Ubiquitin Proteasome Gene Signatures in Ependymoma Molecular Subtypes

The ubiquitin proteasome system (UPS) is critically important for cellular homeostasis and affects virtually all key functions in normal and neoplastic cells. Currently, a comprehensive review of the role of the UPS in ependymoma (EPN) brain tumors is lacking but may provide valuable new information on cellular networks specific to different EPN subtypes and reveal future therapeutic targets. We have reviewed publicly available EPN gene transcription datasets encoding components of the UPS pathway. Reactome analysis of these data revealed genes and pathways that were able to distinguish different EPN subtypes with high significance. We identified differential transcription of several genes encoding ubiquitin E2 conjugases associated with EPN subtypes. The expression of the E2 conjugase genes UBE2C, UBE2S, and UBE2I was elevated in the ST_EPN_RELA subtype. The UBE2C and UBE2S enzymes are associated with the ubiquitin ligase anaphase promoting complex (APC/c), which regulates the degradation of substrates associated with cell cycle progression, whereas UBE2I is a Sumo-conjugating enzyme. Additionally, elevated in ST_EPN_RELA were genes for the E3 ligase and histone deacetylase HDAC4 and the F-box cullin ring ligase adaptor FBX031. Cluster analysis demonstrated several genes encoding E3 ligases and their substrate adaptors as EPN subtype specific genetic markers. The most significant Reactome Pathways associated with differentially expressed genes for E3 ligases and their adaptors included antigen presentation, neddylation, sumoylation, and the APC/c complex. Our analysis provides several UPS associated factors that may be attractive markers and future therapeutic targets for the subtype-specific treatment of EPN patients.


Introduction
Ependymomas (EPN) are glial neoplasms thought to arise from primitive ependymal lining cells of the ventricles in the brain and spinal cord of the central nervous system (CNS) and mainly occur in younger children and younger to mid-age adults. EPN account for 2-3% of primary brain tumors but 8-10% of all pediatric brain tumors. Many EPN patients present with hydrocephalus as this tumor may block the flow of cerebrospinal fluid. Based on their location within the CNS, EPN are divided into supra-tentorial (ST), posterior fossa (PF), and spinal cord (SP) groups. Each of these three EPN groups presents with three different molecular subtypes and some of which are associated with distinct oncogenic mechanisms and drivers. Adding to this complexity, all nine different EPN molecular subtypes have different age and gender distributions and prognostic values [1]. The EPN subtypes may originate from different progenitor cells that includes radial glial cells [2]. The ubiquitin proteasome system (UPS) serves to regulate proteasomal degradation of proteins and can also contribute to transcriptional regulation of genes [3]. The UPS contributes to the regulation of all major cellular processes, including signal transduction, cell cycle, growth, development, differentiation, metabolism, synaptic regulation, antigen presentation, apoptosis, and autophagy. In the present study, we have employed publicly available datasets to investigate the expression profiles of genes encoding components of the UPS in different molecular subtypes of human supratentorial, posterior fossa, and spinal EPN subgroups. Differential gene expression between the EPN molecular subtypes was determined by analysis of variance (ANOVA) through the R2 Genomics site.

Methods
We have analyzed a publicly available gene expression dataset on EPN molecular subtypes through the R2 Genomic analysis and visualization platform (https://r2.amc. nl). The expression of genes encoding UPS associated factors was compared between 8 subtypes of 209 EPN samples in the Pfister dataset (GEO ID: GSE64415). These EPN subtypes were classified as posterior fossa EPN a (PF_EPN_A), posterior fossa EPN b (PF_EPN_B), posterior fossa subependymoma (PF_SE), spinal cord EPN (SP_EPN), spinal cord myxopapillary ependymoma (SP-MPE), supratentorial EPN Rela (ST_EPN_RELA), supratentorial EPN Yap1 (ST_EPN_YAP1), and supratentorial subependymoma (ST_SE) [4]. The Pfister dataset did not include microarray data on the spinal sub-ependymoma (SP-SE) subgroup due to lack of availability of RNA from these samples. Analysis of variance of log2 expression data was used to determine significant differential gene expression. Unsupervised cluster analysis and illustration of clusters of UPS genes encoding E3 ligases and E3 ligase adaptors in the various EPN subtypes was done with the Morpheus heat map program (Morpheus software, https://software.broadinstitute.org/morpheus). The Cytoscape (version 3.8) with ClueGO V2.5.9 plug-in was used to identify pathways overrepresented in the list of genes encoding E3 ligases and E3 ligase adaptors, genes associated with these pathways, and network analysis. Our requirements for the definition of a Gene Ontology (GO) group were a minimum of 3 genes and at least 4% of the genes assigned to a biological pathway. The French glioma dataset, accessible through the R2 genomics site, was used to generate survival curves associated with the expression of selected genes. In the Pfister dataset, poorest overall survival was reported for the EPN-A and ST-EPN-RELA subtypes [4]. Survival data are not available in the R2 genomics dataset for this dataset.

Results
The Pfister EPN dataset included age at diagnosis in 88.5% (185 of 209) of cases (Table 1). For the PF-EPN-A and ST-EPN-RELA subtype, 83% and 75% of cases were between the ages of 0-10 years of age, respectively, whereas for the PF-EPN-B group all cases were >10 years of age. Because of the small sample size of the remaining groups, generalized statements about age of disease manifestation could not be made with confidence. The age distribution in the subtypes of EPN in this dataset is shown in Table 1.

