Introduction

Medulloblastoma (MB) is the most common malignant brain tumor of childhood, with over 500 new diagnoses each year [1,2,3,4,5,6,7,8]. Despite advances in the molecular stratification and understanding of tumor biology, standard of care relies on surgical resection, craniospinal radiation and chemotherapy [7, 9,10,11,12,13,14,15,16,17,18,19]. These are toxic to the normal developing brain and cause long-term quality of life issues and further, are ineffective in the setting of recurrence/metastasis [9,10,11,12,13,14,15, 20]. The 5-year survival for patients diagnosed with MBs is 66%, with molecular subgroups—Wingless (Wnt), Sonic Hedgehog (SHH), group 3, or group 4, serving as the main determinant of outcomes [1, 3, 8, 9, 11, 21,22,23,24,25,26,27,28,29,30,31,32,33]. Patients with group 3 MBs have the worst prognosis, whereas those with SHH and group 4 tumors have intermediate outcomes [1, 3, 8, 9, 11, 21,22,23,24,25,26,27,28,29,30,31,32,33]. While deregulated chromatin remodeling is implicated in MB genesis [21, 26, 32, 34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49], there is also growing evidence of aberrant regulation of protein degradation in SHH-driven MBs [35, 36].

Our previous work implicated the Ubiquitin Specific Protease 37 (USP37)—a deubiquitylase (DUB) and a component of the proteasomal degradation machinery, in MB biology [35, 36]. DUBs catalyze the proteolytic removal of ubiquitin groups from proteins, resulting in changes to their stability, subcellular localization, or function [50,51,52,53,54]. They oppose the functions of enzymes (E1-3), which coordinate the addition of ubiquitin moieties to proteins [55,56,57]. Previously, we found that in SHH-driven MB cells with RE1 silencing transcription factor (REST) elevation, USP37 downregulation prevented the stabilization of the cyclin-dependent kinase inhibitor (CDKI)-p27 and maintained cell proliferation [35]. Pharmacological targeting of G9a histone methyltransferases upregulated USP37 expression and reduced tumor formation of low-USP37 MBs, suggesting an epigenetic basis for USP37 downregulation in MBs [36]. At approximately the same time, work from the Dixit group demonstrated a role for USP37 in controlling S phase entry through its regulation of cyclin A stability in non-neural cells [58]. Since these initial reports, research from other groups has revealed the importance of USP37 in the regulation of DNA replication and cell cycle [35, 59, 60]/cell proliferation [61] in addition to functions in receptor signal transduction [62], homologous recombination [63], DNA damage repair [64], cell migration [65], autophagy [66], epithelial to mesenchymal transition [67] and resistance to chemotherapy [68].

In the current study, we identify Raptor and GLI1 as novel USP37 targets in SHH-MBs and also suggest that USP37 may have context-specific oncogenic roles in addition to its previously described tumor suppressive function in these tumors [35]. We show that REST elevation and low USP37 results in reduced stability of Raptor, a component of the mTORC1 complex, and a regulator of protein translation. In the context of high USP37 expression, we found increased GLI1 protein levels and USP37-dependent GLI1 ubiquitination and protein stabilization. These data suggest a REST-context specific role for USP37 in the regulation of protein translation and in the non-canonical activation of SHH signaling in SHH-MBs. Consistent with these findings, USP37 loss in the context of high-REST expression, as well as USP37 elevation in the background of high-glioma-associated oncogene 1 (GLI1) levels, were both found to be correlated with poor outcomes for patients with SHH-MBs.

Materials and methods

Patient samples

Two publicly available MB patient’s datasets (GSE85217 and GSE124814) were used for gene expression analysis [69, 70]. Differential gene expression analysis was performed using the R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl). A value of p < 0.05 was considered significant. Gene ontology p-values were not corrected for multiple testing. Histopathological analyses was performed in paraffin embedded de-identified MB tumor sections (n = 33) by our collaborating neuropathologist.

Cell culture

293T and SHH-MB cell lines (DAOY, UW228, UW426, and ONS76) are used in this study. DAOY cells were purchased from the American Type Culture Collection (ATCC) (Manassas, VA). UW228 and UW466 cells were a kind gift of Dr. John Silber at the University of Washington. ONS76 cells were purchased from Accegen, NJ, USA. All cell lines were maintained in Dulbecco’s modified Eagle’s medium (Sigma-Aldrich, MO, USA), supplemented with 10% fetal bovine serum (Sigma-Aldrich), 1% antibiotic-antimycotic (Thermo Fisher Scientific, MA, USA), and 1% sodium pyruvate (Thermo Fisher Scientific) and grown at 37 °C with 5% CO2.

Patient-derived xenograft models

Patient-derived xenograft (PDX) models (RCMB-18, 24 and 54) were a kind gift of Dr. Robert Wechsler-Reya (Columbia University). Serial transplantation of tumors was carried out in NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice (Jackson Laboratory, Bar Harbor, ME) by intracranial inoculation of tumor cells using a stereotactic device as described previously [48]. Housing, maintenance, and experiments involving mice were done in compliance with a protocol approved by the University of Texas MD Anderson Cancer Center’s Institutional Animal Care and Use Committee (IACUC). Our study is reported in accordance with ARRIVE guidelines (https://arriveguidelines.org).

