD2/3 pathway but not TGFβR/BMPR pathways\u003c/h3\u003e\n\u003cp\u003eTo explore whether GREM1 serves as a functional ligand for ACVR1C in CRC cells, we first generated GREM1-enriched conditioned medium (GREM1-CM) and control conditioned medium (Vec-CM) using HEK293 cells (Figures \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eA\u0026ndash;C). We then performed RNA-sequencing (RNA-seq) on HCT116 cells treated with GREM1-CM or Vec-CM. Gene Set Enrichment Analysis (GSEA) revealed that SMAD2/3 and EMT pathways were significantly enriched in CRC cells treated with GREM1-CM (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). ACVR1C is one of the receptors of the TGFβ superfamily, and it transduces signals primarily through the phosphorylation of SMAD2/3 (p-SMAD2/3)[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Since commercial antibody for phosphorylated ACVR1C is not available, detection of p-SMAD2/3 serves as an effective proxy to reflect ACVR1C-SMAD2/3 activation. To confirm the effect of secretory GREM1 on the ACVR1C-SMAD2/3 pathway and EMT in CRC cells, we performed immunoblotting and found that GREM1-CM significantly increased p-SMAD2/3 levels in HCT116 and SW480 cells, which was effectively blocked by a GREM1 BAb (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Moreover, immunoblotting and RT\u0026ndash;qPCR analyses also revealed that GREM1-CM induced significant downregulation of E-CADHERIN (E-CAD, encoded by \u003cem\u003eCDH1\u003c/em\u003e)[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] and upregulation of mesenchymal markers, including SNAIL (encoded by \u003cem\u003eSNAI1\u003c/em\u003e), ZEB1, and β-CAT[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] in SW480 and HCT116 cells. GREM1 BAb effectively blocked GREM1-CM-induced EMT activation (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB and S4D, E). Considering that EMT serves as an effective mechanism through which tumor cells acquire stroma-like traits to promote invasion and metastasis, we tested whether blocking GREM1 could inhibit the invasive and migratory capacity of CRC cells. Indeed, \u003cem\u003ein vitro\u003c/em\u003e scratch and transwell assays showed that GREM1-CM significantly enhanced migration and invasion abilities of HCT116 and SW480 cells. Remarkably, these effects were abolished by GREM1 BAb treatment (Figures \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eF\u0026ndash;I). These findings suggest that secretory GREM1 activates the ACVR1C-SMAD2/3 pathway and promotes EMT and subsequent cellular behavior, such as migration and invasion of CRC cells.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe TGFβ superfamily signals through BMPRs, ACVRs, and TGFβRs. Although both ACVRs and TGFβRs converge on the SMAD2/3 axis[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], and GREM1-CM robustly activated TGFβ superfamily signaling in HEK293 cells (Figure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eJ), our co-IP analysis showed no direct interaction between GREM1 and TGFβR1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). To determine whether GREM1 activates SMAD2/3 via TGFβR1 or ACVR1C, we treated SW480 and HCT116 cells with increasing concentrations of recombinant human GREM1 (rhGREM1). This led to a dose-dependent increase in p-SMAD2/3 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC), suggesting activation of a SMAD2/3-coupled receptor. To exclude the possibility that GREM1 indirectly stimulates TGFβ signaling, we examined whether rhGREM1 alters the expression of TGFβ or TGFβR1 in CRC cells. No changes were observed across all doses tested (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC), indicating that GREM1 does not upregulate endogenous TGFβ signaling components. Together, these data suggest that GREM1-induced SMAD2/3 activation is unlikely to be mediated by TGFβR1 and instead proceeds via the ACVR1C pathway (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD).\u003c/p\u003e\u003cp\u003eConsidering that GREM1 is a canonical antagonist of BMP and that BMP receptors (BMPRs) exert their function through the phosphorylation of SMAD1/5/9 (p-SMAD1/5/9), we sought to investigate whether GREM1 regulates EMT via BMPR superfamily pathways. As expected, rhGREM1 suppressed the p-SMAD1/5/9 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC), in keeping with the canonical role of GREM1 as a BMP inhibitor. However, rescue of BMP signaling using the specific agonist sb4, which exclusively increases p-SMAD1/5/9 levels without affecting p-SMAD2/3[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], failed to reverse GREM1-induced EMT marker changes (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE). This definitive exclusion of BMPR-SMAD1/5/9 involvement establishes that GREM1 promotes EMT independently of its classical role as a BMP inhibitor.\u003c/p\u003e\u003cp\u003eSubsequently, to examine whether GREM1 promotes EMT through activation of the ACVR1C-SMAD2/3 pathway, we either stably knocked down ACVR1C (shACVR1C) or inhibited SMAD2/3 phosphorylation using SB505124[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Our immunoblotting and RT\u0026ndash;qPCR analysis revealed that shACVR1C or SB505124 significantly blocked GREM1-CM-induced changes in EMT marker expression (\u003cem\u003ei.e.\u003c/em\u003e E-CAD, ZEB1, β-CAT and SNAIL), indicating that ACVR1C-SMAD2/3 activation is required for GREM1-driven EMT in SW480 and HCT116 CRC cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF, g and S5A, B and S6A, B). In addition, GREM1-CM-induced invasion and migration of these CRC cells was abolished by shACVR1C or SB505124 (Figures \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eC\u0026ndash;G and S6C\u0026ndash;F). Collectively, these findings demonstrate that GREM1 induces EMT, as well as subsequent migration and invasion, by activating the ACVR1C-SMAD2/3 pathway.\u003c/p\u003e\u003cp\u003ep-SMAD2/3 form a complex with SMAD4 that translocates into the nucleus and acts as a transcriptional regulator[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. \u003cem\u003eSNAI1\u003c/em\u003e has been identified as a transcriptional target of the SMAD2/3/4 complex in several cancers, including CRC[\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. To define the direct binding sites involved in this regulation in CRC cells, we queried the JASPAR database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://jaspar.genereg.net/\u003c/span\u003e\u003cspan address=\"http://jaspar.genereg.net/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and identified three candidate SMAD2/3/4 binding sites (\u0026ndash;967, \u0026minus;\u0026thinsp;787, and \u0026minus;\u0026thinsp;186) within the \u003cem\u003eSNAI1\u003c/em\u003e promoter (Figure S6G). Chromatin immunoprecipitation (ChIP) followed by qPCR confirmed specific binding of SMAD2/3/4 to the \u0026minus;\u0026thinsp;787 site. Importantly, inhibition of ACVR1C with SB505124 significantly decreased this binding (Figure S6H). Overall, our data reveal that secretory GREM1 is a specific functional ligand that activates the ACVR1C\u0026ndash;SMAD2/3\u0026ndash;SNAIL signaling axis, thereby promoting EMT, invasion and migration of CRC cells.\u003c/p\u003e\u003cp\u003e\u003cb\u003eExogenous GREM1 induction of endogenous\u003c/b\u003e \u003cb\u003eGREM1\u003c/b\u003e \u003cb\u003etranscription reinforces EMT in CRC cells via the ACVR1C-SMAD2/3 pathway\u003c/b\u003e\u003c/p\u003e\u003cp\u003eOur data above show that there is a marked increase in the proportion of GREM1\u003csup\u003e+\u003c/sup\u003e epithelial tumor cells in stage IV CRC (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC and S2E). To determine whether exogenous GREM1 can regulate endogenous GREM1 expression, we treated SW480 cells with increasing concentrations of rhGREM1. RT-qPCR and immunoblotting analyses revealed a dose-dependent upregulation of tumor GREM1 expression in response to rhGREM1 stimulation (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA\u0026ndash;C). However, the mechanism by which exogenous GREM1 triggers endogenous \u003cem\u003eGREM1\u003c/em\u003e transcription in advanced CRC cells remains unclear. Notably, we also observed that ACVR1C expression was markedly upregulated in stage IV CRC (Figures \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eD, E), coinciding with the emergence of epithelial GREM1 expression, whereas stromal GREM1 was already abundantly present in both stage III and IV tumors (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB, C and S2D, E). This spatiotemporal concordance suggests that epithelial GREM1 induction may depend on elevated ACVR1C expression. To investigate the role of the ACVR1C-SMAD2/3 pathway in the exogenous GREM1-mediated tumor \u003cem\u003eGREM1\u003c/em\u003e transcription, we either overexpressed ACVR1C or inhibited the pathway using SB505124 in SW480 cells following rhGREM1 treatment. Interestingly, ACVR1C overexpression further enhanced endogenous \u003cem\u003eGREM1\u003c/em\u003e expression, whereas SB505124 treatment completely reversed the effect of rhGREM1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). To investigate whether SMAD2/3/4 act as transcription factors for \u003cem\u003eGREM1\u003c/em\u003e, we utilized the JASPAR database and identified five candidate SMAD2/3/4 binding sites (-733, -612, -446, -316, and \u0026minus;\u0026thinsp;3) in the \u003cem\u003eGREM1\u003c/em\u003e promoter region (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE). To validate the predicted results, we performed a ChIP\u0026ndash;qPCR analysis, which revealed that SMAD2/3/4 bound the \u003cem\u003eGREM1\u003c/em\u003e promoter at the \u0026minus;\u0026thinsp;733 and \u0026minus;\u0026thinsp;612 sites. Notably, SB505124 treatment significantly reduced this binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eF). These data demonstrate that exogenous GREM1 efficiently induces endogenous \u003cem\u003eGREM1\u003c/em\u003e transcription in CRC cells via the ACVR1C-SMAD2/3 signaling pathway.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eHaving established that exogenous GREM1 promotes EMT in CRC cells, we sought to determine whether endogenous GREM1 exerts a similar function. In normal epithelial cells, β-CAT is localized at the cellular membrane with the adhesion molecule E-CAD[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. However, during tumor progression and the onset of EMT, E-CAD is gradually lost, and a portion of β-CAT translocates to the cytoplasm and nucleus[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. To delineate the bona fide correlation between epithelial GREM1 expression and EMT hallmarks in CRC, we performed immunofluorescence (IF) staining on stage IV CRC clinical samples, which contained both GREM1\u003csup\u003e+\u003c/sup\u003e and GREM1\u003csup\u003e\u0026minus;\u003c/sup\u003e CRC cells. Remarkably, GREM1\u003csup\u003e+\u003c/sup\u003e CRC cells exhibited a significant loss of E-CAD, along with increased β-CAT expression and its translocation to the cytoplasm and nucleus, compared to adjacent GREM1\u003csup\u003e\u0026minus;\u003c/sup\u003e counterparts within the same tumor (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eG\u0026ndash;J).