Ubiquitin Conjugase (E2) Gene Expression in EPN Subtypes
In the Pfister data set, 31 genes encoding ubiquitin E2 conjugases were differentially expressed (p < 0.0001, log 2 data). Expression of several E2 conjugase encoding genes was elevated in the ST_EPN_RELA subgroup compared to the other seven groups, including UBE2C, UBE2S, UBE2T, UBZE2Q1, UBE2V2, UBE2D2 and UBZE2J2 (p < 0001 by Anova). Expression of UBE2E3 was specifically elevated in PF-EPN_A, while expression of UBE2Z was specifically elevated in SP-MPE (p < 0.0001). Reactome analysis of the 31 differentially expressed ubiquitin E2 conjugase genes identified the immune system as the main Reactome Pathway associations with these genes ( Table 2). The Reactome Pathway terms represented by over 80% (25/31) of these genes included "Antigen processing: ubiquitination and proteasome degradation" and "Class I MHC mediated antigen processing and presentation" (Group p value for over-expression = 4.95 × 10 −35 ). We concluded that the expression of these ubiquitin E2 conjugase genes was differentially stimulated by one or more transcription factors associated with the immune response. Reactome pathway analysis of the EPN data revealed differential expression of four genes encoding ubiquitin E2 conjugases (UBE2D1, UBE2D2, UBE2K, UBE2L6) that were associated with the DDX58/IFIH1 signaling pathway ( Table 2). The expression of the DHX58 gene was specifically elevated in the ST_EPN_RELA subtype (F = 10.82, p = 1.48 × 10 −11 ). DHX58 encodes a regulator for DDX58 (aka DExH-Box Helicase 58 or RIG1) which is an innate immune receptor that initiates proinflammatory and type I interferon induced signaling pathways in response to viral infections [10,11]. IFIH1 (Interferon Induced with Helicase C domain 1) expression was specifically elevated in the PF_EPN_B subtype ( Figure 2) and is part of an interferon response signature [12].

Ubiquitin E2 Conjugases and Patient Survival Times
In addition to UBE2C and UBE2S, the ST_EPN_RELA subtype demonstrated the over-expression of several other ubiquitin E2 conjugases, including UBE2I (p = 3.06         than that of all other groups by at least p < 0.05. The most significant difference was between the PF-EPN-A and the RELA groups (t = 9.88, p < 0.0001). By t-test mean of UBE2Z SP_MPE is significantly greater than that of all other groups, p < 0.0001. By Anova of UBE2Z, F = 28.73, p = 3.14 × 10 −27 . Dataset GSE64415.

Ubiquitin E3 Ligase Gene Expression in EPN Subtypes
Differentially expressed genes encoding E3 ligases were identified through the R2 genomics platform in the Pfister dataset by Anova. Expression of the 75 most significantly different E3 ligase genes was used in the statistical analysis. The heatmap in Figure 5 shows the clusters of E3 ligase genes associated with different EPN subtypes. Tables 4 and 5 summarize the statistically most significant Reactome Pathways and Reactome Reactions associated with the expression of the E3 ligase genes and identified the E3 ligase genes associated with these pathways. We identified several differentially expressed genes encoding ubiquitin E3 ligases by Reactome analysis as part of the immune response pathway in EPN.

Ubiquitin E3 Ligase Gene Expression in EPN Subtypes
Differentially expressed genes encoding E3 ligases were identified through the R2 genomics platform in the Pfister dataset by Anova. Expression of the 75 most significantly different E3 ligase genes was used in the statistical analysis. The heatmap in Figure 5 shows the clusters of E3 ligase genes associated with different EPN subtypes. Tables 4 and  5 summarize the statistically most significant Reactome Pathways and Reactome Reactions associated with the expression of the E3 ligase genes and identified the E3 ligase genes associated with these pathways. We identified several differentially expressed genes encoding ubiquitin E3 ligases by Reactome analysis as part of the immune response pathway in EPN. The most significant Reactome Pathway in the 75 most significant differentially expressed ubiquitin E3 ligases in the Pfister EPN dataset, as determined by the Reactome program ClueGo application, was "Antigen processing: ubiquitination & proteasome degradation". The 13 genes identified in this pathway were the same as the 13 genes identified in the "Class I MHC mediated antigen processing & presentation pathway". Because our cutoff point was 4% of the total genes in the pathway, only the first pathway was included in Table 4. The data suggested a role for several ubiquitin E3 ligases in Class I MHC mediated antigen processing. The Reactome network created by the GlueGo application is shown in Figure 6. Table 4. Main Reactome pathways over-represented in list of E3 ligase genes of Figure 5.

