Valosin-Containing Protein Contributes to Plexiform Neurofibroma Formation and Represents a Novel Therapeutic Target

1. Introduction

Neurofibromatosis type 1 (NF1) is classified as rare, but of the rare diseases is relatively common, affecting 1 in 2500 people worldwide [1,2,3]. One of the hallmarks is that about 30-50% of NF1 patients develop plexiform neurofibromas (PNFs), a benign Schwann cell (SC) tumor. PNFs can transform into malignant peripheral nerve sheath tumors, a lethal sarcoma [4,5,6]. PNFs can be large and locally invasive, and can cause substantial morbidity—including pain, neurologic deficit, motor dysfunction, disfiguration, and even death. The mainstay of PNF therapy, surgery, is sometimes impossible due to tumor integration into nerves in critical anatomic sites. There are two FDA-approved MEK inhibitors (MEKi), selumetinib and mirdametinib, for the treatment of PNFs. These two inhibitors shrink PNFs in 40-70% of NF1 patients, but tumor shrinkage is modest in most tumors, and some tumors regrow when drug is withdrawn [7,8,9]. In addition, patients receiving MEKi treatment often experience side effects—including elevated creatine phosphokinase, acneiform rash, and paronychia [7,8]. Finally, given that PNF onset is during early childhood [10], taking selumetinib or mirdametinib for life can be problematic due to adherence and side effects issues. Therefore, new effective and durable therapeutic strategies are urgently needed.

Valosin-containing protein (VCP, also called p97) is a homohexamer consisting of an N-terminal domain, two ATPase (D1 and D2) domains, a flexible linker region, and a C-terminal domain. VCP is an AAA+ ATPase that, by interaction with a wide variety of partners and cofactors, regulates proteostasis and multiple other cellular processes (e.g., lysosomal dynamic stability) [11,12,13]. VCP, usually through its ATPase activity, regulates ubiquitination [13,14]. VCP also regulates autophagy [15,16]. In neurons, VCP interacts directly with neurofibromin, the NF1 gene encoding protein [17], and heterozygous Nf1 (Nf1^+/-) neurons regulate protein synthesis in endoplasmic reticulum (ER) via VCP [15]. In addition to regulating proteostasis, VCP also suppresses apoptosis by unfold protein response in normal and tumor cells [18,19,20,21,22]. Importantly, in many human tumors, VCP mRNA and protein are elevated, and multiple in vivo studies strongly suggest that VCP promotes tumor growth [13,23,24].

To develop effective and durable therapies for PNF, it is critical to understand the signaling pathways controlling PNF formation and growth — particularly the MEK-independent or MEK synergistic pathways. In this study, we showed that VCP interacts with neurofibromin in SCs and its protein was overexpressed in both mouse and human PNFs. Treatment of primary mouse PNF SCs with CB-5083, a p97/VCP ATPase inhibitor (VCPi), led to accumulation of poly-ubiquitinated proteins and generation of irresolvable proteotoxic stress. Pharmacological or genetic inhibition of VCP reduced mouse neurofibroma-sphere number, and genetic inhibition of Vcp in Schwann cell precursors (SCPs) decreased tumor-like lesion numbers in a cell transplantation model. In vivo treatment with CB-5083 on the Nf1^fl/fl;DhhCre mouse of PNF significantly inhibited cell proliferation, increased cell apoptosis and reduced neurofibroma volume, suggesting targeting VCP might be an alternative therapeutic strategy for patients with PNFs.