\u003c/p\u003e\u003cp\u003eWe previously found that forced \u003cem\u003eGREM1\u003c/em\u003e expression (pLV-GREM1) in CRC cells enhanced their EMT and metastatic traits[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Given the typically low expression of GREM1 in CRC cell lines compared to normal and tumoral fibroblasts (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eK), we sought to model the therapeutic blockade of endogenous GREM1 in advanced tumors. To this end, we used CRISPR/Cas9 techniques to knock out \u003cem\u003eGREM1\u003c/em\u003e in SW480 and HCT116 cells that stably expressed pLV-GREM1. RT\u0026ndash;qPCR analysis revealed that pLV-GREM1 resulted in a significant change in EMT markers, including a marked decrease in \u003cem\u003eCDH1\u003c/em\u003e and an increase in \u003cem\u003eSNAI1, VIM\u003c/em\u003e, and \u003cem\u003eZEB1\u003c/em\u003e, while \u003cem\u003eGREM1\u003c/em\u003e knockout significantly restored the expression of these markers, indicating that endogenous GREM1 promotes EMT within CRC cells (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eL, M). Collectively, these data demonstrate that exogenous GREM1 induces endogenous \u003cem\u003eGREM1\u003c/em\u003e transcription in CRC cells through the ACVR1C\u0026ndash;SMAD2/3 pathway, establishing a self-propelling loop that promotes EMT. Given our observation of a reduction in GREM1⁺ stromal cells in stage IV CRC compared to stage III (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB), we speculate that this self-sustaining GREM1 feedback loop enables CRC cells to maintain EMT independently of stromal inputs, reducing their reliance on stromal GREM1. This autonomous signaling cascade comprises multiple nodes that may be amenable to therapeutic intervention aimed at halting CRC metastasis.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDepletion of Grem1\u003c/b\u003e\u003csup\u003e\u003cb\u003e+\u003c/b\u003e\u003c/sup\u003e \u003cb\u003estromal cells inhibits EMT and metastasis of CRC\u003c/b\u003e \u003cb\u003ein vivo\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo validate the GREM1-ACVR1C-induced autocrine GREM1 feedback loop, we applied genetic and pharmacological strategies (Figure S6I). First, we evaluated the impact of stromal paracrine GREM1 on CRC \u003cem\u003ein vivo\u003c/em\u003e. To block exogenous GREM1, we crossed \u003cem\u003eGrem1-CreER\u003c/em\u003e\u003csup\u003e\u003cem\u003eT2\u003c/em\u003e\u003c/sup\u003e;\u003cem\u003eRosa-LSL-DTA\u003c/em\u003e mice with \u003cem\u003eAPC\u003c/em\u003e\u003csup\u003e\u003cem\u003eMin/+\u003c/em\u003e\u003c/sup\u003e mice[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e], a well-established model for CRC proliferation and EMT that develops intestinal tumors by 10 weeks[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e], generating AGD mice for TMX-induced depletion of GREM1\u003csup\u003e+\u003c/sup\u003e stromal cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). Intriguingly, ablation of Grem1\u003csup\u003e\u003cem\u003e+\u003c/em\u003e\u003c/sup\u003e stromal cells for 6 weeks postnatally did not significantly alter the number or size of intestinal tumors in \u003cem\u003eAPC\u003c/em\u003e\u003csup\u003e\u003cem\u003eMin/+\u003c/em\u003e\u003c/sup\u003e mice (Figures S7A, B). Subsequently, we delved deeper into whether loss of paracrine GREM1 could restrain the malignant potential of intestinal tumor cells. As expected, IF staining in the GREM1\u003csup\u003e\u003cem\u003e+\u003c/em\u003e\u003c/sup\u003e stromal cell infiltration zone of the \u003cem\u003eAPC\u003c/em\u003e\u003csup\u003e\u003cem\u003eMin/+\u003c/em\u003e\u003c/sup\u003e intestines revealed a marked loss of E-cad in tumor cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, upper left panels), accompanied by increased β-cat expression and its translocation to the cytoplasm and nucleus (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, upper right panels). In contrast, upon depletion of GREM1\u003csup\u003e\u003cem\u003e+\u003c/em\u003e\u003c/sup\u003e stromal cells in AGD mice, E-cad expression was significantly elevated (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC), with intense and continuous localization along the tumor cell membrane (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, bottom left). In parallel, β-cat staining was reduced and restricted to the cell membrane (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD, E), co-localizing with E-cad, and rarely observed in the cytoplasm or nucleus (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, bottom right panels). These results suggest a critical role for exogenous Grem1 in orchestrating malignant cell behaviors. Next, to evaluate the impact of paracrine Grem1 on CRC metastasis, we injected luciferase-labeled murine rectal cancer cells (MC38-luc) into Grem1\u003csup\u003e\u003cem\u003e+\u003c/em\u003e\u003c/sup\u003e cell-depleted (GD) or control mice (Figures S7C, D). Cells were administered via the tail vein to induce lung metastasis, or into the spleen or cecum wall to induce liver metastasis[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e] (Figures S7E\u0026ndash;G). Strikingly, we observed a significant reduction in lung and liver metastases of CRC cells in GD mice compared with \u003cem\u003eGrem1-CreER\u003c/em\u003e\u003csup\u003e\u003cem\u003eT2\u003c/em\u003e\u003c/sup\u003e or \u003cem\u003eRosa-LSL-DTA\u003c/em\u003e controls (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF\u0026ndash;I and S7H\u0026ndash;J), suggesting that the stromal factor Grem1 is vital for CRC metastasis.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eInhibition of ACVR1C-SMAD2/3 pathway inhibits EMT and metastasis of CRC\u003c/b\u003e \u003cb\u003ein vivo\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo determine the role of ACVR1C and its downstream SMAD2/3 pathway in GREM1-mediated EMT \u003cem\u003ein vivo\u003c/em\u003e, we first pre-treated HCT116 cells carrying a luciferase reporter (HCT116-luc), stably expressing either shACVR1C or scramble shRNA, with GREM1-CM, followed by subcutaneous transplantation into nude mice. Notably, ACVR1C knockdown significantly inhibited subcutaneous tumor growth (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA\u0026ndash;C), accompanied by a significant increase in epithelial gene expression (e.g. \u003cem\u003eCDH1\u003c/em\u003e) and a marked decrease in mesenchymal gene expression (e.g. \u003cem\u003eSNAI1, CTNNB1\u003c/em\u003e, and \u003cem\u003eZEB1\u003c/em\u003e), as shown by RT\u0026ndash;qPCR analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). These findings were further corroborated by IF staining of EMT markers, including E-CAD and SNAIL (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eE\u0026ndash;G). Meanwhile, we also harnessed SB505124 as a pharmacological alternative to knock down ACVR1C. Consistent with the genetic approach, SB505124 treatment yielded similar effects on local tumor progression, further substantiating the contextual role of the ACVR1C-SMAD2/3 pathway (Figures S8A\u0026ndash;G). Importantly, to investigate the impact of the ACVR1C-SMAD2/3 axis on metastasis, we injected GREM1-CM-pre-treated HCT116-luc cells, stably expressing either shACVR1C or scramble shRNA, into the tail vein of nude mice. Remarkably, lung metastasis was profoundly suppressed in the \u003cem\u003eACVR1C\u003c/em\u003e knockdown group compared with controls (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eH, I). In parallel, we inoculated GREM1-CM-pre-treated HCT116-luc cells into the cecum wall of NOG (NOD/Shi-scid/IL-2Rγ) mice, a new generation of severely immunodeficient mice[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. We found that SB505124 treatment resulted in a significant reduction in liver metastasis (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eI and S8 H, I). Collectively, these findings establish the ACVR1C-SMAD2/3 axis as a critical effector of stroma-derived GREM1, and reveal that its inhibition provides an effective strategy to counteract GREM1-induced EMT and metastasis in CRC \u003cem\u003ein vivo\u003c/em\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eEpithelial GREM1 enhances EMT and metastasis of CRC\u003c/b\u003e \u003cb\u003ein vivo\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFurther, to examine the contribution of tumor-autocrine GREM1 to EMT and metastasis of CRC \u003cem\u003ein vivo\u003c/em\u003e, we first inoculated HCT116 cells carrying pLV-GREM1 into nude mice, which led to significantly enhanced subcutaneous tumor growth (Figures S9A\u0026ndash;C). Subsequent RT\u0026ndash;qPCR analysis demonstrated that autocrine GREM1 upregulated \u003cem\u003eSNAI1, VIM\u003c/em\u003e, and \u003cem\u003eZEB1\u003c/em\u003e while downregulating \u003cem\u003eCDH1\u003c/em\u003e (Figure S9D). These findings were confirmed by IF staining for EMT markers, including E-CAD and SNAIL (Figures S9E\u0026ndash;G), supporting that autocrine GREM1 enhances the EMT process in CRC cells. Next, to confirm the role of tumoral GREM1 in enhancing CRC metastasis, we conducted metastasis assays by injecting HCT116-luc cells expressing either pLV-GREM1 or the control vector via the tail vein or into the cecum wall. We found that overexpression of tumoral GREM1 resulted in a significant increase in lung and liver metastasis compared with controls (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA, B and S9H, I). In summary, these findings demonstrate that tumoral GREM1 promotes both EMT and metastasis of CRC cells \u003cem\u003ein vivo\u003c/em\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eTargeting the GREM1-ACVR1C interaction interface to inhibit metastasis of CRC\u003c/b\u003e \u003cb\u003ein vivo\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn clinical research, the molecular complexity and shared signaling of CRC limit the effective targeted therapy. While the TGFβ superfamily is a promising therapeutic axis[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], its intertwined and overarching roles in development and physiology make selective targeting difficult without systemic toxicity[\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Targeting protein\u0026ndash;protein interaction (PPI) interfaces is thought to further enhance specificity and reduce off-target effects[\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. Based on the principle of GREM1-ACVR1C interface disruption, we designed a peptide inhibitor derived from amino acid residues 84\u0026ndash;102 (AA84-102) of ACVR1C (hereafter referred to as the ACVR1C peptide) (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC). We further examined the binding affinity between GREM1 and the ACVR1C peptide using MST. We found that GREM1 exhibited an affinity for the ACVR1C peptide comparable to that for the ACVR1C-ECD (\u003cem\u003eK\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e = 92.30\u0026thinsp;\u0026plusmn;\u0026thinsp;7.51 nM) (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD). Pull-down assays showed that the introduction of ACVR1C peptide significantly blocked the GREM1-ACVR1C binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eE), suggesting that the peptide disrupts the GREM1-ACVR1C interaction by potently and competitively binding to GREM1. Further, we employed a spleen-to-liver metastasis model to evaluate the functional effect of the ACVR1C peptide in CRC metastasis. Notably, administration of the ACVR1C peptide effectively suppressed the increase in liver metastasis induced by tumor-specific GREM1 overexpression (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eF, G), suggesting that our ACVR1C peptide significantly attenuated CRC progression by blocking the metastasis-promoting effects of tumor-autonomous autocrine GREM1-ACVR1C signaling, laying the foundation for the development of a novel class of peptide-based targeted therapies in CRC.