BRAF ^V600E Expression in c-Kit+ Interstitial Cells of Cajal Drives Gastrointestinal Stromal Tumor Formation in Mice

Abstract

Gastrointestinal stromal tumors (GIST) are rare sarcomas arising predominantly in the stomach and small intestine, with limited incidence in other gastrointestinal sites. GIST originates from c-KIT–/DOG1-positive interstitial cells of Cajal (ICC) most commonly driven by activating gain-of-function mutations in the c-KIT and PDGFRA genes. However, a subset of these malignancies, defined as wild-type GIST (WT-GIST), has mutations in the RAS pathway genes (e.g., NF1, KRAS, and BRAF), succinate dehydrogenase subunit genes, and rare gene fusions. A major obstacle in understanding WT-GIST pathogenesis and treatment resistance has been the lack of robust in vivo models. In this study, we report the development of a fully penetrant mouse model of BRAF-driven GIST by inducing BRAF^V600E expression in c-Kit+ ICC. Unlike previous models which targeted only subsets of ICC leading to hyperplasia, the use of c-KitCreERT2 enables broader targeting, including ICC progenitors, resulting in rapid, multifocal tumor formation primarily in the pyloric region. These tumors express diagnostic GIST markers (c-Kit and DOG1) and show significant response to the BRAF inhibitor dabrafenib. This model recapitulates key histopathologic and molecular features of human BRAF-mutant GIST and provides a valuable platform for studying tumor initiation, progression, and therapeutic resistance. Importantly, it allows for preclinical testing of targeted therapies in BRAF GIST, offering new insights into treatment strategies.

Significance:

This is the first fully penetrant mouse model of BRAF-driven GIST by inducing BRAF^V600E in c-Kit+ ICC. This model faithfully recapitulates human BRAF-mutant GIST, enabling mechanistic studies and preclinical testing of targeted therapies, including response to BRAF inhibition with dabrafenib.

Introduction

Gastrointestinal stromal tumors (GIST) represent a heterogeneous group of mesenchymal neoplasms arising predominantly within the gastrointestinal (GI) tract, particularly the stomach and small intestine. These tumors originate from the interstitial cells of Cajal (ICC), a specialized population of cells within the tunica muscularis responsible for generating the electrical slow waves that regulate gut motility (1–5). ICC express key markers such as c-KIT (CD117) and DOG1 (ANO1), which serve as diagnostic hallmarks in the vast majority of GIST cases (4–7). Approximately 85% of GISTs harbor gain-of-function mutations in the receptor tyrosine kinases KIT (most commonly in exon 11) or PDGFRA (typically in exon 18), resulting in constitutive, ligand-independent activation of downstream signaling pathways (8–10). These molecular insights have led to the development and clinical implementation of targeted therapies, particularly tyrosine kinase inhibitors such as imatinib, sunitinib, regorafenib, and ripretinib (11–14). The remaining ∼15% of cases, known as wild-type GIST (WT-GIST), lack mutations in both KIT and PDGFRA. Genomic studies have revealed diverse molecular alterations among WT-GIST (15–18). A significant proportion of these tumors display mutations in subunits of the mitochondrial enzyme succinate dehydrogenase (SDH) and are classified as SDH-deficient GIST (SDH-GIST; ref. 15). These tumors are especially prevalent in pediatric and young adult populations and are characterized by distinct molecular, clinical, and histopathologic features (6, 18). Other rare subtypes of WT-GIST involve mutations or alterations in the RAS–MEK–MAPK signaling pathway, including genes such as BRAF, NF1, and KRAS (19–25), as well as gene fusions involving NTRK and FGFR (26–28). Among these, activating mutations in BRAF, particularly the BRAF^V600E allele, are found in approximately 1% to 3% of human GIST cases. Despite advances in molecular profiling, few preclinical in vivo models exist for studying WT-GIST. One study conditionally expressed BRAF^V600E in ICC using an inducible Etv1CreERT2 driver (29). Although Etv1 is restricted to myenteric (ICC-MY) and intramuscular (ICC-IM) ICC subsets, BRAF^V600E expression alone resulted only in hyperplasia—not fully developed GIST-like lesions—even in homozygous mice observed for 6 to 15 months. Concurrent loss of Trp53 in this model led to multifocal GIST-like tumors with 100% penetrance, implicating additional tumor suppressor loss as a critical co-factor in GIST pathogenesis. In a separate study, BRAF^V600E was expressed in smooth muscle cells via the Myh11CreERT2 driver (30). This approach yielded GI malignancies with approximately 36% penetrance, suggesting that, under certain conditions, smooth muscle cells could serve as an alternative cell of origin for GIST.