2. Materials and Methods

2.1. Animals

The animal care and use committees of Cincinnati Children’s Hospital Medical Center approved all animal procedures. Institutional Animal Care and Use Committee (IACUC) guidelines were followed with animal subjects. Mice were housed in temperature- and humidity- controlled facilities on 12-hour dark-light cycles with free access to food and water. The floxed -Nf1 mouse (Nf1^fl/fl) has been described previously [25] and was on a mixed 129/BL/6 background. The DhhCre transgenic mouse harbors a nuclear-localized Cre recombinase gene under control of the mouse desert hedgehog (Dhh) promoter/regulatory regions and was maintained on C57/BL/6 background [26]. We bred DhhCre mice onto Nf1^fl/fl mice to obtain the F1 generation (Nf1^fl/+;DhhCre⁺); we bred male F1 mice with female Nf1^fl/fl mice to obtain Nf1^fl/fl;DhhCre mice. Genotyping was performed as described previously [27].

2.2. Mouse MRI and Volumetric Measurements

Mouse MRI and volumetric measurements were performed as described [14]. Specifically, Nf1^fl/fl;DhhCre PNF-bearing mice were scanned by MRI at 5, 7 and 9 months of age to monitor PNFs using 7T Bruker Biospec system equipped with 400 G/cm gradients in the Imaging Resource Center at Cincinnati Children’s Hospital. MRI volumetric measurements were used to determine the tumor volume change as described previously [28]. All volumes were reported as combined tumor volume in an individual mouse. Mouse treatment groups were blind to the person who performed the volumetric measurements.

2.3. Reagents

CB-5083 (ChemGood, Henrico, VA, USA) and PD0325901 (mirdametinib, Selleck Chemicals, Houston, TX, USA) were dissolved in 0.5% [w/v] methylcellulose (Neta Scientific, Hainesport, NJ, USA; #SIAL-M0262) solution with 0.2% [v/v] polysorbate 80 [Tween 80] (Fisher Scientific, Waltham, MA, USA; # BP338-500) for mouse dosing. CB-5083 was sonicated at medium level energy for 2 minutes on ice before use.

2.4. Immunoprecipitation Assay

Passage two embryonic day 12.5 WT mouse SCs (1x10⁸) were lysed in Np-40 buffer as described [29], and 750 µg of the total protein lysate was used for immunoprecipitation with a neurofibromin antibody (Santa Cruz Biotechnology, Dallas, TX, USA; # sc-398267, 1:200) or normal mouse IgG control Ab (Santa Cruz Biotechnology; #sc-2025, 1:200). The immunoprecipitated complexes were separated by SDS-PAGE and transferred to PVDF membranes for immunoblotting with detection anti-NF1 (Cell Signaling Technology, Danvers, MA, USA; #14623, 1:1000) and VCP (Proteintech, Rosemont, IL, USA; # 10736-1-AP, 1:2000) antibodies.

2.5. Western Blots

Mouse spheres, PNFs or WT mouse dorsal root ganglia (DRG)/nerve lysates were used for western blots as described [30]. For immunoblotting, we used the following antibodies: VCP (Proteintech; # 10736-1-AP, 1:2000); pERK (Cell Signaling Technology; #4370, 1:2000), ERK (Cell Signaling Technology; #4695, 1:2000),Ubiquitin (Cell Signaling Technology; #58395, 1:1000), CHOP (Cell Signaling Technology; #5554, 1:1000), GAPDH (Cell Signaling Technology; #3683, 1:3000), or β-actin (Cell Signaling Technology; #5125, 1:5000). Antibody binding to the membrane was visualized using a chemiluminescent detection system (EMD Millipore, Burlington, MA, USA; #WBKLS0500).

2.6. Generation of shRNAs and Lentiviral Transduction

We purchased two VCP specific shRNA-expressing lentiviral plasmids (TRCN0000008410 [#10] and TRCN0000008413[#11]) from Sigma-Aldrich (St. Louis, MO, USA) and produced lentivirus particles by transient co-transfection of 293T cells as described [31]. After concentration of the lentiviral particles using a commercial Lenti-X concentrator (Takara Bio Inc, Shiga, Japan; # 631231), we determined the titer of the concentrated lentivirus.