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eWhether tumor cells can internalize stromal paracrine signals and convert them into autocrine loops during metastasis has remained unclear. Here, we show that in advanced colorectal cancer, GREM1 undergoes a stromal-to-epithelial shunt and binds ACVR1C to activate SMAD2/3, inducing both \u003cem\u003eSNAI1\u003c/em\u003e and \u003cem\u003eGREM1\u003c/em\u003e expression. This establishes a self-amplifying autocrine circuit that drives EMT, conferring signaling autonomy and promoting metastasis (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eH). A functional peptide we designed for targeting the GREM1\u0026ndash;ACVR1C interface effectively disrupts this loop and demonstrates therapeutic potential.\u003c/p\u003e\u003cp\u003eDuring embryonic development and tissue repair, intercellular communication is precisely regulated by paracrine and autocrine signaling pathways[\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Similarly, tumor cells exploit these dual signaling modes to enhance their survival and invasive capacity[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Moreover, during dissemination to distant sites, metastatic tumor cells have been observed to bring along components of the primary tumor stroma, including CAFs, reflecting their persistent reliance on microenvironmental support[\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]. Sporn et al. first proposed that a potential mechanism of malignant transformation is the autocrine production of growth factors to which the cell itself can respond, a process that may originate from the reactivation of autocrine strategies employed during early embryonic development, enabling cells to survive even in the absence of external support[\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e, \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. Weinberg et al. further suggested that paracrine signals from the stroma might trigger the emergence of signaling autonomy in tumor cells. While neuron-secreted NLGN3 has been reported to upregulate NLGN3 expression in glioma cells, it has not been demonstrated whether this process evolves into a self-sustaining autocrine loop [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. Thus, whether tumor cells can transition from paracrine induction to a self-sustaining autocrine circuit, remains unresolved. In this study, we show that during CRC metastasis, GREM1 expression undergoes a stromal-to-epithelial shunt, establishing an autocrine signaling mode initiated by paracrine cues. This transition suggests that metastatic cells acquire signaling autonomy by reactivating developmental autocrine programs, enabling them to survive and disseminate in distant organs with reduced microenvironmental support. Such a shift from \u0026ldquo;dependence on soil\u0026rdquo; to \u0026ldquo;self-construction of soil\u0026rdquo; reflects the remarkable adaptability of tumor cells.\u003c/p\u003e\u003cp\u003eAs a canonical BMP antagonist in the TGF-β superfamily, GREM1 plays a pivotal regulatory role across diverse physiological and pathological contexts. During embryogenesis, GREM1 modulates BMP gradients to orchestrate the development of organs such as the kidney[\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e], skeleton, and gut[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. In the post-developmental setting, GREM1-mediated BMP signaling also plays an important role in benign conditions such as tissue repair[\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e], and fibrosis[\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e]. In the context of cancer, GREM1 displays marked tissue-specific functionality. In hereditary mixed polyposis syndrome (HMPS), duplication of an upstream enhancer leads to aberrant overexpression of GREM1 in epithelial cells, disrupting the BMP-driven stemness gradient and initiating polyp formation[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e]. GREM1 can also enhance stemness and drive cancer progression by suppressing BMP signaling[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. In contrast, in pancreatic ductal adenocarcinoma, GREM1 expression constrains EMT and promotes differentiation toward a less invasive epithelial phenotype[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. While traditionally categorized as a BMP antagonist, portraying GREM1 solely as an inhibitor overlooks its diverse, sometimes paradoxical roles in various biological settings[\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e]. Accumulating evidence suggests that GREM1 possesses non-canonical, BMP-independent activities, engaging multiple receptors and activating diverse downstream signaling pathways. Previous studies have implicated EGFR[\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e] and VEGFR2[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e] as putative GREM1 targets, though their binding affinities have not been determined. Zhu et al. reported that GREM1 binds to FGFR1 with a dissociation constant of \u003cem\u003eK\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e = 10.6 nM[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. In this study, we further identified ACVR1C[\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e] as a novel binding receptor for GREM1, with a dissociation constant of \u003cem\u003eK\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e = 67.67\u0026thinsp;\u0026plusmn;\u0026thinsp;10.35 nM. Notably, this binding affinity is markedly higher than that of the canonical ACVR1C ligand Activin B (\u003cem\u003eK\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e = 854.1\u0026thinsp;\u0026plusmn;\u0026thinsp;127.47 nM), representing a 12.6-fold increase. These results suggest that GREM1 engages ACVR1C with superior affinity, indicating its potential to act as a dominant ligand within this signaling pathway.\u003c/p\u003e\u003cp\u003eUnlike the mechanism described by Lan et al. in pancreatic cancer[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], where GREM1 inhibits EMT via classical BMP antagonism, our findings demonstrate that GREM1 binds ACVR1C and activates the SMAD2/3 axis independently of BMP signaling. This interaction initiates a positive autocrine feedback loop that promotes EMT and metastasis in CRC. These contrasting roles highlight the context-dependent plasticity of GREM1, which engage distinct signaling programs across tissue types to drive divergent outcomes. Although GREM1 is not traditionally classified as a cytokine, emerging evidence suggests it acts as an adipokine in adipose tissue, modulating metabolic homeostasis[\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e]. Our study extends this concept by showing that GREM1 exhibits key cytokine characteristics, including active secretion, receptor engagement, downstream signaling activation, and autocrine feedback amplification. Reframing GREM1 as a cytokine beyond BMP antagonism may provide new insights into its pleiotropic roles in cancer, metabolic disorders, and developmental or regenerative processes.\u003c/p\u003e\u003cp\u003eThe advent of cancer genomics and precision medicine has enabled targeted therapies for advanced CRC, with agents against VEGF, EGFR, BRAF V600E, and HER2 showing efficacy in selected patient subsets. However, issues like limited bioavailability, resistance, and narrow indications restrict their broader use[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Unlike mutation-specific targets, GREM1 is a widely expressed cytokine with sustained, context-dependent activity, making it a promising and potentially broadly applicable therapeutic target in CRC. Therapeutic targeting of GREM1 to date has focused largely on full neutralization strategies using monoclonal antibodies. Fully humanized monoclonal antibodies such as Ginisortamab and TST003 have advanced into phase I/II clinical trials[\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. Although these antibodies have shown efficacy in preclinical models, the indispensable roles of GREM1 in physiological processes such as intestinal homeostasis and bone marrow hematopoiesis raise the risk of adverse effects from systemic inhibition, posing a major challenge to their clinical translation[\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e]. In recent years, PPI have emerged as attractive therapeutic targets, with small peptides showing particular promise due to their high binding affinity and specificity[\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e]. In this study, we exploited the structural interface of the GREM1\u0026ndash;ACVR1C interaction as a therapeutic entry point. Through structure-guided molecular design, we developed a high-affinity peptide capable of specifically disrupting this interaction, thereby markedly suppressing metastatic potential. Unlike conventional antibodies, our approach selectively targets the pathogenic GREM1\u0026ndash;ACVR1C axis, highlighting the translational promise of rational PPI-targeted cancer therapies.\u003c/p\u003e\u003cp\u003eIn summary, our findings reveal that the GREM1\u0026ndash;ACVR1C axis acts as a key mediator of the stromal-to-epithelial shunt and the establishment of autonomous GREM1 signaling in CRC. By uncovering this pathway, we not only deepen our understanding of tumor self-sufficiency mechanisms, but also identify a functionally precise intervention point that opens new avenues for disrupting metastatic competence through targeted dismantling of self-reinforcing oncogenic circuits.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank Ruohan Li and Changxue Li for assisting with IHC data scoring; Guihua Wang for providing the \u003cem\u003eAPC\u003csup\u003eMin/+\u003c/sup\u003e\u003c/em\u003e mouse line; Pengcheng Bu for assisting with the construction of the cecum to liver metastasis mouse model; Lei Zhou for drawing the GREM1-ACVR1C protein docking; and Xingqiao Xie from the Sample Preparation and Analysis Core Facility of Shenzhen Medical Academy of Research and Translation (SMART) for technical support of MST. \u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eH.Z. and N.L. conceived and designed the study and wrote the manuscript. H.Z. performed most of the experiments and analyses. Q.J. and Y.J. helped with RT-qPCR analysis and clinical tissue IF experiments. Z.F. and Y.Y. helped with subcutaneous tumor IF experiments. Y.Y. assisted with the design and drawing of schematics and mechanistic diagrams. Y.G. assisted with histopathological assessment. N.W. and B.Z. performed ACVR1C-ECD and ACVR1C-ECD-doble-mutant proteins expression and purification. J.L. helped with single-cell RNA-seq data analysis. Z.Z. helped with the establishment of the cecum to liver metastasis mouse model. L.G., Y.Z. (Affiliation 1), Y.H., Y.Z. (Affiliation 7), and J.Z. provided intellectual feedback. Note: Y.Z. (Affiliation 1) and Y.Z. (Affiliation 7) are distinct individuals. S.L. and L.F. assisted with data interpretation and analysis and text proofreading. X.W. helped with CHIP experiments. T.W. helped with RNA-seq data analysis. X.S., T.W., Y.J. and N.L. supervised the study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by grants from the National Natural Science Foundation of China (81874176, 82072766,\u0026nbsp;32471261), Guangdong Provincial Key Laboratory of Digestive Cancer Research (2021B1212040006), the Sanming Project of Medicine in Shenzhen (SZSM202111005), Funding of Shenzhen Clinical Research Center for Gastroenterology (Gastrointestinal Surgery) (LCYSSQ20220823091203008), Guangdong Medical Science and Technology Research Foundation (A2022068).