Animals

Generation of Ptch+/−/RESTTG is described in an earlier [48]. The hREST transgene expression was induced by intraperitoneal (ip) injections of (100 µl of 2 mg/ml) tamoxifen (Cat# T5648, Sigma-Aldrich) on post-natal (P) days 2, 3 and 4. Moribund animals were euthanized, and brains were harvested for further analysis. All experiments were compliant with institutional IACUC guidelines and in accordance with ARRIVE guidelines (https://arriveguidelines.org).

Immunohistochemistry

Mouse brain tissues were fixed in 10% buffered formalin phosphate for 48 h (h) and embedded in paraffin. 8 µm thick sections were used for histological analysis using a Gemini AS Automated Stainer (Thermo Fisher Scientific). After overnight incubation with primary antibody at 4 °C, the sections were incubated with biotinylated secondary antibody provided by either the ABC kit or the MOM kit (Vector Laboratories, CA, USA). For detection, VECTASTAIN® Elite® ABC-HRP Kit, Peroxidase (Cat# PK-6101, Vector Laboratories) were used according to the manufacturer’s instructions and developed using the DAB Substrate Kit, Peroxidase (HRP) (Cat# SK-4100, Vector Laboratories) followed by counterstaining with hematoxylin. After dehydration and mounting, slides were visualized under a Nikon ECLIPSE E200 microscope mounted with an Olympus SC100 camera. The list of primary antibodies used for IHC is provided in Supplementary Table S1.

Lentiviral infection and generation of stable cell lines

Embryonic kidney (HEK) 293T cells were co-transfected with either a control or the gene of interest along with packaging plasmid (PAX2) and envelope plasmid (MD2). Lentiviral particles were collected 48 h post-transfection. MB cells were transduced with the collected viral supernatant in the presence of Polybrene (8 μg/mL) and incubated for 48 h. Infected cells were then cultured in medium containing 2 μg/mL puromycin for up to 1 week for selection.

Cycloheximide (CHX) chase assay

SHH-MB cell lines were treated with CHX alone for varying lengths of time (0–240 min) to assess changes in protein levels. To show proteasomal control of protein stability, rescue experiments were conducted, where cells were co-treated with CHX and MG132 for 240 min. Following the above treatment(s), cells were harvested and lysed using RIPA buffer supplemented with protease inhibitors. Protein lysates were quantified and analyzed by Western blotting to evaluate the dynamics of protein degradation.

Western blot analyses

Cell lysates were prepared in lysis buffer (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease/phosphatase inhibitors) and processed for Western blotting as described previously [40] using primary antibodies listed in Supplementary Table S1 followed by HRP-conjugated goat anti-mouse or anti-rabbit secondary antibodies. SuperSignal West Dura Extended Duration Substrate (Cat#34075, Thermo Fisher Scientific) and Western Lightning Plus-ECL, Enhanced Chemiluminescence Substrate (Cat# 50-904-9325, Fisher Scientific, MA, USA) were used to develop the blot and detected using Kodak Medical X-Ray Processor 104 (Eastman Kodak Company) or ChemiDoc Touch Imaging System (Bio-Rad). Images were analyzed using Image Lab software version 5.2.1 (Bio-Rad).

Co-immunoprecipitation

Cell pellets were washed with ice-cold PBS and lysed in mild lysis buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 5 mM EDTA) containing protease inhibitor cocktail (Cat# 78429, Thermo Fisher Scientific) and phosphatase inhibitor (Sigma) and sonicated. Lysates were incubated with control mouse IgG, anti-USP37 (Cat# A300-927A, Thermo Fisher Scientific), anti-4EBP1 (Cat# 9644, Cell Signaling Technology, MA, USA), anti-eIF4E (Cat# 2067, Cell Signaling Technology) primary antibodies overnight at 4 °C and then incubated with Pierce™ Protein A/G UltraLink™ Resin (Cat# 53132, Thermo Fisher Scientific) for 1.5 h at 4 °C. After four washes with lysis buffer, beads were boiled in loading buffer, separated by SDS-PAGE, transferred onto PVDF membranes, and analyzed by Western blotting.

In vitro deubiquitination (DUB) assay

293T cells were co-transfected with pcDNA3-Myc3-Raptor/ pcDNA3-6XHis-GLI1 and HA-Ub or with pDEST26-FLAG-HA-USP37, FLAG-HA-USP37C350S or FLAG-HA-USP1 using jetPRIME (Polyplus, NY, USA). Cells were treated with 20 mM MG132 for 6 h prior to lysis. HA-Ub-Myc-Raptor and HA-Ub-6XHis-GLI1 substrate were purified using EZ view Red Anti-c-Myc Affinity Gel (Cat# E6654, Millipore Sigma) and Ni-NTA Agarose (Cat # R90101, Thermo Fisher Scientific), respectively. Deubiquitinase (DUBs: USP37, USP37C350S or USP1) were immunopurified using anti-FLAG M2 beads (Cat# A2220, Millipore Sigma) and eluted with FLAG peptide (Cat# HY-P0319, MedChem Express, NJ, USA). For DUB assays, equal amounts of substrates and purified DUBs were incubated for 8 h at 37 °C in the presence or absence of 15mM N-ethyl maleimide (NEM). Reactions were terminated by boiling in 2X Laemmli buffer and analyzed by Western blotting.