In this study, we investigated the tumorigenic potential of BRAF^V600E in a broader range of ICC using the validated c-KitCreERT2 mouse model (2, 3, 31). Notably, BRAF^V600E expression alone was sufficient to induce both GI hyperplasia and GIST-like tumors within just 3 weeks after induction. These tumors also demonstrated responsiveness to the BRAF inhibitor dabrafenib, supporting the translational relevance of this model for preclinical therapeutic evaluation. Our findings underscore the utility of this model not only in understanding BRAF^V600E-driven GIST tumorigenesis but also in exploring mechanisms of drug resistance and novel treatment strategies (20, 21). Given the unmet need for robust animal models of WT-GIST, particularly those with defined molecular drivers, these results provide a new in vivo tool for the field.

Materials and Methods

Mouse models

Mouse strains used in this study include c-KitCreERT2 (1); BRAF^V600E (RRID: IMSR, JAX:017837); β-Actin-Cre (RRID: IMSR_JAX:033984); and Rosa26mTmG (RRID: IMSR JAX:007676). Mice were housed under specific pathogen-free conditions with ad libitum access to food and water. Tamoxifen (Sigma-Aldrich T5648) was prepared at a concentration of 10 mg/mL in corn oil (Sigma-Aldrich C8267) as per established protocol in the Jackson Laboratory (Bar Harbor, ME) website. Dabrafenib [HY-14660, MedChemExpress; dissolved in 10% DMSO, 40% polyethylene glycol, 5% polysorbate (Tween 80), and 45% saline] was administered at 100 mg/kg by oral gavage every other day, three times a week (Monday, Wednesday, and Friday) for 4 weeks (12 dosages). All procedures were approved by the NCI-Frederick Animal Care and Use Committee and followed NIH animal welfare guidelines.

Immunofluorescence

Stomachs from c-KitCreERT2/+; Rosa26mTmG/+ and c-KitCreERT2/+ mice were harvested 10 days after tamoxifen (TAM) induction. Tissues were fixed in 4% paraformaldehyde (2 hours at room temperature), cryoprotected overnight in 30% sucrose at 4°C, embedded, and sectioned at 40 μm. Native fluorescence was imaged using a Zeiss LSM 880 confocal microscope without further staining.

IHC and pathology

Complete necropsy was performed on all major organs. Gastric dilation with muscular wall thickening and pyloric obstruction was routinely observed. Tissues (stomach, duodenum, jejunum, ileum, cecum, and colon) were fixed in 10% neutral buffered formalin, paraffin embedded, and sectioned (5 μm) for hematoxylin and eosin staining.

Histologic grading of ICC lesions was performed as described in the “Results” section. IHC was performed on a Leica Bond RX autostainer using the Bond Polymer Refine Detection Kit (#DS9800, Leica Biosystems; minus post-primary). Retrieval conditions and antibodies were

Marker	Retrieval	Antibody	Dilution	RRID
DOG1	Citrate (20′)	Thermo Fisher Scientific MA5-16358	1:25	AB_2537877
c-Kit	EDTA (20′)	Cell Signaling Technology 3074	1:100	AB_1147633

Negative controls used Cell Signaling Technology isotype control IgG (#3900, RRID: AB_1550038). Slides were dehydrated in graded ethanol, cleared with xylene, and coverslipped.

Results

The c-Kit^CreERT2 allele induces cre expression in the c-Kit–positive ICC

As ICCs are the putative GIST cell of origin, we introduced the BRAF^V600E conditional allele in c-kit (KIT)–positive ICC using the validated c-Kit^CreERT2 mouse model (1–3, 32). The efficacy of the c-Kit^CreERT2 allele to induce cre expression was tested with the double-fluorescent R26^mT-mG/+ reporter mouse line (Fig. 1; ref. 1). A single TAM injection (75 mg/kg, i.p.) at postnatal day 20 (P20) induced EGFP expression in ICC of the gastric muscularis by postnatal day 30 (Fig. 1B), suggesting that the c-Kit^CreERT2 allele provides an effective means to induce cre expression in the c-Kit–positive ICC cellular network.