Reactome Term
No.  HDAC4 (Histone deacetylase 4) was the most significantly up-regulated ubiquitin E3 ligase gene in the ST_EPN_RELA subtype (F = 44.88, p = 7.86 × 10 −38 ) ( Figure 7; see also heatmap in Figure 5). Multifunctional HDAC4 encodes a protein that can act as a histone deacetylase and epigenetic regulator of transcription as confirmed by our GO analysis of the EPN dataset, which identified HDAC4 as the most differentially expressed gene of all genes in the GO category of "Epigenetic regulation of gene expression". However, the HDAC4 protein also contains a ubiquitin ligase domain which exhibits SUMO E3 ligase activity and is therefore included in the heatmap of E3 ligases ( Figure 5) [17][18][19].
Illustrated in Figure 6, is the Reactome network "Analysis of the Gene Ontology (GO)" pathways emerging from the list of 75 top ubiquitin E3 ligase genes. It depicts the relationship between Notch signaling, HDAC4, and SUMOylation of intracellular receptors in the cluster of differentially expressed E3 ligase genes of Figure 5.
The expression of several Notch target genes [20,21] was elevated in specific EPN subtypes. Among these Notch target genes, expression of CCND1 (F = 51.  Figure 5). Multifunctional HDAC4 encodes a protein that can act as a histone deacetylase and epigenetic regulator of transcription as confirmed by our GO analysis of the EPN dataset, which identified HDAC4 as the most differentially expressed gene of all genes in the GO category of "Epigenetic regulation of gene expression". However, the HDAC4 protein also contains a ubiquitin ligase domain which exhibits SUMO E3 ligase activity and is therefore included in the heatmap of E3 ligases ( Figure 5) [17][18][19]. Illustrated in Figure 6, is the Reactome network "Analysis of the Gene Ontology (GO)" pathways emerging from the list of 75 top ubiquitin E3 ligase genes. It depicts the relationship between Notch signaling, HDAC4, and SUMOylation of intracellular receptors in the cluster of differentially expressed E3 ligase genes of Figure 5.  Our analysis of the Pfister EPN dataset on proteins with E3 ligase domains suggests a subtype-specific contribution of Notch signaling to the development of EPN. In support of these data, a recent study of de Almeida Magalhaes et al. [22] has documented the overexpression of NOTCH1, JAG1, JAG2, and HES4 in the ST_EPN_RELA subgroup.

Differentiation and DNA Repair Factors Are UPS Targets in Selected EPN Subtypes
The Pfister EPN dataset showed a selective increase in expression of the tumor invasion-promoting L1CAM gene [23] in the ST_EPN_RELA subtype at a very high level of significance (by Anova, F = 175, p = 1.35 × 10 −81 ) ( Figure 9). The L1CAM encodes a protein which may serve as a marker for the SP_EPN_RELA subtype in the Pfister dataset [24]. Ubiquitination facilitates L1CAM protein degradation at the cell surface [25]. Additionally, over-expressed in ST_EPN_RELA was the gene encoding the stem cell marker and non-canonical Notch inhibitor, DLK1 (F = 6.81, p = 2.96 × 10 −7 ). Our analysis of the Pfister EPN dataset on proteins with E3 ligase domains suggests a subtype-specific contribution of Notch signaling to the development of EPN. In support of these data, a recent study of de Almeida Magalhaes et al. [22] has documented the overexpression of NOTCH1, JAG1, JAG2, and HES4 in the ST_EPN_RELA subgroup.

