Meningiomas are the most common benign brain tumors. Mutations of the E3 ubiquitin ligase TRAF7 occur in 25% of meningiomas and commonly cooccur with mutations in KLF4, yet the functional link between TRAF7 and KLF4 mutations remains unclear. By generating an in vitro meningioma model derived from primary meningeal cells, we elucidated the cooperative interactions that promote meningioma development. By integrating TRAF7-driven ubiquitinome and proteome alterations in meningeal cells and the TRAF7 interactome, we identified TRAF7 as a proteostatic regulator of RAS-related small GTPases. Meningioma-associated TRAF7 mutations disrupted either its catalytic activity or its interaction with RAS GTPases. TRAF7 loss in meningeal cells altered actin dynamics and promoted anchorage-independent growth by inducing CDC42 and RAS signaling. TRAF deficiency–driven activation of the RAS/MAPK pathway promoted KLF4-dependent transcription that led to upregulation of the tumor-suppressive Semaphorin pathway, a negative regulator of small GTPases. KLF4 loss of function disrupted this negative feedback loop and enhanced mutant TRAF7-mediated cell transformation. Overall, this study provides new mechanistic insights into meningioma development, which could lead to novel treatment strategies.

Significance:

The intricate molecular cross-talk between the ubiquitin ligase TRAF7 and the transcription factor KLF4 provides a first step toward the identification of new therapies for patients with meningioma.

Meningiomas are the most common benign intracranial tumors, accounting for about 35% of all primary brain tumors (1). Although surgery is the preferred treatment for the majority of patients with meningioma, maximal surgical resection without compromising neurologic functions can be challenging, and residual meningiomas following surgical resection have a high likelihood of recurrence. A comprehensive understanding of the functional basis underlying meningioma development may lead to more effective treatments of patients with meningioma (2).

The most well-studied genomic alterations in meningiomas are monosomy of chromosome 22 and inactivating mutations of Neurofibromatosis 2 gene (NF2). Mutations in the TNF receptor–associated factor 7 gene TRAF7 are the second most frequent genetic alterations, found in about 25% of all grade I and grade II meningiomas (Fig. 1A). TRAF7 mutations commonly cooccur with mutations in Kruppel-like factor 4 (KLF4), v-Akt murine thymoma viral oncogene homolog 1 (AKT1), or phosphatidylinositol-4,5-biphosphate 3-kinase catalytic subunit alpha (PIK3CA) but are mutually exclusive with mutations in the Hedgehog pathway member Smoothened (SMO) and in NF2 (3).

Figure 1.

Integrated analysis of TRAF7-mediated alterations of ubiquitome, proteome and transcriptome, and TRAF7 interactome. A, Schematic diagram of the TRAF7 and KLF4 domains showing the mutated residues in meningiomas; mutations analyzed in this study are indicated. RING, the ubiquitin ligase RING catalytic domain; WD40, WD40 repeats domain; ZNF, zinc finger domain. B, AI growth of HMCs expressing shRNA targeting GFP or TRAF7. Values are means ± SEM; n ≥ 3; P values were calculated by two-sided Student t test. C, Immunoblotting analysis and AI growth of HMCs expressing an empty vector (EV), wt-TRAF7, or the indicated TRAF7 mutants. Values are means ± SEM; n = 3; P values were calculated by one-way ANOVA. D, Growth of CH157-MN xenografts expressing shGFP or shTRAF7–2. n = 6; P value was assessed by two-way ANOVA. E, Tumor weight of CH157-MN xenografts expressing shGFP or shTRAF7–2 at the end point is presented as mean ± SEM, n = 6 per group. P value was assessed by Mann–Whitney unpaired t test. F, A workflow for studying TRAF7-mediated alterations and the TRAF7 interactome. G, Ingenuity Pathway Analysis of TRAF7-mediated alterations and TRAF7 interactome. Top 10 significantly enriched pathways for each of the datasets are visualized as a network using the EnrichmentMap Cytoscape App. The pie charts indicate datasets that showed significant pathway enrichment (P < 0.01) of the indicated canonical pathways. The pathways that shared the same genes are connected by lines.

Figure 1.

Integrated analysis of TRAF7-mediated alterations of ubiquitome, proteome and transcriptome, and TRAF7 interactome. A, Schematic diagram of the TRAF7 and KLF4 domains showing the mutated residues in meningiomas; mutations analyzed in this study are indicated. RING, the ubiquitin ligase RING catalytic domain; WD40, WD40 repeats domain; ZNF, zinc finger domain. B, AI growth of HMCs expressing shRNA targeting GFP or TRAF7. Values are means ± SEM; n ≥ 3; P values were calculated by two-sided Student t test. C, Immunoblotting analysis and AI growth of HMCs expressing an empty vector (EV), wt-TRAF7, or the indicated TRAF7 mutants. Values are means ± SEM; n = 3; P values were calculated by one-way ANOVA. D, Growth of CH157-MN xenografts expressing shGFP or shTRAF7–2. n = 6; P value was assessed by two-way ANOVA. E, Tumor weight of CH157-MN xenografts expressing shGFP or shTRAF7–2 at the end point is presented as mean ± SEM, n = 6 per group. P value was assessed by Mann–Whitney unpaired t test. F, A workflow for studying TRAF7-mediated alterations and the TRAF7 interactome. G, Ingenuity Pathway Analysis of TRAF7-mediated alterations and TRAF7 interactome. Top 10 significantly enriched pathways for each of the datasets are visualized as a network using the EnrichmentMap Cytoscape App. The pie charts indicate datasets that showed significant pathway enrichment (P < 0.01) of the indicated canonical pathways. The pathways that shared the same genes are connected by lines.

Close modal

TRAF7 is an E3 ubiquitin ligase containing an N-terminal RING finger domain and a C-terminal WD40 repeat domain responsible for the substrate recognition (Fig. 1A). Even though TRAF7 was initially classified as a member of the TNF receptor–associated factor (TRAF) family, the lack of the classical TRAF domain suggests a misclassification (4). In addition to meningioma, TRAF7 mutations have been identified at a high frequency in adenomatoid tumors of the genital tract and intraneural perineuriomas, a rare nerve sheath tumor derived from perineurial cells (5, 6). TRAF7 mutations, mutually exclusive with NF2 alterations, have also been identified in a subset of pleural mesotheliomas (7, 8).

About 40% of TRAF7-mutated meningiomas harbor a recurrent KLF4-K409Q mutation, which occurs almost exclusively in meningiomas (Fig. 1A). KLF4 belongs to a family of transcriptional regulators involved in proliferation, differentiation, migration, inflammation, and pluripotency (3). In cancer, KLF4 has been reported to act as both a tumor suppressor and an oncogene (9–11). A recent study demonstrated that decreased KLF4 expression in anaplastic meningioma stem-like cells promotes their tumorigenicity, suggesting a tumor-suppressive role of KLF4 in anaplastic meningiomas (12). Nonetheless, it is still unclear how the K409Q mutation alters the regulation of KLF4′s target genes, eventually contributing to meningioma development (13, 14). Approximately, a third of TRAF7-mutated meningiomas have AKT1 mutations. The only recurrent oncogenic AKT1-E17K mutation is well known to activate the PI3K/AKT pathway (1). The PI3K/AKT pathway could be also upregulated in TRAF7-mutated meningiomas due to mutations in PI3KCA (3).

The majority of TRAF7-mutated meningiomas are located at the skull base and are known to invade the neighboring brain parenchyma due to its intrusive growth and intimately associated with the surrounding tissues (2), challenging a complete surgical resection with no alternative treatment options. Understanding the mechanisms leading to non–NF2-mutated meningioma development may have a considerable translational impact by increasing therapeutic options for meningioma treatment.

Cell culture and reagents

HEK293T cells and CH157-MN cells (a kind gift of Dr. Ian Dunn, Brigham and Women's Hospital, Boston, MA) were cultured in DMEM (GIBCO) supplemented with 10% FBS and 1% penicillin/streptomycin. Meningioma HBL52 cells (Cell Lines Service) and primary Human Meningeal Cells (Sciencell Research Laboratories) were cultured according to the manufacturer's protocols. Cells were tested with a Mycoplasma Testing Kit (Lonza MycoAlert, LT07–318) every 2 weeks.

Lentiviral infections were performed as described previously by the RNAi Consortium (TRC; http://www.broadinstitute.org/rnai/public/resources/protocols). Infected cells were selected by treatment with 2 μg/mL Puromycin (InvivoGen), 400 μg/mL of Hygromycin-B (Thermo Fisher Scientific), or 500 μg/mL G418 (InvivoGen) and then recovered for 2–3 weeks. Transient transfections were performed using GeneJuice (Millipore), Turbofect (Thermo Fisher Scientific), or Lipofectamine 3000 (Thermo Fisher Scientific).