Figure 1.

Targeting c-Kit–positive ICC. A, Timing of the protocol used to target mouse ICC with the c-Kit^CreERT2 driver. Mice were injected intraperitoneal (i.p.) with TAM (75 mg/kg) at postnatal day 20 (P20) and analyzed at postnatal day 30 (P30). B, Representative confocal images of the stomach wall of uninjected (−TAM) or TAM injected (+TAM) mice with the c-Kit^CreERT2/+ and the R26^mT-mG reporter allele. The R26^mT-mG allele induces tdTomato expression ubiquitously and after cre-mediated removal of tdTomato promotes EGFP expression. Note, activation of EGFP-expressing c-Kit–positive ICC within the tunica muscularis of c-Kit^CreERT2/+; R26^mT-mG TAM-injected mice, suggesting ICC are efficiently targeted. DIC, differential interference contrast.

BRAF ^V600E activation in c-Kit+ ICC drives hyperplasia and GIST-like lesions

The BRAF^V600E mutation is a rare genetic alteration affecting the RAS/RAF/MAPK pathway in WT-GIST. To determine whether BRAF^V600E is sufficient to drive GIST, we expressed the allele in c-Kit+ ICCs via c-KitCreERT2. Prior models using Etv1- or Lrig1-driven expression produced hyperplasia without full transformation unless combined with Trp53 loss or smooth muscle–specific promoters (29, 30). To test whether (i) BRAF^V600E was insufficient to induce GIST, (ii) the specific ICC population leading to tumor formation was not targeted in the previous studies or (iii) these tumors derive from an alternative cell of origin, we used the c-Kit^CreERT2 allele to induce cre expression in a broader population of ICC, including ICC precursor cells. Strikingly, all BRAF^V600E;c-Kit^CreERT2 mice developed severe signs of distress, including immobility, hunched posture, and abnormal abdomen shape and had to be sacrificed at about 3 weeks after TAM administration (Supplementary Table S1). At necropsy, all animals showed an extremely dilated and distended stomach filled with air and fluids (Fig. 2A). Pathologic and histologic assessment revealed that the stomach dilation was caused by a pyloric annular constriction which highly reduced gastric outflow into the small intestine, promoting accumulation of fluids and air in the lumen of the stomach. Further analysis showed nodular formation of infiltrative neoplastic cells that circumferentially expanded the tunica muscularis at the region of the gastroduodenal junction in all animals, suggesting that this was most likely the cause of the gastroduodenal occlusion (Fig. 2B). In about 40% of the animals, the neoplastic aggregates were consistent with GIST-like lesions that displayed pleomorphic and occasionally spindloid cells. Other features consisting of moderate levels of anisokaryosis, anisocytosis, and karyomegalic cells containing open chromatin and distinct nucleoli were also noticeable within the neoplastic lesions (Fig. 2C and D). Although the pyloric region was by far the most consistently and severely affected site, many nodular aggregates of hyperplastic/preneoplastic cells were also present in the tunica muscularis of the stomach or small and large intestine wall. These nodular aggregates were often infiltrating the adjacent muscle fibers, with focal, multifocal, and coalescing nodular proliferative lesions (Fig. 2E and F), although complete necropsy revealed no signs of metastasis. Tumor cells also expressed c-KIT (KIT) and DOG1 (ANO1), a key feature of human GIST diagnosis, from the initial stages of nodular cell aggregates to more advanced stages of GIST (Fig. 3; Supplementary Fig. S1). These findings indicate that BRAF^V600E activation in ICC is sufficient to initiate GIST-like tumors, with lethality likely limiting full progression in some cases.

Figure 2.