2.7. Mouse PNF Cell-Derived Sphere Culture and Treatment

Mouse PNF cell-derived sphere culture was performed as described [30,32]. Briefly, we cut PNFs into around 1 mm³ piece and dissociated them in 20 ml L-15 (Mediatech, Herndon, VA, USA; #10-045-CV ) containing 0.5 mg/ml collagenase type 1 (Worthington Biochemical, Lakewood, NJ, USA; #LS004196) and 2.5 mg/ml Dispase protease type II (Roche, Indianapolis, IN,USA; #C756V28) at 37 °C for 3 hours with 200rpm shaking. We plated trypan blue negative single cells at 4 x 10⁴ cells per well in 24-well low-binding plates (Fisher Scientific, # Corning 3473) with 1mL medium containing DMEM:F-12 (3:1) containing 20 ng/ml rhEGF (R&D Systems, Minneapolis, MN,USA; # 236-EG-200), 20 ng/ml rh bFGF (PeproTech, Cranbury, NJ; #450-33-50µg), 1% B-27 supplement (Life Technologies, Carlsbad, CA, USA; #17504-044), and 2 mg/ml heparin (Sigma-Aldrich; #H3149-10KU). Spheres were cultured at 37 °C at 5% CO₂. We used secondary spheres for each experiment.

For treatment, we seeded dissociated live cells from primary spheres into 24-well ultra-low attachment plates at a density of 4×10⁴ cells/well for 48-72 hours and then treated them with vehicle, CB-5083, shVcp or shNT lentivirus. For drug treatment, CB-5083 at 0nM, 1nM, 10nM, 100nM, 1,000nM, and 10,000nM were used. After 4 days, the sphere numbers were counted using a phase-contrast microscopy (Olympus CKX41, Tokyo, Japan). For lentivirus treatment, we transduced secondary PNF cell- derived spheres with purified shVcp or shNT (Sigma-Aldrich) at the multiplicity of infection of 1:10 for 3 days and collect sphere for western blot, and at 4 days for sphere number counting.

2.8. Tumorigenesis Assay in NSG Mice

We treated Nf1^fl/fl;DhhCre mouse PNF- derived secondary spheres with shNT or shVcp (#11) lentivirus at multiplicity of infection of 1:10 for 3 days and dissociated them into single cells. We subcutaneously injected 5 × 10⁵ dissociated sphere cells/injection within 33% Matrigel (Fisher Scientific; #CB-40234) into right flanks of NSG mice (males and females, Harlan, Indianapolis, IN) as described [30]. After 3 months, we dissected lesions under a dissection microscope to confirm visible tumors and counted tumor number.

2.9. Quantitative Real-Time PCR (qRT-PCR)

Total RNA was isolated from mouse tissues using the Qiagen RNeasy Kit (Hilden, Germany). The cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Thermo Scientific, Waltham, MA, USA; #4368814). QRT-PCR was performed using SYBR Green (Applied Biosystems, Foster City, CA, USA; #4368708). The ΔΔCt method was used for calculation. Data represented three independent experiments with triplicate in each sample. The following mouse primers were used: Svip: forward: 5’-GCTTCCAGACTTAGTGCCTTTA-3’, reverse; 5’-CGTCCTACGGACATGTAACTT-3’. Tublin: forward: 5’-CATTCAGGGCTCCATCAAAT-3’, reverse :5’-GCCCTACAACTCCATCCTCA-3’