Differentiation and DNA Repair Factors Are UPS Targets in Selected EPN Subtypes
The Pfister EPN dataset showed a selective increase in expression of the tumor invasion-promoting L1CAM gene [23] in the ST_EPN_RELA subtype at a very high level of significance (by Anova, F = 175, p = 1.35 × 10 −81 ) ( Figure 9). The L1CAM encodes a protein which may serve as a marker for the SP_EPN_RELA subtype in the Pfister dataset [24]. Ubiquitination facilitates L1CAM protein degradation at the cell surface [25]. Additionally, over-expressed in ST_EPN_RELA was the gene encoding the stem cell marker and non-canonical Notch inhibitor, DLK1 (F = 6.81, p = 2.96 × 10 −7 ).
The R2 program identified three genes in the homeobox cluster of chromosome 17 (at 17q21.32) that were highly over-expressed selectively in SP_MPE. This included HOXB9 (F = 25.80, p = 5.26 × 10 −25 ), HOXB13 (F = 160.28, p = 1.21 × 10 −78 ; >150-fold), and PRAC1 (F = 953.53, p = 1.96 × 10 −150 ; >150 fold). Expression of the homeobox genes is shown in Figure 11 and may provide suitable new developmental markers of the SP_MPE subtype. HNF1B (Hepatic Nuclear Fact 1B, aka HNF1 homeobox B) is a gene located on mosome 17 (17q12) and encodes a transcription factor that regulates Notch signaling and facilitates stem cell differentiation [27]. In the EPN dataset, HNF1B expression selectively elevated by approximately 20-fold (F = 151.68, p = 1.27 × 10 −76 ) in the SP_ subgroup ( Figure 10). Acetylation of HNF1B attenuates ubiquitin-proteasomal degr tion [28].  HNF1B (Hepatic Nuclear Fact 1B, aka HNF1 homeobox B) is a gene located on chro mosome 17 (17q12) and encodes a transcription factor that regulates Notch signaling [26 and facilitates stem cell differentiation [27]. In the EPN dataset, HNF1B expression wa selectively elevated by approximately 20-fold (F = 151.68, p = 1.27 × 10 −76 ) in the SP_MPE subgroup ( Figure 10). Acetylation of HNF1B attenuates ubiquitin-proteasomal degrada tion [28]. The R2 program identified three genes in the homeobox cluster of chromosome 17 (at 17q21.32) that were highly over-expressed selectively in SP_MPE. This included HOXB9 (F = 25.80, p = 5.26 × 10 −25 ), HOXB13 (F = 160.28, p = 1.21 × 10 −78 ; >150-fold), and PRAC1 (F = 953.53, p = 1.96 × 10 −150 ; >150 fold). Expression of the homeobox genes is shown in Figure 11 and may provide suitable new developmental markers of the SP_MPE subtype.  Additonal homeobox genes, HOXC13 (F = 32.41, p = 7.04 × 10 −30 ) and HOXA13 (F = 66.90, p = 4.13 × 10 −49 ), located on the homebox clusters of chromosome 12 (12q13.13) and chromosome 7 (7p15.2), respectively, were also over-expressed in SP_MPE. The E3 ligases RNF20 and RNF40 have been shown to regulate many homeobox genes [29]. Expression of several other markers for the SP_MPE subtype were highly correlated with the expression of the homeobox genes. The NEFL (Neurofilament Light Chain) gene was found over-expressed in the SP_MPE subtype by more than 150-fold (F = 1505.18, p = 7.05 × 10 −170 ) compared ot the other groups ( Figure 12) and confirmed a previous report by Barton et al. [30]. The NEFL protein is a substrate of the ubiquitin ligase Trim2 [31,32]. Additionally, substantially increased in the SP_MPE subtype was the expression of the genes EPHB2 (F = 64.970, p = 3.13 × 10 −48 ) and P2RX5 (F = 86.573, p = 3.35 × 10 −57 ). The ubiquitin ligase SPSB4 regulates EPHB2 [33], while the UPS regulation of P2RX5 has not been reported. In summary, the data suggest that NEFL, EPHB2, and P2RX5 are useful genetic markers for the SP_MPE group and their proteins should be investigated further as to their potential as therapeutic targets. RNF20 and RNF40 have been shown to regulate many homeobox genes [29]. Expressio of several other markers for the SP_MPE subtype were highly correlated with the expres sion of the homeobox genes. The NEFL (Neurofilament Light Chain) gene was found over expressed in the SP_MPE subtype by more than 150-fold (F = 1505.18, p = 7.05 × 10 −17 compared ot the other groups ( Figure 12) and confirmed a previous report by Barton e al. [30]. The NEFL protein is a substrate of the ubiquitin ligase Trim2 [31,32]. Additionally substantially increased in the SP_MPE subtype was the expression of the genes EPHB2 ( = 64.970, p = 3.13 × 10 −48 ) and P2RX5 (F = 86.573, p = 3.35 × 10 −57 ). The ubiquitin ligase SPSB regulates EPHB2 [33], while the UPS regulation of P2RX5 has not been reported. In sum mary, the data suggest that NEFL, EPHB2, and P2RX5 are useful genetic markers for th SP_MPE group and their proteins should be investigated further as to their potential a therapeutic targets. Located on chromosome 17 at 17q21.32, IGF2BP1 is a regulator of RNA stability an over-expression of IGF2BP1 is associated with tumor progression and poor prognosis i a variety of cancers [34]. Of all EPN subtypes, IGF2BP1 expression was selectively elevate in PF_EPN_A (F = 30.48, p = 1.66 × 10 −28 ) (Figure 13). IGF2BP1 is a substrate of the E3 ligas adaptor FBXO45 [35] and potential cancer therapeutic target [36]. Elevated protein level of the ubiquitin E3 ligase adaptor FBXO45 promote tumor progression by stimulatin ubiquitination of IGF2BP1 in hepatic cancer [35]. Over-expression of IGF2BP1 is assoc ated with poor prognosis in a variety of cancers [34]. FBXO45 is a therapeutic target i cancer [36] and IGF2BP1 may also be considered a potential therapeutic target i PF_EPN_A. Located on chromosome 17 at 17q21.32, IGF2BP1 is a regulator of RNA stability and over-expression of IGF2BP1 is associated with tumor progression and poor prognosis in a variety of cancers [34]. Of all EPN subtypes, IGF2BP1 expression was selectively elevated in PF_EPN_A (F = 30.48, p = 1.66 × 10 −28 ) (Figure 13). IGF2BP1 is a substrate of the E3 ligase adaptor FBXO45 [35] and potential cancer therapeutic target [36]. Elevated protein levels of the ubiquitin E3 ligase adaptor FBXO45 promote tumor progression by stimulating ubiquitination of IGF2BP1 in hepatic cancer [35]. Over-expression of IGF2BP1 is associated with poor prognosis in a variety of cancers [34]. FBXO45 is a therapeutic target in cancer [36] and IGF2BP1 may also be considered a potential therapeutic target in PF_EPN_A.
Expression of the transcription factor GATA4 was selectively elevated in the ST_SE subgroup ( Figure 13). GATA4 is regulated by sumolyation via the E3 ligase PIAS1 [37].
The ubiquitin E3 ligase RNF8 plays an important role in response to DNA repair and interacts with E2 conjugase, UBE2N (UBC13), in response to DNA damage [38]. RNF8 is recruited to double strand breaks where it functions as the major E3 ligase for polyubiquitination of DNA damage related proteins [38,39]. In the Pfister EPN dataset, RNF8 expression was elevated in the SP_EPN subtype (F = 11.86, p = 1.28 × 10 −12 ), providing evidence for a transcriptional EPN subtype specific UPS gene signature on DNA damage response pathways (Figure 14). Expression of the transcription factor GATA4 was selectively elevated in the ST_SE subgroup ( Figure 13). GATA4 is regulated by sumolyation via the E3 ligase PIAS1 [37].
The ubiquitin E3 ligase RNF8 plays an important role in response to DNA repair and interacts with E2 conjugase, UBE2N (UBC13), in response to DNA damage [38]. RNF8 is recruited to double strand breaks where it functions as the major E3 ligase for poly-ubiquitination of DNA damage related proteins [38,39]. In the Pfister EPN dataset, RNF8 expression was elevated in the SP_EPN subtype (F = 11.86, p = 1.28 × 10 −12 ), providing evidence for a transcriptional EPN subtype specific UPS gene signature on DNA damage response pathways ( Figure 14).   Expression of the transcription factor GATA4 was selectively elevated in the ST_SE subgroup ( Figure 13). GATA4 is regulated by sumolyation via the E3 ligase PIAS1 [37].
The ubiquitin E3 ligase RNF8 plays an important role in response to DNA repair and interacts with E2 conjugase, UBE2N (UBC13), in response to DNA damage [38]. RNF8 is recruited to double strand breaks where it functions as the major E3 ligase for poly-ubiquitination of DNA damage related proteins [38,39]. In the Pfister EPN dataset, RNF8 expression was elevated in the SP_EPN subtype (F = 11.86, p = 1.28 × 10 −12 ), providing evidence for a transcriptional EPN subtype specific UPS gene signature on DNA damage response pathways (Figure 14).

