Abstract
Meningiomas are the most common primary tumors of the central nervous system and are typically treated with surgery or radiation, as targeted therapies remain limited. Despite extensive study, seventeen percent of meningiomas lack known genetic drivers. Our analysis of meningiomas without driver mutations or major chromosomal alterations identifies a subset with recurrent genomic rearrangements involving the FOS and FOSB genes. These tumors exhibit elevated FOS/FOSB protein levels and retain meningothelial lineage. Here we show that FOS/FOSB fusion-positive meningiomas represent a distinct molecular subgroup, defined by unique gene expression patterns, including activation of AP-1 target genes and signatures resembling preadipocyte-like and mast cell–associated profiles. Clinically, these tumors display low-grade behavior and DNA methylation profiles consistent with benign subtypes. Our findings identify a meningioma subgroup with distinct genetic, transcriptomic, and clinical features, expanding the molecular classification of meningiomas and opening new avenues for targeted treatment strategies.
Similar content being viewed by others
Introduction
Meningiomas are the most common type of primary central nervous system (CNS) tumors, characterized by diverse genomic, histological, and clinical subtypes and predominantly treated through surgical resection and radiation therapy. Despite their prevalence, the lack of approved targeted therapies underscores the necessity for a deeper understanding of their molecular mechanisms.
Meningiomas forming along the dura, which is the outermost layer of the meninges, display markedly high inter-tumoral heterogeneity. This heterogeneity spans molecular, histologic, and clinical domains, posing challenges to the development of targeted therapies. According to WHO CNS5 (2021) guidelines, meningiomas are classified into 15 distinct histological subtypes across three malignancy grades1. These tumors are further categorized into over 14 subtypes based on their anatomical location2,3.
To date, genomic studies have uncovered driver events in approximately 83% of sporadic meningiomas, delineating mutually exclusive molecular subgroups characterized by somatic mutations, with distinctive genomic, transcriptomic, and clinical features3,4. Nearly half of all meningiomas have somatic bi-allelic loss of NF2, often arising through a combination of copy number variations (CNV), single nucleotide variations (SNV), or short insertions/deletions (INDELs)5,6, sometimes co-occuring with mutations in SMARCB1. Another major subgroup, representing approximately 20% of meningiomas, is driven by mutations in TRAF7. These tumors present with or without co-mutations in PI3K pathway genes such as AKT1 or PIK3CA, or in KLF4, with these mutations occurring in a mutually exclusive manner4,7,8. Meningiomas harboring alterations in the Hedgehog (HH) pathway, constituting 3–5% of all cases, are typically associated with somatic mutations in SMO, SUFU, and PRKAR1A9,10. We have recently identified structural events on chromosomes 2q35 and 7q36 leading to enhancer hijacking and activation of IHH and SHH, ligands activating the HH pathway, further characterizing HH-driven meningiomas11. We and others have also identified rare subtypes driven by mutations in POLR2A, SMARCE1, VHL, etc., seen in approximately 1% of all meningiomas12,13. BAP1 alterations and chromosome 3p-loss are also reported in aggressive meningiomas11,14. These findings underscore the remarkable heterogeneity of meningiomas, as these mutually exclusive genomic subgroups fail to converge on a single pathway, illustrating the complexity of meningioma pathogenesis. In addition to somatic mutations, CNVs play a pivotal role in both driving tumorigenesis and impacting clinical behavior in meningiomas. For instance, deletions of chromosome 22 are a well-established hallmark of meningiomas, while CNVs involving chromosomes 1p, 6p, and 14q are associated with more aggressive tumor behavior and poorer prognosis15,16,17. This intricate interplay between genomic alterations and clinical features underscores the need for continued research to unravel the biological diversity of meningiomas and to refine prognostic and therapeutic strategies.
Methylation and transcriptome-based studies have focused on stratifying meningiomas into clusters that correlate with clinical outcomes. These studies often converge on three to six main clusters, such as Baylor transcriptomic classifier (A-C)18, UCSF methylation classifier (Hypermitotic, immune-enriched and merlin-intact)19, Toronto methylation groups (MG1-4)20, and DKFZ methylation classification (MCben1-3, MCintA-B, MCmal)21. Although these methylomic and transcriptomic classifiers have proven valuable for prognostication and clinical decision making20,22, they remain limited in addressing the full genomic diversity in meningiomas and identifying genomic drivers.
The extensive intertumoral heterogeneity across molecular, histological, and clinical domains highlights the necessity of in-depth characterization for each meningioma. Comprehensive profiling is crucial to understanding the unique biology and clinical trajectory of each tumor, paving the way for the development of targeted treatment modalities.
Despite significant advances in understanding the molecular diversity of meningiomas, approximately 17% of sporadic cases still lack identifiable driver mutations, leaving a gap in our comprehensive characterization of this prevalent tumor type. This study aims to address this gap in genomic characterization of meningiomas by investigating structural variations in a cohort of meningiomas lacking known genomic drivers. By characterizing these previously unexplored alterations, correlating with transcriptomic, histologic, and clinical features, we seek to uncover subgroups that may contribute to meningioma pathogenesis and potentially reveal new therapeutic targets.
We present fusion events occurring within a subgroup of meningiomas, affecting the Fos proto-oncogene, activator protein 1 (AP-1) transcription factor subunit (FOS) gene and its paralogue, FosB Proto-Oncogene, AP-1 Transcription Factor Subunit (FOSB) gene. FOS was first characterized as the cellular counterpart of two viral v-FOS oncogenes, identified to lead to osteosarcoma in rats and mice23. FOS and FOSB, along with other members of the FOS family genes (FOSL1, FOSL2), constitute the AP-1 transcription factor, which regulates cellular proliferation, differentiation, and survival.
Gene fusions involving the FOS and FOSB have been previously identified in tumor types with very distinct histological characteristics, such as bone and soft tissue tumors, including epithelioid hemangioma, pseudomyogenic hemangioendothelioma, osteoid osteoma, osteoblastoma, cementoblastoma, and proliferative myositis and fasciitis24,25,26,27,28. While these gene fusions have been recognized as driver events in these tumors, their detailed transcriptomic characterization and potential as therapeutic targets remain largely unexplored. Though FOSB fusions have been listed in a small number of meningioma cases previously, including by our group11,29, a systematic analysis of the involvement of FOS and FOSB in meningioma tumorigenesis has not yet been performed.
Structural variations (SVs), particularly gene fusions, are known to drive transcriptional dysregulation through rearrangement of regulatory elements or alterations in copy number across a wide range of tumor types. The same SVs, resulting in identical gene fusions or fusions between a common gene and various partners, are often observed across different cancer types. However, the downstream functional effects of these fusions can vary significantly by tumor type, highlighting the need for comprehensive, tumor-specific transcriptomic characterization of each fusion event30.
In this work, we identify a subtype of meningiomas characterized by fusions impacting FOS/FOSB genes and activating the targets of the AP-1 complex. Ultimately, our findings hold the promise of improved characterization and guiding the development of precision therapies tailored to the molecular profiles of individual meningioma subtypes, thereby enhancing patient outcomes and laying the groundwork for personalized management strategies for this prevalent brain tumor.
Results
We investigated structural variation (SV) events in 105 meningioma cases from Yale, which had no alterations on the previously identified meningioma driver genes from whole exome sequencing (WES) data. We define meningiomas drivers as follows based on previously published work3 :(1) NF2 mutations with or without Chr22q loss, (2) TRAF7 mutations with or without KLF4 (p.K409Q) or with or without PI3K pathway gene mutations (AKT1, PI3KCA, PI3KR1, etc.) 3) POLR2A:p.Q403K/L438_H439del mutations and (4) HH pathway alterations including recurrent mutations in SMO or PRKAR1A, damaging mutations in SUFU, or structural variations impacting chr2q34 (IHH) or chr7q36.3 (SHH)11. While the majority of the 105 “mutation unknown” meningiomas were Grade 1, 20% were classified as Grade 2 or 3 (n = 21). 12 of these cases had no CNVs or had CNVs impacting less than 1% of the genome (Supplementary Data 1). The remaining nine cases with Grade 2 or 3 histology had an average of 16% genome alteration by CNV/LOH events. Only two of these nine Grade 2/3 cases harbored chr1p deletion overlapping with loss-of-heterozygosity, with no chr22q deletion. There were two cases with chr3p deletion, with no chr1p/chr22q deletion events. Overall, this cohort of 105 meningiomas displayed genomic characteristics similar to the previously reported profile of meningiomas with an average tumor mutation burden (TMB) of 1.38 mutations per megabase and a mean percentage of genome alteration by CNV/LOH events of 6.07%. The anatomical distribution of the cases in this cohort was also consistent with prior reports, with 48.6% located in convexity and 46.7% located at the skull base2 (Supplementary Data 1).
Recurrent FOS and FOSB rearrangements are identified in meningiomas with WES
Detailed investigation of the SV events of the WES data in the cohort of meningiomas with no known driver events in previously reported genes, revealed a recurrent LMNA-FOSB fusion in four of the cases (MN-52391, MN-70002, MN-70003, and MN-62278) and a FOS-SLC4A4 fusion in a single case (MN-70001) (Fig. 1, Supplementary Data 2). Notably, all identified cases exhibited no alterations in known meningioma driver genes and had stable genomes with no other SV or CNV events.
a–d (left panels), Magnetic resonance imaging (MRI) and circos plots depicting the fusion events for patients MN-70001, MN-52391, MN-70002, and MN-70003 (a-d respectively). The MRI panels demonstrate axial T1 post-contrasted images with the tumor depicted by the red asterisk (left); corresponding fluid attenuated inversion recovery (FLAIR) sequences (center) reveal minimal, if any, edema associated with all tumors. The right image is another plane of T1 post-contrasted sequence. a–d (right panels), Circos plots displaying FOS-SLC4A4 a (right panel), and LMNA-FOSB fusions b–d (right panels). Circos plot outer tracks represent the chromosomes, gene labels, and points indicate somatic SNV/INDELs identified, orange: loss of function, blue: missense. Links are the breakpoints identified for the fusion events.
Proteins encoded by FOS and FOSB genes harbor a conserved basic region Leucine Zipper (bZIP) domain, which is essential for their dimerization31,32. The breakpoint identified in MN-70001 (FOS-SLC4A4 fusion) occurred in exon-4 of the FOS gene, similar to breakpoints identified in other tumor types25,26,27. This breakpoint consequently resulted in a premature stop codon, losing the regulatory component for degradation yet preserving the bZIP domain (Fig. 2a, b). The genomic breakpoints in all LMNA-FOSB fusion events were clustered in intron-1 and intron-2 of FOSB and resulted in the promoter switch between LMNA and FOSB, preserving the bZIP domain (Fig. 2c, d). Breakpoints in both FOS and FOSB were similar to the previous reports in other tumor types31. Fusion events were confirmed with PCR using Sanger sequencing for the cases with available DNA, i.e., MN-70001, MN-52391, MN-70002 and MN-70003 (Fig. 2b, d, Supplementary Fig. 1).
a Breakpoints on FOS gene identified in meningiomas (MN-70001) and other tumor types (A, B, C, D and E)24. b (top), Fusion event representation between FOS-SLC4A4 forming a premature stop codon b (bottom), trace file image showing PCR (sanger sequencing) confirmation. c Breakpoints on FOSB gene identified in meningiomas (MN-62278, MN-52391, MN-70002, MN-70003) and other tumor types (F, G, H, I). d (top), fusion event representation between LMNA-FOSB leads to a promoter switch. d (bottom), trace file image showing PCR (Sanger sequencing) confirmation.
Meningiomas with FOS/FOSB fusion events display low-grade clinical features
In all cases with FOS/FOSB fusions, surgical impressions indicated that the tumors originated from the dura rather than being centered within the bone. While the cohort of meningiomas harboring FOS/FOSB gene fusions with detailed clinical data from Yale is undeniably small, some clinical patterns emerge (Table 1). Notably, all tumors had involvement of the associated bone and had intimate involvement with the brain parenchyma, which was noted during surgery. Despite this, all meningiomas were histologically classified as WHO Grade 1. Interestingly, all tumors, regardless of size and mass effect, were associated with minimal, if any, brain edema (Fig. 1). All patients underwent maximal safe surgical resection without evidence of recurrent disease at the time of last follow-up (29.75 months).
Meningiomas with FOS/FOSB fusion events display distinctive mRNA and immunohistochemistry features
To investigate the histological features of FOS/FOSB fusion cases, we performed immunohistochemical (IHC) studies including Epithelial membrane antigen (EMA) and somatostatin receptor 2A (SSTR2A), a sensitive marker of meningothelial differentiation. Indeed, all tumors showed widespread expression of both EMA and SSTR2A, consistent with meningothelial differentiation. We also performed IHC studies for markers of vascular differentiation (ERG and CD31). Both ERG and CD31 staining were negative in the tumor cells, but labeled endothelial nuclei of admixed blood vessels as an appropriate internal control (Supplementary Fig. 3). Overall, these findings indicate that these tumors are morphologically meningiomas from a diagnostic standpoint, and their histologic appearance is typical of meningothelial and microcystic/angiomatous subtypes.
Real-time quantitative PCR was performed on the FOS/FOSB fusion specimens with available RNA to show the increased expression for the preserved exons of FOS and FOSB (Supplementary Table 1, Fig. 3a). To further investigate the implications of identified genomic events, and their protein level impact, we performed immunohistochemistry (IHC) analysis with c-FOS and delta-FOSB antibodies. The staining patterns and intensity levels demonstrated that cases with FOS/FOSB fusion had increased expression of FOS and FOSB as compared to the meningiomas that do not harbor these fusions (n = 3) (Fig. 3b–g). Quantitative IHC analysis further demonstrated elevated expression of c-FOS and delta-FOSB in respective fusion-positive meningiomas. Specifically, the ratio of delta-FOSB positively stained nuclei exhibited a statistically significant difference in distribution between the FOSB fusion-positive group and the control group (Fig. 3h). Statistical analysis was not applied to the FOS fusion-positive case due to the single sample size; however, a positive staining quantification of 69.26% for the FOS fusion-positive case versus 1.9% positivity for fusion-negative cases suggests elevated FOS expression levels (Supplementary Fig. 4).
a Quantitative RT–PCR results for FOS and FOSB fusion positive cases, MN-70001 and MN-70002, respectively, and control cases (NF2-loss meningiomas) for corresponding exons (Supplementary Table 1). b–g Immunohistochemistry stains of meningiomas using delta-FOSB and Anti-c-FOS antibodies. b NF2-driven meningioma stained with the antibody Anti-c-FOS shows absent nuclear staining. c, NF2-driven meningioma labeled with antibody delta-FOSB showed weak cytoplasmic but absent nuclear staining in the negative control. d Patchy positive nuclear c-FOS staining (~60% nuclei) in the FOS-SLC4A4 fusion tumor. e-g Strong diffuse nuclear delta-FOSB staining in LMNA-FOSB fusion tumors. h Violin plot showing significantly higher percentages of positively stained nuclei counts between the FOSB fusion-positive case group and the fusion-negative control group. Values from each tumor are shown as individual points. An independent t-test was performed (t = 5.68, p = 0.0051) (Control n = 3, n = 3) (Supplementary Fig. 3). Boxplots represent the interquartile (IQR range, center line representing median, and whiskers extending to the lowest or highest value with 1.5 * IQR range; individual values are shown with points. Scale bars (bottom left) represent 100 µm.
Meningiomas with FOS/FOSB fusion events define a new meningioma subgroup with a distinct genomic, transcriptomic profile
We performed RNA-sequencing on two samples with available RNA (MN-70001, MN-70002) and four cases with NF2-loss as a control group to investigate the transcriptomic profile of FOS/FOSB fusion events. To extend the findings from the Yale cohort, we utilized a recent study presenting a reference transcriptomic map of meningiomas using a large cohort33. We integrated 1,226 adult meningiomas from published studies across eight different centers20,33,34,35,36,37,38,39, including 42 previously published Yale samples11,12 (Supplementary Data 3). All 1232 meningiomas, including the six from this study, underwent integrative analysis using batch correction to evaluate their transcriptional characteristics. Dimensionality reduction on 1232 samples followed by density-based spatial clustering (DBSCAN) revealed 12 distinct clusters (Fig. 4a, b) without a clear institutional batch effect. Detailed genomic data available on Yale samples (n = 48) and genomic driver information from other samples in publications enabled characterization of these clusters based on underlying and previously reported genomic events: NF2-loss (clusters-2-4), HH-driven (cluster-3), POLR2A (cluster-8), TRAF7/PI3K (cluster-1), TRAF7/KLF4 (cluster-7), chr3p deleted (cluster-6) (Fig. 4c). Samples with available matching methylation classifiers from corresponding publications, chr1p/chr22q deletion and structural variation/fusion status calculated from RNA-seq data together with clinical features such as age, gender and WHO Grade were also annotated to further characterize these clusters (Fig. 4, Supplementary Fig. 5-6). Notably, all Yale FOS/FOSB fusion cases were clustered in cluster-11, which is comprised of nine meningiomas, eight of which were WHO-Grade-1. This cluster included cases with matching methylation cluster annotations classified as “Merlin-intact” and “A-NF2-intact-benign”, for UCSF and Baylor classifiers, respectively18,19. CNV analysis revealed that none of the samples in this cluster exhibited chr1p/chr22q deletions. However, structural variation analysis identified five additional meningiomas with LMNA-FOSB fusion, all clustered within cluster-11 and originating from four different centers, FHCC (n = 2), Baylor (n = 1) and Palacky (n = 2). This brought the total number of FOS/FOSB fusion cases to seven, including two cases from this study, all of which were Grade 1 tumors (Supplementary Data 4, 5). Interestingly, one case in this cluster, with no FOS/FOSB fusion harbored an inter-chromosomal translocation involving NR4A1 (Fig. 4d). NR4A1 is reported to regulate the expression of immediate early expression genes, such as FOS and FOSB, in differentiated cancers with a favorable prognosis40. To investigate possible associations of these genes, we analyzed the co-expression profiles of NR4A1 and FOS, FOSB using the BrainSpan data for the developmental transcriptome map of human brain from 42 specimens41. Both FOS and FOSB showed strong positive correlation with NR4A1 (R = 0.82 and R = 0.73, respectively, p < 2.2e−16). To further explore this co-expression pattern across different tissue and cell types in human, we utilized a human and mouse gene co-expression database, COXPRESdb v842. This analysis confirmed that FOSB and FOS were among the top co-expressed genes with NR4A1, with FOSB ranking third, after NR4A3 and NR4A2 (z-score=6.33, Supplementary Table 2, Supplementary Fig. 7). The identification of a new structural event involving NR4A1, along with its similar transcriptomic profile to FOS/FOSB fusion cases, provides further evidence that this subgroup of meningiomas is distinctly characterized by genomic events primarily targeting the FOS/FOSB transcription factors. The only sample with no FOS/FOSB fusion was also the only WHO Grade 2 tumor in cluster-11. Further whole genome/exome sequencing is warranted to reveal the genomic driver of this specific tumor that results in similar transcriptomic profile as the cases with FOS/FOSB fusion.
a Batch corrected cohort of 1232 cases from five centers. b 12 clusters identified using DBSCAN. c Clusters are overlapping with genomic drivers (for the cases with available information in corresponding publications) and revealing FOS/FOSB samples as s distinct cluster d Cluster-11 is composed of six LMNA-FOSB fusion cases, one FOS-SLC4A4 and one NR4A1 translocation, depicted in circos plot with breakpoints. e WHO Grade and f Methylation clusters, for cases with available information, g chr1p and h chr22q deletion status.
Given that meningiomas arise from the dura, the outermost layer of the meninges, we investigated the cellular composition of our meningioma cohort to assess whether the distribution of dural cell populations was preserved in the FOS/FOSB fusion-positive cases compared to the rest of the cohort. To this end, we performed deconvolution analysis using single-cell reference data from the developing human spine, comprising 66 annotated cell types, including dural populations43,44 (Supplementary Fig. 8). Notably, dura and endothelium, were the most abundant cell types across the whole meningiomas cohort, including the FOS/FOSB altered meningiomas. While there were variations across the cohort, possibly due to tumor purity, histological grade, subtype and microenvironment features, the dural cell population was predicted to account for an average of 18.8% in the fusion-positive tumors, closely aligning with 17.8% in the broader meningioma cohort (Fig. 5a, Supplementary Fig. 8). Together, the comparable enrichment of dural cell types in the fusion-positive tumors further supports their meningeal (dural) origin.
a Predicted dura and endothelium cell populations of FOS/FOSB fusion cases (n = 7), compared to the rest of the meningiomas (n = 1165). Boxplots represent the interquartile (IQR range, center line representing the median, and whiskers extending to the lowest or highest value with 1.5 * IQR range; values for outliers, as defined by lying outside the whiskers, are shown with points. b Volcano plot of all DEGs (|log2FC | >2, p.adj <0.05) for cluster-11 vs all other meningiomas in the cohort using a two-tailed Wald-test adjusted using Benjamini-Hochberg procedure (Control n = 1223, Case n = 9). c Manhattan plot showing enrichment analysis results for DEG (Control n = 1163, n = 9) Fisher's exact test was performed, adjusted by the Benjamini-Hochberg procedure (upregulated, log2FC > 2, p.adj <0.05) for gene sets using GO:BP, KEGG pathways, and transcription target databases. d Significantly enriched GO:BP terms (n = 171) in DEG set (log2FC > 0, p.adj <0.05) are clustered based on their semantic similarity and keyword-based enrichment results (from simplifyEnrichment R package using Sim_XGraSM_2013 algorithm). e Cell type enrichment analysis results predicting increased mast cells. Beta‑distribution–based BH adjusted p‑value was used, calibrated to random mixtures of gene expression data. A two-sided Wilcoxon Rank Sum test was used for group-wise comparisons of results. (p = 0.048) (Control n = 1163, Case n = 9) and f preadipocyte cell population in FOS/FOSB fusion cases. Beta‑distribution–based BH adjusted p‑value was used, calibrated to random mixtures of gene expression data. A two-sided Wilcoxon Rank Sum test was used for group-wise comparisons of results. (p = 3.8e−05) (Control n = 1163, n = 9). Boxplots represent the interquartile (IQR range, the center line representing the median, and whiskers extending to the lowest or highest value with 1.5 * IQR range; values for outliers, as defined by lying outside the whiskers, are shown with points.
To further characterize the transcriptomic features of FOS/FOSB cluster, we performed differential gene expression (DGE) analysis based on the DBSCAN clusters. DGE analysis between FOS/FOSB cluster, i.e., cluster-11 vs. the rest of the meningiomas identified 1276 up-regulated and 3150 down-regulated genes (Fig. 5b, Supplementary Data 6).
To gain a deeper understanding of the downstream mechanisms associated with FOS/FOSB fusions in meningiomas, we conducted a comprehensive analysis of differentially expressed genes (DEG) within the FOS/FOSB cluster. Notably, analysis of all the statistically significantly upregulated genes in FOS/FOSB cluster identified enrichment of targets of AP-1 member transcription factors, c-JUN, FOSL1 and FOS (Fig. 5c). Pathway enrichment analysis of all DEGs indicated involvement in key cellular pathways, including “neuroactive ligand-receptor activity”, “cAMP signaling pathway”, “ECM-receptor interaction” and “calcium signaling pathway”. Additionally, the analysis highlighted biological processes such as “positive regulation of cell population proliferation” (GO:0008284) and “regulation of wound healing” (GO:0061041) (Fig. 5c).
Further semantic similarity analysis of the significantly enriched Gene Ontology (GO) biological processes identified clusters of related processes and associated keywords, highlighting ‘wound healing’ as a significant keyword (Fig. 5d). These computational predictions pointing to possible involvement of immune response prompted us to characterize the immune cell type abundance and tumor purity of FOS/FOSB cases using the transcriptomic signatures. Tumor purity, as estimated by the stromal and immune cell population characteristics of the transcriptomic data by ESTIMATE45, showed no significant difference between FOS/FOSB cluster and rest of the tumors (Supplementary Fig. 9). However, analysis of immune cell type distribution using two methods with distinct reference datasets, (i) deconvolution of 22 immune cell type expression signature matrix with CIBERSORTx46, and (ii) enrichment of 64 immune and stromal cell signature using xCell47, revealed that the FOS/FOSB cluster was notably enriched for mast cell population signatures compared to the rest of the meningiomas (Fig. 5e, Supplementary Fig. 9-10). Cell type enrichment analysis using the stromal cell signatures also predicted a statistically significant enrichment of preadipocytes in the FOS/FOSB cluster compared to other clusters (Fig. 5f). Although the cranial dura and the meninges more broadly, lack mature adipocytes and preadipocytes, they harbor fibroblasts, which, like osteoblasts, originate from a common mesenchymal stem cell lineage. Thus, the deconvolution results likely reflect a mesenchymal cell population enriched in preadipocyte-like transcriptional features.
To evaluate the prediction of increased mast cell involvement in FOS/FOSB fusion cases, we performed mast cell tryptase (MCT) IHC staining on four FOS/FOSB fusion-positive cases and nine fusion-negative controls, (NF2 altered meningiomas; Fig. 6a). The cases with FOS/FOSB fusion exhibited elevated staining for MCT compared to the fusion-negative control group (p-value = 0.0069, median= 1.01%, range: 0.63–44.84%, vs. median = 0.06%, range: 0.0049– 0.45%) (Fig. 6b).
a High-power and stitched MCT staining images of tumor tissue centers using nine 11.1x magnification with (left, middle panel) and without (right panel) FOS/FOSB fusion. b Violin plot depicting the statistically significant difference of positively staining cells between cases with (n = 4, FOS fusion positive meningiomas) and without FOS/FOSB fusion (n = 9, NF2-altered meningiomas). A two-sided Mann-Whitney U test was performed (p = 0.0069). Boxplots represent the interquartile (IQR range, center line representing median and whiskers extending to the lowest or highest value with 1.5 * IQR range. Scale bars (bottom left) represent 1000 µm.
Discussion
In this study, we report a molecular subgroup of meningiomas characterized by FOS and FOSB gene fusions, exhibiting distinct genomic and transcriptomic characteristics. While FOS and FOSB gene fusions have been documented in other tumor types, including epithelioid hemangioma, pseudomyogenic hemangioendothelioma, osteoid osteoma, and osteoblastoma24,25,26,27,28, their downstream mechanisms and transcriptomic profiles have not been thoroughly investigated. Moreover, the prevalence of FOS/FOSB fusions in meningiomas, along with the distinct molecular, clinical, and histological profile of cases with these fusions, and their presentation as a distinct meningioma subgroup, had not been explored until now.
Proteins encoded by the FOS and FOSB genes harbor a conserved basic region Leucine Zipper (bZIP) domain, which is essential for their dimerization. Strikingly, our genomic analysis revealed that while the promoter or regulatory elements controlling the transcription or degradation of the protein were altered, the bZIP domain was preserved, underscoring the mechanistic importance of dimerization in FOS/FOSB activation. Indeed, the truncated form of FOSB is reported to act as an auto-regulated active transcription factor and upregulate PDGFRA and FLT1 expression in a Pseudomyogenic hemangioendothelioma (PH) study48. The use of targeted treatment with telatinib, a multi-tyrosine kinase inhibitor, displayed a complete response in a PH case, with in-vitro studies demonstrating increased apoptosis through VEGFR downstream signaling blocking48. These findings suggest a potential targeted treatment approach for a subgroup of meningiomas with FOS/FOSB fusion, for which no approved targeted therapies are currently available.
Our comprehensive transcriptomic analysis, integrating 1232 meningiomas revealed that meningiomas with FOS and FOSB fusions form a subgroup characterized by distinct genomic and transcriptomic features. This group of meningiomas, clustering separately from the rest of the meningiomas, distinctively displayed increased activation of AP-1 complex member target genes compared to the rest of meningiomas with other known genomic drivers. Further computational analysis of the transcriptomic profile using immune and stromal cell population characteristics predicted that meningiomas with FOS/FOSB fusions had a higher abundance of mast cells compared to other meningioma subtypes. Interestingly, meningiomas with FOS/FOSB fusions reported in this study were all Grade-1 and had no recurrence at the time of reporting. Overall, the distinctive immunogenic feature, characterized by increased mast cell populations, may explain the non-aggressive nature of the tumors with FOS/FOSB fusion and could also present an opportunity for immunotherapy in cases of tumor progression or when surgery is not feasible. Additionally, the computational analysis predicted that these tumors exhibit an expansion of a less differentiated mesenchymal cell population characterized by transcriptional signatures resembling preadipocytes—likely reflecting shifts in mesenchymal stem cell fate. Mesenchymal stem cells are multipotent progenitors capable of differentiating into specialized cell types, including osteoblasts, chondrocytes, adipocytes, and fibroblasts—the predominant cell type in the dura49,50,51,52. These findings suggest a potential reprogramming mechanism through which FOS/FOSB fusion events may alter mesenchymal lineage commitment, ultimately influencing tumor phenotype and behavior. Supporting this, increased expression of FOSB has been shown to enhance osteoblast formation while suppressing adipogenesis in mouse models, indicating a shift in mesenchymal stem cell fate53. Importantly, this mechanism may extend beyond meningiomas, offering a unifying model for tumorigenesis in a broader spectrum of bone and soft tissue neoplasms harboring FOS/FOSB fusions. Further investigation is warranted to validate this mechanism and to explore whether dysregulation of mesenchymal stem cell fate decisions represents a shared pathogenic pathway across diverse tumor types.
These findings enhance our understanding of heterogeneous molecular profile of meningiomas and highlight potential targets for therapeutic interventions in a subtype. Continued efforts in genomic characterization are crucial for the development of targeted therapies that could significantly improve the prognosis and quality of life for patients with meningiomas.
Methods
This study complies with all relevant ethical regulations and was approved by the Yale Institutional Review Board (HIC Protocol #9406007680). Informed consent was obtained from patients under institutional review board approval. Sex and/or gender were not considered in the study design. The sex of participants was determined based on self-report.
We have performed comprehensive genomic analysis on the WES on 105 meningioma samples lacking known driver alterations, currently predicted be constitute 17% of the meningiomas.
Cohort selection
Tumor samples were obtained during surgical resection for consenting patients and processed for genomic analysis. Meningothelial lineage was confirmed by positive staining of Epithelial membrane antigen (EMA) and somatostatin receptor 2 A (SSTR2A) IHC staining. Markers of vascular differentiation (ERG and CD31) were negative in the tumor cells but labeled endothelial nuclei of admixed blood vessels (as an appropriate internal control). Detailed clinical and pathological data, including tumor grade and location, were prospectively collected and retrospectively analyzed from electronic medical records. A total of 105 meningioma cases were identified based on the genomic profile lacking previously identified genomic drivers. Cases with multiple tumors and cases with prior treatment or recurrence were excluded from this study. All tumors were classified according to the 5th edition of the WHO CNS classification system54 and were reviewed by a board-certified neuropathologist (Supplementary Fig. 3).
Whole Exome Sequencing (WES)
Sequencing of the libraries was carried out by Yale Center for Genome Analysis using the standard protocol. Coding sequences were captured using the xGEN Exome Research Panel v1.0 kit (Integrated DNA Technologies) with additional spike-in regions and sequenced using the Illumina NovaSeq6000 sequencing platform, resulting in 2x100bp reads. Mean coverage of 115.2x and 235.8x was achieved for blood and tumor tissue, respectively. Reads were aligned to the GRCh37 human reference genome using BWA-MEM (version 0.7.10). The aligned reads were then processed with Picard (version 1.18) to mark polymerase chain reaction duplicates. Next, the Genome Analysis Toolkit (GATK; version 3.2-2, Broad Institute) was used for multisequence local realignment, base quality recalibration. Somatic single-nucleotide variations (SNV) and insertions or deletions (INDEL) were detected using GATK Mutect2 (version 4.2.0.0). The raw somatic SNVs were filtered using GATK FilterMutectCalls followed by annotation using ANNOVAR (version 2015-12-14) and VEP (version 83). Somatic variants with Minor Allele Frequency <= 5% and with frequency > 1% in control databases such as ExAC, 1000Genomes, and NHLBI were filtered out. Copy number variations (CNV) were determined using Genome Analysis ToolKit (GATK) version 4.2.0.0, mostly adhering to the “GATK Best Practices” guidelines the exception of two parameters for denoising, i.e., smoothing iterations per fit set to 1 from default value of 0, and number of changepoints penalty factor set to 5 from the default setting of 1.
Structural variant analysis from WES data
We used Manta (version 1.6.0) to identify breakpoints using default parameters, with the --exome flag55. Initial filtering was performed on Manta VCF outputs, specifically removing events with the IMPRECISE flag, which means the exact breakpoint location is not resolved. Events that failed MANTA internal filtering were excluded. After initial filtering, Manta VCF outputs were then converted to ANNOVAR format using convert2annovar.pl script and then annotated using ANNOVAR (version 2019-10-24). Both gene-based and regional annotations were performed, and events that overlap with segmental duplication regions were removed. After annotation, breakpoint events that are not deletions or duplications were matched with their mates, and orphan events (events with missing pairs) were removed. Merged events were further filtered by an event size threshold of 5000 bases.
Confirmation of fusion events
Four cases with reported fusion events identified with WES were validated with PCR followed by Sanger sequencing (Fig. 2, Supplementary Fig. 1). MN-62778 did not have enough tissue for confirmation. DNA samples were isolated from formalin-fixed paraffin-embedded (FFPE) patient tumor tissues using Qiagen QIAamp DNA FFPE Advanced kit, and DNA repair was performed by NEBNext FFPE DNA repair V2 kit. Primers were designed using Primer3, and the specificity of each primer was confirmed by ENSEMBL’s BLAST tool (Supplementary Table 1). PCR was performed on 25 ng of DNA from each patient’s tumor samples with 10 µM of forward and reverse primers and KAPA Taq ReadyMix (KAPA, Wilmington, MA) according to the manufacturer’s protocol. The PCR products of patients’ samples and negative control (human DNA without expected fusions) were analyzed on a 1.8% agarose gel and sequenced at the W. M. Keck Center. The sequence data were analyzed using Sequencher 4.9 (Gene Codes) to confirm the fusion sites.
Real-time Quantitative PCR
Total mRNA was extracted from tumors per the vendor's instructions (Qiagen). cDNA generation was done using Superscript II Reverse Transcriptase from Invitrogen (#18064014). All Primers are designed using Primer3 (Supplementary Table 1). RT-qPCR experiments were performed for the assessment of differential expression of FOS and FOSB in the tumor samples. iTaq™ Universal SYBR® Green Supermix (# 1708890) was used for qPCR reaction, and Bio-Rad CFX384 Real Time System was used for cDNA amplification and quantification. Samples were run in 384-well plates in triplicate as multiplexed reactions with a normalizing internal control (actin). All PCR assays were performed at least three times in three independent experiments. The relative fold change in gene expression was calculated as 2-ΔΔCT.
Immunohistochemistry (IHC)
IHC staining for c-FOS, delta-FOSB, mast cell tryptase (TPSAB1), CD31, somatostatin receptor 2 A (SSTR2A), epithelial membrane antigen (EMA), and erythroblast transformation-specific regulated gene-1 (ERG) proteins was performed manually using optimized protocols. Briefly, formalin-fixed paraffin-embedded (FFPE) tumor tissue sections were cut to a thickness of approximately 5 µm. The sections were dewaxed, rehydrated, and washed, followed by quenching of endogenous hydrogen peroxide (H2O2) with 3% H2O2for 5 minutes. Polyclonal Anti-c-FOS (ABE457; Sigma-Aldrich) staining was performed at a 1:2000 dilution. Recombinant delta-FOSB (D3S8R; Cell Signaling Technology) staining was performed at a 1:150 dilution. Monoclonal mast cell tryptase (MCT) (M7052; Dako) staining was performed at a 1:100 dilution. Monoclonal CD31 (CMC131; Biocare Medical) staining was performed at a 1:75 dilution. Monoclonal SSTR2A (UMB1; Abcam) staining was performed at a 1:1000 dilution. Monoclonal EMA (MS741; Thermo Fisher Scientific) staining was performed at a 1:400 dilution. Monoclonal ERG (CM421C; Biocore Medical) staining was performed at a 1:100 dilution. Heat-mediated antigen retrieval method was used with tris/EDTA buffer pH ~9.0 for c-FOS, delta-FOSB, ERG, citrate buffer pH ~6.0 for MCT, CD31, SSTR2A, and EMA staining. The secondary antibody used for all primary antibodies was the MACH 2 Rabbit HRP-Polymer (RHRP520; Biocare Medical) with a 30-minute incubation period. Chromogen development was performed using Betazoid DAB buffer (DS900; Biocare Medical) with a 5-minute incubation period. The slides were then counterstained with hematoxylin Gill No. 3 (Avantik; RS4021). H&E staining was performed with deparaffinization in a standard fashion using hematoxylin Gill No. 3 (Avantik; RS4021) and Eosin Y (Avantik; RS4017). Board-certified pathologists assessed the staining patterns and intensity levels blinded to the study design.
Quantitative immunohistochemistry (IHC) analysis via QuPath
Quantification of c-FOS, delta-FOSB, and MCT immunohistochemical staining was performed using QuPath29 digital pathology software (Version 0.5.1). For segmentation and analysis, QuPath’s built-in functions were employed. Briefly, the “Stain Vector Estimation” followed by “Cell detection” was carried out using QuPath’s default “Optical Density Sum (ODS)” image segmentation. Positive staining cells were identified with “Positive Cell Detection” function. Detection thresholds were set at 0.05 for c-FOS, 0.4 for delta-FOSB, and 0.2 for MCT (200x magnification images) (0.16 on stitched images), adjusted as needed based on background staining intensity for each marker. For statistical analysis, an independent t-test was performed to compare staining quantifications between the FOSB fusion-positive case group and a fusion-negative control group. All analyses and results were reviewed and validated by a board-certified neuropathologist.
RNA-sequencing
RNA isolation was performed on fresh frozen tissue. The RNA Integrity Number (RIN) for the six samples was calculated by Bioanalyzer 2100 (Agilent; RRID: SCR_019389). Each specimen underwent ribosomal RNA depletion library prep before sequencing on Ithe llumina NovaSeq platform using paired-end, 100 base pair reads.
Analysis of RNA-Seq data
Raw paired-end RNA-seq reads were pre-processed using TrimGalore v0.6.7 (https://github.com/FelixKrueger/TrimGalore). Filtered reads were aligned to the human reference genome (GRCh38) using STAR v2.7.11a56. Gene quantification was performed with HTSeq v0.13.5, utilizing the htseq-count command with the following parameters: BAM format (-f bam), unstranded data (-s no), and exon feature counting (-t exon), with gene annotations identified by gene_id in the provided GTF file57.
ComBat-seq was applied to correct for batch effect, using the center of sample origination as a batch58. Normalization was performed using the variance-stabilizing transformation (VST). From the normalized and batch-corrected data, we computed the principal components, which were used to calculate the UMAP coordinates with n_neighbors = 13, min_dist = 0.01, n_components =2 parameters. The results were then projected on a 2-dimensional plot. We performed density-based spatial clustering of applications with noise (DBSCAN)59, on the UMAP reduction to identify the individual clusters based on their transcriptomic profile using the following parameters eps = 0.28, minPts =8, which revealed 12 distinct clusters, excluding the cases identified as noise. Other dimension reduction techniques (PCA, tSNE) were also tested, revealing the presented UMAP reduction best option for visualization and clustering of the presented cohort (Supplementary Figs. 11 and 12).
Differential expression analysis was carried out on non-normalized data using DESeq2 across clusters, and genes were filtered based on a threshold of adjusted p-value (padj < 0.05) and absolute log2 fold change (|log2FC | > 2) to identify significantly DEGs60.
Enrichment analysis on DEG set was subsequently conducted using R package enrichR, utilizing the EnrichR database, specifically with “KEGG_2021_Human” and “ChEA_2016,” and “TRANSFAC_and_JASPAR_PWMs” databases61. Gene ontology overrepresentation of GO: Biological Processes (GO Source Data: 2023-01-01) was performed using ClusterProfiler with Benjamini-Hochberg multiple correction62. Semantic similarity analysis on the GO: Biological processes found to be overrepresented and enrichment of keywords in the clustering of the semantic similarity matrix was performed using the simplifyEnrichment R package using default parameters63. Tumor purity was assessed using ESTIMATE45 (Supplementary Fig. 9). To estimate immune cell abundances, CIBERSORTx46 was used using the LM22 signature matrix of 22 functionally defined human immune cell subsets64 (Supplementary Fig. 9). To validate and extend these findings, the xCell method was then performed to assess the enrichment of a broader repertoire of 64 immune and stromal cell populations47. (Supplementary Fig. 10). xCell was performed on Combat‑batch‑corrected, transcripts‑per‑million (TPM)–normalized data, excluding the mitochondrial content. Pairwise comparisons across clusters were performed using the Wilcoxon rank-sum test, with p-values adjusted for multiple testing using the Benjamini-Hochberg (BH). Only statistically significant results with adjusted p-values (p.adj < 0.05) were considered. The deconvolution of developing human spinal cord single cell data44 was performed by MuSIC to assess the dural cell population of presented cases compared to the broader meningioma cohort43 (Supplementary Fig. 8). Outlier samples identified in DBSCAN were excluded from all deconvolution and cell type enrichment analysis.
Co-expression analysis for NR4A1 and FOS/FOSB was carried out using BrainSpan, a developmental transcriptomic survey of gene expression profiles across different developmental stages in specific brain regions using RNA sequencing and exon microarray techniques41. Data were downloaded for these genes for 524 specimens from 42 donors, and the co-expression score was calculated using Pearson correlation. We also investigated the co-expression pattern of NR4A1 with FOS and FOSB using COXPRESdb v8, a tool suite providing analysis tools to investigate gene co-expression patterns in a curated database from publicly available published gene expression data42. NR4A1 and FOS/FOSB co-expression score was calculated using 17 and 29 published cohorts, respectively (Supplementary Table 2).
Identifying gene fusions from RNA-Seq data
Arriba (v2.1.0)65 was utilized to identify fusion events on the STAR-aligned, sorted, and indexed BAM files of all 1,232 cases. A blacklist file, as provided by Arriba, was used to remove recurrent alignment artifacts and chimeric transcripts that are present in healthy tissue, such as read-through fusions between neighboring genes, circular RNAs, and other non-canonically spliced transcripts. Annotation (-g) was performed using the GENCODE19 gene annotation database. While “High Confidence” calls were prioritized, “Medium/Low Confidence” calls were also investigated for samples clustering with FOS/FOSB fusion cases, revealing an additional case with a LMNA-FOSB fusion. Any “High Confidence” call overlapping with chr2q.35 or chr7q.36 is identified as a structural variation leading to Hedgehog activation, as reported before11.
Identifying arm-level CNVs from RNA-Seq data
To assess large-scale and chromosome arm-level copy number alterations in all 1232 samples with bulk RNA-seq data, we utilized the CaSpER package66. B-allele frequencies were extracted from aligned BAM files and used as input for the CaSpER algorithm, together with transcript quantifications. Control samples used in this analysis were obtained from previously published studies (SRR3995987, SRR3995989, SRR3995995, SRR3995998, SRR3996001, SRR3996002, SRR3996004, SRR3996005). Additionally, hg38 cytoband and centromere information were obtained from the UCSC genome browser. Cutoff of for arm-level events was set to 75%, expression cut-off was set to 1.5, and smoothing value was set to 3.
Statistics and reproducibility
All statistical analyses were conducted using R. Boxplots represent the interquartile (IQR range, central line representing the median, and whiskers extending to the lowest or highest value with 1.5 * IQR range; values for outliers, as defined by lying outside the whiskers, are shown with points. Statistical significance for differential gene expression analysis testing for log2 fold changes greater in absolute value than a given threshold (default, 0) was performed using a two-tailed Wald test. For comparisons between two groups (FOS/FOSB fusion vs others), an independent t-test or Mann-Whitney U test was performed. For gene set enrichment analysis, Fisher's exact test was performed, adjusted by the Benjamini-Hochberg procedure. For cell type enrichment analysis, a beta‑distribution–based BH-adjusted p‑value was used, and a two-sided Wilcoxon Rank Sum test was used for group-wise comparisons of results. For the deconvolution of 22 immune cell type characterizations, a two-sided Student’s t-test was used to determine the statistical significance of the deconvolution results, and a two-sided Wilcoxon Rank Sum test was used for group-wise comparisons of the results. Significance levels are represented as follows: ***p < 0.001, **p < 0.01, *p < 0.05 and ‘ns’ as not significant (p > 0.05). For clustering-based analysis, the samples which are designated as noise by DBSCAN (cluster0, n = 60) were excluded for group-wise comparisons.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The data generated and analyzed during this study for cases with FOS/FOSB fusion are available in the European Nucleotide Archive (ENA) under accession number PRJEB81142. The following publicly available datasets were also utilized in this study: Baylor College of Medicine (160 cases; GEO accession number GSE136661)[https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE136661], University of California San Francisco (185 cases; GEO accession number GSE183653) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE183653], Palacky University and University Hospital (70 cases; ENA accession number PRJNA705586) [https://www.ncbi.nlm.nih.gov/sra/?term=PRJNA705586], Fred Hutchinson Cancer Center (279 cases; GEO accession number GSE252291) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE252291], Yale University School of Medicine—Nature Genetics (19 cases; GEO accession number GSE85133) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE85133], and Yale University School of Medicine—Nature Communications (23 cases; ENA accession number PRJEB55424) [https://www.ncbi.nlm.nih.gov/sra/?term=PRJEB55424], UCSF-2020 (13 cases, GSE151921) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE151921], UCSF-2018 (42 cases, GSE101638)[https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE101638], UCSD (10 cases, GSE139652) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE139652], HKU/UCSF (302 cases, GSE212666) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE212666] and University of Toronto (123 cases, EGAD00001007494) [https://ega-archive.org/search/EGAD00001007494]. All datasets are publicly available, and detailed information on data access can be found in the respective repositories. Source data are provided with this paper.
Code availability
Scripts used to perform the analysis presented in this study are available on github (https://github.com/ErsonLab/FOS_FOSB_Paper) (https://doi.org/10.5281/zenodo.16615390).Source data are provided with this paper.
References
Torp, S. H., Solheim, O. & Skjulsvik, A. J. The WHO 2021 Classification of Central Nervous System tumours: a practical update on what neurosurgeons need to know-a minireview. Acta Neurochir. 164, 2453–2464 (2022).
Sun, C. et al. The Preferred Locations of Meningioma According to Different Biological Characteristics Based on Voxel-Wise Analysis. Front Oncol. 10, 1412 (2020).
Youngblood, M. W. et al. Correlations between genomic subgroup and clinical features in a cohort of more than 3000 meningiomas. J. Neurosurg. 133, 1345–1354 (2020).
Clark, V. E. et al. Genomic analysis of non-NF2 Meningiomas reveals mutations in TRAF7, KLF4, AKT1, and SMO. Science 339, 1077–1080 (2013).
Papi, L. et al. Somatic mutations in the neurofibromatosis type 2 gene in sporadic meningiomas. Hum. Genet. 95, 347–351 (1995).
Dumanski, J. P., Rouleau, G. A., Nordenskjold, M. & Collins, V. P. Molecular genetic analysis of chromosome 22 in 81 cases of meningioma. Cancer Res. 50, 5863–5867 (1990).
Abedalthagafi, M. et al. Oncogenic PI3K mutations are as common as AKT1 and SMO mutations in meningioma. Neuro Oncol. 18, 649–655 (2016).
Reuss, D. E. et al. Secretory meningiomas are defined by combined KLF4 K409Q and TRAF7 mutations. Acta Neuropathol. 125, 351–358 (2013).
Boetto, J., Bielle, F., Sanson, M., Peyre, M. & Kalamarides, M. SMO mutation status defines a distinct and frequent molecular subgroup in olfactory groove meningiomas. Neuro Oncol. 19, 345–351 (2017).
Brastianos, P. K. et al. Genomic sequencing of meningiomas identifies oncogenic SMO and AKT1 mutations. Nat. Genet 45, 285–289 (2013).
Youngblood, M. W. et al. Super-enhancer hijacking drives ectopic expression of hedgehog pathway ligands in meningiomas. Nat. Commun. 14, 6279 (2023).
Clark, V. E. et al. Recurrent somatic mutations in POLR2A define a distinct subset of meningiomas. Nat. Genet 48, 1253–1259 (2016).
Shankar, G. M. et al. Germline and somatic BAP1 mutations in high-grade rhabdoid meningiomas. Neuro Oncol. 19, 535–545 (2017).
Sievers, P. et al. Molecular signatures define BAP1-altered meningioma as a distinct CNS tumor with deregulation of Polycomb repressive complex target genes. Neuro Oncol. https://doi.org/10.1093/neuonc/noaf105 (2025).
Vaubel, R. A. et al. Genomic markers of recurrence risk in atypical meningioma following gross total resection. Neurooncol. Adv. 5, vdad004 (2023).
Driver, J. et al. A molecularly integrated grade for meningioma. Neuro Oncol. 24, 796–808 (2022).
Maas, S. L. N. et al. Loss over 5% of chromosome 1p is a clinically relevant and applicable cut-off for increased risk of recurrence in meningioma. Acta Neuropathol. 148, 17 (2024).
Bayley, J. C. T. et al. Multiple approaches converge on three biological subtypes of meningioma and extract new insights from published studies. Sci. Adv. 8, eabm6247 (2022).
Choudhury, A. et al. Hypermitotic meningiomas harbor DNA methylation subgroups with distinct biological and clinical features. Neuro Oncol. 25, 520–530 (2023).
Nassiri, F. et al. A clinically applicable integrative molecular classification of meningiomas. Nature 597, 119–125 (2021).
Sahm, F. et al. DNA methylation-based classification and grading system for meningioma: a multicentre, retrospective analysis. Lancet Oncol. 18, 682–694 (2017).
Wang, J. Z. et al. Molecular classification to refine surgical and radiotherapeutic decision-making in meningioma. Nat Med., https://doi.org/10.1038/s41591-024-03167-4 (2024).
Wagner, E. F. AP-1-Introductory remarks. Oncogene 20, 2334–2335 (2001).
Antonescu, C. R. et al. ZFP36-FOSB fusion defines a subset of epithelioid hemangioma with atypical features. Genes Chromosomes Cancer 53, 951–959 (2014).
Fittall, M. W. et al. Recurrent rearrangements of FOS and FOSB define osteoblastoma. Nat. Commun. 9, 2150 (2018).
Huang, S. C. et al. Frequent FOS Gene Rearrangements in Epithelioid Hemangioma: A molecular study of 58 cases with morphologic reappraisal. Am. J. Surg. Pathol. 39, 1313–1321 (2015).
Makise, N. et al. Recurrent FOS rearrangement in proliferative fasciitis/proliferative myositis. Mod. Pathol. 34, 942–950 (2021).
Walther, C. et al. A novel SERPINE1-FOSB fusion gene results in transcriptional up-regulation of FOSB in pseudomyogenic haemangioendothelioma. J. Pathol. 232, 534–540 (2014).
Bankhead, P. et al. QuPath: Open source software for digital pathology image analysis. Sci. Rep. 7, 16878 (2017).
Gillani, R. et al. Gene Fusions Create Partner and Collateral Dependencies Essential to Cancer Cell Survival. Cancer Res. 81, 3971–3984 (2021).
Cordier, F. & Creytens, D. New kids on the block: FOS and FOSB gene. J. Clin. Pathol. 76, 721–726 (2023).
Manna, P. R., Dyson, M. T. & Stocco, D. M. Role of basic leucine zipper proteins in transcriptional regulation of the steroidogenic acute regulatory protein gene. Mol. Cell Endocrinol. 302, 1–11 (2009).
Thirimanne, H. N. et al. Meningioma transcriptomic landscape demonstrates novel subtypes with regional associated biology and patient outcome. Cell Genom. 4, 100566 (2024).
Patel, A. J. et al. Molecular profiling predicts meningioma recurrence and reveals loss of DREAM complex repression in aggressive tumors. Proc. Natl. Acad. Sci. USA 116, 21715–21726 (2019).
Choudhury, A. et al. Meningioma DNA methylation groups identify biological drivers and therapeutic vulnerabilities. Nat. Genet. 54, 649–659 (2022).
Slavik, H. et al. Transcriptomic profiling revealed Lnc-GOLGA6A-1 as a novel prognostic biomarker of meningioma recurrence. Neurosurgery 91, 360–369 (2022).
Vasudevan, H. N. et al. Comprehensive molecular profiling identifies FOXM1 as a key transcription factor for Meningioma Proliferation. Cell Rep. 22, 3672–3683 (2018).
Prager, B. C. et al. The Meningioma enhancer landscape delineates novel subgroups and drives druggable dependencies. Cancer Discov. 10, 1722–1741 (2020).
Magill, S. T. et al. Multiplatform genomic profiling and magnetic resonance imaging identify mechanisms underlying intratumor heterogeneity in meningioma. Nat. Commun. 11, 4803 (2020).
Guo, H. et al. NR4A1 regulates expression of immediate early genes, suppressing replication stress in cancer. Mol. Cell 81, 4041–4058.e4015 (2021).
Miller, J. A. et al. Transcriptional landscape of the prenatal human brain. Nature 508, 199–206 (2014).
Obayashi, T., Kodate, S., Hibara, H., Kagaya, Y. & Kinoshita, K. COXPRESdb v8: an animal gene coexpression database navigating from a global view to detailed investigations. Nucleic Acids Res. 51, D80–D87 (2023).
Wang, X., Park, J., Susztak, K., Zhang, N. R. & Li, M. Bulk tissue cell type deconvolution with multi-subject single-cell expression reference. Nat. Commun. 10, 380 (2019).
Lawrence, J. E. G. et al. HOX gene expression in the developing human spine. Nat. Commun. 15, 10023 (2024).
Yoshihara, K. et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat. Commun. 4, 2612 (2013).
Newman, A. M. et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat. Biotechnol. 37, 773–782 (2019).
Aran, D., Hu, Z. & Butte, A. J. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 18, 220 (2017).
van, I. D. G. P. et al. Telatinib Is an Effective Targeted Therapy for Pseudomyogenic Hemangioendothelioma. Clin. Cancer Res. 24, 2678–2687 (2018).
Pittenger, M. F. et al. Multilineage potential of adult human mesenchymal stem cells. Science 284, 143–147 (1999).
Uccelli, A., Moretta, L. & Pistoia, V. Mesenchymal stem cells in health and disease. Nat. Rev. Immunol. 8, 726–736 (2008).
Hayakawa, K., Snyder, E. Y. & Lo, E. H. Meningeal multipotent cells: a hidden target for CNS Repair?. Neuromol. Med. 23, 339–343 (2021).
DeSisto, J. et al. Single-cell transcriptomic analyses of the developing meninges reveal meningeal fibroblast diversity and function. Dev. Cell 54, 43–59.e44 (2020).
Sabatakos, G. et al. Overexpression of DeltaFosB transcription factor(s) increases bone formation and inhibits adipogenesis. Nat. Med 6, 985–990 (2000).
Board, W. C. o. T. E. Central Nervous System Tumours. (International Agency for Research on Cancer, 2021).
Chen, X. et al. Manta: rapid detection of structural variants and indels for germline and cancer sequencing applications. Bioinformatics 32, 1220–1222 (2016).
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Putri, G. H., Anders, S., Pyl, P. T., Pimanda, J. E. & Zanini, F. Analysing high-throughput sequencing data in Python with HTSeq 2.0. Bioinformatics 38, 2943–2945 (2022).
Zhang, Y., Parmigiani, G. & Johnson, W. E. ComBat-seq: batch effect adjustment for RNA-seq count data. NAR Genom. Bioinform. 2, lqaa078 (2020).
Ester, M., Kriegel, H.-P., Sander, J. & Xu, X. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, 226–231 (AAAI Press, Portland, Oregon, 1996).
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Kuleshov, M. V. et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 44, W90–W97 (2016).
Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovations 2, 100141 (2021).
Gu, Z. & Hubschmann, D. simplifyEnrichment: A Bioconductor Package for Clustering and Visualizing Functional Enrichment Results. Genom. Proteom. Bioinforma. 21, 190–202 (2023).
Newman, A. M. et al. Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods 12, 453–457 (2015).
Uhrig, S. et al. Accurate and efficient detection of gene fusions from RNA sequencing data. Genome Res. 31, 448–460 (2021).
Serin Harmanci, A., Harmanci, A. O. & Zhou, X. CaSpER identifies and visualizes CNV events by integrative analysis of single-cell or bulk RNA-sequencing data. Nat. Commun. 11, 89 (2020).
Acknowledgements
The authors would like to thank the patients and families who contributed to this study. The authors gratefully acknowledge the contributions of Boris Krischek (1972–2021), who contributed resources for this study but unfortunately passed away before publication. This work is supported by Yale School of Medicine, Department of Neurosurgery Clinical Sequencing Funds to cover the whole exome sequencing of the specimens, and by Connecticut Brain Tumor Alliance and Gregory M. Kiez and Mehmet Kutman Foundation for other research-related costs. All sequencing was performed at Yale Center of Genomics, at Yale Cancer Center, which is supported in part by NIH Research Grant no. P30CA016359, from the National Cancer Institute.
Author information
Authors and Affiliations
Contributions
Conceptualization (C), Methodology (M), Formal Analysis (FA), Validation (V), Data Curation (DC), Writing Original Draft (WOD), Writing - Review & Editing (WRE), Figure Preparation (FP), Supervision (Sup), Project Administration (PA), Funding Acquisition (F). KYalcin: C., M., F.A., D.C., W.O.D., F.P.; H.A.: M., F.A., D.C., F.P.; B.G.: M., F.A., D.C., F.P.; D.S.F.: V., F.P.; T.B.: M., Sup; D.F.M.: M., D.C.; M.W.Y.: F.A., D.C.; M.G.H.: D.C.; A.B.: V.; N.S.L.: D.C.; J.O.B.: D.C.; M.M.C.: D.C.; A.D.: D.C.; H.W.: F.A., S.B.O.: D.C., Sup; KYasuno: F.A., K.B.: D.C., F.A.; A.G.E.S.: M., V., F.P., Sup; K.M.G.: M., V., F.P., Sup; D.M.: M., F.A., V., Sup; J.M.: M., F.A., D.C., W.R.E., Sup; M.G.: C., M., F.A., D.C., W.O.D., W.R.E., F.P., Sup, P.A., F.; E.Z.E.O.: C., M., F.A., V., D.C., W.O.D., W.R.E., F.P., Sup, P.A.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks anonymous reviewer(s) for their contribution to the peer review of this work. [A peer review file is available].
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Yalcin, K., Alanya, H., Gultekin, B. et al. Subgroup of meningiomas involving FOS and FOSB gene fusions. Nat Commun 16, 10532 (2025). https://doi.org/10.1038/s41467-025-65549-7
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41467-025-65549-7