\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe use of human CRC tissues was approved by the Ethics Committee of Sun Yat-sen University (Approval No. KY-2023-071-01). Informed consent was obtained from all participants, and the study adhered to the principles of the 1975 Declaration of Helsinki. All animal experiments were conducted in accordance with institutional guidelines and approved by the Animal Ethics Committee of Sun Yat-sen University (Approval No. SYSU-IACUC-2024-001631).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConsent was obtained from each patient.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors declare that they have no competing interests.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eCiardiello F, Ciardiello D, Martini G, Napolitano S, Tabernero J, Cervantes A: \u003cstrong\u003eClinical management of metastatic colorectal cancer in the era of precision medicine.\u003c/strong\u003e \u003cem\u003eCA Cancer J Clin \u003c/em\u003e2022, \u003cstrong\u003e72:\u003c/strong\u003e372-401.\u003c/li\u003e\n\u003cli\u003ede Visser KE, Joyce JA: \u003cstrong\u003eThe evolving tumor microenvironment: From cancer initiation to metastatic outgrowth.\u003c/strong\u003e \u003cem\u003eCancer Cell \u003c/em\u003e2023, \u003cstrong\u003e41:\u003c/strong\u003e374-403.\u003c/li\u003e\n\u003cli\u003eLambert AW, Zhang Y, Weinberg RA: \u003cstrong\u003eCell-intrinsic and microenvironmental determinants of metastatic colonization.\u003c/strong\u003e \u003cem\u003eNat Cell Biol \u003c/em\u003e2024, \u003cstrong\u003e26:\u003c/strong\u003e687-697.\u003c/li\u003e\n\u003cli\u003eHanahan D, Weinberg RA: \u003cstrong\u003eThe hallmarks of cancer.\u003c/strong\u003e \u003cem\u003eCell \u003c/em\u003e2000, \u003cstrong\u003e100:\u003c/strong\u003e57-70.\u003c/li\u003e\n\u003cli\u003eCords L, de Souza N, Bodenmiller B: \u003cstrong\u003eClassifying cancer-associated fibroblasts-The good, the bad, and the target.\u003c/strong\u003e \u003cem\u003eCancer Cell \u003c/em\u003e2024, \u003cstrong\u003e42:\u003c/strong\u003e1480-1485.\u003c/li\u003e\n\u003cli\u003eKobayashi H, Gieniec KA, Wright JA, Wang T, Asai N, Mizutani Y, Lida T, Ando R, Suzuki N, Lannagan TRM, et al: \u003cstrong\u003eThe Balance of Stromal BMP Signaling Mediated by GREM1 and ISLR Drives Colorectal Carcinogenesis.\u003c/strong\u003e \u003cem\u003eGastroenterology \u003c/em\u003e2021, \u003cstrong\u003e160:\u003c/strong\u003e1224-1239 e1230.\u003c/li\u003e\n\u003cli\u003eLi T, Fei Y, Sun X, Zhu J, Fang X, Meng F, Wang D, Zhang X, Liu C, Du Q: \u003cstrong\u003eOver-expressed Phospholamban in Cancer-associated Fibroblasts Is a Driver for CMS4 Subtype Colorectal Cancer Formation.\u003c/strong\u003e \u003cem\u003eCell Mol Gastroenterol Hepatol \u003c/em\u003e2025, \u003cstrong\u003e19:\u003c/strong\u003e101524.\u003c/li\u003e\n\u003cli\u003eJiang Y, Fu Z, Chen Y, Jin Q, Yang Y, Lin Z, Li C, Gao Y, Dong Z, He Y, et al: \u003cstrong\u003eMapping and tracing Grem1(+) stromal cells in an Apc (Min/+) mouse utilizing cryopreserved intestinal sections prepared via modified Swiss-roll technique.\u003c/strong\u003e \u003cem\u003eiScience \u003c/em\u003e2024, \u003cstrong\u003e27:\u003c/strong\u003e111173.\u003c/li\u003e\n\u003cli\u003eRen J, Smid M, Iaria J, Salvatori DCF, van Dam H, Zhu HJ, Martens JWM, Ten Dijke P: \u003cstrong\u003eCancer-associated fibroblast-derived Gremlin 1 promotes breast cancer progression.\u003c/strong\u003e \u003cem\u003eBreast Cancer Res \u003c/em\u003e2019, \u003cstrong\u003e21:\u003c/strong\u003e109.\u003c/li\u003e\n\u003cli\u003eLan L, Evan T, Li H, Hussain A, Ruiz EJ, Zaw Thin M, Ferreira RMM, Ps H, Riising EM, Zen Y, et al: \u003cstrong\u003eGREM1 is required to maintain cellular heterogeneity in pancreatic cancer.\u003c/strong\u003e \u003cem\u003eNature \u003c/em\u003e2022, \u003cstrong\u003e607:\u003c/strong\u003e163-168.\u003c/li\u003e\n\u003cli\u003eCheng C, Wang J, Xu P, Zhang K, Xin Z, Zhao H, Ji Z, Zhang M, Wang D, He Y, et al: \u003cstrong\u003eGremlin1 is a therapeutically targetable FGFR1 ligand that regulates lineage plasticity and castration resistance in prostate cancer.\u003c/strong\u003e \u003cem\u003eNat Cancer \u003c/em\u003e2022, \u003cstrong\u003e3:\u003c/strong\u003e565-580.\u003c/li\u003e\n\u003cli\u003eJaeger E, Leedham S, Lewis A, Segditsas S, Becker M, Cuadrado PR, Davis H, Kaur K, Heinimann K, Howarth K, et al: \u003cstrong\u003eHereditary mixed polyposis syndrome is caused by a 40-kb upstream duplication that leads to increased and ectopic expression of the BMP antagonist GREM1.\u003c/strong\u003e \u003cem\u003eNat Genet \u003c/em\u003e2012, \u003cstrong\u003e44:\u003c/strong\u003e699-703.\u003c/li\u003e\n\u003cli\u003eLi R, Zhou H, Li M, Mai Q, Fu Z, Jiang Y, Li C, Gao Y, Fan Y, Wu K, et al: \u003cstrong\u003eGremlin-1 Promotes Colorectal Cancer Cell Metastasis by Activating ATF6 and Inhibiting ATF4 Pathways.\u003c/strong\u003e \u003cem\u003eCells \u003c/em\u003e2022, \u003cstrong\u003e11\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eLiu Y, Li Y, Hou R, Shu Z: \u003cstrong\u003eKnockdown GREM1 suppresses cell growth, angiogenesis, and epithelial-mesenchymal transition in colon cancer.\u003c/strong\u003e \u003cem\u003eJ Cell Biochem \u003c/em\u003e2019, \u003cstrong\u003e120:\u003c/strong\u003e5583-5596.\u003c/li\u003e\n\u003cli\u003eBrazil DP, Church RH, Surae S, Godson C, Martin F: \u003cstrong\u003eBMP signalling: agony and antagony in the family.\u003c/strong\u003e \u003cem\u003eTrends Cell Biol \u003c/em\u003e2015, \u003cstrong\u003e25:\u003c/strong\u003e249-264.\u003c/li\u003e\n\u003cli\u003eMitola S, Ravelli C, Moroni E, Salvi V, Leali D, Ballmer-Hofer K, Zammataro L, Presta M: \u003cstrong\u003eGremlin is a novel agonist of the major proangiogenic receptor VEGFR2.\u003c/strong\u003e \u003cem\u003eBlood \u003c/em\u003e2010, \u003cstrong\u003e116:\u003c/strong\u003e3677-3680.\u003c/li\u003e\n\u003cli\u003ePark SA, Sung NJ, Choi BJ, Kim W, Kim SH, Surh YJ: \u003cstrong\u003eGremlin-1 augments the oestrogen-related receptor alpha signalling through EGFR activation: implications for the progression of breast cancer.\u003c/strong\u003e \u003cem\u003eBr J Cancer \u003c/em\u003e2020, \u003cstrong\u003e123:\u003c/strong\u003e988-999.\u003c/li\u003e\n\u003cli\u003eDerynck R, Zhang YE: \u003cstrong\u003eSmad-dependent and Smad-independent pathways in TGF-beta family signalling.\u003c/strong\u003e \u003cem\u003eNature \u003c/em\u003e2003, \u003cstrong\u003e425:\u003c/strong\u003e577-584.\u003c/li\u003e\n\u003cli\u003eWorthley DL, Churchill M, Compton JT, Tailor Y, Rao M, Si Y, Levin D, Schwartz MG, Uygur A, Hayakawa Y, et al: \u003cstrong\u003eGremlin 1 identifies a skeletal stem cell with bone, cartilage, and reticular stromal potential.\u003c/strong\u003e \u003cem\u003eCell \u003c/em\u003e2015, \u003cstrong\u003e160:\u003c/strong\u003e269-284.\u003c/li\u003e\n\u003cli\u003eKalluri R, Zeisberg M: \u003cstrong\u003eFibroblasts in cancer.\u003c/strong\u003e \u003cem\u003eNat Rev Cancer \u003c/em\u003e2006, \u003cstrong\u003e6:\u003c/strong\u003e392-401.\u003c/li\u003e\n\u003cli\u003ePastushenko I, Brisebarre A, Sifrim A, Fioramonti M, Revenco T, Boumahdi S, Van Keymeulen A, Brown D, Moers V, Lemaire S, et al: \u003cstrong\u003eIdentification of the tumour transition states occurring during EMT.\u003c/strong\u003e \u003cem\u003eNature \u003c/em\u003e2018, \u003cstrong\u003e556:\u003c/strong\u003e463-468.\u003c/li\u003e\n\u003cli\u003eZhao H, Ming T, Tang S, Ren S, Yang H, Liu M, Tao Q, Xu H: \u003cstrong\u003eWnt signaling in colorectal cancer: pathogenic role and therapeutic target.\u003c/strong\u003e \u003cem\u003eMol Cancer \u003c/em\u003e2022, \u003cstrong\u003e21:\u003c/strong\u003e144.\u003c/li\u003e\n\u003cli\u003eTsuchida K, Nakatani M, Yamakawa N, Hashimoto O, Hasegawa Y, Sugino H: \u003cstrong\u003eActivin isoforms signal through type I receptor serine/threonine kinase ALK7.\u003c/strong\u003e \u003cem\u003eMol Cell Endocrinol \u003c/em\u003e2004, \u003cstrong\u003e220:\u003c/strong\u003e59-65.\u003c/li\u003e\n\u003cli\u003eKisonaite M, Wang X, Hyvonen M: \u003cstrong\u003eStructure of Gremlin-1 and analysis of its interaction with BMP-2.\u003c/strong\u003e \u003cem\u003eBiochem J \u003c/em\u003e2016, \u003cstrong\u003e473:\u003c/strong\u003e1593-1604.\u003c/li\u003e\n\u003cli\u003eDongre A, Weinberg RA: \u003cstrong\u003eNew insights into the mechanisms of epithelial-mesenchymal transition and implications for cancer.\u003c/strong\u003e \u003cem\u003eNat Rev Mol Cell Biol \u003c/em\u003e2019, \u003cstrong\u003e20:\u003c/strong\u003e69-84.\u003c/li\u003e\n\u003cli\u003eStemmler MP, Eccles RL, Brabletz S, Brabletz T: \u003cstrong\u003eNon-redundant functions of EMT transcription factors.\u003c/strong\u003e \u003cem\u003eNat Cell Biol \u003c/em\u003e2019, \u003cstrong\u003e21:\u003c/strong\u003e102-112.\u003c/li\u003e\n\u003cli\u003eBradford STJ, Ranghini EJ, Grimley E, Lee PH, Dressler GR: \u003cstrong\u003eHigh-throughput screens for agonists of bone morphogenetic protein (BMP) signaling identify potent benzoxazole compounds.\u003c/strong\u003e \u003cem\u003eJ Biol Chem \u003c/em\u003e2019, \u003cstrong\u003e294:\u003c/strong\u003e3125-3136.\u003c/li\u003e\n\u003cli\u003eDaCosta Byfield S, Major C, Laping NJ, Roberts AB: \u003cstrong\u003eSB-505124 is a selective inhibitor of transforming growth factor-beta type I receptors ALK4, ALK5, and ALK7.\u003c/strong\u003e \u003cem\u003eMol Pharmacol \u003c/em\u003e2004, \u003cstrong\u003e65:\u003c/strong\u003e744-752.\u003c/li\u003e\n\u003cli\u003eShi Y, Massagu\u0026eacute; J: \u003cstrong\u003eMechanisms of TGF-beta signaling from cell membrane to the nucleus.\u003c/strong\u003e \u003cem\u003eCell \u003c/em\u003e2003, \u003cstrong\u003e113:\u003c/strong\u003e685-700.\u003c/li\u003e\n\u003cli\u003eAsnaghi L, White DT, Key N, Choi J, Mahale A, Alkatan H, Edward DP, Elkhamary SM, Al-Mesfer S, Maktabi A, et al: \u003cstrong\u003eACVR1C/SMAD2 signaling promotes invasion and growth in retinoblastoma.\u003c/strong\u003e \u003cem\u003eOncogene \u003c/em\u003e2019, \u003cstrong\u003e38:\u003c/strong\u003e2056-2075.\u003c/li\u003e\n\u003cli\u003eYang S, Zhang H, Yang H, Zhang J, Wang J, Luo T, Jiang Y, Hua H: \u003cstrong\u003eSEPHS1 promotes SMAD2/3/4 expression and hepatocellular carcinoma cells invasion.\u003c/strong\u003e \u003cem\u003eExp Hematol Oncol \u003c/em\u003e2021, \u003cstrong\u003e10:\u003c/strong\u003e17.