Ubiquitin Ligase E3 Adaptor Gene Expression in EPN Subtypes
Cluster analysis revealed that ubiquitin E3 ligase adaptor gene expression was distinctly different in EPN subtypes ( Figure 15). Reactome Pathway analysis identified a significant association of ubiquitin E3 ligase adaptor genes depicted in the heatmap with the Neddylation pathway (12 genes) and genes associated with the APC/c (Anaphase promoting complex/cyclosome) E3 ligase complex (3 genes) ( Table 6). Figure 16 illustrates the various steps in the neddylation cycle.

Ubiquitin Ligase E3 Adaptor Gene Expression in EPN Subtypes
Cluster analysis revealed that ubiquitin E3 ligase adaptor gene expression was distinctly different in EPN subtypes ( Figure 15). Reactome Pathway analysis identified a significant association of ubiquitin E3 ligase adaptor genes depicted in the heatmap with the Neddylation pathway (12 genes) and genes associated with the APC/c (Anaphase promoting complex/cyclosome) E3 ligase complex (3 genes) ( Table 6). Figure 16 illustrates the various steps in the neddylation cycle.   Table 6. Main Reactome Pathways overrepresented by ubiquitin E3 ligase adaptor genes depicted in Figure 15. Neddylation is essential for cell cycle progression [40][41][42] and dysfunction of neddylation regulation of E3 ligase complexes controlling the cell cycle contributes to one or more EPN subtypes. Neddylation reactions that were most significantly associated with the E3 ligase adaptor genes of Figure 15 are listed in Table 7. This includes DCAF7, DCAF10, DCAF16, DDB2 genes, which encode proteins that serve as substrate recognition adaptors to CUL4 E3 ligase complexes. In particular, the DCAF genes encode substrate receptor proteins for CUL4-DDB1 E3 ligase complexes, whereas DDB2 (Damage Specific DNA Binding Protein 2) encodes a substrate adaptor protein which is also a DNA damage sensor and part of the nucleotide excision repair (NER) pathway) [43][44][45]. Another group of neddylation associated genes encodes substrate recognition adaptors for Cullin 1 E3 ligase complexes (FBXL13, FBXL14, FBX015, FBXO31) and for Cullin Ring 4 and Cullin Ring 5 E3 ligase adaptors (CISH, GAN, KCTD6 and KLHL42) ( Table 7). Expression of the gene encoding the neddylation activating enzyme UBA3 (aka NAE2) was elevated in the ST_EPN_RELA subtype, suggesting enhanced neddylation as a relevant biological process in this EPN subtype. Elevated expression of several E3 adaptors associated with the Neddylation Reactome Pathway (Table 6) was identified in ST_EPN_RELA (FBXO31, KLHL42, and CISH), in PF_EPN_A (KCTD6), and in PF_SE (GAN) (Figures 17-19   DNA Binding Protein 2) encodes a substrate adaptor protein which is also a DNA damage sensor and part of the nucleotide excision repair (NER) pathway) [43][44][45]. Another group of neddylation associated genes encodes substrate recognition adaptors for Cullin 1 E3 ligase complexes (FBXL13, FBXL14, FBX015, FBXO31) and for Cullin Ring 4 and Cullin Ring 5 E3 ligase adaptors (CISH, GAN, KCTD6 and KLHL42) ( Table 7). Expression of the gene encoding the neddylation activating enzyme UBA3 (aka NAE2) was elevated in the ST_EPN_RELA subtype, suggesting enhanced neddylation as a relevant biological process in this EPN subtype. Elevated expression of several E3 adaptors associated with the Neddylation Reactome Pathway (Table 6) was identified in ST_EPN_RELA (FBXO31, KLHL42, and CISH), in PF_EPN_A (KCTD6), and in PF_SE (GAN) (Figures 17-19).     Currently, little is known about the regulation and functional impact of neddylation of E3 ligases and adaptors in EPN. An increase in the expression of the protein SCCRO3 encoded by DCUN1D3 (Defective in Neddylation 1D3) was shown to reduce neddylation of F-Box ubiquitin ligases (Cullin 1 ligases). SCCRO3 regulates neddylation by binding to CRL1 ligases and the protein CAND1 to block the nuclear translocation of E3 ligases [46], which contributes to cell cycle progression upon UV damage [47]. In the EPN dataset, expression of the DCUN1D3 gene was elevated in the SP_MPE subgroup (F = 11.62, p = Currently, little is known about the regulation and functional impact of neddylation of E3 ligases and adaptors in EPN. An increase in the expression of the protein SCCRO3 encoded by DCUN1D3 (Defective in Neddylation 1D3) was shown to reduce neddylation of F-Box ubiquitin ligases (Cullin 1 ligases). SCCRO3 regulates neddylation by binding to CRL1 ligases and the protein CAND1 to block the nuclear translocation of E3 ligases [46], which contributes to cell cycle progression upon UV damage [47]. In the EPN dataset, expression of the DCUN1D3 gene was elevated in the SP_MPE subgroup (F = 11.62, p = 2.22 × 10 −12 ), suggesting a dysregulation of neddylation of E3 ligases by SCCRO3 protein in this EPN subtype.
The several-fold elevated expression of FBXO31 and its substrate CCND1 selectively in ST_EPN_RELA (F = 51.39, p = 1.77 × 10 −41 ) may contribute to cell cycle dysregulation in ST_EPN_RELA ( Figure 17). The E3 ligase adaptor protein FBXO31 was identified as a tumor suppressor in breast cancer [48] and was shown to stimulate cell proliferation, invasion and metastasis in lung cancer [49]. FBXO31 binds to the cell cycle regulator cyclin D1 (CCND1) to cause the ubiquitination and degradation of CCND1 [50]. Over-expression of FBXO31 after DNA damage results in cell cycle arrest in G1 [51].