In vivo DUB assay

MB cells were transiently transfected with plasmids expressing (Flag-HA-USP37WT and Flag-HA-USP37C350S (Addgene, MA, USA). Cells were treated with 20 mM MG132 for 6 h prior to lysis. Samples were heat-denatured and subjected to SDS-PAGE and Western blotting using anti-Raptor (Cat# 2280, Cell Signaling Technology), anti-GLI1 (Cat# ab134906, Abcam, MA, USA), anti-FLAG (Cat# F1804, Millipore Sigma) and anti-Ubiquitin (Cat# 3933, Cell Signaling Technology) antibodies.

Statistical analysis

Statistical analyses were performed using GraphPad Prism version 10. Data are presented as mean ± standard deviation (SD), derived from at least three independent biological replicates. Comparisons of means between groups were conducted using unpaired Student’s t tests. For survival analysis, the Kaplan–Meier method was employed, and statistical significance between survival curves was assessed using the log-rank (Mantel–Cox) test. To analyze transcriptomic data from MB patients, Welch’s t test was used to account for unequal variances and sample sizes. Statistical significance was defined as p < 0.05, and results are denoted as: ****p < 0.0001, ***p = 0.0001–0.001, **p = 0.001–0.01, *p = 0.01–0.05, ns = not significant.

Results

Low- and high-USP37 expression in SHH-MBs is correlated with poor patient outcomes

Our previous work reported USP37 as a novel target of REST in SHH-MBs and also studied its expression in MB cell lines compared to normal cerebella [35, 36]. Here, we assessed the levels of USP37 in human MB samples (n = 33) (Fig. 1A). Samples were graded as negative (-) or positive (+/++/ +++/ ++++) for USP37 expression (Fig. 1A). Approximately, 33.3% of the samples were negative for USP37, whereas weak and focal staining (+) was noted in 30% of the samples. Weak diffuse and multifocal focal staining (++) was noted in 6.7% samples (Fig. 1A, B). The remaining 30% of samples exhibited strong expression of USP37 (23.3% - strong and focal staining (+++) and (6.7% - strong diffuse or focal (++++)) (Fig. 1A, B). Further, samples with negative or weak expression of USP37 (-/+/++) exhibited high levels of REST (++/+++/++++), while samples with high USP37 (+++/++++) were negative or expressed low levels of REST (-/+) (Fig. 1C).

Fig. 1: High- and low-USP37 in SHH-driven MBs is associated with poor patient outcomes.
figure 1

A Human MB samples (n = 33) were stained with anti-USP37 antibodies and graded on a scale from (-) to (++++) based on the level of USP37 expression (magnification: 40×). B Quantitation of data shown in (A) to show the distribution of tumors with different grades and patterns of USP37 staining. C Graphical representation of data from (A) and (B) to show the inverse correlation between REST and USP37 levels. Each dot represents a tumor. Significance and correlation were measured using the unpaired t-test and Pearson correlation coefficient (r = 0.478 and P value = <0.0001). D USP37 gene expression in a microarray data set (GSE85217) [69] of human SHH- MB samples. Hierarchical clustering based on the expression of neuronal differentiation markers divided the SHH-type MB patient samples into six distinct clusters [48]. Each dot corresponds to an individual patient. Statistical significance was assessed using Welch’s t-test. E Kaplan–Meier plot to demonstrate significant differences in the overall survival of SHH-MB patients divided into four cohorts based on the relative expression of REST and USP37 in their tumors (GSE85217) [69]. Statistical significance between survival curves was assessed using the log-rank (Mantel–Cox) test. F Pathway analysis to show significantly enriched pathways in tumors with high-REST/low-USP37 expression relative to samples with low-REST/high-USP37 expression using the GSE124814 dataset [70].