2.10. Immunohistochemistry and Immunofluorescence

Paraformaldehyde -fixed, paraffin-embedded tissue blocks were sectioned at 10 μm. Tissue sections were de-paraffinized in xylene, rehydrated in serial concentrations of ethanol, and then boiled in citrate buffer (pH 6.0) for antigen retrieval. Tissues were permeabilized with 0.2% triton-X100/PBS (TBST) for 30 minutes and washed with PBS three times. After blocking with 10% normal goat serum (Jackson ImmunoResearch, West Grove, PA, USA; #005-000-121) in TBST, the sections were incubated with primary antibodies: Ki67 (Cell Signaling Technology, # 12202S, 1:200), VCP (Proteintech, # 10736-1-AP, 1:200), and SVIP (Abcam, Cambridge, MA, USA; #ab122590, 1:250) overnight at 4 °C and then treated with biotinylated or Fluorescence dye-conjugated secondary antibodies for 1 hr. The biotinylated antibody was detected with horseradish peroxidase-conjugated streptavidin (Elite ABC, Vector Laboratories, Burlingame, CA, USA; #PK-6100) and cell nuclei were stained with hematoxylin. Fluorescently labeled sections were stained with 0.1 μg/ml DAPI (Fisher Scientific; #62248) for nuclei counting and then mounted in Fluoromount G slide mounting medium (Electron Microscopy Sciences, Hatfield, PA, USA; 17984-25). Images were captured on a Nikon C2 Confocal microscope.

2.11. TUNEL Assay

TUNEL assay was performed according to Roche In situ cell death detection kit instruction (TMR red, Roche Diagnostics Gmbh, Mannheim, Germany; Ref# 12156792910), on deparaffinized sections. We stained nuclei with DAPI. We counted TUNEL⁺ cells and DAPI nuclei in at least three cross sections per sample. Data were presented as an average percentage of TUNEL⁺ cells per sample.

2.12. Statistical Analysis

We used unpaired 2-tailed Student’s t-tests to analyze the significance of cell population percentages or relative mRNA expression when two samples were compared. We used one-way ANOVA to analyze the significance of cell proliferation, apoptosis and sphere numbers per well. Data was reported as the mean ± SD in at least three independent experiments or samples. P<0.05 was considered significant. We used a mixed-effects model to analyze PNF growth. P values were generated with a random-effects model analysis on log-transformed tumor volume data using the SAS mixed procedure [33]. For the in vivo cell transplantation experiment, we used Fisher’s exact test for p-value.

3. Results

3.1. VCP Interacts with Neurofibromin in SCs and Is Overexpressed in Mouse and Human PNFs

We previously showed that ER stress protein homoeostasis is unbalanced in PNF SC/SCPs [29]. Given the increased protein synthesis and activation of ER stress pathways observed in PNF cells, we decided to examine upstream components of the ER proteostasis machinery. By comparing PNF RNA sequencing and PNF SC RUNX1-CHIP sequencing data on mouse PNF, we found that transcript encoding the NF1 interacting p97/VCP gene was overexpressed in PNF. To determine if VCP interacts with neurofibromin in SCs, we performed immunoprecipitation on mouse WT SCs. Western blot confirmed that anti-NF1 immunoprecipates contained VCP while was not present in IgG control (Figure 1A), indicating VCP specifically interacts with neurofibromin in SCs. To check if VCP expression is altered upon loss of Nf1, we performed western blots on PNFs from 7-month-old Nf1^fl/fl;DhhCre mice and age matched WT DRG. We detected increased expression of VCP in PNFs compared to WT mouse DRG (Figure 1B). Immunofluorescence staining on human PNF showed that there was significantly increased VCP expression cells compared to normal human nerve controls (Figure 1C, 1D). These data suggests that VCP is present in SCs.

3.2. RUNX1 Binding to SVIP Indirectly Regulates VCP Expression

We decided to determine how VCP expression is increased in PNFs.