Ubiquitin E3 Ligase Adaptors and Viral Markers
A role for human cytomegalovirus (HCMV) in a variety of tumors has recently been proposed due to the widespread detection of this virus in human cancers [53]. Table 7 summarizes evidence for an involvement of several different ubiquitin E3 ligase adaptors (DYNC1I1, DYNC1I2, NUP37, RAE1) associated with HCMV infection. Some viruses can facilitate the transport of their genome into the nucleus through nuclear pore complexes [54,55] and the Reactome program identified the HCMV nuclear pore docking pathway in the EPN dataset. NUP37 is a marker for various cancers and is required for the assembly of the nuclear pore complex and mitosis [56,57]. In the EPN dataset, NUP37 expression was elevated in the PF_EPN_B and SP_EPN subtypes (Figure 20). Expression of DYNC1I1 was elevated in SP_EPN and expression of DYNC1I2 was depressed in PF_EPN_B ( Figure 21). DYNC1I1 and DYNC1I2 encode E3 ligase adaptor proteins that are part of the Dynein transport system along microtubules [58]. Dynein has been shown to interact with components of the nuclear pore.

Ubiquitin E3 Ligase Adaptors and Viral Markers
A role for human cytomegalovirus (HCMV) in a variety of tumors has recently been proposed due to the widespread detection of this virus in human cancers [53]. Table 7 summarizes evidence for an involvement of several different ubiquitin E3 ligase adaptors (DYNC1I1, DYNC1I2, NUP37, RAE1) associated with HCMV infection. Some viruses can facilitate the transport of their genome into the nucleus through nuclear pore complexes [54,55] and the Reactome program identified the HCMV nuclear pore docking pathway in the EPN dataset. NUP37 is a marker for various cancers and is required for the assembly of the nuclear pore complex and mitosis [56,57]. In the EPN dataset, NUP37 expression was elevated in the PF_EPN_B and SP_EPN subtypes (Figure 20). Expression of DYNC1I1 was elevated in SP_EPN and expression of DYNC1I2 was depressed in PF_EPN_B ( Figure  21). DYNC1I1 and DYNC1I2 encode E3 ligase adaptor proteins that are part of the Dynein transport system along microtubules [58]. Dynein has been shown to interact with components of the nuclear pore.

Anaphase Promoting Complex/Cyclosome (APC/c) E3 Ligase Adaptors
Three genes encoding for E3 ligase adaptors were identified by the Reactome program as genes in the heatmap of Figure 15 that encode for components of the APC/c ubiquitin ligase complex, including ANAPC1, ANAPC5, and ANAPC7. These genes are associated with the degradation of cell cycle proteins Cyclin A, Cyclin B, and securin. We sug-

Anaphase Promoting Complex/Cyclosome (APC/c) E3 Ligase Adaptors
Three genes encoding for E3 ligase adaptors were identified by the Reactome program as genes in the heatmap of Figure 15 that encode for components of the APC/c ubiquitin ligase complex, including ANAPC1, ANAPC5, and ANAPC7. These genes are associated with the degradation of cell cycle proteins Cyclin A, Cyclin B, and securin. We suggest a role for the APC/c complex in EPN oncogenesis and implicate dysregulation of the APC/c complex as a contributor towards a subtype dependent EPN pathogenesis. This is supported by data showing that UBCH10 (UBE2C) initiates ubiquitination of the APC/c ubiquitin E3 ligase complex [14][15][16] and that the expression of UBE2C, UBE2S and ANAPC1, are elevated in the ST_EPN_RELA subtype (Figure 3).

Discussion
Our analysis of the EPN data sets identified several genes encoding ubiquitin E1 activators, E2 conjugases, and E3 ligases and adaptors of the UPS pathway that were highly differentially expressed in specific EPN subtypes. GO and Reactome analyses provided valuable information on a diverse set of probable biological pathways supported by the UPS genes in EPN subtypes.