Full size image

Transcriptomic data from MB patient samples revealed that USP37 levels were significantly lower in WNT, SHH and group 3 tumors compared to group 4 samples (Fig. S1A). Since our previous work implicated USP37 in SHH-MB pathology [35], we further investigated USP37 mRNA expression in the six differentiation-based clusters of SHH-MB patients described in our earlier studies [48, 69]. As expected, a significant reduction in USP37 expression was observed in clusters 2 and 5, which we previously showed to exhibit elevated REST expression and to be associated with poor outcomes (Fig. 1D). Kaplan Meier analyses confirmed that SHH-MB patients with high-REST/low-USP37 (n = 69) did indeed exhibit poor survival compared to patients with high-REST/high-USP37 (n = 27) and low-REST/low-USP37 (Fig. 1E). Unexpectedly, a cohort of SHH-MB patients with low-REST/high-USP37 (n = 19) in their tumors also exhibited poor outcomes (Fig. 1E). Further evaluation of the high-REST/low-USP37 and low-REST/high-USP37 cohorts of patients from the GSE124814 and GSE85217 datasets by pathway analyses revealed a significant enrichment of processes known to regulate cerebellar and hind brain development, including those regulating proliferation and maintenance of stem-cells and neuronal precursor cells, neurogenesis, and neuronal differentiation, Smoothened, Wnt and Hippo signaling (Fig. 1F and S1B). Interestingly, pathways which we previously discovered as REST-regulated, such as vasculogenesis, angiogenesis, hypoxia, cell-cycle and cell migration were enriched in the high-REST/low-USP37 cohort of patient samples compared to the low-REST/high-USP37 group (Figs. 1F and S1B). Apoptosis, protein translation, and mTOR were also differentially enriched in those two patient cohorts (Fig. 1F).

Animal models of SHH-MBs express variable levels of USP37 and REST

Immunohistochemical staining of patient-derived orthotropic xenograft (PDOXs) sections of SHH-MB tumors was performed to assess the levels of USP37 and REST (Fig. 2A). Of the four PDOX tumors, two showed strong Ki67 staining, while all four showed some positivity for TUBB3. Strong REST staining was observed in all four samples. Varying levels of USP37 staining were observed in the tumors, with RCMB-018 showing minimal staining and Med1712 exhibiting maximum cytoplasmic USP37 expression among the four samples (Fig. 2A). Cerebellar sections of Ptch+/-/RESTTG animals bearing tumors were also histologically analyzed, which revealed strong Rest staining as expected (Fig. 2B). Usp37 was low in two of the three samples studied (Fig. 2B).

Fig. 2: H&E and IHC analyses of mouse models of SHH-driven MBs.
figure 2

Staining of cerebellar sections from A mice with SHH-MB PDOX and B tumor-bearing Ptch+/−/RESTTG animals (n = 3) for Ki67, TUBB3, REST, USP37, and Rest and Usp37 respectively (scale bar: 200 µm for H&E and 20 µm for IHC).

Full size image

Raptor is a novel target of USP37

In previous work by Dobson et al., we found that REST elevation in SHH-MB tumors to correlate with AKT activation as measured by its phosphorylation at Ser473 (pAKTSer473), a modification brought about by the mammalian target of rapamycin complex 2 (mTORC2) [48]. Western blotting was used to confirm these findings in human SHH-MB cell lines expressing varying levels of REST and USP37 (Figs. S2A and 3A). As shown in Fig. 3A, cells with higher levels of REST (293 and DAOY) had higher p-AKTSer473 levels and that of its known targets p-GSK-3βSer9 and p-p27Thr157 compared to lower REST expressing cell line (UW228) [71, 72]. Total GSK-3β and Actin were included as controls (Fig. 3A). p27 levels were lower in DAOY cells consistent with our previous report showing reduced USP37 levels prevented the stabilization of the cell cycle regulatory protein [35]. In line with this finding, the kinase activity of Rictor was higher in 293 and DAOY cells compared to UW228 and UW426 cells as measured by reduction in the levels of the Thr1135 phosphorylation, an event known to inhibit Rictor activity [73, 74] (Fig. 3B). Total Rictor as well as MLST8 levels were not different in these cells (Fig. 3B). Interestingly, mTOR expression was lower in UW228 cells whereas expression of mSIN1, a structural component of the mTORC2 complex, was lower in UW228 and UW426 cells compared to DAOY and 293 cells (Fig. 3B). β-actin was used as the loading control (Fig. 3B). These results show that REST elevation is associated with increased mTORC2 complex activity in some SHH-MB cells.

Fig. 3: Raptor is a target of USP37.
figure 3

Western blot analyses to assess levels of A AKT, GSK3β, and p27 and their respective phosphorylated forms in 293 and different MB cell lines with varying levels of REST, B mTORC2 components -p-Rictor, Rictor, mTOR, mLST8, and mSIN1 in SHH-MB cell lines with varying levels of REST, and C mTORC1 components - Raptor p-Raptor, PRAS40, Deptor, and the mTORC1 target - 4EBP1 and its phosphorylated form. D Western blot analysis to demonstrate post-translation and proteasomal control of Raptor in DAOY cells following treatment with cycloheximide (CHX) alone or in combination with the proteasome inhibitor, MG132 (shown by *). E Co-immunoprecipitation assay using control IgG or anti-USP37 antibodies followed by Western blotting to assess interaction between USP37 and mTORC1/2 components. Input shows 2% of total lysate. F In vivo DUB assay using transiently transfected FLAG (Fl)-tagged wildtype (WT) USP37 and mutant USP37 (USP37C350S) to study longitudinal changes in endogenous Raptor levels in DAOY cells. G In vitro DUB assay was done by co-incubation of purified MYC tagged-Ub-Raptor with Flag-WT-USP37 (lane 2), Flag-WT-USP37 in the presence of NEM (lane3), Flag-USP37C350S (lane 4) and Flag-USP1 (lane 5). The reaction containing MYC tagged-Ub-Raptor substrate alone is included as a control (lane 1). IHC to show Raptor expression in cerebellar sections of mice with H human SHH-MB PDOX tumors and I Ptch+/−/RESTTG tumors (scale bar: 20 µm). Statistical data are presented for three independent biological replicates as the means ± SDs. ns = non-significant, *p < 0.05, **p < 0.01, and ***p < 0.001 by Student’s t test.