We previously showed that RUNX1 contributes to PNF formation [34]. Western blot of VCP on Nf1^fl/fl;DhhCre and Runx1^fl/fl;Runx3^fl/fl;Nf1^fl/fl;DhhCre (Runx1/3-cKO; Nf1^fl/fl;DhhCre) mouse PNFs showed decreased VCP protein expression in Runx1/3-cKO; Nf1^fl/fl;DhhCre (Figure 2A). Because RUNX1 has not been noted to interact with VCP, we explored potential downstream targets of RUNX1 that might interact with VCP, by integrating 3 published data sets: Runx1/3-cKO; Nf1^fl/fl;DhhCre RNAseq differential gene expression [34], Runx1 CHIP-seq peaks [34]. and known VCP-interacting proteins [12]. The small VCP interacting protein (Svip) was the only gene common to all 3 data sets (Figure 2B). This suggests that loss of NF1 increases RUNX1 expression, which in turn promotes RUNX1 binding to SVIP, thereby indirectly regulating VCP expression. In fact, Svip mRNA expression was significantly increased in Runx1/3-cKO; Nf1^fl/fl;DhhCre PNFs (vs Nf1^fl/fl;DhhCre) (Figure 2C), and increased relative mRNA expression data was validated by qRT-PCR (Figure 2D). We were unbale to identify an anti-SVIP antibody for immunofluorescence staining in mouse PNF. Therefore, we checked SVIP expression levels in human PNFs. Consistent with the anti-correlated expression of SVIP and VCP [11], immunofluorescence staining on human PNFs showed decreased SVIP protein expression compared with normal human nerve controls (Figure 2E).

3.3. Overwhelming Proteotoxic Stress by Targeting VCP Inhibits Cell Growth and Induces Protein Ubiquitination in PNF SCs

Since the anti-tumor activity of MEK inhibitors (PD0325901 or selumetinib) is not cytotoxic and is not durable post-treatment [7,8,35,36], we tested whether targeting VCP is cytotoxic for NF1 PNF cells. The D2 domain of VCP is the primary driver of ATPase activity and conformational changes, and CB-5083 is a selective VCP D2 domain inhibitor [37]. CB-5083 treatment led to a dramatic dose-dependent growth inhibition in PNF-derived SCs compared to WT SCs; importantly, the cytotoxic effect on WT SCs was relatively modest (Figure 3A). Similarly, protein ubiquitination increased in CB-5083-treated PNF derived SCs 6 hours after treatment and was also dose dependent (Figure 3B); at 500 nM the difference in protein ubiquitination between WT and PNF cells was most dramatic, suggesting a potential therapeutic dose window. Interestingly, CB-5083 did not change p-ERK expression (an MAPK inhibition readout) even at 700nM, suggesting that VCP might be MEK/ERK pathway independent or regulated by MEK/ERK pathway. To verify that CB-5083 selectively affects the VCP ATPase activity in this setting, we performed ATPase quantification assays on CB-5083 treated mouse WT and PNF SCs. ATPase activity inhibition was similar in both 0.5µm and 1µM CB-5083 treated WT SCs. On the contrary, ATPase activity was significantly reduced in 1µM CB-5083 treated PNF SCs compared to vehicle control (Figure 3C).

We next tested if VCPi + MEKi combination treatment leads to additive effects. We treated WT and PNF SCs with control, PD0325901 (1µM), CB-5083 (0.5µM), or the combination for 6 hours. Western blot showed that the combination treatment inhibited p-ERK expression and induced ubiquitination simultaneously; MEKi reduced p-ERK expression while CB-5083 induced protein ubiquitination. Consistently, CB-5083 +/- PD0325901 showed stronger effects on protein ubiquitination in PNF SCs vs WT SCs (Figure 3D).

3.4. Pharmacological and Genetic Inhibition of VCP Decreases Mouse PNF Cell-Derived Sphere Number In Vitro

Schwann cell precursors (SCPs) are the cell of origin for PNF. To test if inhibition of VCP affects SCP growth and/or tumorigenesis, we used a neurofibroma sphere culture system, in which SCPs grow as self-renewing spheres that can be passaged in vitro for 2-3 passages [38]. CB-5083 inhibited sphere formation (numbers) in a dose-dependent manner (Figure 4A). Eexposure (4 days) to CB-5083 at 100nM led accumulation of protein ubiquitination (Figure 4B), which was not observed after short term (6 hours) treatment at this dose (Figure 3B). To verify that this inhibitory effect was specifically caused by VCP, we treated Nf1^fl/fl;DhhCre mouse PNF SC-derived secondary spheres with two different Vcp sh-RNAs (shVcp#10 and shVcp#11) for 4 days. Each shVcp significantly decreased sphere numbers compared to shNT control (Figure 4C). VCP knockdown was confirmed by western blot (Figure 4D). These results indicates that VCP is important for SCP growth.

3.5. VCP Contributes to Neurofibroma Initiation

To test whether VCP reduction in neurofibroma sphere numbers decreases tumorigenic potential in vivo, we dissociated 3-day shVcp#11 or shNT lentivirus-treated Nf1^fl/fl;DhhCre mouse PNF-derived secondary tumor sphere cells into single cell and injected them subcutaneously into NSG mice. Three months after transplantation, all mice (10 of 10) grafted with shNT-treated spheres showed neurofibroma-like micro-lesions, while only 5 of 10 NSG mice showed detectable lesions in mice transplanted with shVcp-treated sphere cells (p< 0.05) (Figure 4F). Thus, VCP loss compromised neurofibroma initiation in this assay. Global Vcp knockout mice die during embryonic development [39]. We tried to delete VCP in SCc and SCPs using conditional VCP^fl/fl mice [40] to determine the role of Vcp in PNF initiation and/or maintenance, unfortunately the VCP^fl/fl;Nf1^fl/fl;DhhCre mice died before post-natal day 2 (not shown). Therefore, genetic deletion of Vcp in SCs and SCPs in PNF-bearing mice is not feasible for evaluating its role in PNF development.

3.6. Short- Term CB-5083 + PD0325901 Treatment Significantly Inhibites Cell Proliferation and Induces Cell Death

Treatment of MEK inhibitors, mirdametinib (PD0325901) and selumetinib, delays PNF growth, but the effects are modest on most of the PNFs and some PNFs re-grow after stopping treatment [7,8,35,36]. Based on our in vitro combination results (Figure3D), we tested if VCPi might provide additional benefit in vivo. We first determined the maximum tolerated dose for CB-5083. Prior reports showed that CB-5083 at 100 mg/kg daily treatment for one week induces protein ubiquitination and cell apoptosis and reduced tumor growth in vivo [41,42]. This dose (in single or combination treatment) was toxic to Nf1^fl/fl;DhhCre mouse after daily treatment for 2 months. After dose de-escalation, we confirmed that the maximum tolerated dose for CB-5083 was 80mg/kg in this model. All mice survived with less than 20% weight loss in either single reagent or combination treatment (not shown). To evaluate in vivo efficacy, we treated Nf1^fl/fl;DhhCre mouse with vehicle control, CB-5083 (80 mg/kg), PD0325901 (1.5 mg/kg), or combination [CB-5083 (80mg/kg) + PD0325901 (1.5mg/kg)] for 5 days by oral gavage, the PNFs were removed 2 hours after last dose for analysis. Ki67 immunolabeling revealed decreased cell proliferation following either PD0325901 or combination treatment, but not in VCB-5083 only treatment group (Figure 5A, 5B). There was no significant difference in cell death between control and PD0325901 alone; in fact, we have never observed increased cell death following PD0325901 treatment either in vivo or in vitro [35,36]. Importantly, the combination treatment showed significantly more TUNEL⁺ cells than either CB-5083 or PD0325901 alone (Figure 5C, 5D).

3.7. Long-Term CB-5083 or CB-5083 + PD0325901 Treatment Reduces Tumor Volume, and Inhibits Cell Proliferation, and Induces Cell Death/Protein Ubiquitination

Finally, we assessed the long-term in vivo effects of inhibiting VCP activity with CB-5083 on PNF growth. We treated the Nf1^fl/fl;DhhCre mice with CB-5083 (80mg/kg), PD0325901 (1.5mg/kg), combination, or vehicle control using a 5 on, 2 off regimen for 8 weeks. A few mice lost <10% weight during the first two weeks but all regained weight thereafter (data not shown). All mice survived and there were no obvious side morbidities. AS predicted, PD0325901 treatment significantly reduced tumor volume (p<0.0001). Volumetric measurement also showed that the percent tumor volume changes also significantly decreased in CB-5083 or combination treated mice compared to vehicle controls (p=0.037, p<0.0001 respectively) (Figure 6A). However, no difference was observed in the CB-5083 + PD0325901 treated group when compared with CB-5083 (p=0.492) or PD0325901 (p=0.435) monotherapy. CB-5083, PD0325901, or combination treatment all significantly reduced the number of Ki67+ proliferating cells compared to vehicle, with the CB-5083 + PD0325901 combination showing the greatest effect. (Figure 6B, 6C). TUNEL staining showed CB-5083 or CB-5083 + PD0325901 treatment significantly increased cell death while no difference was detected in PD0325901 treatment group compared to vehicle (Figure 6D, 6E). Western blot revealed unfold protein response and ER stress response as evidenced by increased CHOP and protein ubiquitination signal following treatment with VCPi alone, or in combination with MEKi. Importantly, the western blot data also showed that p-ERK was strongly inhibited by the combination treatment (Figure 6F).

4. Discussion

VCP is a highly conserved AAA+ ATPase. It has many functions including protein quality control, ER associated degradation and the ubiquitin–proteasome system maintenance. VCP plays a key role in protein homeostasis and cell survival in normal cells. It is believed that cancer cells can hijack this function to survive cell stress. Thus, VCP is increasingly recognized as a pro-tumorigenic factor and biomarker in many cancers, including non-small cell lung cancer, pancreatic cancer, colorectal cancer, bladder cancer, multiple myeloma, melanoma, breast cancer, hepatocellular carcinoma, and ovarian cancer [43,44]. In this study, we showed that VCP interacts with neurofibromin, the NF1 gene encoding protein, in SCs and that its expression is increased in both mouse and human PNFs. The data are consistent with the idea that RAS–MAPK activation in Nf1^-/- cells induce proteostatic stress that increases reliance on VCP to maintain protein homeostasis and cell survival, and thereby supporting PNF growth.

VCP interacts with a wide variety of cofactors to regulate proteostasis. More than 40 proteins that bind to VCP’s domains have been identified. These proteins regulate VCP ATPase activity and/or direct VCP to specific location to carry out functions such as DNA repair, membrane trafficking, or cell cycle control [12]. SVIP, a small regulatory protein, localizes on the ER membrane and acts as a key regulator of VCP-mediated processes, including unfold protein response and ER associated degradation. lt has been described as having a tumor suppressor role in glioblastoma [45]. We showed that SVIP expression is reduced in human PNF (Figure 2E) and might act as a negative regulator of VCP to contribute to PNF growth, consistent with a tumor suppressor role. Our results suggest that VCP might be either directly regulated by neurofibromin, or that loss of Nf1 increases Runx1 expression, which in turn represses Svip expression, leading to activation of Vcp. It will be important to define which pathway affects VCP expression in PNF in the future.

In PNFs, SCPs contribute to tumor formation. We showed that pharmacological and genetic inhibition of VCP decreases mouse neurofibroma sphere number in vitro. Genetic knock-down Vcp significantly reduced tumor like lesion formation in NSG mice (Figure 4E), suggesting VCP itself or in coordination with RAS-MAPK signaling contributes to PNF initiation.

VCP contributes to tumor growth suggested that targeting VCP might be a good therapeutic approach. VCP uses energy from ATP hydrolysis to unfold and segregate ubiquitinated protein substrates for degradation. When VCP’s ATPase activity is inhibited, ubiquitinated proteins accumulate, leading to cell stress and often initiating programmed cell death (apoptosis). CB-5083, a potent with moderate oral bioavailability VCPi, was identified in 2015. CB-5083 bound to the D2 domain of VCP to inhibit VCP ATPase activity and led to retention of ER associated degradation and generation of irresolvable proteotoxic stress, therefore induced apoptosis. It inhibited tumor growth in hematological xenograft models [41,46]. CB-5083 suppressed osteosarcoma growth and stem cell properties by altering protein homeostasis [47]. Zhang et al. show that pharmacological inhibition of VCP by CB-5083 partially abolished the oncogenic effects of TMEM63A on triple-negative breast cancer progression both in vitro and in vivo [48]. Single-agent CB-5083 inhibited xenografted bladder cancer growth and CB-5083 effectively kills tumor cells and acted as a radio sensitizing agent both in vitro and in vivo [49] and multiple myeloma [50]. Targeting VCP enhanced colorectal cancer therapy through STING stabilization [51]. The differential inhibitory effects between WT SCs and PNF SCs on cell growth (Figure 3A) and protein ubiquitination (Figure 3B) indicated a potential therapeutic efficacy for PNFs by targeting VCP. We showed that inhibition of VCP significantly reduced PNF growth compared to control (Figure 6A), suggesting that VCP might be an alternative therapeutic target.

In vitro CB-5083 + PD0325901 combination treatment resulted in protein ubiquitination and p-ERK inhibition, while single reagent only led to protein ubiquitination (CB-5083) or p-ERK inhibition (PD0325901) (Figure 3D), suggesting co-targeting VCP and MEK might provide synthetic lethality. Importantly, we detected increased cell death in either at 5-day or 60-day combination treatment group. Although combination showed a trend of better efficacy, we did not detect significant difference in reducing PNF volume percentage change compared to single reagent PD0325901 or CB-5083 (Figure6A). It is possible that: 1) PD0325901 or CB-5083 each has very strong effects on reducing tumor volume, so we did not have enough power (i.e., mouse number) to reach statistical difference. 2) VCP functions in parallel to, or is modulated by, RAS–MAPK signaling under stress or oncogenic condition. 3) Elevated VCP expression is a secondary adaptive response to RAS-MAPK driven stress rather than a direct signaling target. 4) Other unidentified pathway(s) might be involved or compensated in PNF initiation/maintenance upon inhibition of VCP and MEK. 5) The combination did not kill SCPs, the PNF-initiating cells. Future work should focus on identifying the specific cell populations that die in response to CB-5083 or combination treatment.

The first generation VCPi (CB-5083) tested here showed some off-target effect: two phase I clinical trials (NCT02243917 and NCT02223598) were terminated due to adverse effects on vision resulting from inhibition of phosphodiesterase-6 [52]. In 2021, a second-generation VCPi, CB-5339, was developed and it showed efficacy in several multiple AML models [53]. This compound has progressed through early clinical evaluation with the successful completion of a Phase I study in acute myeloid leukemia or myelodysplastic syndrome (NCT04402541), providing important data on safety, tolerability, and pharmacokinetics. A second phase I study on testing the safety of CB-5339 in patients with solid tumors that have spread to other places in the body (advanced) or lymphomas (NCT04372641) was withdrawn for unreported reasons. Overall, these early clinical experiences have informed dose selection and study design, supporting the continued advancement of this inhibitor. It will be interesting to test CB-5339 in our PNF mouse model and compare its effect on PNF growth with that of CB-5083 in the future.

5. Conclusions

It is well established that loss of Nf1 increases RAS activity, thereby activating the MEK/ERK signaling pathway and contributes to PNF formation. We demonstrate that VCP interacts with neurofibromin. Furthermore, loss of Nf1 upregulates Runx1 expression, which binds to SVIP to enhance VCP expression, thereby further contributing to PNF pathogenesis. (Figure 7). VCP expression is increased in both mouse and human PNFs. It might be regulated by and functionally coupled to RAS–MAPK driven SC states. The significant effects of VCP inhibition in this pre-clinical study suggest a potential novel therapy for patients with PNFs.