Setting the UPS Stage: Ubiquitin E1 Activator
Forming a heterodimer with NAE1 that activates and transfers NEDD8 to ubiquitin conjugases (Figure 16), the ubiquitin activator UBA3 (aka Neddylation activating enzyme NAE2) was significantly elevated in the ST_EPN_RELA subtype ( Figure 1). This may suggest a particular involvement of UBA3/NAE2 gene product in enhanced activation of NEDD8 in the ST_EPN_RELA subtype. The notion that neddylation is a major pathway distinguishing the EPN subtypes was further supported by our analysis of ubiquitin ligase adaptors (Tables 6 and 7), which were also associated with an aberrant neddylation pathway in some EPN subtypes. In the order of significance, high gene expression of UBA1, UBA6, and UBA3/NAE2 were associated with significantly shorter survival times in the French glioma dataset (chi = 26.79, p = 2.3 × 10 −7 , chi = 21.04, p = 4.5 × 10 −6 , chi = 8.79, p = 3.0 × 10 −3 , respectively). TAK-243 is a small molecular inhibitor of the proteins UBA1, UBA6 and NAE, used for cancer treatment in pre-clinical animal models [59,60]. TAK-243 is also undergoing Phase 1 clinical trials in patients with refractory acute myeloid leukemia (AML) to block UBA1, UBA6, and NAE targeted proteins involved in tumor cell division (ClinicalTrials.gov Identifier: NCT03816319). A neddylation inhibitor MLN4924 (aka Pevonedistat) that interacts with UBA3/NAE2 [61] blocks activation of many cullin ring ligases. Pevonedistat has been shown to have therapeutic effects in several, but not all, cancer clinical trials [62][63][64], and is undergoing clinical trials for glioblastoma [65,66]. The E2 conjugase protein UBE2Z (aka UBA6-specific enzyme) is a substrate of the UBA6 E1 activator. The expression of UBE2Z gene was elevated in the SP-MPE subtype relative to the other EPN subtypes (F = 28.73, p = 3.14 × 10 −27 ), highlighting UBE2Z as a possible specific therapeutic target for the SP_MPE subytpe.

Immune Recognition and Inflammation
The UPS plays a critical role in regulation of the immune response [67] and is required to process cytoplasmic proteins into MHC class I antigenic peptides prior to surface presentation on antigen presenting cells for recognition by T cells [68,69]. The proteasome responds to immunological challenges with specific modifications that result in the formation of an immunoproteasome, in which the 3 catalytic subunits, β1, β2, and β7 have been replaced by catalytic immuno-subunits LMP2, MECL-2, and LMP7 under the influence of interferon gamma (IFN-γ) [68]. A major Reactome pathway associated with differential expression of E2 ubiquitin conjugase genes (other than the E2 category itself) was "Antigen processing: ubiquitination and proteasome degradation" ( Table 2). Both ubiquitin E2 conjugases, UBE2H (aka UBC8) and UBE2L6 (important in ISGylation), are interferon-inducible ubiquitin E2 conjugases and may be involved in the process of antigen presentation and EPN differentiation (Table 2) [70][71][72]. The UPS can affect antigen presentation to attenuate immune recognition and prevent destruction by reactive humoral and cellular immune responses [73]. Coinciding with our finding of differential expression of genes encoding ubiquitin E2 conjugases in this pathway (Table 2), several ubiquitin E3 ligases in EPN subtypes were also associated with the Reactome Pathway of Antigen processing: ubiquitination & proteasome degradation (Table 4). Ubiquitination of the MHC class I heavy chain is required for translocation into the cytosol [74,75], where antigenic peptides are produced by the 26S proteasome. Membrane-anchored E3 ligases of the Modulator of Immune Recognition (MIR) family of E3 ligases affect antigen presentation by regulating lysosomal degradation of MHC class I antigens [76]. Multiple E3 ubiquitin ligase genes of the TRIM family are located in the MHC Class I region of chromosome 6 [77]. This includes TRIM27, a gene included in the heatmap of differentially expressed E3 ligases in EPN ( Figure 5). The network analysis of Figure 6 illustrates the interaction of other E3 ligases on TRIM27 expression. The expression is related to inflammatory disesases and viral infections [77]. While some TRIM proteins have been associated with cancer [78], we could find no previous reports relating TRIM27 to EPN tumorigenesis.
Several lines of evidence in the UPS data suggested an association between cellular responses to viral infection and inflammation in EPN and warrant further investigations into viral contributions to EPN subtypes. Primarily with the PF_EPN_B subtype, we identified the Reactome Pathway "HCMV nuclear pore docking pathway" associated with differential expression of genes encoding E3 adaptors that are part of the dynein motor complex (Table 7). Interferon gamma (IFNγ) stimulates many genes associated with the MHC class 1 immune response [79]. Induction of IFNγ promotes immunoproteasome functionalities to facilitate the degradation of endogenous proteins into antigenic peptides for binding to MHC class I proteins prior to translocation to the cell membrane and presentation of these MHC class I-peptide complexes to CD8 T cells [68]. The induction of IFNβ-and IFNγ-inducible genes is regulated by the ubiquitin E3 ligase PIAS1 (Protein Inhibitor of activated Stats 1) [80]. Expression changes in UPS genes identified in both ST_EPN_RELA and PF_EPN_B included several differentially expressed UPS genes associated with interferon gamma signaling (MID1, PIAS1, TRIM2, TRIM22, TRIM45) that may promote a proinflammatory milieu similar to that observed upon viral infections (Table 4). INFγ stimulates the transcription of PSMB8 (LMP7), PSMB9 (LMP2), PSMB10 (LMP10) immunoproteasome subunit genes, proteasome activator genes PSME1 (PA28α) and PSME2 (PA28β), the transporters associated with antigen processing, TAP1 and TAP2, and the MHC class 1 genes, HLA.A, HLA.B, and HLA.C [79]. A number of TRIM E3 ligase proteins, including TRIM2 and TRIM22, were specifically down-regulated in PF-EPN_A but not PF-EPB_B. TRM2 and TRIM22 play a role in antiviral immunity [81][82][83] and are either induced and/or activated by interferon gamma [84]. The UPS also appears to have functions associated with early cellular warning systems of viral infections. The IFIH1 gene product, melanoma differentiation associated gene-5 (MDA5), and the DDX58 gene product retinoic acid inducible gene-I (RIG-I) are intracellular RNA sensors and part of an elaborate cellular defense system against viral challenges that stimulates type 1 interferon production [85]. This includes endoplasmic reticulum (ER)-associated virus inhibitory protein RSAD2, radical S-adenosyl methionine domain containing 2, which is a downstream target of pattern RNA sensing Toll-like receptor (TLR) 7, and also selectively increased in PF-EPB_B. These data suggest that the expression changes in UPS genes identified in both ST_EPN_RELA and PF_EPN_B may promote a proinflammatory milieu similar to that observed upon viral infection.