The BRAF^V600E mutation in c-Kit–positive ICCs leads to severe hyperplasia and GIST formation. A, Representative image of the distended stomach developed by the c-Kit^CreERT2/+; BRAF^V600E/+-mutant animals about 3 weeks after 75 mg/kg TAM administration. B, Low-power image of a hematoxylin and eosin–stained section of the pyloric region from a TAM-induced animal as in A. Note the nodular aggregates of cells circumferentially expanding the tunica muscularis at the pyloric region. Black arrows indicate the gastroduodenal constriction. C, Low-power image of a hematoxylin and eosin–stained section of the pyloric region of TAM-induced BRAF^V600E animal showing a GIST in the tunica muscularis (arrowhead). D, High-power image of a GIST-like lesion showing pleomorphic and spindeloid cells. Note the moderate anisokaryosis and anisocytosis and karyomegalic cells containing open chromatin and distinct nucleoli (arrow heads). E and F, Representative images of hematoxylin and eosin sections from BRAF^V600E mice showing nodular (E) and multinodular (F) aggregates of hyperplastic cells within the tunica muscularis along the stomach and intestine (arrow heads). H&E, hematoxylin and eosin.

Figure 3.

Mouse GIST lesions express the typical molecular markers of human GIST, Dog1 and c-Kit. A–D, Representative histologic images of GIST-like lesions from the pyloric area of BRAF^V600E-mutant mice stained with hematoxylin and eosin (A and C) showing aggregates of coalescing hyperplastic cells. Note, these lesions express the typical molecular markers of human GIST, Dog1 (B) and c-Kit (D). H&E, hematoxylin and eosin.

We focused on the pyloric region because it is the most consistent site of lesions. Pylori from mutant mice were evaluated using a 0 to 5 grading system (Fig. 4). Grade 0 indicates no abnormalities detected in the pyloric sections. Grade 1, 2, and 3 indicate, respectively, focal, multifocal, and coalescing nodular hyperplasia of ICC populations associated with a minimal (1), mild (2), or moderate (3) grade of hyperplasia. Grade 4 corresponds to a severe ICC hyperplasia and more complex GIST-like lesions caused by proliferative ICC coalescing and effacing the tunica muscularis and displaying cells arranged in bundles or sheets and minimal stroma. Grade 5 indicates GIST malignancies with clear tumor masses expanding the tunica muscularis and including cellular atypia and pleomorphic morphology. This histologic grading showed that grade 4 to 5 GIST were present in 41% (five of 12) of mice (Fig. 5A).

Figure 4.

Tumor histologic grading of the pyloric regions from mutant mice. Low (left)- and high (right)-magnification images from representative hematoxylin and eosin staining of sections from the pyloric region of BRAF^V600E-mutant mice 3 weeks after TAM (75 mg/kg) injection showing the typical morphology used to grade the severity of the GIST lesions. Grade 0, normal pylorus. Grades 1 and 2 represent focal and multifocal nodular hyperplasia of ICC, respectively (minimal and mild ICC hyperplasia). Grade 3 indicates coalescing nodular proliferative lesions (moderate ICC hyperplasia). Grade 4 indicates proliferative ICC coalescing and effacing the tunica muscularis with cells arranged in bundles or sheets with minimal stroma (severe ICC hyperplasia – GIST-like lesions). In grade 5, there are GIST tumors expanding from the tunica muscularis with frequent cells showing atypia and pleomorphic morphology.

Figure 5.

Activation of the BRAF^V600E allele in c-Kit–positive ICCs originates hyperplasia and GIST. A, Violin plot reporting the number of c-Kit^CreERT2/+; BRAF^V600E/+ mice with lesions of the indicated histopathologic grade as in Fig. 4 following induction with 75 or 7.5 mg/Kg TAM. B, Kaplan–Meier survival curve of c-Kit^CreERT2/+ (Ctrl) and c-Kit^CreERT2/+; BRAF^V600E/+ (BRAF) mice induced with 75 mg/kg TAM (red line) or 7.5 mg/kg TAM (blue line) at 3 weeks.

As most mice die within 3 weeks after induction because of the pyloric restriction, we decided to test whether reducing TAM dose could influence the number of cells with the B-raf activation and consequently the severity of tumor development and mouse viability. We reasoned that this could help gain insights into tumor formation and progression over time. Thus, a second cohort of mice received a 10 times lower TAM dose (7.5 mg/kg). At 3 months after injection, survival was 75% in the BRAF low-dose TAM animals (Fig. 5B). Moreover, histopathologic analysis of the stomach and pyloric region of sick mice showed that animals in the low-dose group stratified over milder pathology grades when compared with the higher dose (Fig. 5A; Supplementary Table S1). These data indicate that dosing TAM can allow for control of tumor penetrance and expand our temporal analysis of BRAF^V600E-driven GIST lesions (Fig. 5).