Anchorage-independent growth in soft agar was performed as described previously (15). Briefly, 1 × 104 cells were plated in 0.35% Noble agar covered over a 0.5% agar bottom layer. Two or 3 weeks after plating, several random areas were imaged, and the colony number was quantified using the ImageJ Software (NIH).

All drugs, plasmids, and antibodies used in this study are listed in Supplementary Table S1.

Tumor xenografts

All xenograft experiments were approved and performed according to the KU Leuven IACUC guidelines (P117/2016). A total of 2.5 × 105 of CH157-MN cells were injected subcutaneously in the flank of Rj:NMRI-Foxn1nu/nu mice. Tumor volumes were monitored every 2 days. The volume of each tumor was measured with a caliper using the formula: V = π × d2 × D/6, where D is the major tumor axis and d is the minor tumor axis.

Analysis of protein expression, activity, and interactions

Cells were washed in cold PBS and scraped on ice in the IP lysis buffer (50 mmol/L Tris-HCl pH 7.5, 150 mmol/L NaCl, 1% NP-40; Thermo Fisher Scientific) containing protease inhibitor and phosphatase inhibitor cocktails (Roche). Cell lysates were centrifugated for 10 minutes 16,000 × g at 4°C, and proteins were immunoprecipitated either using anti-Flag (M2) or anti-HA agarose beads (Sigma-Aldrich) overnight at 4°C, or protein A/G magnetic beads (Thermo Fisher Scientific, #88803) premixed with primary antibody, or mouse IgG1 for 4 hours at 4°C, and washed with ice-cold IP lysis buffer. Proteins were finally eluted with 3×Flag or HA Peptides (Sigma-Aldrich) or by boiling for 10 minutes in 2× NuPAGE LDS Sample Buffer (Thermo Fisher Scientific, #NP0007).

Ubiquitinated proteins were purified as described previously (16). Subcellular fractionation was performed according to the manufacturer's protocol (Thermo Fisher Scientific). Protein stability in living cells was assessed by the global protein stability (GPS) approach (17, 18). Twenty-four hours after plating, HEK293T cells were transfected with the GPS reporter and shRNA constructs in a ratio 1:1. Forty-eight hours after transfection, cells were harvested and analyzed using the MACSQuant VYB Flow Cytometer (Miltenyi Biotec). Raw data were analyzed using FlowJo Software (BD Biosciences). Live single cells were monitored for the expression of GFP and DsRed, and the GFP/DsRed ratio was counted for measuring the relative protein stability. RAS activity was assessed using the RAS activation ELISA Assay Kit (EMD Millipore). CDC42 activation assay was performed according to the manufacturer's protocol (Cytoskeleton, Inc). The proteomic and MS-based approaches are described in detail in Supplementary Materials and Methods.

Immunostaining and live imaging

For immunostaining, 4 × 104 cells per well were plated on 8-well chamber glass slide (Corning, # 354118) and fixed with 4% paraformaldehyde. Immunofluorescence was performed as described previously (19). Briefly, cells were permeabilized in PBS-0.15% Triton-X100 and then blocked with 1% BSA-10% goat serum for 60 minutes. Primary antibodies, goat Alexa-conjugated secondary antibodies, Phalloidin-Atto 488 (Sigma-Aldrich, 49409) were diluted in blocking buffer and applied before final mounting with Vectashield (Vector Laboratories). Cross-reactivity between antibodies was avoided by preblocking secondary antimouse antibodies with 10% rat serum. Images were captured using the Nikon C2 or the Nikon A1R confocal microscopes. Image analysis was performed using the software ImageJ (NIH).

For image acquisition of focal adhesions, a Nikon TiE inverted A1R (+HD resonant scanning upgrade) microscope in combination with a 60× Oil objective (NA 1,4) The setup was controlled by NIS-Elements (NIS 5.11.01 build 1368a, Nikon Instruments Europe B.V.). The number and size of the focal adhesions per cell was measured using a general analysis protocol (GA3) in NIS-elements (NIS 5.11.01 build 1368a, Nikon Instruments Europe B.V.). Individual cells were manually selected as regions of interest (ROI). Within the ROIs, the focal adhesions were detected via a bright-spot detection method after Gaussian filtering of the image (kernel size 2). After removing artifacts, only focal adhesion sites inside the individual cells were measured. Live imaging is described in Supplementary Materials and Methods.

mRNA expression

RNA extraction was performed using the NucleoSpin RNA Kit (Machery Nagel). For qRT-PCR, 500 ng of tRNA was used to prepare cDNA for mRNA analysis. cDNA was reverse-transcribed with the Sensifast cDNA Synthesis Kit (Bioline). qRT-PCR was performed with the LightCycler 480 SYBR Green I Master reagent using the LightCycler 480 System (Roche). All primer sequences are reported in Supplementary Table S1.

DNA sequencing

Original human tumor biopsies from benign meningioma were obtained fresh from the operating theater, and all patients gave written informed consent for use of their biopsy samples. All procedures involving human samples were approved by the UZ Leuven/KU Leuven medical ethics committee (ML8713/S54185). All procedures involving animals were approved and performed by the PDX platform in accordance with the guidelines of the KU Leuven IACUC (P147/2012). DNA extraction was performed on the PDX-frozen tissues using DNeasy Blood & Tissue Kit (Qiagen).

To detect KLF4 indels, genomic DNA of the cells was extracted using the NucleoSpin Tissue Kit (Macherey-Nagel, #740952), the loci encompassing gKLF4–1 (KLF4 exon3) and gKLF4–2 (KLF4 exon4) were amplified, and the amplicons were sequenced using their respective forward primers. The presence of InDels was determined using the Synthego ICE analysis tool (https://ice.synthego.com).

Statistical and bioinformatic analyses

All data were analyzed using GraphPad prism 8. Bioinformatic analyses are described in Supplementary Materials and Methods.

A multiomics analysis reveals TRAF7-mediated alterations in human meningeal cells

Some of the meningioma-associated mutations of TRAF7 present heterozygous indels (Fig. 1A), suggesting that a partial loss of TRAF7 function could promote meningioma development. To model meningioma development, we used primary human meningeal cells (HMC) immortalized by overexpressing the SV40 large T antigen. A partial suppression of TRAF7 in HMCs promoted their anchorage-independent (AI) growth, whereas overexpression of wt-TRAF7 in HMCs inhibited their ability to form soft-agar colonies (Fig. 1B and C; Supplementary Fig. S1A–S1D). In contrast, overexpression of the most recurrent meningioma-associated TRAF7 mutants, TRAF7-T391I, TRAF7-N520S, and TRAF7-G536S, or the truncated TRAF7-E353* mutant did not affect the AI growth (Fig. 1C; Supplementary Fig. S1C and S1D). Similarly, ectopic expression of wt-TRAF7 inhibited the AI growth of meningioma CH157-MN cells (Supplementary Fig. S1E). Furthermore, suppression of TRAF7 in CH157-MN cells promoted the xenograft growth (Fig. 1D and E; Supplementary Fig. S1F). These results indicate a tumor-suppressive function of TRAF7 in meningeal cells.

To uncover the molecular mechanism by which TRAF7 contributes to meningioma development, we interrogated TRAF7-mediated alterations in meningeal cells by integrating ubiquitinome, proteome, and transcriptome analyses (Fig. 1F; Supplementary Fig. S1G). In parallel, we characterized TRAF7 interactome by employing a mass spectrometry (MS)-based Virotrap approach, which allows trapping protein complexes from intact mammalian cells under native conditions (Fig. 1F; ref. 20). To identify the functional biological processes modulated by TRAF7, we performed an Ingenuity Pathway Analysis (IPA) in all datasets. We found that the ubiquitome and proteome datasets shared most of the top-ranked pathways, but showed a limited overlap with the pathways altered at mRNA level (Fig. 1G; Supplementary Fig. S2). Thus, a multiomics analysis of meningeal cells revealed a global TRAF7-mediated proteome rewiring, with the majority of dynamic protein expression being regulated posttranslationally.

Integrating TRAF7-driven alterations and TRAF7 interactome point out the role of TRAF7 in the regulation of actin cytoskeleton (Fig. 1G; Fig. 2A). Consistently, we found that TRAF7 is localized in the cytoplasm and at the leading edge of meningeal cells (Fig. 2B). Moreover, immunostaining analysis of HMCs showed colocalization of TRAF7 with actin cytoskeleton at the leading edges, but no colocalization with tubulin (Supplementary Fig. S3A). Next, we examined the effect of TRAF7 on the dynamics of the actin cytoskeleton. Time-lapse video of HMCs showed that TRAF7 suppression promoted actin dynamics, inducing contraction, retraction, and protrusions at both leading and rear edges (Fig. 2C; Supplementary Movies S1 and S2). TRAF7 depletion also led to an increased number of filopodia and slowed down filopodial tip retraction, hence enhanced stability of nonstationary filopodia (Fig. 2C and D). More stable filopodia in TRAF7-depleted cells were associated with an increased number of focal adhesion (Fig. 2E). In contrast, wt-TRAF7 suppressed actin dynamics (Supplementary Fig. S3B; Supplementary Movies S3 and S4). We observed decreased filopodia number and stability as well as a reduced number of focal adhesions in HMCs overexpressing wt-TRAF7 (Fig. 2F and G). Altogether, these results implicate TRAF7 in the regulation of actin dynamics. Similarly, the proteomic analysis of NF2-mutated meningiomas revealed focal adhesion and actin cytoskeleton signaling among the most significantly enriched pathways (21). This strongly indicates that mutations of both NF2 and TRAF7 converge on alterations of the actin cytoskeleton.

Figure 2.

TRAF7 is a regulator of cytoskeleton organization. A, Overall enrichment scores of the canonical IPA pathways identified across all datasets are shown in Fig. 1F and 1G. B, Immunostaining of HMCs expressing HA-tagged TRAF7 with anti-HA antibody. Scale bar, 10 μm. C, Representative still images of live imaging of actin dynamics in HMCs expressing shGFP or shTRAF7. Acquisition every 2 minutes for 10 minutes. Scale bar, 10 μm. D, Actin dynamics in HMCs expressing shGFP or shTRAF7. The number of filopodia per cell is shown as a whisker plot. P values were calculated by a two-sided Student t test. The filopodia length is shown as mean ± SEM; P values were detected by two-way ANOVA. E, Immunostaining of HMCs expressing shGFP or shTRAF7 with antivinculin antibody. Scale bar, 10 μm. The number of focal adhesions per cells is shown as a whisker plot. P values were calculated by a two-sided Student t test. F, Imaging of actin dynamics in HMCs expressing an empty vector (EV) or HA-TRAF7. The filopodia length is shown as mean± SEM; P values were detected by two-way ANOVA. The number of filopodia per cell is shown as a whisker plot. P values were calculated by a two-sided Student t test. G, Immunostaining of HMCs expressing an empty vector or HA-TRAF7 with antivinculin antibody. Scale bar, 10 μm. The number of focal adhesions per cell is shown as mean± SEM; P values were calculated by a two-sided Student t test. B and D–G, Number of analyzed cells ≥50. A whisker plot depicting the 5th, 25th, 75th, and 95th percentiles with outliers.

Figure 2.

TRAF7 is a regulator of cytoskeleton organization. A, Overall enrichment scores of the canonical IPA pathways identified across all datasets are shown in Fig. 1F and 1G. B, Immunostaining of HMCs expressing HA-tagged TRAF7 with anti-HA antibody. Scale bar, 10 μm. C, Representative still images of live imaging of actin dynamics in HMCs expressing shGFP or shTRAF7. Acquisition every 2 minutes for 10 minutes. Scale bar, 10 μm. D, Actin dynamics in HMCs expressing shGFP or shTRAF7. The number of filopodia per cell is shown as a whisker plot. P values were calculated by a two-sided Student t test. The filopodia length is shown as mean ± SEM; P values were detected by two-way ANOVA. E, Immunostaining of HMCs expressing shGFP or shTRAF7 with antivinculin antibody. Scale bar, 10 μm. The number of focal adhesions per cells is shown as a whisker plot. P values were calculated by a two-sided Student t test. F, Imaging of actin dynamics in HMCs expressing an empty vector (EV) or HA-TRAF7. The filopodia length is shown as mean± SEM; P values were detected by two-way ANOVA. The number of filopodia per cell is shown as a whisker plot. P values were calculated by a two-sided Student t test. G, Immunostaining of HMCs expressing an empty vector or HA-TRAF7 with antivinculin antibody. Scale bar, 10 μm. The number of focal adhesions per cell is shown as mean± SEM; P values were calculated by a two-sided Student t test. B and D–G, Number of analyzed cells ≥50. A whisker plot depicting the 5th, 25th, 75th, and 95th percentiles with outliers.

Close modal

TRAF7 interacts with RAS-related GTPases

The actin cytoskeleton pathway was among the top 10 pathways enriched in the TRAF7 interactome (Fig. 2A; Supplementary Fig. S2). This pathway was also highly enriched among potential TRAF7 interactors identified by the BioPlex network (P = 6.29e–3; ref. 22). Among putative TRAF7 interactors involved in the regulation of actin cytoskeleton, we identified several RAS-related GTPases, such as cell division cycle 42 (CDC42), Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS), Harvey Rat Sarcoma Viral Oncogene Homolog (HRAS), RAC Family Small GTPase 1/2/3 (RAC1/RAC2/RAC3), Related RAS Viral Oncogene Homolog 2 (RRAS2), and RAS Homolog Family Member A (RHOA; Fig. 3A). For further studies, we decided to focus on CDC42, a key regulator of filopodia formation, and on the RAS proto-oncogenes. Activating mutations in CDC42 were found be to mutually exclusive with TRAF7 mutations in differentiated papillary mesotheliomas (23), suggesting that TRAF7 and CDC42 might belong to the same signaling network. On the other hand, we previously identified TRAF7 as a potential HRAS interactor in a Virotrap screen, in which HRAS served as bait (19). Moreover, germline TRAF7 mutations in the TRAF7 syndrome patients phenotypically overlap with RASopathy, a group of genetic syndromes caused by changes in genes that are part of the RAS pathway (24, 25).

Figure 3.

TRAF7 forms a complex with the RAS-related GTPases. A, A Virotrap screen performed in HEK293T cells using group-specific antigen (GAG)–TRAF7 as bait. Escherichia coli dihydrofolate reductase (eDHFR) fused to GAG was used as a negative control. A heatmap showing the RAS-related GTPases differentially interacting with eDHFR and TRAF7. The scale shows Z-scored site intensity values. B, Flag-tagged TRAF7 was immunoprecipitated from HEK293T cell lysates using anti-FLAG (M2) resin. CDC42 and panRAS were detected by immunoblotting. C, panRAS or Flag-tagged CDC42 was immunoprecipitated from HEK293T cells overexpressing FLAG-tagged or HA-tagged TRAF7 using anti-panRAS antibody coupled to protein A/G agarose resin or anti-FLAG (M2) resin. Flag or HA-tagged TRAF7 was detected by immunoblotting. D, HA-tagged RAS proteins were immunoprecipitated from HEK293T cells overexpressing Flag-tagged TRAF7 using anti-HA resin. TRAF7 was detected by immunoblotting with anti-Flag-antibody. E, Flag-tagged TRAF7 was immunoprecipitated from HEK293T cells overexpressing HA-tagged RAS using anti-FLAG (M2) resin. RAS isoforms were detected by immunoblotting with anti-HA antibody. F, Immunostaining of HMCs expressing HA-TRAF7 and mCherry-CDC42 or mCherry-NRAS with anti-HA antibody. Scale bar, 10 μm. G, HA-tagged CDC42 or RAS proteins were immunoprecipitated from HEK293T cells overexpressing Flag-tagged TRAF7 using anti-HA resin. TRAF7 was detected by immunoblotting with anti-Flag-antibody.

Figure 3.

TRAF7 forms a complex with the RAS-related GTPases. A, A Virotrap screen performed in HEK293T cells using group-specific antigen (GAG)–TRAF7 as bait. Escherichia coli dihydrofolate reductase (eDHFR) fused to GAG was used as a negative control. A heatmap showing the RAS-related GTPases differentially interacting with eDHFR and TRAF7. The scale shows Z-scored site intensity values. B, Flag-tagged TRAF7 was immunoprecipitated from HEK293T cell lysates using anti-FLAG (M2) resin. CDC42 and panRAS were detected by immunoblotting. C, panRAS or Flag-tagged CDC42 was immunoprecipitated from HEK293T cells overexpressing FLAG-tagged or HA-tagged TRAF7 using anti-panRAS antibody coupled to protein A/G agarose resin or anti-FLAG (M2) resin. Flag or HA-tagged TRAF7 was detected by immunoblotting. D, HA-tagged RAS proteins were immunoprecipitated from HEK293T cells overexpressing Flag-tagged TRAF7 using anti-HA resin. TRAF7 was detected by immunoblotting with anti-Flag-antibody. E, Flag-tagged TRAF7 was immunoprecipitated from HEK293T cells overexpressing HA-tagged RAS using anti-FLAG (M2) resin. RAS isoforms were detected by immunoblotting with anti-HA antibody. F, Immunostaining of HMCs expressing HA-TRAF7 and mCherry-CDC42 or mCherry-NRAS with anti-HA antibody. Scale bar, 10 μm. G, HA-tagged CDC42 or RAS proteins were immunoprecipitated from HEK293T cells overexpressing Flag-tagged TRAF7 using anti-HA resin. TRAF7 was detected by immunoblotting with anti-Flag-antibody.

Close modal

To confirm the interaction between TRAF7 and RAS proteins and CDC42, we performed a set of immunoprecipitations. We found that Flag-tagged TRAF7 coimmunoprecipitated with endogenous RAS proteins and CDC42 (Fig. 3B and C). Reciprocal coimmunoprecipitations demonstrated that TRAF7 interacted with all RAS isoforms and CDC42 (Fig. 3C–E). Moreover, immunostainings analysis showed colocalization of TRAF7 with NRAS and CDC42, mostly at the leading edge of meningeal cells (Fig. 3F). These results indicate that TRAF7 forms a complex with small GTPases, RAS and CDC42.

Given that the majority of recurrent TRAF7 mutations in meningioma are located within the WD40 repeat domain responsible for the substrate recognition (Fig. 1A; ref. 13), we assessed whether these mutations affect the ability of TRAF7 to bind RAS and/or CDC42 (Fig. 3G). We found that the truncated TRAF7-E353* and the most recurrent WD40 domain TRAF7 mutants showed reduced binding to either NRAS or CDC42. On the other hand, the TRAF7-R153S mutation located in the RING catalytic domain, which we identified in our cohort of patients with meningioma, did not affect binding to NRAS or CDC42 (Fig. 1A; Fig. 3G). The results suggest that TRAF7 mutations in the WD40 domain could promote meningioma development by disrupting TRAF7 binding to the RAS-like GTPases.

TRAF7 loss of function promotes the AI growth of human meningeal cells

The ubiquitome analysis revealed that TRAF7 suppression decreased CDC42 ubiquitination at different sites with the most pronounced effect at K163 (Fig. 4A). We also observed a tendency to lower levels of ubiquitination of several RAS-related GTPases in TRAF7-depleted HMCs (Fig. 4A). His-ubiquitin pull-down confirmed that TRAF7 suppression led to decreased ubiquitination of NRAS (Fig. 4B; Supplementary Fig. S4A). In contrast, overexpression of wt-TRAF7, but not meningioma-associated TRAF7 mutants, increased ubiquitination of both CDC42 and NRAS (Fig. 4C and D). These results indicate that TRAF7 controls ubiquitination of several RAS-related GTPases, while meningioma-associated TRAF7 mutants lose this function.

Figure 4.

TRAF7 controls proteostasis of the RAS-related GTPases. A, A heatmap showing the change in ubiquitination of RAS-related GTPases in HMCs expressing shGFP or shTRAF7. The scale shows log2 (fold of change) of site intensity values. n = 3 B, Ubiquitinated Flag-tagged NRAS was purified from HEK293T cells expressing the indicated constructs by Co2+ metal affinity chromatography and detected by immunoblotting. C, Ubiquitinated Flag-tagged CDC42 was purified from HEK293T cells expressing Flag-tagged wt-TRAF7 or TRAF7 mutants by Co2+ metal affinity chromatography and detected by immunoblotting. D, Ubiquitinated Flag-tagged NRAS was purified from HEK293T cells expressing Flag-tagged wt-TRAF7 or TRAF7 mutants by Co2+ metal affinity chromatography and detected by immunoblotting. E, A heatmap showing protein expression of RAS-related GTPases in HMCs expressing shGFP and shTRAF7. The scale shows Z-scored protein intensity values. n = 3 F, Immunoblotting analysis of CDC42 and NRAS protein levels in HMCs expressing either shGFP or shTRAF7. G, A heatmap showing the change in mRNA expression of RAS-related GTPases in HMCs expressing shGFP and shTRAF7. The scale shows log2 (fold of change) of TPM. n = 3. H, Global protein stability approach measuring NRAS, KRAS, HRAS, and CDC42 relative protein stability in HEK293T cells expressing shGFP or shTRAF7. Values are means ± SEM; n ≥ 4. P values are from a two-sided Student t test. I, Immunoblotting of the membrane and cytoplasmic fractions isolated from HMCs expressing shGFP or shTRAF7. B–D,n = 3. Numbers on the left are molecular masses in kDa. 2xUb, two Ub; 1xUb, one Ub; EV, empty vector; WCL, whole-cell lysate.

Figure 4.

TRAF7 controls proteostasis of the RAS-related GTPases. A, A heatmap showing the change in ubiquitination of RAS-related GTPases in HMCs expressing shGFP or shTRAF7. The scale shows log2 (fold of change) of site intensity values. n = 3 B, Ubiquitinated Flag-tagged NRAS was purified from HEK293T cells expressing the indicated constructs by Co2+ metal affinity chromatography and detected by immunoblotting. C, Ubiquitinated Flag-tagged CDC42 was purified from HEK293T cells expressing Flag-tagged wt-TRAF7 or TRAF7 mutants by Co2+ metal affinity chromatography and detected by immunoblotting. D, Ubiquitinated Flag-tagged NRAS was purified from HEK293T cells expressing Flag-tagged wt-TRAF7 or TRAF7 mutants by Co2+ metal affinity chromatography and detected by immunoblotting. E, A heatmap showing protein expression of RAS-related GTPases in HMCs expressing shGFP and shTRAF7. The scale shows Z-scored protein intensity values. n = 3 F, Immunoblotting analysis of CDC42 and NRAS protein levels in HMCs expressing either shGFP or shTRAF7. G, A heatmap showing the change in mRNA expression of RAS-related GTPases in HMCs expressing shGFP and shTRAF7. The scale shows log2 (fold of change) of TPM. n = 3. H, Global protein stability approach measuring NRAS, KRAS, HRAS, and CDC42 relative protein stability in HEK293T cells expressing shGFP or shTRAF7. Values are means ± SEM; n ≥ 4. P values are from a two-sided Student t test. I, Immunoblotting of the membrane and cytoplasmic fractions isolated from HMCs expressing shGFP or shTRAF7. B–D,n = 3. Numbers on the left are molecular masses in kDa. 2xUb, two Ub; 1xUb, one Ub; EV, empty vector; WCL, whole-cell lysate.

Close modal

The proteome analysis demonstrated that suppression of TRAF7 slightly increased expression of the RAS-related GTPases (Fig. 4E). Immunoblotting corroborated increased protein expression of panRAS and CDC42 in TRAF7-depleted HMCs, whereas we did not observe an increase of RAS-related GTPases at mRNA level (Fig. 4F and G). This suggests that TRAF7 could regulate the protein stability of RAS-like small GTPases. To test this idea, we utilized the GPS approach, a fluorescence-based system for monitoring protein stability in living cells (17, 18). We generated RAS and CDC42 stability reporter systems that contain DsRed followed by an internal ribosome entry site 2 (IRES2) and GFP-fused RAS GTPases (Supplementary Fig. S4B). The increase in GFP-RAS signals positively correlated with the presence of the proteasome inhibitor MG132, whereas the DsRed signal remained constant, indicating that the GFP/DsRed ratio reflects the stability of the RAS-related GTPases (Supplementary Fig. S4C). The GPS analysis showed that TRAF7 suppression led to increased stability of all RAS isoforms and CDC42 (Fig. 4H). Previous studies also demonstrated that ubiquitination of the RAS GTPases affects their subcellular localization (19). Consistent with these reports, we found that, in addition to increasing the stability of the RAS-related GTPases, TRAF7 depletion also led to an accumulation of RAS and CDC42 in the membrane fraction, indicating that TRAF7 controls subcellular localization of RAS and CDC42 (Fig. 4I). Altogether, these results implicate TRAF7 in the proteostatic regulation of the RAS-related small GTPases.

Live imaging implicated TRAF7 in regulating the dynamics of the actin cytoskeleton (Fig. 2C–G; Supplementary Fig. S3B), suggesting that TRAF7-mediated ubiquitination alters the activity of the RAS-like GTPases. Consistently, we found that loss of TRAF7 in meningeal cells promoted the activity of CDC42 and p21-activated kinase (PAK1; Fig. 5A and B; Supplementary Fig. S4D). TRAF7 knockdown in HMCs or CH157-MN cells also led to enhanced RAS activity and upregulation of the MAPK cascade (Fig. 5C and D; Supplementary Fig. S4E and S4F). In contrast, overexpression of wt-TRAF7, but not TRAF7 mutants, suppressed the activity of the CDC42/PAK and RAS/MAPK signaling cascades (Fig. 5E–G; Supplementary Fig. S4G and S4H). Altogether, these results indicate that TRAF7 loss of function induces the activity of both CDC42/PAK and RAS/MAPK pathways.

Figure 5.

TRAF7 loss of function promotes tumorigenic transformation by inducing CDC42 activity. A and B, Serum-starved HMCs expressing shGFP or shTRAF7 were stimulated with Meningeal Cell Medium (MenCM). The amount of GTP-bound CDC42 in human meningeal cells expressing shGFP or shTRAF7 was determined by pull down with RAC/CDC42–binding domain (PBD) conjugated to agarose beads. Data are shown as mean of GTP-bound CDC42 normalized to total CDC42 expression (A). Immunoblotting analysis using the indicated antibodies (B). C and D, Serum-starved HMCs expressing shGFP or shTRAF7 were stimulated with 2% serum. The amount of GTP-bound RAS in HMCs expressing the indicated vectors was determined by RAS GTPase Activation ELISA kit (C). Immunoblotting analysis using the indicated antibodies (D). E, Serum-starved HMCs expressing an empty vector (EV), wt-TRAF7, or TRAF7-E353* mutant were stimulated with 2% serum for the indicated time periods. The amount of GTP-bound RAS in HMCs expressing the indicated vectors was determined by RAS GTPase Activation ELISA kit. F, Serum-starved HMCs expressing an empty vector wt-TRAF7, TRAF7-T391I, or TRAF7-N520S were stimulated with 2% serum for the indicated time periods. Immunoblotting analysis using the indicated antibodies. G, Serum-starved HMCs expressing empty vector, wt-TRAF7, TRAF7-T391I, or TRAF7-N520S were stimulated with MenCM for the indicated time periods. Immunoblotting analysis using the indicated antibodies. H, AI growth of HMCs expressing an empty vector, CDC42-K163R, or CDC42-G12V mutants. I, AI growth of HMCs expressing shRNAs targeting GFP or TRAF7 and treated with DMSO or 0.1 μmol/L FRAX597 for 7 days. Values are means ± SEM; P values were detected by one-way ANOVA. n ≥ 3. J, AI growth of HMCs expressing an empty vector or ubiquitination-deficient mutant NRAS (UbDef-NRAS) and shRNAs targeting GFP or TRAF7. K, AI growth of HMCs expressing an empty vector or NRAS-Q61K. A, C, and E, Values are means ± SEM; P values were detected by two-way ANOVA. n = 3. H, J, and K, Values are means ± SEM. n ≥ 3. P values are from a two-sided Student t test.

Figure 5.

TRAF7 loss of function promotes tumorigenic transformation by inducing CDC42 activity. A and B, Serum-starved HMCs expressing shGFP or shTRAF7 were stimulated with Meningeal Cell Medium (MenCM). The amount of GTP-bound CDC42 in human meningeal cells expressing shGFP or shTRAF7 was determined by pull down with RAC/CDC42–binding domain (PBD) conjugated to agarose beads. Data are shown as mean of GTP-bound CDC42 normalized to total CDC42 expression (A). Immunoblotting analysis using the indicated antibodies (B). C and D, Serum-starved HMCs expressing shGFP or shTRAF7 were stimulated with 2% serum. The amount of GTP-bound RAS in HMCs expressing the indicated vectors was determined by RAS GTPase Activation ELISA kit (C). Immunoblotting analysis using the indicated antibodies (D). E, Serum-starved HMCs expressing an empty vector (EV), wt-TRAF7, or TRAF7-E353* mutant were stimulated with 2% serum for the indicated time periods. The amount of GTP-bound RAS in HMCs expressing the indicated vectors was determined by RAS GTPase Activation ELISA kit. F, Serum-starved HMCs expressing an empty vector wt-TRAF7, TRAF7-T391I, or TRAF7-N520S were stimulated with 2% serum for the indicated time periods. Immunoblotting analysis using the indicated antibodies. G, Serum-starved HMCs expressing empty vector, wt-TRAF7, TRAF7-T391I, or TRAF7-N520S were stimulated with MenCM for the indicated time periods. Immunoblotting analysis using the indicated antibodies. H, AI growth of HMCs expressing an empty vector, CDC42-K163R, or CDC42-G12V mutants. I, AI growth of HMCs expressing shRNAs targeting GFP or TRAF7 and treated with DMSO or 0.1 μmol/L FRAX597 for 7 days. Values are means ± SEM; P values were detected by one-way ANOVA. n ≥ 3. J, AI growth of HMCs expressing an empty vector or ubiquitination-deficient mutant NRAS (UbDef-NRAS) and shRNAs targeting GFP or TRAF7. K, AI growth of HMCs expressing an empty vector or NRAS-Q61K. A, C, and E, Values are means ± SEM; P values were detected by two-way ANOVA. n = 3. H, J, and K, Values are means ± SEM. n ≥ 3. P values are from a two-sided Student t test.

Close modal

We next validated the contribution of TRAF7-mediated ubiquitination of the RAS-like GTPases to tumorigenic transformation of HMCs. We found that either the ubiquitination-deficient CDC42-K163R, or the constitutively active CDC42-G12V mutant promoted the AI growth of HMCs (Fig. 4A; Fig. 5H). Moreover, the PAK inhibitor FRAX597 diminished AI growth of TRAF7-depleted HMCs, but did not affect the growth of HMCs expressing shGFP, indicating that activation of the CDC42/PAK pathway could contribute to TRAF7-mediated cell transformation (Fig. 5I).

We also generated ubiquitination-deficient NRAS (UbDef-NRAS) by mutating 5 lysine residues, K104, K117, K128, K135, and K147, found to be ubiquitinated in the MS analyses. Overexpression of the ubiquitination-deficient NRAS mutant promoted the AI growth of HMCs expressing shGFP, whereas the UbDef-NRAS mutant did not significantly increase the growth of TRAF7-depleted HMCs (Fig. 5J), suggesting that dysregulation of RAS ubiquitination could contribute to transformation driven by TRAF7 loss. On the other hand, the oncogenic NRAS-Q61K mutant suppressed the ability of HMCs to form colonies in soft agar (Fig. 5K). Together, these results suggest that upregulation of the RAS pathway might have a dual effect on tumorigenic transformation of HMCs.

TRAF7 loss promotes the tumor-suppressive function of KLF4

Inhibition of the AI growth of meningeal cells by the oncogenic RAS (Fig. 5K) suggests that TRAF7-mediated activation of RAS/MAPK signaling may turn on tumor-suppressive mechanisms. Given that KLF4 mutations are observed in 40% of TRAF7-mutated meningiomas and KLF4 expression is upregulated by the RAS/MAPK pathway (26, 27), we hypothesized that TRAF7 loss might promote the tumor-suppressive function of KLF4.

Consistently, we found an increased KLF4 expression in TRAF7-mutated meningiomas compared with adult meninges and non–TRAF7-mutated meningiomas (Fig. 6A). TRAF7 depletion in HMCs or CH157-MN cells also led to increased KLF4 expression (Fig. 6B and C; Supplementary Fig. S5A and S5B). Furthermore, the IPA upstream regulator analysis predicted KLF4 activation in TRAF7-depleted HMCs (activation z-score = 2.074). Inhibition of the MAPK pathway by the MEK inhibitor pimasertib abolished TRAF7-mediated KLF4 upregulation (Fig. 6D; Supplementary Fig. S5B). Moreover, KLF4 expression was induced by the NRAS-Q61K mutant, but not by the constitutively active CDC42-G12V mutant (Fig. 6E). These results indicate that KLF4 upregulation is triggered by TRAF7-mediated activation of the RAS/MAPK pathway.

Figure 6.

Cooperation of KLF4 and TRAF7 loss of function to promote tumorigenic transformation of human meningeal cells. A,KLF4 (probe ILMN_1779857) expression in adult meninges and meningioma subgroups (14). Boxplot shows the 25th, 50th, and 75th percentile of expression values with whiskers expanding to 1.5 times the interquartile range of the data. P value is from a two-sided Student t test. B, qRT-PCR analysis of KLF4 mRNA in HMCs expressing shGFP or shTRAF7. C, Immunoblotting analysis of KLF4 expression in HMCs expressing shGFP or shTRAF7. D, qRT-PCR analysis of TRAF7 and KLF4 expression in HMCs expressing shGFP or shTRAF7 treated with DMSO or pimasertib (10 μmol/L, 48 hours). Values are means ± SEM. n ≥ 3. P values are from a two-way ANOVA E, qRT-PCR analysis of KLF4 expression in HMCs expressing an empty vector (EV), CDC42-G12V, or NRAS-Q61K. F, Overlap of KLF4 target genes differentially expressed among HMCs overexpressing GFP or wt-KLF4, HMCs expressing shGFP or shTRAF7, and HBL52 expressing shGFP or shKLF4. KLF4 target genes were identified by the IPA upstream regulator analysis. G and I, qRT-PCR analysis of CRABP2 and SEMA3F in HMCs expressing the indicated vectors. J, AI growth of HMCs expressing an empty vector, CRABP2, or SEMA3F. K, Enrichment of the canonical IPA pathways in transcriptome analyses of HMCs overexpressing an empty vector or wt-KLF4; HMCs expressing shGFP or shTRAF7; HBL52 expressing shGFP or shKLF4; and KLF4/TRAF7-mutated meningiomas compared with wt-KLF4/mutant TRAF7 meningiomas. L, AI growth and the IPA upstream regulator analysis of HMCs expressing GFP, wt-KLF4, or KLF4-K409Q. M–O, AI growth of HMCs expressing the indicated constructs. gSCR, scramble gRNA; gKLF4, gRNA targeting KLF4. P, AI growth of CH157-MN cells expressing the indicated constructs. B, E, G, H, I, J, and L–P, Values are means ± SEM. n ≥ 3. P values are from a two-sided Student t test.

Figure 6.

Cooperation of KLF4 and TRAF7 loss of function to promote tumorigenic transformation of human meningeal cells. A,KLF4 (probe ILMN_1779857) expression in adult meninges and meningioma subgroups (14). Boxplot shows the 25th, 50th, and 75th percentile of expression values with whiskers expanding to 1.5 times the interquartile range of the data. P value is from a two-sided Student t test. B, qRT-PCR analysis of KLF4 mRNA in HMCs expressing shGFP or shTRAF7. C, Immunoblotting analysis of KLF4 expression in HMCs expressing shGFP or shTRAF7. D, qRT-PCR analysis of TRAF7 and KLF4 expression in HMCs expressing shGFP or shTRAF7 treated with DMSO or pimasertib (10 μmol/L, 48 hours). Values are means ± SEM. n ≥ 3. P values are from a two-way ANOVA E, qRT-PCR analysis of KLF4 expression in HMCs expressing an empty vector (EV), CDC42-G12V, or NRAS-Q61K. F, Overlap of KLF4 target genes differentially expressed among HMCs overexpressing GFP or wt-KLF4, HMCs expressing shGFP or shTRAF7, and HBL52 expressing shGFP or shKLF4. KLF4 target genes were identified by the IPA upstream regulator analysis. G and I, qRT-PCR analysis of CRABP2 and SEMA3F in HMCs expressing the indicated vectors. J, AI growth of HMCs expressing an empty vector, CRABP2, or SEMA3F. K, Enrichment of the canonical IPA pathways in transcriptome analyses of HMCs overexpressing an empty vector or wt-KLF4; HMCs expressing shGFP or shTRAF7; HBL52 expressing shGFP or shKLF4; and KLF4/TRAF7-mutated meningiomas compared with wt-KLF4/mutant TRAF7 meningiomas. L, AI growth and the IPA upstream regulator analysis of HMCs expressing GFP, wt-KLF4, or KLF4-K409Q. M–O, AI growth of HMCs expressing the indicated constructs. gSCR, scramble gRNA; gKLF4, gRNA targeting KLF4. P, AI growth of CH157-MN cells expressing the indicated constructs. B, E, G, H, I, J, and L–P, Values are means ± SEM. n ≥ 3. P values are from a two-sided Student t test.

Close modal

The most well-described KLF4 transcriptional target involved in growth inhibition is cyclin dependent kinase inhibitor 1A (CDKN1A; ref. 28). However, wt-KLF4 overexpression in meningeal cells did not induce CDKN1A expression, suggesting KLF4 regulates CDKN1A expression in a tissue-specific manner (Supplementary Fig. S5C). To elucidate the tumor-suppressive mechanism of KLF4 during meningioma development, we searched for KLF4 target genes upregulated in TRAF7-depleted or wt-KLF4–overexpressing HMCs and downregulated in KLF4-depleted HBL52 cells. HBL52 is the benign meningioma cell line harboring TRAF7-G536S mutation and wt-KLF4 (Supplementary Fig. S5D; ref. 29). Among the identified KLF4 target genes, Semaphorin 3F (SEMA3F) and cellular retinoic acid binding protein 2 (CRABP2) were dysregulated in all three datasets (Fig. 6F). qRT-PCR analysis confirmed the upregulation of CRABP2 and SEMA3F by wt-KLF4 overexpression or TRAF7 suppression in HMCs (Fig. 6G and H). In contrast, KLF4 knockdown in HBL52 cells decreased expression of CRABP2 and SEMA3F (Fig. 6I). Moreover, overexpression of SEMA3F, but not CRABP2, inhibited the AI growth of meningeal cells (Fig. 6J). These results indicate that TRAF7-mediated upregulation of KLF4 might suppress meningioma development by inducing the Semaphorin pathway. The Semaphorin pathway is an indispensable regulator of neuron-axonal guidance (30), whereas axonal guidance signaling was highly enriched in our model systems as well as in double KLF4/TRAF7-mutant meningiomas compared with wt-KLF4/mutant TRAF7 meningiomas (Fig. 6K; Supplementary Fig. S5E). The axonal guidance pathway was also the top enriched pathway in TRAF7-mutated fibroblasts of the TRAF7 syndrome patients (25) and in the proteome analysis of TRAF7-depleted HMCs (Fig. 1G; Supplementary Fig. S2), further confirming the contribution of Semaphorin signaling in meningioma development. Given that Semaphorin signaling suppresses the activity of the RAS and RHO family members, TRAF7-mediated activation of the Semaphorin pathway might represent a negative feedback loop to fine tune the GTPase activity.

TRAF7 and KLF4 cooperate to promote transformation of human meningeal cells

We next explored the cooperation between TRAF7 and KLF4 to promote meningioma development. TRAF7-mutated meningiomas harbor a recurrent KLF4-K409Q mutation, which exclusively occurs in meningiomas with extremely rare occurrences in other tumors. K409 residue in the first zinc finger domain is involved in direct contact with DNA (Fig. 1A; ref. 31). In silico modeling indicated that K409Q mutations might disrupt KLF4 interaction with DNA (Supplementary Fig. S6A), suggesting that K409Q mutation is a loss of function. The IPA upstream regulator analysis also predicted upregulation of KLF4 activity in HMCs overexpressing wt-KLF4, but not the KLF4-K409Q mutant (Fig. 6L; Supplementary Fig. S5C). KLF4-K409Q did not also induce expression of SEMA3F (Fig. 6G). Furthermore, overexpression of wt-KLF4 suppressed AI growth of HMCs, whereas the KLF4-K409Q mutant slightly promoted the ability of HMCs to grow in soft agar (Fig. 6L).

Loss-of-function KLF4 mutation might abrogate the tumor-suppressive function of wt-KLF4 induced by TRAF7 loss. Concordantly, KLF4 loss in TRAF7-depleted HMCs or TRAF7-mutated HBL52 cells promoted their AI growth, whereas such effect was not in HMCs with intact TRAF7 expression (Fig. 6M–O; Supplementary Fig. S6B–S6F). Moreover, we observed a cooperation between loss of TRAF7 and KLF4 in short-term cultured meningioma CH157-MN cells. Whereas suppression of either KLF4 or TRAF7 did not dramatically affect their growth in soft agars, suppression of both KLF4 and TRAF7 induced AI growth (Fig. 6P; Supplementary Fig. S7A and S7B). Overexpression of either TRAF7 or KLF4 partially rescued the increased AI growth of HMCs expressing a combination of shTRAF7 and shKLF4 (Supplementary Fig. S7C and S7D). Together, these results indicate that KLF4 loss of function breaks the Semaphorin-induced inhibition loop down and cooperates with the loss of TRAF7 to enhance tumorigenic transformation of meningeal cells.

Recent advances in next-generation sequencing have uncovered the mutational signature of meningiomas. Proteomic profiling and functional studies are the next critical steps in advancing our understanding of meningioma biology. Integrative analysis of TRAF7 loss of function in human meningeal cells demonstrated that dysregulation of TRAF7 activity leads to alterations in actin cytoskeleton at the posttranslational level due to the activation of RAS-related GTPases. Even though early studies linked TRAF7 to the modulation of the NF-κΒ transcription factor and TNFα-mediated activation of cellular stress pathways (32), TRAF7 lacks the TRAF homology domain responsible for its association with the cytoplasmic regions of receptors of the TNF-R superfamily. Instead of the TRAF domain, the C‐terminus of TRAF7 contains the WD40 repeats domain. The WD40 domain was reported to have a scaffolding structure known to be involved in the interaction with small GTPases (33, 34).

Similar to TRAF7-depleted meningeal cells, NF2-deficient meningioma cells also show dramatic differences in actin cytoskeletal organization (35). The NF2 protein Merlin, an ezrin-moesin-radixin family member, is a tumor suppressor that links the actin cytoskeleton with plasma membrane proteins and mediates contact-dependent cell growth inhibition. Emerging evidences indicate the interconnection between Merlin and RAS-related GTPases. RAC and CDC42 induce PAK-mediated inactivating phosphorylation of Merlin (36), whereas Merlin inhibits the activity of the RHO-related GTPases and modulates the activity of RAS signaling (37). These functional interactions between Merlin and the RAS-like GTPases indicate a feed-forward mechanism for ensuring their coordinated activation and inactivation. This strongly indicates that mutations of both NF2 and TRAF7 converge on hyperactivation of RAS-related GTPases and alterations of the actin cytoskeleton.

TRAF7 loss of function leads to dysregulation of proteostasis and activation of several RAS-related small GTPases, such as CDC42 and all RAS isoforms. Previous studies demonstrated that increased RAS stability could drive oncogenic transformation. Higher stability of RAS proteins in colorectal cancer could be due to decreased expression of the E3 ubiquitin ligase NEDD4 (38) or dysregulation of β-TrCP–mediated RAS degradation in APC-mutated tumors (39, 40). Another proteostatic regulator of RAS proteins is the leucine zipper–like transcriptional regulator 1 (LZTR1)/CUL3 ubiquitin–ligase complex that controls localization and expression levels of RAS proteins (19, 41). LZTR1 mutations are associated with Noonan syndrome, glioblastoma, and schwannomatosis (42–46). Interestingly, the TRAF7 syndrome resulting from germline TRAF7 mutations phenotypically overlaps with RASopathy disorders (24, 25).

On the other hand, dysregulation of X-linked inhibitor of apoptosis protein (XIAP)-dependent CDC42 ubiquitination enhances filopodia formation and promotes lung colonization of tumor cells in mice in a Cdc42-dependent manner (47). Of note, activating mutations in CDC42 are mutually exclusive with TRAF7 mutations in differentiated papillary mesotheliomas (23), suggesting that TRAF7 and CDC42 could act within the same signaling network. Altogether, these studies highlight the importance of ubiquitination in the regulation of small GTPase.

Whereas activation of CDC42 in meningeal cells promotes AI growth, RAS activation leads to paradoxical growth inhibition. RAS-mediated induction of KLF4 and activation of Semaphorin signaling appear to be responsible for this growth inhibition. Semaphorins have been found to inhibit proliferation and metastatic spread of tumor cells as well as suppress tumor progression by modulating tumor microenvironment (48, 49). Semaphorins transduce the signal through their interaction with members of the plexin and neuropilin families of transmembrane receptors. A hallmark of Semaphorin receptors is their ability to interact directly with small GTPases of the RAS and RHO families through their intracellular region and to act as GTPase-activating proteins (GAP). This suggests that TRAF7-mediated activation of the Semaphorin pathway might represent a negative feedback loop to fine tune the GTPase activity, whereas KLF4 mutation might breakdown this feedback to fully activate small GTPases.

The emerging molecular knowledge of meningioma development provides a tremendous opportunity to explore novel therapeutic strategies for patients with meningioma. Our results suggest that treating meningiomas with MEK or BRAF inhibitors could moderate the tumor-suppressive function of KLF4. On the other hand, the use of PAK inhibitors, such as FRAX597, could be beneficial for treating TRAF7-mutated meningiomas. PAKs have increasingly attracted attention as potential therapeutic targets due to their central roles in oncogenic signaling pathways. Both pan-PAK inhibitors and group-specific PAK inhibitors have been extensively tested for their anticancerous efficacy in different cancer models (50). Thus, integrated omics studies that outline clinically important genetic changes of meningioma will facilitate comprehensive and personalized care of patients with meningioma.

No disclosures were reported.

P. Najm: Data curation, investigation, methodology, writing–review and editing. P. Zhao: Data curation, software, investigation, methodology, writing–review and editing. M. Steklov: Software, investigation, methodology, writing–review and editing. R.N. Sewduth: Investigation. M.F. Baietti: Investigation, methodology, writing–review and editing. S. Pandolfi: Investigation, methodology, writing–review and editing. N. Criem: Investigation, methodology, writing–review and editing. B. Lechat: Investigation. T.M. Maia: Formal analysis, investigation, methodology, writing–review and editing. D. Van Haver: Data curation, formal analysis, investigation, visualization. N. Corhout: Software, formal analysis, supervision, investigation, visualization, methodology. S. Eyckerman: Data curation, formal analysis, supervision, investigation, visualization, methodology. F. Impens: Data curation, formal analysis, supervision, investigation, visualization. A.A. Sablina: Data curation, supervision, methodology, visualization, formal analysis, writing–review and editing.

The work was supported by H2020 European Research Council (ub-RASDisease, 772649).

The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Bi
WL
,
Zhang
M
,
Wu
WW
,
Mei
Y
,
Dunn
IF
. 
Meningioma genomics: diagnostic, prognostic, and therapeutic applications
.
Front Surg
2016
;
3
:
40
.
2.
Preusser
M
,
Brastianos
PK
,
Mawrin
C
. 
Advances in meningioma genetics: novel therapeutic opportunities
.
Nat Rev Neurol
2018
;
14
:
106
15
.
3.
Yuzawa
S
,
Nishihara
H
,
Tanaka
S
. 
Genetic landscape of meningioma
.
Brain Tumor Pathol
2016
;
33
:
237
47
.
4.
Zotti
T
,
Scudiero
I
,
Vito
P
,
Stilo
R
. 
The emerging role of TRAF7 in tumor development
.
J Cell Physiol
2017
;
232
:
1233
8
.
5.
Klein
CJ
,
Wu
Y
,
Jentoft
ME
,
Mer
G
,
Spinner
RJ
,
Dyck
PJ
, et al
Genomic analysis reveals frequent TRAF7 mutations in intraneural perineuriomas
.
Ann Neurol
2017
;
81
:
316
21
.
6.
Goode
B
,
Joseph
NM
,
Stevers
M
,
Van Ziffle
J
,
Onodera
C
,
Talevich
E
, et al
Adenomatoid tumors of the male and female genital tract are defined by TRAF7 mutations that drive aberrant NF-kB pathway activation
.
Mod Pathol
2018
;
31
:
660
73
.
7.
Bueno
R
,
Stawiski
EW
,
Goldstein
LD
,
Durinck
S
,
De Rienzo
A
,
Modrusan
Z
, et al
Comprehensive genomic analysis of malignant pleural mesothelioma identifies recurrent mutations, gene fusions and splicing alterations
.
Nat Genet
2016
;
48
:
407
16
.
8.
Hung
YP
,
Dong
F
,
Dubuc
AM
,
Dal Cin
P
,
Bueno
R
,
Chirieac
LR
. 
Molecular characterization of localized pleural mesothelioma
.
Mod Pathol
2020
;
33
:
271
80
.
9.
Zammarchi
F
,
Morelli
M
,
Menicagli
M
,
Di Cristofano
C
,
Zavaglia
K
,
Paolucci
A
, et al
KLF4 is a novel candidate tumor suppressor gene in pancreatic ductal carcinoma
.
Am J Pathol
2011
;
178
:
361
72
.
10.
Yu
F
,
Li
J
,
Chen
H
,
Fu
J
,
Ray
S
,
Huang
S
, et al
Kruppel-like factor 4 (KLF4) is required for maintenance of breast cancer stem cells and for cell migration and invasion
.
Oncogene
2011
;
30
:
2161
72
.
11.
Yu
T
,
Chen
X
,
Zhang
W
,
Liu
J
,
Avdiushko
R
,
Napier
DL
, et al
KLF4 regulates adult lung tumor-initiating cells and represses K-Ras-mediated lung cancer
.
Cell Death Differ
2016
;
23
:
207
15
.
12.
Tang
H
,
Zhu
H
,
Wang
X
,
Hua
L
,
Li
J
,
Xie
Q
, et al
KLF4 is a tumor suppressor in anaplastic meningioma stem-like cells and human meningiomas
.
J Mol Cell Biol
2017
;
9
:
315
24
.
13.
Clark
VE
,
Erson-Omay
EZ
,
Serin
A
,
Yin
J
,
Cotney
J
,
Ozduman
K
, et al
Genomic analysis of non-NF2 meningiomas reveals mutations in TRAF7, KLF4, AKT1, and SMO
.
Science
2013
;
339
:
1077
80
.
14.
Clark
VE
,
Harmancı
AS
,
Bai
H
,
Youngblood
MW
,
Lee
TI
,
Baranoski
JF
, et al
Recurrent somatic mutations in POLR2A define a distinct subset of meningiomas
.
Nat Genet
2016
;
48
:
1253
9
.
15.
Boehm
JS
,
Zhao
JJ
,
Yao
J
,
Kim
SY
,
Firestein
R
,
Dunn
IF
, et al
Integrative genomic approaches identify IKBKE as a breast cancer oncogene
.
Cell
2007
;
129
:
1065
79
.
16.
Simicek
M
,
Lievens
S
,
Laga
M
,
Guzenko
D
,
Aushev
VN
,
Kalev
P
, et al
The deubiquitylase USP33 discriminates between RALB functions in autophagy and innate immune response
.
Nat Cell Biol
2013
;
15
:
1220
30
.
17.
Yen
HC
,
Xu
Q
,
Chou
DM
,
Zhao
Z
,
Elledge
SJ
. 
Global protein stability profiling in mammalian cells
.
Science
2008
;
322
:
918
23
.
18.
Yen
HC
,
Elledge
SJ
. 
Identification of SCF ubiquitin ligase substrates by global protein stability profiling
.
Science
2008
;
322
:
923
9
.
19.
Steklov
M
,
Pandolfi
S
,
Baietti
MF
,
Batiuk
A
,
Carai
P
,
Najm
P
, et al
Mutations in LZTR1 drive human disease by dysregulating RAS ubiquitination
.
Science
2018
;
362
:
1177
82
.
20.
Eyckerman
S
,
Titeca
K
,
Van Quickelberghe
E
,
Cloots
E
,
Verhee
A
,
Samyn
N
, et al
Trapping mammalian protein complexes in viral particles
.
Nat Commun
2016
;
7
:
11416
.
21.
Bassiri
K
,
Ferluga
S
,
Sharma
V
,
Syed
N
,
Adams
CL
,
Lasonder
E
, et al
Global proteome and phospho-proteome analysis of merlin-deficient meningioma and schwannoma identifies PDLIM2 as a novel therapeutic target
.
EBioMedicine
2017
;
16
:
76
86
.
22.
Huttlin
EL
,
Bruckner
RJ
,
Paulo
JA
,
Cannon
JR
,
Ting
L
,
Baltier
K
, et al
Architecture of the human interactome defines protein communities and disease networks
.
Nature
2017
;
545
:
505
9
.
23.
Stevers
M
,
Rabban
JT
,
Garg
K
,
Van Ziffle
J
,
Onodera
C
,
Grenert
JP
, et al
Well-differentiated papillary mesothelioma of the peritoneum is genetically defined by mutually exclusive mutations in TRAF7 and CDC42
.
Mod Pathol
2019
;
32
:
88
99
.
24.
Tokita
MJ
,
Chen
CA
,
Chitayat
D
,
Macnamara
E
,
Rosenfeld
JA
,
Hanchard
N
, et al
De novo missense variants in TRAF7 cause developmental delay, Congenital Anomalies, and Dysmorphic Features
.
Am J Hum Genet
2018
;
103
:
154
62
.
25.
Castilla-Vallmanya
L
,
Selmer
KK
,
Dimartino
C
,
Rabionet
R
,
Blanco-Sánchez
B
,
Yang
S
, et al
Phenotypic spectrum and transcriptomic profile associated with germline variants in TRAF7
.
Genet Med
2020
;
22
:
1215
26
.
26.
Kim
MO
,
Kim
SH
,
Cho
YY
,
Nadas
J
,
Jeong
CH
,
Yao
K
, et al
ERK1 and ERK2 regulate embryonic stem cell self-renewal through phosphorylation of Klf4
.
Nat Struct Mol Biol
2012
;
19
:
283
90
.
27.
Riverso
M
,
Montagnani
V
,
Stecca
B
. 
KLF4 is regulated by RAS/RAF/MEK/ERK signaling through E2F1 and promotes melanoma cell growth
.
Oncogene
2017
;
36
:
3322
33
.
28.
Rowland
BD
,
Peeper
DS
. 
KLF4, p21 and context-dependent opposing forces in cancer
.
Nat Rev Cancer
2006
;
6
:
11
23
.
29.
Mei
Y
,
Bi
WL
,
Greenwald
NF
,
Agar
NY
,
Beroukhim
R
,
Dunn
GP
, et al
Genomic profile of human meningioma cell lines
.
PLoS One
2017
;
12
:
e0178322
.
30.
Kolodkin
AL
,
Matthes
DJ
,
Goodman
CS
. 
The semaphorin genes encode a family of transmembrane and secreted growth cone guidance molecules
.
Cell
1993
;
75
:
1389
99
.
31.
Schuetz
A
,
Nana
D
,
Rose
C
,
Zocher
G
,
Milanovic
M
,
Koenigsmann
J
, et al
The structure of the Klf4 DNA-binding domain links to self-renewal and macrophage differentiation
.
Cell Mol Life Sci
2011
;
68
:
3121
31
.
32.
Zotti
T
,
Vito
P
,
Stilo
R
. 
The seventh ring: exploring TRAF7 functions
.
J Cell Physiol
2012
;
227
:
1280
4
.
33.
Schapira
M
,
Tyers
M
,
Torrent
M
,
Arrowsmith
CH
. 
WD40 repeat domain proteins: a novel target class?
Nat Rev Drug Discov
2017
;
16
:
773
86
.
34.
Shen
W
,
Kilander
MBC
,
Bridi
MS
,
Frei
JA
,
Niescier
RF
,
Huang
S
, et al
Tomosyn regulates the small RhoA GTPase to control the dendritic stability of neurons and the surface expression of AMPA receptors
.
J Neurosci Res
2020
;
98
:
1213
31
.
35.
James
MF
,
Lelke
JM
,
Maccollin
M
,
Plotkin
SR
,
Stemmer-Rachamimov
AO
,
Ramesh
V
, et al
Modeling NF2 with human arachnoidal and meningioma cell culture systems: NF2 silencing reflects the benign character of tumor growth
.
Neurobiol Dis
2008
;
29
:
278
92
.
36.
McClatchey
AI
. 
Merlin and ERM proteins: unappreciated roles in cancer development?
Nat Rev Cancer
2003
;
3
:
877
83
.
37.
Cui
Y
,
Groth
S
,
Troutman
S
,
Carlstedt
A
,
Sperka
T
,
Riecken
LB
, et al
The NF2 tumor suppressor merlin interacts with Ras and RasGAP, which may modulate Ras signaling
.
Oncogene
2019
;
38
:
6370
81
.
38.
Zeng
T
,
Wang
Q
,
Fu
J
,
Lin
Q
,
Bi
J
,
Ding
W
, et al
Impeded Nedd4–1-mediated Ras degradation underlies Ras-driven tumorigenesis
.
Cell Rep
2014
;
7
:
871
82
.
39.
Kim
SE
,
Yoon
JY
,
Jeong
WJ
,
Jeon
SH
,
Park
Y
,
Yoon
JB
, et al
H-Ras is degraded by Wnt/beta-catenin signaling via beta-TrCP-mediated polyubiquitylation
.
J Cell Sci
2009
;
122
:
842
8
.
40.
Moon
BS
,
Jeong
WJ
,
Park
J
,
Kim
TI
,
dS
M
,
Choi
KY
. 
Role of oncogenic K-Ras in cancer stem cell activation by aberrant Wnt/β-catenin signaling
.
J Natl Cancer Inst
2014
;
106
:
djt373
.
41.
Bigenzahn
JW
,
Collu
GM
,
Kartnig
F
,
Pieraks
M
,
Vladimer
GI
,
Heinz
LX
, et al
LZTR1 is a regulator of RAS ubiquitination and signaling
.
Science
2018
;
362
:
1171
7
.
42.
Paganini
I
,
Chang
VY
,
Capone
GL
,
Vitte
J
,
Benelli
M
,
Barbetti
L
, et al
Expanding the mutational spectrum of LZTR1 in schwannomatosis
.
Eur J Hum Genet
2015
;
23
:
963
8
.
43.
Piotrowski
A
,
Xie
J
,
Liu
YF
,
Poplawski
AB
,
Gomes
AR
,
Madanecki
P
, et al
Germline loss-of-function mutations in LZTR1 predispose to an inherited disorder of multiple schwannomas
.
Nat Genet
2014
;
46
:
182
7
.
44.
Smith
MJ
,
Isidor
B
,
Beetz
C
,
Williams
SG
,
Bhaskar
SS
,
Richer
W
, et al
Mutations in LZTR1 add to the complex heterogeneity of schwannomatosis
.
Neurology
2015
;
84
:
141
7
.
45.
Yamamoto
GL
,
Aguena
M
,
Gos
M
,
Hung
C
,
Pilch
J
,
Fahiminiya
S
, et al
Rare variants in SOS2 and LZTR1 are associated with Noonan syndrome
.
J Med Genet
2015
;
52
:
413
21
.
46.
Johnston
JJ
,
van der Smagt
JJ
,
Rosenfeld
JA
,
Pagnamenta
AT
,
Alswaid
A
,
Baker
EH
, et al
Autosomal recessive Noonan syndrome associated with biallelic LZTR1 variants
.
Genet Med
2018
;
20
:
1175
85
.
47.
Murali
A
,
Shin
J
,
Yurugi
H
,
Krishnan
A
,
Akutsu
M
,
Carpy
A
, et al
Ubiquitin-dependent regulation of Cdc42 by XIAP
.
Cell Death Dis
2017
;
8
:
e2900
.
48.
Capparuccia
L
,
Tamagnone
L
. 
Semaphorin signaling in cancer cells and in cells of the tumor microenvironment–two sides of a coin
.
J Cell Sci
2009
;
122
:
1723
36
.
49.
Neufeld
G
,
Mumblat
Y
,
Smolkin
T
,
Toledano
S
,
Nir-Zvi
I
,
Ziv
K
, et al
The role of the semaphorins in cancer
.
Cell Adh Migr
2016
;
10
:
652
74
.
50.
Radu
M
,
Semenova
G
,
Kosoff
R
,
Chernoff
J
. 
PAK signalling during the development and progression of cancer
.
Nat Rev Cancer
2014
;
14
:
13
25
.