\u003c/li\u003e\n\u003cli\u003eCai J, Xia L, Li J, Ni S, Song H, Wu X: \u003cstrong\u003eTumor-Associated Macrophages Derived TGF-beta\u003c/strong\u003e\u003cstrong\u003e‒Induced Epithelial to Mesenchymal Transition in Colorectal Cancer Cells through Smad2,3-4/Snail Signaling Pathway.\u003c/strong\u003e \u003cem\u003eCancer Res Treat \u003c/em\u003e2019, \u003cstrong\u003e51:\u003c/strong\u003e252-266.\u003c/li\u003e\n\u003cli\u003eClevers H: \u003cstrong\u003eWnt/beta-catenin signaling in development and disease.\u003c/strong\u003e \u003cem\u003eCell \u003c/em\u003e2006, \u003cstrong\u003e127:\u003c/strong\u003e469-480.\u003c/li\u003e\n\u003cli\u003eS\u0026aacute;nchez-Till\u0026oacute; E, de Barrios O, Siles L, Cuatrecasas M, Castells A, Postigo A: \u003cstrong\u003e\u0026beta;-catenin/TCF4 complex induces the epithelial-to-mesenchymal transition (EMT)-activator ZEB1 to regulate tumor invasiveness.\u003c/strong\u003e \u003cem\u003eProc Natl Acad Sci U S A \u003c/em\u003e2011, \u003cstrong\u003e108:\u003c/strong\u003e19204-19209.\u003c/li\u003e\n\u003cli\u003eInomata M, Ochiai A, Akimoto S, Kitano S, Hirohashi S: \u003cstrong\u003eAlteration of beta-catenin expression in colonic epithelial cells of familial adenomatous polyposis patients.\u003c/strong\u003e \u003cem\u003eCancer Res \u003c/em\u003e1996, \u003cstrong\u003e56:\u003c/strong\u003e2213-2217.\u003c/li\u003e\n\u003cli\u003eMoser AR, Pitot HC, Dove WF: \u003cstrong\u003eA dominant mutation that predisposes to multiple intestinal neoplasia in the mouse.\u003c/strong\u003e \u003cem\u003eScience \u003c/em\u003e1990, \u003cstrong\u003e247:\u003c/strong\u003e322-324.\u003c/li\u003e\n\u003cli\u003eRen J, Sui H, Fang F, Li Q, Li B: \u003cstrong\u003eThe application of Apc(Min/+) mouse model in colorectal tumor researches.\u003c/strong\u003e \u003cem\u003eJ Cancer Res Clin Oncol \u003c/em\u003e2019, \u003cstrong\u003e145:\u003c/strong\u003e1111-1122.\u003c/li\u003e\n\u003cli\u003eBabaei-Jadidi R, Li N, Saadeddin A, Spencer-Dene B, Jandke A, Muhammad B, Ibrahim EE, Muraleedharan R, Abuzinadah M, Davis H, et al: \u003cstrong\u003eFBXW7 influences murine intestinal homeostasis and cancer, targeting Notch, Jun, and DEK for degradation.\u003c/strong\u003e \u003cem\u003eJ Exp Med \u003c/em\u003e2011, \u003cstrong\u003e208:\u003c/strong\u003e295-312.\u003c/li\u003e\n\u003cli\u003eZhang L, Zhu Z, Yan H, Wang W, Wu Z, Zhang F, Zhang Q, Shi G, Du J, Cai H, et al: \u003cstrong\u003eCreatine promotes cancer metastasis through activation of Smad2/3.\u003c/strong\u003e \u003cem\u003eCell Metab \u003c/em\u003e2021, \u003cstrong\u003e33:\u003c/strong\u003e1111-1123.e1114.\u003c/li\u003e\n\u003cli\u003eIto M, Hiramatsu H, Kobayashi K, Suzue K, Kawahata M, Hioki K, Ueyama Y, Koyanagi Y, Sugamura K, Tsuji K, et al: \u003cstrong\u003eNOD/SCID/gamma(c)(null) mouse: an excellent recipient mouse model for engraftment of human cells.\u003c/strong\u003e \u003cem\u003eBlood \u003c/em\u003e2002, \u003cstrong\u003e100:\u003c/strong\u003e3175-3182.\u003c/li\u003e\n\u003cli\u003eJung B, Staudacher JJ, Beauchamp D: \u003cstrong\u003eTransforming Growth Factor \u003c/strong\u003e\u003cstrong\u003e\u0026beta; Superfamily Signaling in Development of Colorectal Cancer.\u003c/strong\u003e \u003cem\u003eGastroenterology \u003c/em\u003e2017, \u003cstrong\u003e152:\u003c/strong\u003e36-52.\u003c/li\u003e\n\u003cli\u003eLiu S, Ren J, Ten Dijke P: \u003cstrong\u003eTargeting TGF\u003c/strong\u003e\u003cstrong\u003e\u0026beta; signal transduction for cancer therapy.\u003c/strong\u003e \u003cem\u003eSignal Transduct Target Ther \u003c/em\u003e2021, \u003cstrong\u003e6:\u003c/strong\u003e8.\u003c/li\u003e\n\u003cli\u003ePeng D, Fu M, Wang M, Wei Y, Wei X: \u003cstrong\u003eTargeting TGF-\u003c/strong\u003e\u003cstrong\u003e\u0026beta; signal transduction for fibrosis and cancer therapy.\u003c/strong\u003e \u003cem\u003eMol Cancer \u003c/em\u003e2022, \u003cstrong\u003e21:\u003c/strong\u003e104.\u003c/li\u003e\n\u003cli\u003eModell AE, Blosser SL, Arora PS: \u003cstrong\u003eSystematic Targeting of Protein-Protein Interactions.\u003c/strong\u003e \u003cem\u003eTrends Pharmacol Sci \u003c/em\u003e2016, \u003cstrong\u003e37:\u003c/strong\u003e702-713.\u003c/li\u003e\n\u003cli\u003eWang Y, Bikle DD, Chang W: \u003cstrong\u003eAutocrine and Paracrine Actions of IGF-I Signaling in Skeletal Development.\u003c/strong\u003e \u003cem\u003eBone Res \u003c/em\u003e2013, \u003cstrong\u003e1:\u003c/strong\u003e249-259.\u003c/li\u003e\n\u003cli\u003eLiu S, Hur YH, Cai X, Cong Q, Yang Y, Xu C, Bilate AM, Gonzales KAU, Parigi SM, Cowley CJ, et al: \u003cstrong\u003eA tissue injury sensing and repair pathway distinct from host pathogen defense.\u003c/strong\u003e \u003cem\u003eCell \u003c/em\u003e2023, \u003cstrong\u003e186:\u003c/strong\u003e2127-2143 e2122.\u003c/li\u003e\n\u003cli\u003eBasu R, Boguszewski CL, Kopchick JJ: \u003cstrong\u003eGrowth Hormone Action as a Target in Cancer: Significance, Mechanisms, and Possible Therapies.\u003c/strong\u003e \u003cem\u003eEndocr Rev \u003c/em\u003e2025, \u003cstrong\u003e46:\u003c/strong\u003e224-280.\u003c/li\u003e\n\u003cli\u003eAo Z, Shah SH, Machlin LM, Parajuli R, Miller PC, Rawal S, Williams AJ, Cote RJ, Lippman ME, Datar RH, El-Ashry D: \u003cstrong\u003eIdentification of Cancer-Associated Fibroblasts in Circulating Blood from Patients with Metastatic Breast Cancer.\u003c/strong\u003e \u003cem\u003eCancer Res \u003c/em\u003e2015, \u003cstrong\u003e75:\u003c/strong\u003e4681-4687.\u003c/li\u003e\n\u003cli\u003eHurtado P, Martinez-Pena I, Pineiro R: \u003cstrong\u003eDangerous Liaisons: Circulating Tumor Cells (CTCs) and Cancer-Associated Fibroblasts (CAFs).\u003c/strong\u003e \u003cem\u003eCancers (Basel) \u003c/em\u003e2020, \u003cstrong\u003e12\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eSporn MB, Todaro GJ: \u003cstrong\u003eAutocrine secretion and malignant transformation of cells.\u003c/strong\u003e \u003cem\u003eN Engl J Med \u003c/em\u003e1980, \u003cstrong\u003e303:\u003c/strong\u003e878-880.\u003c/li\u003e\n\u003cli\u003eWydooghe E, Vandaele L, Heras S, De Sutter P, Deforce D, Peelman L, De Schauwer C, Van Soom A: \u003cstrong\u003eAutocrine embryotropins revisited: how do embryos communicate with each other in vitro when cultured in groups?\u003c/strong\u003e \u003cem\u003eBiol Rev Camb Philos Soc \u003c/em\u003e2017, \u003cstrong\u003e92:\u003c/strong\u003e505-520.\u003c/li\u003e\n\u003cli\u003eVenkatesh HS, Johung TB, Caretti V, Noll A, Tang Y, Nagaraja S, Gibson EM, Mount CW, Polepalli J, Mitra SS, et al: \u003cstrong\u003eNeuronal Activity Promotes Glioma Growth through Neuroligin-3 Secretion.\u003c/strong\u003e \u003cem\u003eCell \u003c/em\u003e2015, \u003cstrong\u003e161:\u003c/strong\u003e803-816.\u003c/li\u003e\n\u003cli\u003eMichos O, Panman L, Vintersten K, Beier K, Zeller R, Zuniga A: \u003cstrong\u003eGremlin-mediated BMP antagonism induces the epithelial-mesenchymal feedback signaling controlling metanephric kidney and limb organogenesis.\u003c/strong\u003e \u003cem\u003eDevelopment \u003c/em\u003e2004, \u003cstrong\u003e131:\u003c/strong\u003e3401-3410.\u003c/li\u003e\n\u003cli\u003eKoppens MAJ, Davis H, Valbuena GN, Mulholland EJ, Nasreddin N, Colombe M, Antanaviciute A, Biswas S, Friedrich M, Lee L, et al: \u003cstrong\u003eBone Morphogenetic Protein Pathway Antagonism by Grem1 Regulates Epithelial Cell Fate in Intestinal Regeneration.\u003c/strong\u003e \u003cem\u003eGastroenterology \u003c/em\u003e2021, \u003cstrong\u003e161:\u003c/strong\u003e239-254.e239.\u003c/li\u003e\n\u003cli\u003eBaboota RK, Rawshani A, Bonnet L, Li X, Yang H, Mardinoglu A, Tchkonia T, Kirkland JL, Hoffmann A, Dietrich A, et al: \u003cstrong\u003eBMP4 and Gremlin 1 regulate hepatic cell senescence during clinical progression of NAFLD/NASH.\u003c/strong\u003e \u003cem\u003eNat Metab \u003c/em\u003e2022, \u003cstrong\u003e4:\u003c/strong\u003e1007-1021.\u003c/li\u003e\n\u003cli\u003eMulholland EJ, Belnoue-Davis HL, Valbuena GN, Gunduz N, Ligeza A, Lin M, Biswas S, Gil Vasquez E, Omwenga S, Nasreddin N, et al: \u003cstrong\u003eEpithelial GREMLIN1 disrupts intestinal epithelial-mesenchymal crosstalk to induce a wnt-dependent ectopic stem cell niche through stromal remodelling.\u003c/strong\u003e \u003cem\u003eNat Commun \u003c/em\u003e2025, \u003cstrong\u003e16:\u003c/strong\u003e5167.\u003c/li\u003e\n\u003cli\u003eGao Z, Houthuijzen JM, Ten Dijke P, Brazil DP: \u003cstrong\u003eGREM1 signaling in cancer: tumor promotor and suppressor?\u003c/strong\u003e \u003cem\u003eJ Cell Commun Signal \u003c/em\u003e2023, \u003cstrong\u003e17:\u003c/strong\u003e1517-1526.\u003c/li\u003e\n\u003cli\u003ePark SA, Sung NJ, Choi BJ, Kim W, Kim SH, Surh YJ: \u003cstrong\u003eGremlin-1 augments the oestrogen-related receptor \u003c/strong\u003e\u003cstrong\u003e\u0026alpha; signalling through EGFR activation: implications for the progression of breast cancer.\u003c/strong\u003e \u003cem\u003eBr J Cancer \u003c/em\u003e2020, \u003cstrong\u003e123:\u003c/strong\u003e988-999.\u003c/li\u003e\n\u003cli\u003eTsuchida K, Sawchenko PE, Nishikawa S, Vale WW: \u003cstrong\u003eMolecular cloning of a novel type I receptor serine/threonine kinase for the TGF beta superfamily from rat brain.\u003c/strong\u003e \u003cem\u003eMol Cell Neurosci \u003c/em\u003e1996, \u003cstrong\u003e7:\u003c/strong\u003e467-478.\u003c/li\u003e\n\u003cli\u003eHedjazifar S, Khatib Shahidi R, Hammarstedt A, Bonnet L, Church C, Boucher J, Bl\u0026uuml;her M, Smith U: \u003cstrong\u003eThe Novel Adipokine Gremlin 1 Antagonizes Insulin Action and Is Increased in Type 2 Diabetes and NAFLD/NASH.\u003c/strong\u003e \u003cem\u003eDiabetes \u003c/em\u003e2020, \u003cstrong\u003e69:\u003c/strong\u003e331-341.\u003c/li\u003e\n\u003cli\u003eSarker D, Blagden S, Cook N, Evans TR, Plummer R, Angelakas A, Bashir S, Badri SE, Mavroeidis L, Morton C, et al: \u003cstrong\u003eAbstract CT094: Phase I dose-escalation study evaluating the safety an d tolerability of ginisortamab (UCB6114), a first-in-class anti-gremli n-1 monoclonal antibody (mAb), as monotherapy in advanced solid tumors.\u003c/strong\u003e \u003cem\u003eCancer Research\u003c/em\u003e, \u003cstrong\u003e83:\u003c/strong\u003eCT094-CT094.\u003c/li\u003e\n\u003cli\u003eRivera IR, Shen L, Kummar S, Barve M, Germa C, Qi C, Chen L, Milata J, Yao J, Shen L, Zhu X: \u003cstrong\u003eAbstract 5977: A first-in-human, open-label, multi-center phase 1 stud y of TST003, a GREM 1 inhibitor, in subjects with locally advanced or metastatic solid tumors.\u003c/strong\u003e \u003cem\u003eCancer Research\u003c/em\u003e, \u003cstrong\u003e84:\u003c/strong\u003e5977-5977.\u003c/li\u003e\n\u003cli\u003eRowan SC, Jahns H, Mthunzi L, Piouceau L, Cornwell J, Doody R, Frohlich S, Callanan JJ, McLoughlin P: \u003cstrong\u003eGremlin 1 depletion in vivo causes severe enteropathy and bone marrow failure.\u003c/strong\u003e \u003cem\u003eJ Pathol \u003c/em\u003e2020, \u003cstrong\u003e251:\u003c/strong\u003e117-122.\u003c/li\u003e\n\u003cli\u003eLu H, Zhou Q, He J, Jiang Z, Peng C, Tong R, Shi J: \u003cstrong\u003eRecent advances in the development of protein-protein interactions modulators: mechanisms and clinical trials.\u003c/strong\u003e \u003cem\u003eSignal Transduct Target Ther \u003c/em\u003e2020, \u003cstrong\u003e5:\u003c/strong\u003e213.\u003c/li\u003e\n\u003cli\u003eGibson DG, Young L, Chuang RY, Venter JC, Hutchison CA, 3rd, Smith HO: \u003cstrong\u003eEnzymatic assembly of DNA molecules up to several hundred kilobases.