Dabrafenib partially rescues the GIST phenotype in mutant mice

Given the clinical use of dabrafenib in BRAF^V600E-driven cancers, including non–small cell lung cancer, melanoma, GIST, and anaplastic thyroid cancer, we assessed its efficacy in our model (20, 21, 33–35). We chose a drug administration protocol that partially inhibits BRAF activity to test whether it could extend the life span of mice with tumors. BRAF^V600E animals first received a single TAM dose (75 mg/kg) at P20 to activate the mutation in ICC. Ten days after TAM, they received dabrafenib (100 mg/kg by oral gavage) or vehicle treatment three times a week for 4 weeks (12 dosages; Fig. 6A). Although all animals receiving vehicle treatment became sick and had to be sacrificed at 3 weeks after TAM, as previously observed, mice that received dabrafenib had an overall better outcome as they survived to the end of the dabrafenib treatment (Fig. 6A). Histopathology revealed marked reduction in lesion severity, suggesting that they could be employed in studies of tumor resistance of BRAF GIST (Fig. 6B and C).

Figure 6.

BRAF-driven lesions are sensitive to the B-Raf inhibitor dabrafenib. A, Kaplan–Meier survival curve of c-Kit^CreERT2/+; BRAF^V600E/+ (BRAF)-mutant animals injected at P20 with vehicle or 75 mg/kg TAM, followed by dabrafenib (DAB) treatment three times a week for 4 weeks (100 mg/kg) starting at P30. B, Histopathologic evaluation of the pyloric regions of mice in A according to the classification in Fig. 4. C, Representative images of hematoxylin and eosin–stained sections of the pyloric region of vehicle- and dabrafenib-treated c-Kit^CreERT2/+; BRAF^V600E/+ animals as in A and B. Note the decreased tumor burden in the tunica muscularis of dabrafenib-treated mice (scale bar, 500 μm).

Discussion

WT-GIST represent a heterogeneous subgroup of mesenchymal malignancies of the GI tract for which there are limited therapeutic opportunities. The lack of robust genetic in vivo models has significantly limited our understanding of their molecular pathogenesis, mechanisms of resistance, and responses to targeted therapies.

In this study, we demonstrate that the expression of the oncogenic BRAF^V600E allele in ICC, the putative cells of origin for GIST, is sufficient to initiate GIST formation. Importantly, we show that the resulting GIST-like lesions are responsive to treatment with the BRAF inhibitor dabrafenib. This reproduces clinical observations in GIST with a BRAF^V600E mutation that responds to BRAF inhibitors (20, 21). This finding establishes a valuable preclinical platform for studying both tumor initiation and therapeutic response in the context of BRAF-mutant GIST.

Models of GIST driven by an activating mutation in KIT have been successfully developed through the introduction of an activating KIT mutation in the germline and expanded through a second-site KIT mutation targeting ETV+ cells (36, 37). However, previous attempts to generate GIST models through activation of the RAS–RAF–MAPK pathway by introducing BRAF^V600E targeting ETV+ or Lrig1+ cells have been limited in their success. In these earlier models, the expression of BRAF^V600E led only to focal hyperplasia, without full transformation to GIST, and lesions occurred only after prolonged latency periods ranging from 3 to 15 months (29, 30). A key distinction in our study is the choice of promoter used to drive Cre-mediated recombination. Although prior models utilized the Etv1CreERT2 and Lrig1CreERT2 lines, both of which target only subsets of ICC, we employed a c-KitCreERT2 allele, which has been shown to broadly and effectively target the entire ICC population along the GI tract (1, 32). The success of this strategy suggests that it could be applied to generate additional molecular subtypes of WT-GIST models, including SDH-deficient GIST and other rare subtypes such as NF1-null or KRAS-mutated GIST.