Full size image

AKT signaling is also known to negatively control mTORC1 activity. Indeed, Western blot analyses revealed lower mTORC1 activity in DAOY cells compared to UW228 and UW426 cells based on levels of phosphorylated 4EBP1, although total 4EBP1 levels were not different between these cells (Fig. 3C). Deptor, a component of both mTORC1 and mTORC2 complexes, was absent in UW228 cells and lower in DAOY cells compared to 293 and UW426 cells (Fig. 3C). PRAS40 levels were similar between the cells (Fig. 3C). However, Raptor levels were significantly reduced in DAOY cells and was also associated with higher levels of the inhibitory Ser792 phosphorylation and a laddering pattern of migration, suggestive of post-translational modification such as ubiquitination [75]. A significant decrease in Raptor levels in DAOY cells following treatment with cycloheximide (CHX) suggested that its expression is post-transcriptionally controlled in SHH-MB cells (Fig. 3D). Further, co-treatment with the proteasome inhibitor, MG132, countered the reduction in Raptor levels suggesting that Raptor may be subjected to ubiquitin-mediated proteasome degradation (Fig. 3D). Other mTORC1 components did not reveal a similar MG132-dependent change in levels (Fig. S2B). The CHX-induced reduction in Raptor levels, along with its reversal upon co-treatment with MG132, was also validated in ONS76 cells (Fig. S2C). Immunoprecipitation assays were performed following co-transfection of DAOY cells with plasmids expressing Myc-Raptor and HA-Ubiquitin. Pull down with anti-HA antibody revealed strong ubiquitination of Raptor (Fig. S2D).

As a first step in investigating if the stability of Raptor is regulated by USP37, we performed co-immunoprecipitation assays using anti-USP37 antibody to demonstrate an interaction between endogenous USP37 and components of mTORC1 and mTORC2 in MB cells (Fig. 3E). Strong interaction between USP37 and Raptor was observed, but not between USP37 and other components of the mTORC complex such as Rictor, Deptor, mTOR, pRAS40, and mSIN1 (Fig. 3E). Studies with ONS76 cells also confirmed the interaction between USP37 and Raptor (Fig. S2E). Next, we assessed the stability of Raptor in response to USP37 overexpression. In vivo DUB assays confirmed an increase in Raptor levels and a decrease in high-molecular-weight bands of Raptor following USP37 overexpression in DAOY cells and ONS76 cells (Figs. 3F and S2F). Raptor levels were not discernably different in cells overexpressing a catalytically inactive mutant of USP37 (USP37C350S) (Figs. 3F and S2F). We also performed in vitro DUB assay using purified tagged proteins to ask if USP37 directly deubiquitinates Raptor. Addition of USP37 to the reaction containing Myc-Ub-Raptor resulted in a significant increase in Raptor levels compared to that containing USP37 and a protease inhibitor N-ethyl maleimide (NEM) or USP37C350S or USP1 (Fig. 3G). Thus, these findings indicate that REST-dependent destabilization of Raptor is caused by reduction in USP37 levels. Immunohistochemical staining of SHH-MB PDOXs and Ptch+/-/RESTTG tumor sections was also carried out to demonstrate a positive correlation between USP37 and Raptor levels (Figs. 2A, B, 3H, I). Together, these data identify Raptor as a novel target of USP37.

REST and USP37 converge on the cellular cap-dependent protein translational machinery

USP37 overexpression-dependent increase in Raptor levels in MB cells was accompanied by an increase in phospho-4EBP1 (Fig. 4A). However, total 4EBP1 levels were unaffected (Fig. 4A). p27 levels were also increased under these conditions, as expected, and served as a positive control (Fig. 4A). Actin was used as a loading control (Fig. 4A). mTORC1 kinase controls protein translation initiation through phosphorylation of 4EBP1, which inhibits its binding with eIF4E and favors cap-dependent translation initiation (Fig. 4B) [76, 77]. As seen in Fig. 4A, C, USP37 overexpression in DAOY and ONS76 cells caused an increase in phospho-4EBP1 levels, and co-immunoprecipitation assays using anti-4EBP1 antibodies showed an USP37-dependent decrease in interaction between 4EBP1 and eIF4E. An increase in the phosphorylated form of 4EBP1 was also noted in the immunoprecipitate, validating the decreased interaction between 4EBP1 and eIF4E and confirming increased activation of mTORC1 complex under conditions of USP37 elevation (Fig. 4A, C). As expected, 4EBP1 did not interact with eIF4G, a scaffolding component of the translational initiation machinery (Fig. 4C). These results suggest that cap-dependent translation initiation may be favored under conditions of constitutive USP37 expression. In contrast, under conditions of REST overexpression, interaction between eIF4E and 4EBP1 was increased and that between IF4E and eIF4G was decreased, suggesting an inhibition of cap-dependent translation initiation (Fig. 4D).

Fig. 4: REST and USP37 modulate cap-dependent protein translation initiation machinery.
figure 4

A Western blotting to show changes in the levels of Raptor, 4EBP1, p-4EBP1 and p27 in DAOY and ONS76 MB cell lines with constitutive over expression of USP37. B Schema to illustrate expected effect of mTORC1 kinase activity on phosphorylation of 4EBP1 and its interaction with eIF4E. Phosphorylated 4EBP1 releases eIF4E, allowing its binding to the CAP structure of mRNA and thereby enabling canonical protein translation. Conversely, mTORC1 inactivity results in the binding of unphosphorylated 4EBP1 to eIF4E, preventing its association with the CAP structure of mRNA. This is expected to prevent canonical protein translation initiation and cause a potential shift to CAP-independent protein translation. Co-immunoprecipitation assays to study C interaction of 4EBP1 with eIF4E and eIF4G using control IgG or anti-4EBP1 antibody and D eIF4E with 4EBP1, p-4EBP1, and eIF4G using control IgG or anti-eIF4E antibody, in parental and isogenic USP37- and REST- overexpressing DAOY and ONS76 cells, respectively. Statistical data are presented for three independent biological replicates as the means ± SDs. ns = non-significant, *p < 0.05, **p < 0.01, and ****p < 0.0001 by Student’s t test.

Full size image

USP37 promotes GLI1 stability

As shown in Fig. 1E, patients with high-REST/low-USP37 as well as low-REST/high-USP37 expression in their tumors exhibited poor survival. Enrichment of pathways associated with SHH signaling led us to investigate if the underlying mechanism may involve USP37-mediated regulation of one or more molecules driving SHH pathway activity. Based on a previous report that GLI1 is a target of USP37 in breast cancer cells - a non-neural cell type, we explored if a similar USP37-dependent stabilization of GLI1 and consequent hyperactivation of SHH signaling may drive SHH MB tumor growth [67]. Consistent with this possibility, patients with high-GLI1 expression, and low or high-USP37 expression, in their tumors exhibited poor overall survival (Fig. 5A). Pathways known to be controlled by GLI1 including apoptosis and cell migration, ciliary assembly, transport and organization, early-stage brain development, and development of metencephalon, hindbrain, cerebellum, and neuronal differentiation, were enriched in patient tumors with high GLI1 expression in the GSE124814 dataset (Fig. 5B). Immunohistochemical staining for GLI1 in SHH-MB PDOX brain sections showed higher USP37 levels to be associated with higher GLI1 protein levels (Figs. 2A and 5C). CCND1, a known GLI1 target, was also elevated in high USP37/high-GLI1 expressing tumors along with low levels of cleaved Caspase 3, indicating an increase in proliferation and decrease in apoptosis under these conditions (Fig. 5C). Tumor sections from Ptch+/- SHH-MB mice also showed a positive association between Usp37 and Gli1 protein levels (Figs. 2B and 5D).

Fig. 5: High-USP37 and high-GLI1 expression in SHH-MBs correlates with poor overall patient survival.
figure 5

A Kaplan–Meier plot to show a significant correlation between reduced overall survival of SHH-MB patients with high expression of USP37 and GLI1 in their tumors (GSE85217) [69]. Statistical significance was assessed using the log-rank (Mantel–Cox) test. B Bubble plot to show significantly enriched pathways in tumors with high-USP37/high-GLI1 compared to low-USP37/high-GLI1 expressing samples in the GSE124814 dataset [70]. IHC staining of cerebellar sections with C human SHH-MB PDOX and D Ptch+/− SHH-MB tumors (n = 3) for GLI1, CCND1 and Cleaved caspase3 and Usp37 and Gli1, respectively. (scale bar: 200 µm for H&E (D) and 20 µm for IHC).