\u003c/strong\u003e \u003cem\u003eNat Methods \u003c/em\u003e2009, \u003cstrong\u003e6:\u003c/strong\u003e343-345.\u003c/li\u003e\n\u003cli\u003eFlanagan SP: \u003cstrong\u003e\u0026apos;Nude\u0026apos;, a new hairless gene with pleiotropic effects in the mouse.\u003c/strong\u003e \u003cem\u003eGenet Res \u003c/em\u003e1966, \u003cstrong\u003e8:\u003c/strong\u003e295-309.\u003c/li\u003e\n\u003cli\u003eMuzumdar MD, Tasic B, Miyamichi K, Li L, Luo L: \u003cstrong\u003eA global double-fluorescent Cre reporter mouse.\u003c/strong\u003e \u003cem\u003eGenesis \u003c/em\u003e2007, \u003cstrong\u003e45:\u003c/strong\u003e593-605.\u003c/li\u003e\n\u003cli\u003eVoehringer D, Liang HE, Locksley RM: \u003cstrong\u003eHomeostasis and effector function of lymphopenia-induced \u0026quot;memory-like\u0026quot; T cells in constitutively T cell-depleted mice.\u003c/strong\u003e \u003cem\u003eJ Immunol \u003c/em\u003e2008, \u003cstrong\u003e180:\u003c/strong\u003e4742-4753.\u003c/li\u003e\n\u003cli\u003eYan Y, Zhang D, Zhou P, Li B, Huang SY: \u003cstrong\u003eHDOCK: a web server for protein-protein and protein-DNA/RNA docking based on a hybrid strategy.\u003c/strong\u003e \u003cem\u003eNucleic Acids Res \u003c/em\u003e2017, \u003cstrong\u003e45:\u003c/strong\u003eW365-w373.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eCell lines \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuman colon cancer cell lines (HCT116, SW480, DLD1, LOVO, LS174T, and HT29), mouse colon cancer cell line (MC38) and human embryonic kidney cell lines (HEK293 and its derivative 293T) were purchased from ATCC and tested negative for mycoplasma contamination. Colon cancer cells, HEK293 cells and 293T cells were cultured in RPMI1640 (HyClone, SH30809.01) and DMEM (SH30022.01), respectively, supplemented with 10% fetal bovine serum (Biological Industries, 04-001-1ACS) and 1% Penicillin/Streptomycin (Biological Industries, 03-031-1B) at 37\u0026deg;C, 5% CO\u003csub\u003e2\u003c/sub\u003e. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePatients and tissue samples\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 106 archived human colorectal cancer specimens were obtained from the colorectal cancer database and tissue bank of the First Affiliated Hospital of Sun Yat-sen University (SYSU). These tissues were collected from patients who underwent radical resection for colorectal cancer between 2008 and 2015, and followed up until December 2017. All patients had provided written informed consent, and the use of these samples was approved by the Institutional Review Board of the First Affiliated Hospital, SYSU. In addition, a commercial human CRC tissue microarray containing primary tumor tissues from patients with stage I\u0026ndash;III CRC (n = 93) was purchased from Shanghai Outdo Biotech Company (Shanghai, China), with ethics approval documented by the company. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConditioned medium (CM)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo investigate the effect of secreted GREM1 on CRC cells, an \u003cem\u003ein vitro\u003c/em\u003e GREM1 secretion system was established using HEK293 cells. Stably GREM1-expressing cells (pcDNA3.1-GREM1) and empty-vector transfected cells (pcDNA3.1) were cultured in DMEM/F12 medium with 10% FBS until 40% confluence. After complete removal of the normal culture medium, HEK293-GREM1 and HEK293-Vec cells were continuously cultured in DMEM/F12 medium without FBS for 5 days before medium collection. GREM1 or Vector conditioned medium (GREM1-CM/Vec-CM) was then centrifuged at 1000 rpm for 30 min and the supernatant was collected for further study. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIsolation of fibroblasts from normal and tumoral human intestinal tissues \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFibroblasts were isolated from normal and tumoral human intestinal tissues. Briefly, a cell dissociation buffer was prepared using DMEM supplemented with 10% FBS, 1% Penicillin/Streptomycin, collagenase type D (1 mg/mL, Roche, 11088866001), and DNase I (20 \u0026micro;g/mL, Roche, 10104159001). Tumor tissues were washed twice with DMEM or phosphate-buffered saline (PBS), transferred to a 100-mm culture dish containing 15 mL of cell dissociation buffer, and minced into fragments (\u0026lt; 1 mm\u0026sup3;) using sterile razor blades. The tissue fragments were enzymatically digested at 37 \u0026deg;C for 30 min. Following digestion, the cell suspension was filtered through a 70-\u0026micro;m cell strainer to obtain a single-cell suspension. The cell suspension was centrifuged at 1000 rpm for 5 min, and the pellet was resuspended in DMEM. This step was repeated, and the final pellet was resuspended in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37 \u0026deg;C in a humidified incubator with 5% CO₂. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLentiviral plasmid construction, lentivirus production and infection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuman \u003cem\u003eGREM1\u003c/em\u003e CDS (NM_013372.7) with a HA-mCherry tag or human \u003cem\u003eACVR1C\u003c/em\u003e CDS (NM_145259.3) with a Flag-GFP tag was cloned into a pCDH-CMV-MCS-EF1-puro vector. Truncated or point mutations of GREM1 or ACVR1C were cloned from entire GREM1- or ACVR1C-expressing plasmids by PCR. Human \u003cem\u003eGREM1\u003c/em\u003e CDS (NM_013372.7) was cloned into the pLV-EF1a-IRES-Puro lentiviral vector. CRISPR-mediated gene knockout: The sequences targeting GREM1 were GREM1 KO27 (gRNA1: 5\u0026prime;\u0026ndash; GCAAATACCTGAAGCGAGAC \u0026ndash;3\u0026prime;) and GREM1 KO28 (g\u003cem\u003eRNA2: \u003c/em\u003e5\u0026prime;\u0026ndash; AAGCAGACCATCCACGAGGA \u0026ndash;3\u0026prime;). The Cas9 lentivirus and gRNA1/2 lentivirus were purchased from GenePharma. shRNA-mediated silencing: The human ACVR1C shRNA target sequences are listed as follows: shACVR1C#1 (5\u0026prime;\u0026ndash; CGGAGGAATTGTTGAGGAGTA \u0026ndash;3\u0026prime;); shACVR1C#2 (5\u0026prime;\u0026ndash; GCAACACCTCAACTCATCTTT \u0026ndash;3\u0026prime;). All inserts and vectors were purified from agarose gel using the FastPure \u0026reg; Gel DNA Extraction Mini Kit (Vazyme, DC301-01) and assembled with Gibson Assembly Master Mix[65] (NEB, E2611) according to the manufacturers\u0026rsquo; protocols. All plasmids were verified by Sanger sequencing. HEK 293T cells were seeded at a density expected to reach 70-80% confluence at the time of transfection.\u003c/p\u003e\n\u003cp\u003eTo produce lentivirus, plasmids mentioned above together with packaging plasmid (\u003cem\u003epsPAX2\u003c/em\u003e) and envelope plasmid (\u003cem\u003epMD2.G\u003c/em\u003e) were mixed in a 3.9:2.1:1 ratio and transfected into the cells using polyethylenimine (PEI). After 48-72 hours, supernatant containing lentivirus was collected, filtered, and either used immediately or stored at -80℃ for later applications.\u003c/p\u003e\n\u003cp\u003eHCT116 and SW480 tumor cells were infected with lentiviral particles in the presence of 5 \u0026mu;g/mL polybrene. To establish stable cell lines, these infected cells were selected with 1.25 \u0026mu;g/mL puromycin for 2 weeks. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAnimal experiments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe immunocompromised nude[66] and NOG[40] female mice (6 weeks old) were purchased from Guangzhou Vital River Laboratory Animal Technology Co., Ltd. \u003cem\u003eGrem1-CreER\u003csup\u003eT2 \u003c/sup\u003e\u003c/em\u003e(stock no. 027039)[19], \u003cem\u003eRosa-mTmG \u003c/em\u003e(stock no. 007576)[67], \u003cem\u003eRosa-LSL-DTA\u003c/em\u003e (stock no. 007900) mice[68] were obtained from the Jackson Laboratory. \u003cem\u003eAPC\u003csup\u003eMin/+\u003c/sup\u003e\u003c/em\u003e mice were obtained from the Gempharmatech Co., Ltd (stock no. 002020)[36]. \u003cem\u003eGrem1-CreER\u003csup\u003eT2\u003c/sup\u003e\u003c/em\u003emice were crossed with \u003cem\u003eRosa-mTmG\u003c/em\u003e mice\u003cem\u003e \u003c/em\u003eor \u003cem\u003eRosa-LSL-DTA\u003c/em\u003e mice to generate GR or GD mice, respectively. The Cre recombinase activity was induced by the ER antagonist tamoxifen (TMX), allowing Grem1\u003csup\u003e+\u003c/sup\u003e cells to express GFP in GD mice. GD mice were further crossed with \u003cem\u003eAPC\u003csup\u003eMin/+\u003c/sup\u003e\u003c/em\u003e mice to generate AGD mice. AGD and control mice were administered with 100mg/kg tamoxifen (TMX) through oral gavage at 4-week-old time, when tumor initiates. In AGD and its control mice, activation of Cre lead to the expression of DTA (diphtheria toxin A chain), which removed the population of Grem1\u003csup\u003e+\u003c/sup\u003e cells from the \u003cem\u003eAPC\u003csup\u003eMin/+\u003c/sup\u003e\u003c/em\u003e mice. All animals were maintained at the Animal Experiment Center of Sun-Yat-Sen University, and all procedures were approved by the Animal Care and Use Committee of Sun-Yat-Sen University. Mice were randomized at the beginning of each experiment. \u003c/p\u003e\n\u003cp\u003eFor tail vein-to-lung metastasis model, 5\u0026times;10\u003csup\u003e5\u003c/sup\u003e MC38-luc cells were resuspended in 100 \u0026mu;l of PBS and injected into the tail veins of GD mice or \u003cem\u003eGrem1-CreER\u003csup\u003eT2\u003c/sup\u003e/Rosa-LSL-DTA\u003c/em\u003e mice (n = 5 mice in each group); 1\u0026times;10\u003csup\u003e6\u003c/sup\u003e HCT116-luc cells, transduced with lentivirus carrying a control shRNA or two ACVR1C shRNAs or carrying a pLV or pLV-GREM1, were injected into the tail veins of nude mice (n = 5 mice in each group). \u003c/p\u003e\n\u003cp\u003eFor spleen-to-liver metastasis model, 5\u0026times;10\u003csup\u003e5\u003c/sup\u003e MC38-luc cells were resuspended in 50 \u0026mu;l of PBS and injected into the spleen of GD mice or \u003cem\u003eGrem1-CreER\u003csup\u003eT2\u003c/sup\u003e/Rosa-LSL-DTA\u003c/em\u003e mice (n = 5 mice in each group). 5\u0026times;10\u003csup\u003e5\u003c/sup\u003e HCT116-luc cells carrying a pLV or pLV-GREM1 were resuspended in 50 \u0026mu;l of PBS and intrasplenically injected in NOG mice (n = 5 mice in each group). ACVR1C peptide was administered via tail vein injection at a dose of 10 mg/kg once every other day. \u003c/p\u003e\n\u003cp\u003eFor cecum-to-liver metastasis model, 1\u0026times;10\u003csup\u003e6\u003c/sup\u003e MC38-luc cells were resuspended in 50 \u0026mu;l of PBS and injected into the cecum of GD mice or \u003cem\u003eGrem1-CreER\u003csup\u003eT2\u003c/sup\u003e/Rosa-LSL-DTA\u003c/em\u003e mice (n = 6 mice in each group); 5\u0026times;10\u003csup\u003e6\u003c/sup\u003e HCT116-luc cells transduced with or without lentivirus carrying a pLV or pLV-GREM1 were injected into the cecum of NOG mice (n = 5 mice in each group). SB505124 was administered via intraperitoneal injection at a dose of 10 mg/kg once every other day. The metastases were examined every 5 days post injection using an IVIS Lumina Imaging System. Mice were euthanized between 2-6 weeks after injection. For subcutaneous transplantation, 1\u0026times;10\u003csup\u003e6\u003c/sup\u003e HCT116 cells, either unmodified or transduced with lentivirus carrying a control shRNA, two ACVR1C shRNAs, pLV, or pLV-GREM1, were subcutaneously injected into mice (n = 6-8 mice in each group). Mice were euthanized 4 weeks after injection. The tumor tissues were collected for further evaluation. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImmunohistochemical\u003c/strong\u003e\u003cstrong\u003e (IHC) staining \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eImmunohistochemical staining of GREM1 (1:50, Biorbyt, orb10741), ACVR1C (1:50, Thermo, PA587475) and Ki67 (1:100, Servicebio, GB111499) was performed on primary tumors tissues. After dewaxing, hydration, and antigen retrieval, the rest of the experimental procedures were performed according to the instructions of the SP Immunohistochemistry Kit (ZSBIO, PV9000). Finally, after DAB staining, hematoxylin re-staining, and neutral resin sealing, the sections were observed under a microscope. Images were taken with a Slide Scanning Imaging System (SQS-1000, sqray). Quantification of positive staining was performed using Fiji (ImageJ). \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImmunofluorescence (IF) staining\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTissue was fixed in 4% paraformaldehyde (Thermo Scientific, I28800) for 24 h at 4\u0026thinsp;\u0026deg;C, washed with PBS, embedded in paraffin, and sectioned at 5 \u0026mu;m thickness. Antigen retrieval was performed using target retrieval solution, pH 9.0 in a pressure cooker for 15\u0026ndash;20 min. Non-specific binding was then blocked with 10% normal donkey serum (Abcam, ab7475) and 0.3% Triton X-100 in PBS for 30 min at room temperature. Cells for IF were fixed with 4% paraformaldehyde for 20 min at room temperature, washed with PBS, and permeabilized with 0.2% Triton X-100 in PBS for 20 min. Cells were then blocked in PBS with 5% BSA for 30 min at room temperature. Subsequently, the samples were incubated with goat anti-GREM1 (3 \u0026micro;g/mL, R\u0026amp;D, AF956), mouse anti-\u0026beta;-Catenin (1:100, BD, 610154), rabbit anti-\u0026beta;-Catenin (1:100, Absea, RC-6352), rabbit anti-Vimentin (1:100, CST, 5741), rabbit anti-CD68 (1:100, CST, 26042), rabbit anti-FAP (1:50, Proteintech, 15384-1-AP), rabbit anti-\u0026alpha;-SMA (1:100, Abcam, ab5694), rabbit anti-E-Cadherin (1:200, CST, 3195), rabbit anti-ACVR1C (1:50, Thermo, PA587475), rabbit anti-Snail (1:200, Abcam, ab224731) overnight at 4\u0026thinsp;\u0026deg;C. The tissues were incubated with Alexa-Fluor-conjugated secondary antibodies (Invitrogen) in PBS with 1 % normal donkey serum for 1 h at room temperature. DAPI was then used for counterstaining the nuclei, and images were obtained by a laser scanning confocal microscope (LSM880, Zeiss). \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAnalysis of scRNA-seq data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSingle-cell RNA sequencing (scRNA-seq) data from the Gene Expression Omnibus (GEO) database colorectal cancer datasets (GSE200997 and GSE221575) were processed using the R \u0026apos;Seurat\u0026apos; package (v4.4). Initial quality control involved rigorous filtering of low-quality cells: Cells expressing fewer than 200 genes or more than 10,000 genes were excluded, and cells with mitochondrial gene content exceeding 25% were discarded to remove potential apoptotic cells or debris. After quality control, a total of 34,675 high-quality cells were retained for downstream analysis. Gene expression matrices were normalized using the \u0026quot;LogNormalize\u0026quot; method implemented in the NormalizeData function, which scales feature counts per cell by total expression and multiplies by a scale factor (10,000), followed by natural log transformation. To identify biologically relevant features, the FindVariableFeatures function was employed to select the top 2,000 highly variable genes (HVGs) exhibiting the highest cell-to-cell variation. Dimensionality reduction was performed using principal component analysis (PCA) on scaled expression data of the identified HVGs. To address technical batch effects between samples and datasets, we applied multiple Canonical Correlation Analysis (CCA) as implemented in Seurat\u0026apos;s integration workflow. Cell clustering was performed using a graph-based approach: The FindNeighbors function constructed a shared nearest neighbor (SNN) graph based on the first 30 principal components, followed by the FindClusters function using the Louvain algorithm at a resolution of 0.8 to identify distinct cell subpopulations. Finally, non-linear dimensionality reduction was achieved through t-distributed Stochastic Neighbor Embedding (t-SNE) using the same principal components. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImmunoprecipitation (IP)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHCT116 cells were transfected with the indicated plasmids and lysed in NP40 lysis buffer (Beyotime, P0013F) supplemented with protease inhibitor cocktail (Thermo, 78446). Lysates were incubated with the indicated Anti-Flag nanobody magarose beads (Ktsm-life, KTSM1338), Anti-HA nanobody magarose beads (Ktsm-life, KTSM1335) or Anti-GFP nanobody magarose beads (Ktsm-life, KTSM1334) overnight at 4\u0026thinsp;\u0026deg;C. The protein complex was washed four times with the NP40 lysis buffer, eluted with 1\u0026times;loading buffer (Beyotime, P0015) by boiling for 5 min, followed by mass spectrometry and immunoblotting with the indicated antibodies. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMass spectrometry (MS) analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eProteins were separated by 10% SDS-PAGE and visualized using Coomassie Brilliant Blue staining before mass spectrometry analysis. The stained gel bands were excised (~1\u0026ndash;2 mm), washed with MilliQ water, and destained using 25 mM NH₄HCO₃ and 50% acetonitrile (ACN) at 37\u0026deg;C. The gel pieces were dehydrated with ACN, reduced with 10 mM dithiothreitol (DTT) in 25 mM NH₄HCO₃ at 37\u0026deg;C for 1 h, and alkylated with 30 mM iodoacetamide (IAA) in 25 mM NH₄HCO₃ in the dark for 45 min. After sequential washing with MilliQ water and 50% ACN, the gel pieces were dehydrated with ACN and digested overnight at 37\u0026deg;C with trypsin (20 ng/\u0026mu;L) in 25 mM NH₄HCO₃. Peptides were extracted using 60% ACN followed by pure ACN, pooled, lyophilized, resuspended in 0.1% formic acid (FA), and purified using ZipTip C18 before analysis. Mass spectrometry was performed using a Thermo Fisher Orbitrap HF-X coupled with an Easy-nLC 1200 system and a C18 column, employing a 90-min gradient of 5\u0026ndash;35% ACN in 0.1% FA at a flow rate of 300 nL/min. MS1 scans were acquired at a resolution of 60,000 with an AGC target of 3 \u0026times; 10⁶, a maximum injection time of 20 ms, and a scan range of m/z 350\u0026ndash;1800. MS2 scans were performed at a resolution of 15,000 with an AGC target of 2 \u0026times; 10⁵, a maximum injection time of 100 ms, TopN of 20, and a normalized collision energy (NCE) of 32. Raw MS data were analyzed using Proteome Discoverer 2.4, with protein identification performed against the SwissProt human database using trypsin specificity (allowing one missed cleavage site), cysteine alkylation with MMTS, a precursor mass tolerance of 10 ppm, a fragment mass tolerance of 0.02 Da, and a false discovery rate (FDR) threshold of \u0026lt;1%. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImmunoblotting (IB)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eProtein was extracted from the cells with RIPA buffer (Beyotime, P0018) or NP40 lysis buffer (Beyotime, P0013F) and separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Primary antibodies against GREM1 (1:1,000, SinoBiological, 50016-R117), ACVR1C (1:1,000, Thermo, PA587475), Flag-tag (1:1,000, CST, 14793), HA-tag (1:1,000, CST, 3724), E-Cadherin (1:1,000, CST, 3195), \u0026beta;-Catenin (1:1,000, CST, 8480), ZEB1 (1:1,000, CST, 3396), Snail (1:1,000, CST, 3879), SMAD2/3 (1:1,000, CST, 8685), p-SMAD2/3 (1:1,000, CST, 8828), SMAD1 (1:1,000, CST, 6944), p-SMAD1/5/9 (1:1,000, CST, 13820), TGF\u0026beta; (1:1,000, CST, 3709), TGF\u0026beta;R1 (0.3\u0026micro;g/mL, R\u0026amp;D, AF3025), \u0026beta;-actin (1:5,000, Beyotime, AF0003) and GAPDH (1:5,000, Beyotime, AF0006) were used in this study. Peroxidase-conjugated secondary antibody (1:10,000, Cell Signaling Technology, 7074, 7076) was used and signal was visualized using an enhanced chemiluminescence assay (ECL, Thermo), according to the manufacturer\u0026rsquo;s protocol. Band intensity was quantified using Fiji (ImageJ) by grayscale analysis. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRecombinant protein production and purification\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eExpi293F cells were transfected with a \u003cem\u003epcDNA3.4-ACVR1C-ECD-Fc\u003c/em\u003e and \u003cem\u003epcDNA3.4-ACVR1C-ECD-Fc-double mutant (E85A/T101A)\u003c/em\u003e expression vector to produce the target protein, which was subsequently purified using a Protein G column. Briefly, the coding sequence (CDS) of the human \u003cem\u003eACVR1C\u003c/em\u003e extracellular domain (\u003cem\u003eACVR1C-ECD\u003c/em\u003e, NM_145259.3, residues 1-339) fused to an Fc tag was cloned into the pcDNA3.4 vector. Expi293F cells were transfected with this construct, the supernatant was harvested 5 days post transfection.\u003c/p\u003e\n\u003cp\u003eThe supernatant was first centrifuged at 1000 rpm for 20 minutes to remove cell debris, and the supernatant was further centrifuged at 8000 rpm for 30 minutes, followed by filtering with a 0.45 \u0026mu;m PES filter. The protein in the supernatant was then purified using a Protein G column equilibrated with binding buffer (0.15 M NaCl, 20 mM Na₂HPO₄, pH 7.0). The target protein was eluted with 0.1 M glycine (pH 2.5) and immediately neutralized with 1 M Tris-HCl (pH 8.5).\u003c/p\u003e\n\u003cp\u003eSubsequently, the protein buffer was exchanged into a 20 mM Tris-HCl (pH 7.5) system. To further purify the sample, it was centrifuged at 12000 rpm and 4\u0026deg;C for 10 minutes to remove impurities and precipitates. The clarified sample was then loaded onto an ion exchange column (HiTrap\u0026trade; Capto\u0026trade; Q ImpRes) equilibrated with binding buffer (20 mM Tris-HCl, pH 7.5). The ACVR1C-ECD-Fc protein, having an opposite charge to the resin, was bound to the column. Finally, the target protein was eluted with a linear gradient (0-100%) of elution buffer (20 mM Tris-HCl, 1 M NaCl, pH 7.5) over 6 column volumes. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProtein pull-down assay\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eProtein pull-down assay was performed using purified recombinant human His-tagged GREM1 protein and recombinant human ACVR1C-ECD and Fc chimera protein. Protein was enriched by Pierce Protein A magnetic beads (MCE, HY-K0202) or Ni Sepharose 6 Fast Flow (GE, 17531801) following the manufacturer\u0026rsquo;s instructions. Pulled-down proteins were detected by Coomassie Brilliant Blue staining. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMicroScale thermophoresis (MST)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMST was carried out on a Monolith NT.115 instrument (NanoTemper Technologies GmbH). To evaluate ACVR1C-ECD or ACVR1C peptide or ACVR1C-ECD-double mutant (E85A/T101A) binding to GREM1-His or ACTIVINB-His, an increasing concentration of purified ACVR1C-ECD-Fc protein (0\u0026ndash;27.5 \u0026mu;M) or ACVR1C peptide (0\u0026ndash;2.3 \u0026mu;M) or ACVR1C-ECD-double mutant (0\u0026ndash;27.5 \u0026mu;M) was incubated with 50 nM RED-labeled (NanoTemper Technologies GmbH) GREM1-His protein (R\u0026amp;D, 5190-GR) or ACTIVINB-His protein (SinoBiological, 10814-H08H). Experiments were carried out in a PBS buffer pH 7.4 using premium capillaries. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProtein-protein interaction docking study\u003c/strong\u003e \u003c/p\u003e\n\u003cp\u003eGREM1 (PDB: 5AEJ) was selected as the ligand and ACVR1C (PDB: AF-Q8NER5-F1) as the receptor for protein-protein docking. The HDOCK web service was used for docking with default parameters (http://hdock.phys.hust.edu.cn/). Key amino acid residues in the binding pocket between GREM1 and ACVR1C were further identified based on the docking module[69]. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRNA-seq and gene set enrichment analysis (GSEA)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTotal RNA was extracted using Trizol reagent (Invitrogen, 15596026) and quantified with a NanoDrop spectrophotometer (Thermo Fisher Scientific). RNA integrity was assessed using an Agilent 2100 Bioanalyzer. mRNA was enriched using oligo(dT) magnetic beads, fragmented, and reverse-transcribed into cDNA. After adapter ligation and PCR amplification, libraries were sequenced on an Illumina platform, generating 150-bp paired-end reads. Raw reads were trimmed and aligned to the human reference genome (GRCh38) using STAR. Differential gene expression analysis was conducted using Limma, with significance thresholds set at |log2FoldChange| \u0026gt;1.5 and adjusted P-value \u0026lt;0.05. Gene set enrichment analysis (GSEA) was performed using the GSEA software (Broad Institute) with the MSigDB gene sets to identify enriched biological pathways, employing 1,000 permutations and FDR \u0026lt;0.25 as the cutoff for significance. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRT\u0026ndash;qPCR\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTotal RNA was extracted using Trizol reagent (Invitrogen, 15596026). According to the instruction, cDNA was generated using the PrimeScript RT reagent Kit with gDNA Eraser (Accurate Biology, AG11706). The SYBR Green Premix Pro Taq HS qPCR Kit (Accurate Biology, AG11701) was then used to quantify mRNA expression according to the manufacturer\u0026rsquo;s instruction. All results were calculated using the 2\u003csup\u003e-\u0026Delta;\u0026Delta;ct\u003c/sup\u003e method. Primers used in the study are listed in Supplementary Table 1. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eChIP\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSW480 cells were starved in DMEM with 1% FCS overnight before treatment with vehicle, 10 \u0026mu;M SB505124 for 24 hours. Cells were fixed in 1 % paraformaldehyde for 10 min at RT for DNA-protein cross-linking, followed by quenching with glycine. Cross-linking chromatin was prepared using the SimpleChIP\u0026reg; Enzymatic Chromatin IP Kit (CST, 9002) according to the manufacturer\u0026rsquo;s instructions. For immunoprecipitation, 10 \u0026mu;g chromatin was incubated with 10 \u0026mu;L anti-histone H3 rabbit IgG (CST, 14269, positive control), 2 \u0026mu;L normal Rabbit IgG (CST, 2729) or 5 \u0026mu;L anti-SMAD2/3 rabbit IgG (CST, 8685) at 4 \u0026deg;C overnight. 2% chromatin prior to immunoprecipitation was used as input. Chromatin-protein-antibody complex was captured by protein G magnetic beads, and chromatin was released by reversal of cross-links and purified using the SimpleChIP\u0026reg; Enzymatic Chromatin IP Kit (CST, 9002) according to the manufacturer\u0026rsquo;s instructions. DNA was quantified by qPCR with primers targeting predicted SMAD2/3/4 binding regions on \u003cem\u003eGREM1\u003c/em\u003e or \u003cem\u003eSNAI1\u003c/em\u003e promoters. DNA levels were normalised to the input, and the fold-change of enrichment was calculated over the control. ChIP\u0026ndash;qPCR primers are listed in Supplementary Table 2. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eScratch assay\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCells were seeded into 6-well plates after centrifugation and digestion with 0.05% trypsin. When the cell density reached 90%, three vertical lines were scratched in each well with a 10 \u0026mu;L pipette tip and the floating cells were gently washed away with 1\u0026times;PBS. Complete medium was added, and images of the scratch area were taken at 0 h. Three different fields of view were selected for each well. After photography, the medium was replaced with serum-free medium. Wound healing was documented at the same location after 24 h or 48 h of incubation. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTranswell invasion assay\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCells (1\u0026times;10\u003csup\u003e5\u003c/sup\u003e) were seeded in serum-free medium in the Matrigel-coated (Corning, 354480) transwell chambers (24-well insert, 8-\u0026mu;m pore size; BD Biosciences) for invasion experiments. The lower chamber was filled with RPMI1640 or DMEM containing 20% FBS. The migration of HCT116 and SW480 cells was measured in three random visual fields and quantified by microscopy after 48 h of incubation, followed by staining with DAPI or crystal violet. The invasive capacity of the cells was assessed using ImageJ software for quantification. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll the statistical analyses were performed using GraphPad Prism 9, and error bars indicate s.e.m. Student\u0026rsquo;s t-test assuming equal variance and one-way analysis of variance for independent variance were used. Growth curves were generated using ANOVA for repeated measurement. \u003cem\u003eP\u003c/em\u003e\u0026lt;0.05 was considered significant. The number of independent experiments, the number of events and information about the statistical details and methods are indicated in the relevant figure legends. \u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"molecular-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"molc","sideBox":"Learn more about [Molecular Cancer](http://gsejournal.biomedcentral.com/)","snPcode":"12943","submissionUrl":"https://submission.nature.com/new-submission/12943/3","title":"Molecular Cancer","twitterHandle":"@SN_Oncology","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Colorectal cancer, GREM1–ACVR1C axis, paracrine-to-autocrine shift, signaling autonomy, EMT","lastPublishedDoi":"10.21203/rs.3.rs-7484753/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7484753/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground:\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTumor cells typically rely on paracrine stromal signals to guide malignant behavior, yet whether they gain signaling autonomy and thereby reduce microenvironment dependency during metastasis remains unclear.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods:\u003c/b\u003e\u003c/p\u003e\u003cp\u003eColorectal cancer (CRC) specimens from stages I\u0026ndash;IV were analyzed by immunohistochemistry and single-cell transcriptomics to assess GREM1 and ACVR1C expression and localization. The GREM1\u0026ndash;ACVR1C interaction was validated by interaction proteomics, co-immunoprecipitation, immunofluorescence, and microscale thermophoresis (MST). Functional roles of the axis in metastasis were examined by transcriptomic profiling, pathway analysis, immunoblotting, RT\u0026ndash;qPCR, scratch and transwell assays, and genetically engineered and xenograft mouse models. An inhibitory peptide targeting the GREM1\u0026ndash;ACVR1C interface was designed and evaluated.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults:\u003c/b\u003e\u003c/p\u003e\u003cp\u003eWhile GREM1 remains restricted to stromal cells in earlier-stage (I\u0026ndash;III) CRC, its ectopic expression in tumor epithelium increases markedly in stage IV. Mechanistically, we identify activin A receptor type 1C (ACVR1C) as a direct, high-affinity epithelial receptor for GREM1. Their interaction, independent of canonical TGFβR and BMP signaling, activates SMAD2/3, which in turn induces the transcription of \u003cem\u003eSNAI1\u003c/em\u003e and \u003cem\u003eGREM1\u003c/em\u003e, thereby establishing a self-sustaining feedback loop that amplifies epithelial-mesenchymal transition (EMT). Disrupting this loop via stromal GREM1 deletion, epithelial ACVR1C knockdown, kinase inhibition, or a novel GREM1-blocking peptide targeting the GREM1-ACVR1C binding interface significantly impairs CRC metastasis \u003cem\u003ein vivo\u003c/em\u003e. Clinically, epithelial GREM1 or ACVR1C expression predicts aggressive disease and poor survival.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions:\u003c/b\u003e\u003c/p\u003e\u003cp\u003eOur findings define a paradigm in which tumor cells hijack stromal GREM1 to establish a GREM1\u0026ndash;ACVR1C autocrine loop that sustains EMT and metastasis, marking a shift toward signaling autonomy and revealing a targetable vulnerability in advanced CRC.\u003c/p\u003e","manuscriptTitle":"A Paracrine-to-Autocrine Shunt of GREM1 Fuels Colorectal Cancer Metastasis via ACVR1C","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-17 13:04:34","doi":"10.21203/rs.3.rs-7484753/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-01T02:23:24+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-26T14:23:41+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-26T08:29:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"284432269376984615364829670475934462245","date":"2025-09-22T08:26:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"11408380184543376920058962728812874463","date":"2025-09-22T01:25:05+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"6460647665454953748255479659266157418","date":"2025-09-21T03:13:19+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"71683942688463997716291094492539335586","date":"2025-09-21T01:31:23+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"273266150230461538959450977165723869187","date":"2025-09-20T19:33:20+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-09T23:18:56+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-02T01:13:06+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-02T01:12:09+00:00","index":"","fulltext":""},{"type":"submitted","content":"Molecular Cancer","date":"2025-08-29T04:21:30+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"molecular-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"molc","sideBox":"Learn more about [Molecular Cancer](http://gsejournal.biomedcentral.com/)","snPcode":"12943","submissionUrl":"https://submission.nature.com/new-submission/12943/3","title":"Molecular Cancer","twitterHandle":"@SN_Oncology","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"45263e45-5266-4227-8646-044f34009164","owner":[],"postedDate":"September 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-26T16:08:30+00:00","versionOfRecord":{"articleIdentity":"rs-7484753","link":"https://doi.org/10.1186/s12943-025-02554-w","journal":{"identity":"molecular-cancer","isVorOnly":false,"title":"Molecular Cancer"},"publishedOn":"2026-01-24 15:57:27","publishedOnDateReadable":"January 24th, 2026"},"versionCreatedAt":"2025-09-17 13:04:34","video":"","vorDoi":"10.1186/s12943-025-02554-w","vorDoiUrl":"https://doi.org/10.1186/s12943-025-02554-w","workflowStages":[]},"version":"v1","identity":"rs-7484753","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7484753","identity":"rs-7484753","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}