The spatial and developmental specificity of these promoters likely underlies the differing outcomes. Etv1 expression is limited to ICC-MY and ICC-IM subtypes and is excluded from the deep muscular (ICC-DMP) and submucosal plexus ICC (ICC-SMP; refs. 38, 39). In contrast, Lrig1 expression is largely restricted to ICC-DMP and ICC-SMP (40). Thus, both previous models may have targeted more differentiated or regionally restricted ICC subtypes, which may be less susceptible to oncogenic transformation. On the other hand, c-Kit expression spans a broader range of ICC, including precursor and progenitor populations derived from mesenchymal stem cells (41). These c-Kit–positive progenitors are likely to represent the critical population that, upon acquiring BRAF^V600E, undergoes oncogenic transformation into GIST cells.

Supporting this interpretation is the well-established use of c-KIT and DOG1 (ANO1) as both lineage markers for ICC and diagnostic markers for GIST in clinical pathology (5, 42, 43). Our IHC analysis confirmed the expression of c-KIT and DOG1 in the BRAF lesions, further validating their identity as bona fide GIST. Thus, we propose that the broader targeting enabled by the c-KitCreERT2 allele is a crucial factor underlying the observed differences in tumor latency, distribution, and penetrance compared with previous models.

Our model demonstrates that BRAF^V600E expression in c-Kit–positive ICC leads to rapid and multifocal hyperplasia, with the earliest and most prominent lesions occurring at the pyloric junction, a region enriched in smooth muscles and ICC (44). This regional susceptibility likely accounts for the reproducible formation of pyloric tumors and the early lethality observed in mice. These findings suggest that the pylorus may serve as a reliable and anatomically consistent site for tumor evaluation and therapeutic testing. Therapeutically, we show that treatment with dabrafenib significantly reduces tumor burden in mice harboring BRAF-mutant lesions. Here, we tested a single treatment protocol of 4 weeks (Fig. 6), but future studies should allow us to evaluate whether different dabrafenib dosages and/or treatment durations can induce tumor resistance mechanisms. Furthermore, this model should enable exploration of the combinatorial effects of other drugs such as trametinib or other MEK inhibitors (35). Together, such studies, coupled with varying TAM concentrations, should allow for extended temporal analysis of these malignancies, potentially enabling these lesions to acquire metastatic properties over time. This highlights the translational relevance of our model and suggests that it could be instrumental in testing new targeted therapies for BRAF-mutant GIST that have developed drug resistance.

In summary, we have developed a novel, genetically defined and fully penetrant mouse model of WT-GIST driven by BRAF^V600E. This is the first in vivo model to recapitulate the full spectrum of key molecular and histopathologic features of human BRAF-mutant GIST. Its rapid and reproducible tumor development, especially in the pyloric region, provides a robust platform for investigating tumor initiation, progression, and therapeutic resistance. Furthermore, this model opens the door for preclinical testing of novel therapies and combination strategies targeting the diverse molecular landscape of WT-GIST, including RAS-/MAPK-driven subtypes.

Supplementary Material

Supplementary Figure 1

Supplementary Figure 1. Expression of Dog1 and c-kit in developing mouse GIST lesions.

Supplementary Table 1

Supplementary Table 1. Summary of the mouse data from the experiments using tamoxifen at the concentration of 75 and 7.5 mg/kg and the BRAF inhibitor 'Dabrafenib'.

Data Availability

The data generated in this study are available upon request from the corresponding author.

Authors’ Disclosures

D. Saur reports grants from German Cancer Consortium (DKTK), Deutsche Forschungsgemeinschaft, and Deutsche Krebshilfe during the conduct of the study, as well as grants from German Cancer Consortium (DKTK), Joint Funding, outside the submitted work. J.W. Glod reports other support from US WorldMeds and Inhibrx outside the submitted work. No disclosures were reported by the other authors.

Authors’ Contributions

F. Tomassoni-Ardori: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. K.M. Reilly: Conceptualization, resources, funding acquisition. C. Barrick: Investigation. R. Koogle: Investigation. E.F. Edmondson: Data curation, formal analysis. S. Yanpallewar: Investigation. D. Saur: Resources. B.C. Widemann: Conceptualization, funding acquisition. J.W. Glod: Conceptualization, data curation, formal analysis, validation, writing–review and editing. L. Tessarollo: Conceptualization, resources, data curation, supervision, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing.