Results
Case 1, a 16 year-old female with a 2.5 cm rapidly growing subcutaneous thigh mass was diagnosed with a malignant histiocytic neoplasm (“M group”)8, with high-grade morphologic features and a phenotype spanning histiocytic sarcoma (CD163/CD14/CD68+) and Langerhans cell sarcoma (CD1a/Langerin/S100) with a modestly elevated Ki-67 proliferation index (up to 20%) (Figure 1A-F). Targeted RNA-sequencing identified a MTAP-BRAF fusion transcript. Resection margins were negative. The patient is disease-free three years post-resection. Case 2, a 12-year-old female with a 5.3 cm rapidly enlarging heel mass invading the calcaneus was originally diagnosed with a juvenile xanthogranuloma (JXG) family lesion (CD163/CD68/CD14/fascin/Factor XIIIa+) (Figure 1G-M). Despite low-grade morphologic features and lack of cytologic atypia or increased mitotic rate by H&E stain, a high proliferation rate (up to 40%) was noted by Ki-67 proliferation index stain. Targeted RNA-seq identified a MS4A6A-BRAF fusion transcript. During staging, the patient was found to have PET-avid dissemination to lymph nodes and lung (Figure 1N-P). While the initial morphologic features were consistent with a low-grade histiocytic lesion of JXG phenotype, the integration of the high Ki-67 proliferation index and aggressive clinical behavior with lymphatic/metastatic-like spread, along with a novel molecular BRAF-fusion, at the time of diagnosis, suggested an atypical JXG family neoplasm with uncertain biological behavior. The patient was treated with 12 cycles of clofarabine with clinical remission of metastatic sites and near clinical remission at primary site now 18 months off therapy.
MTAP-BRAF and MS4A6A-BRAF fusions are predicted to contain all functional domains of MTAP and MS4A6A, respectively, along with the BRAF kinase domain but no N-terminal regulatory, RAS-binding domain (Figure 1Q). For molecular and therapeutic characterization, MTAP-BRAFand MS4A6A-BRAF were cloned and stably expressed in a heterologous cell model since patient-derived cell lines were lacking. The NIH/3T3 cells model system was utilized for its ability to reliably discern oncogenic fusion profiles9-11. In soft agar assays, both MTAP-BRAF and MS4A6A-BRAF expressing NIH3T3 showed a significant increase in colony count over control (p<0.0001, Figure 1R). Next, we tested activation of downstream MAPK and PI3K/mTOR pathways. Upon serum starvation, we observed elevated levels of phosphorylated-ERK and -S6 in both BRAF- fusion expressing cells compared to controls, indicating aberrant activation of the MAPK and PI3K/mTOR pathways, respectively (Figure 1S). Slightly higher PI3K/mTOR pathway activation levels in MTAP-BRAF versus MS4A6A-BRAF cells are partly explained by higher MTAP-BRAF protein expression (Figure 1S, Myc-tag blot).
A single report on BRAF- fusions in LCH4 has shown unresponsiveness to first generation specific BRAF-V600E inhibitors (RAFi) such as vemurafenib, but observed suppression by second-generation RAFi, PLX8394, and downstream MEK inhibition, similar to other pediatric glioma studies on BRAF-fusions9,11. Herein, we evaluated the responsiveness of MTAP-BRAF and MS4A6A-BRAF to such targeted inhibitors. Upon targeting the NIH3T3 models with first-generation RAFi PLX4720, as expected, no suppression of BRAF-fusion driven signaling or growth was observed (Supplemental Figure S1A). Interestingly, second-generation RAFi PLX8394 also showed no suppression in MTAP- or MS4A6A-BRAF driven soft agar growth despite targeting MAPK/PI3K signaling (Figure 2A-B). This is in contrast to PLX8394-mediated suppression of BRAF-fusion driven growth in the previously described LCH4 and other cancers, such as the KIAA1549-BRAF fusion in pediatric glioma9-11. PLX8394 suppressed FAM131B-BRAF (a pediatric glioma derived fusion12,13) and BRAF-V600E driven growth and signaling as well as actively disrupted FAM131B-BRAF dimers (Supplemental Figures S1B-D), highlighting therapeutic differences between MTAP-/MS4A6A-BRAF, BRAF-V600E and other BRAF-fusions.
BRAF- fusions function as active homo- and heterodimers (with wild-type BRAF) to mediate cell signaling9,11. We found that MTAP- and MS4A6A-BRAF also mediate such protein-protein interactions in co-immunoprecipitation assays (Figure 2C-D, DMSO lanes). PLX8394 blocks BRAF kinase activity via disrupting BRAF dimerization14 but we observed no disruption of MTAP- and MS4A6A-BRAF fusion dimerization with PLX8394 (Figure 2C-D, PLX8394 lanes), thereby providing a plausible explanation for PLX8394 unresponsiveness in soft agar assays though MAPK/PI3K signaling remains discordantly suppressed by some unknown mechanism. This distinct unresponsiveness to pan-RAFi represents a significant departure from the current view that BRAF -fusions and other BRAF mutations should respond to second-generation RAFi such as PLX83949,15. We found that this difference arises due to the contribution of N-terminal partners, MTAP (exons 1-7) and MS4A6A (exons 1-6), to respective fusion dimerization that is unaffected by PLX8394 (Figure 2C-D, lanes 3,7). Similar role of N-terminal partner accounts for differential response of CRAF- fusions to PLX839410. Furthermore, we observed that Trunc-MTAP (exons 1-7) competitively substituted MTAP-BRAF homo-dimerization in a dose-dependent manner, suggesting preferential and potent protein interactions mediated by the N-terminal partner in these histiocytic-specific BRAF-fusions (Figure 2E).
To target dimerization-dependent oncogenicity of MTAP- and MS4A6A-BRAF via a different mechanism, we tested LY3009120, a pan-RAF dimer inhibitor that binds and stabilizes the BRAF dimer in an inactive conformation16. LY3009120 showed robust suppression of both fusion-mediated signaling and colony transformation (Figure 2B) while stabilizing the MTAP- or MS4A6A-BRAF in inactive conformation (Figure 2C-D, respectively, lanes 9-11). We also tested the effect of FDA-approved MEK inhibitors (MEKi)17, selumetinib and trametinib. We observed dose-dependent decrease in phospho-ERK and growth with trametinib (Figure 2F) and selumetinib (Supplemental Figure S2) in both BRAF-fusion models suggesting downstream MEKi as a therapeutic alternative to RAFi.