Full size image

Next, biochemical analyses were performed to investigate if GLI1 is a target of USP37 in MB cells. We first examined the expression of REST, USP37, and GLI1 across four SHH-MB cell lines (Fig. S3A). DAOY and UW228 cells, which have low USP37 expression, also exhibited reduced GLI1 levels. In contrast, ONS76 cells, with higher USP37 expression, demonstrated elevated GLI1 levels (Fig. S3A). Further, in ONS76 cells, GLI1 levels were markedly reduced following CHX treatment, but were restored upon co-treatment with MG132 suggesting that GLI1 is subjected to proteasomal degradation, likely in a ubiquitination-dependent process. (Fig. 6A). In UW228 cells, CHX treatment led to a reduction in GLI1 levels. Although co-treatment with MG132 significantly increased GLI1 levels compared to 240 h of CHX alone, the levels could not be restored to that in untreated cells (Fig. 6A). Next, co-immunoprecipitation assays were done to show a strong interaction of endogenous USP37 with GLI1 in UW228 and ONS76 cells (Fig. 6B). Overexpression of USP37 in both SHH-MB cell lines promoted a significant increase in the levels of GLI1 and its target, CCND1 (Fig. 6C). In vivo DUB assays showed an increased GLI1 levels at 16 and 24 h post-transient transfection of WT, but not the catalytically impaired USP37 (USP37C350S) mutant in ONS76 and UW228 cells (Figs. 6D and S3B). Anti-FLAG antibody was used to confirm expression of both USP37 transgenes (Figs. 6D and S3B). Finally, in vitro DUB assays were performed using tagged proteins affinity-purified from transiently transfected 293 cells to assess if GLI1 is a direct target of USP37. Incubation with WT USP37 caused a significant increase in GLI1 protein at the expected molecular weight of ~150 kDa (Fig. 6E). In contrast, incubation with USP37C530S, USP1 or WT-USP37 in the presence of NEM did not support GLI1 stabilization (Fig. 6E). The above data confirm that GLI1 protein level is controlled by USP37-mediated deubiquitination in SHH-MB cells.

Fig. 6: GLI1 is a target of USP37 in SHH-MB.
figure 6

A Western blot analyses to show changes in the levels of GLI1 in UW228 and ONS76 cells after treatment with CHX for 0 min to 240 min and co-treatment with CHX and MG132 for 240 min (*). B Co-immunoprecipitation assay using control IgG or anti-USP37 antibody to show interaction between endogenous USP37 and GLI1 proteins in UW228 and ONS76 cells. C Western blot analysis to demonstrate an increase in the levels of GLI1 and its target gene product, CCND1 in UW228 and ONS76 cells following overexpression of transiently transfected USP37. D In vivo DUB assay to show a longitudinal increase in GLI1 protein levels after transient expression of Fl-USP37WT but not with Fl- USP37C350S mutant proteins in ONS76 cells. Overexpressed USP37 proteins are shown using anti-FLAG antibody. E In vitro DUB assay was performed by co-incubation of purified HA-Ub-6XHis-GLI1 with Flag-WT-USP37 (lane 2), Flag-WT USP37 in the presence of NEM (lane3), Flag-USP37C350S (lane 4) and Flag-USP1 (lane 5). The reaction containing purified HA-Ub-6XHis-GLI1 alone is shown in lane 1. Statistical data are presented for three independent biological replicates as the means ± SDs. ns = non-significant, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 by Student’s t test.

Full size image

Discussion

DUBs are a class of catalytic enzymes that remove ubiquitination moieties from proteins to regulate their stability and or function [78]. Deregulation of DUBs is linked to critical cellular processes such as cell cycle control, apoptosis, and DNA repair, contributing to the development and progression of various cancers [79, 80]. USP37 is a cysteine protease that has been reported to control biological processes such as chromosomal cohesion and mitotic progression, cell cycle progression, stemness and cancer cell therapeutic sensitivity, metastasis, DNA damage response and epithelial to mesenchymal transition [58, 60, 63, 65, 67, 68, 81]. Our previous work had implicated USP37 in the stabilization of the CDKi–p27 and showed that USP37 downregulation in REST-driven SHH-MBs drives cell proliferation and prevents cell cycle exit [35]. In follow up studies, we demonstrated that REST-associated G9a histone methyltransferase was involved in the epigenetically silencing of USP37 in the REST-high SHH-MB cell, DAOY [36]. Overall, these studies suggested a tumor suppressive role for USP37 in REST-driven SHH-MBs. However, in non-neural cells, USP37 was also shown to exhibit oncogenic properties [61, 62, 82, 83], which suggested that USP37 may have cell and context-specific tumor suppressive and oncogenic roles.

In the current study, the identification of Raptor as a novel target of USP37 links it to the mTORC1 complex, and through it to the regulation of protein translation and potential management of cellular energetics and metabolism [84,85,86]. mTORC1 directly controls protein translation through its kinase activity and phosphorylation of S6 kinase (S6K) and 4EBP1, and downregulation of their activity under conditions of low USP37 would be expected to significantly impact cap-dependent protein translation [85]. However, REST elevation in SHH-MB cells is associated with increased cell proliferation, which requires protein translation [39, 48]. This raises questions regarding how cancer cells meet the demand for new protein synthesis associated with rapidly dividing cells. A possible scenario is that in context of REST elevation and USP37 downregulation forces SHH-MB cells to switch to non-canonical cap-independent protein translation initiation mechanisms to meet the increased cellular demand for proteins necessary for stress adaptation and survival [87, 88]. Indeed, MB cells have been shown to employ internal ribosome entry site (IRES) elements to initiate protein translation without relying on the 5’ cap structure, and to selectively elevate the translation of oncoproteins essential for tumor cell growth and survival [89, 90]. However, additional studies are needed to demonstrate that indeed USP37-deficient cells rely on non-canonical translational initiation mechanisms, as well as identify the specific molecular mechanisms deployed under conditions of low mTORC1 activity.

Impaired mTORC1 complex activity is also known to reduce anabolic processes such as protein and lipid biosynthesis and decrease translation of mitochondrial proteins and biogenesis to modulate cellular energy production [85, 86]. Under these conditions, cells exhibit an increased reliance on catabolic and scavenging mechanisms such as autophagy for survival [91, 92]. In fact, in previous work we showed that REST elevation in SHH-MB cells upregulates HIF1α expression and autophagy [40]. Interestingly, in earlier work we also identified a role for REST in the phosphorylation of AKT, tumor cell infiltration and leptomeningeal dissemination [48]. AKT is also shown to modulate mTORC1 complex activity through TSC1/2 and PRAS40 [93, 94]. Thus, it is possible that REST elevation and consequent silencing of USP37 expression may allow the survival of infiltrative and metastatic tumor cells through engagement of the cap-independent protein translation machinery as well as co-opting catabolic processes. These data point to a role for USP37 in the adaptation of REST-driven SHH MBs to cell stress, allowing for cell survival.

In contrast, in the absence of perturbations in REST levels in SHH-MBs or when USP37 expression is elevated and GLI1 is stabilized in a USP37-dependent manner, the upregulation of GLI1 target oncogene product, CCND1 drives proliferation of SHH-MB cells [95]. Work from other groups has shown that GLI1 is subject to both transcriptional and post-translational control by ubiquitination [96,97,98,99]. Our findings are also consistent with work by Qin et al, which showed that USP37 mediated the increase in GLI1 stability in breast cancer cells [67]. Although not tested, it may also facilitate an auto feedback loop to increase GLI1 transcription. Here, USP37 appears to function as an oncogene. Paradoxically, when REST is elevated and USP37 is downregulated, and presumably GLI1 levels are reduced, SHH signaling continues to be maintained by mechanisms that are yet to be elucidated. GLI1 is one of three GLI transcription factors, and its functional overlap with GLI2 may help maintain SHH pathway activity. When GLI1 is downregulated, GLI2 may partially compensate due to its role as an important activator of SHH signaling, a topic of ongoing investigation in our group [100,101,102]. It is also important to note that USP37 is not the only GLI1-specific DUB and HAUSP/USP7, USP48 and USP21, which have been identified as GLI1 deubiquitinating enzymes in non-MB systems, and whether these compensate for USP37 loss to allow proliferation of REST-high SHH-MBs remains to be investigated [103,104,105]

This work also highlights the paradoxical duality of USP37 function in MBs. This is not unique to USP37 since multiple publications in the literature support the functional duality of proteins as tumor promoters or suppressors [106]. SHH-MBs tumors have their origins in cerebellar granule precursor cells (CGNPs), and potentially cells at different stages of neuronal lineage commitment [107, 108]. REST elevation in these different neuronal lineage commitment states may give rise to the clusters of SHH-MB cells described in Fig. 1D and create unique cellular contexts in which USP37 interaction with various substrate(s) may occur. Alternatively spliced isoforms of USP37 are also suggested to exist, and their preferential expression under normal or elevated REST expression conditions may be an additional avenue for context-specific USP37-substrate interactions, which is also being actively explored in our laboratory.

Our findings described here support a growing body of literature which suggest that non-canonical activation of GLI1 may turn on SHH pathway signaling independent of SMO activation [96, 109, 110]. Critically, tumors (including MBs), with non-canonical activation of GLI signaling are less sensitive to SMO inhibitors [111,112,113]. GLI inhibitors are under consideration for other cancers, yet targeting GLI proteins is thought to be challenging. GANT61, which interferes with GLI-DNA binding, is the most promising of GLI antagonists, but its clinical use is restricted by its pharmacological properties [114]. SRI-38832, a derivative of GANT61 is more potent but is not commercially available [115]. ATO, which is FDA-approved against promyelocytic leukemia, has shown GLI inhibition in mouse models of MB and basal cell carcinoma [116]. If effective, GLI inhibitors may have promise against USP37-high SHH-MBs. While G9a is involved in the silencing of USP37 expression, the mechanisms underlying UPS37 upregulation in SHH-MBs are unclear and defining this process may allow for further therapeutic advances. However, the duality of USP37 function in SHH-MBs (Fig. 7), which our work clearly highlights, must be taken into consideration while effective therapeutic strategies are being developed.

Fig. 7: Model to explain the context-specific role of USP37 in SHH-MB cells.
figure 7

Under high REST conditions, USP37 is downregulated, leading to destabilization of Raptor and subsequent inactivation of mTORC1. This inactivation decreases cap-dependent protein translation initiation. Conversely, when USP37 levels are elevated, it stabilizes GLI1 protein to promote proliferation of SHH-MB cells.

Full size image