UPS and NOTCH Signaling
HDAC4 protein stimulates IFNγ (Type 2 interferon signaling) [86] and HDAC inhibition is considered a potential therapeutic strategy in pediatric brain cancers [87,88]. Although the effect of HDAC inhibitors on acetylation-dependent and acetylation-independent mechanisms is well documented, we were unable to identify studies that determined the effect of HDAC inhibitors on ubiquitin E3 ligase/SUMOylation activity of HDAC4. The regulation of HDAC4 gene expression is incompletely understood but emerging evidence suggests an involvement of post-translational protein modifications, because neddylation deficiency was reported to enhance the HDAC4 mRNA and protein expression in myoblasts [88]. Importantly, HDAC enzymes can inhibit Notch signaling but a role of HDAC4 ubiquitin E3 ligase activity in this signaling process remains largely unclear [89,90]. Our Reactome network analysis identified a relationship of differentially expressed E3 ligases and Notch signaling pathway ( Figure 6). Many Notch target genes have been identified that show cell-type specific expression [91] and contribute to unique roles in Notch directed context-dependent development and differentiation in different cell types [21]. Notch nuclear signaling via its intracellular domain (NICD) stimulates transcription of developmental genes linked to cancer [91][92][93]. Ubiquitin E3 ligases continue to emerge as important post-translational modulators of NOTCH signaling. Deltex E3 ligase proteins have key functions in the cleavage and endocytosis of the Notch receptor and the release of its NICD domain [94][95][96]. These Deltex E3 ligase proteins are encoded by the Deltex genes DTX1, DTX2, DTX3, DTX3L, and DTX4. SUMOylation was reported to promote enhanced nuclear localization of NICD but decrease expression of some Notch target genes during in vitro cell stress [97]. HDAC4 is attracted to the NICD transcriptional complex during stress-induced SUMOylation and has SUMO E3 ligase activity [18,19]. Phosphorylation of NICD followed by ubiquitination by an E3 ligase complex leads to NICD degradation [91]. The ubiquitin ligase adaptor for this complex, FBXW7 (aka FBW7) [98], is one of the most commonly dysregulated UPS proteins in human cancers [91,99]. Inhibition of FBXW7 would be expected to attenuate the degradation of NICD protein. Our analysis implicate over-expression of a subset of Notch target genes as part of an oncogenic molecular signature of ST_EPN_RELA. In addition, in supratentorial (ST) EPN, NOTCH1 expression was associated with cancer stem cell (CSC) markers, VEGFA and L1CAM, and the ST_EPN_RELA subtype showed the activation of selected key members of the Notch signaling pathway (NOTCH1, JAG1, JAG2, HES4) [22]. The L1CAM gene was strong marker for the ST_EPN_RELA subtype ( Figure 9). Ubiquitination of full length L1CAM protein regulates its lysosomal degradation and intracellular trafficking [25], while SUMOylation of a smaller 70 kDa intracellular fragment is required for nuclear translocation [100]. Lutz et al. [100] found that the L1CAM 70 kDa fragment was generated in the mouse primarily before postnatal Day 3.

The Anaphase Promoting Complex/Cyclosome (APC/c)
The role of the APC/c complex in cancer development and progression has been reviewed by VanGendersen and colleagues [101]. Dysregulated gene transcription of E2 conjugases and E3 ligase adaptors of the APC/c complex was detected in the ST-EPN-RELA subgroup. The APC/c complex is an ubiquitin E3 ligase complex that regulates mitotic exit by stimulating the transition from metaphase to anaphase by ubiquitination and proteasomal degradation of securin and cyclin B [102]. UBE2C and UBE2S ubiquitinate the APC/c ubiquitin ligase complex, facilitating the conversion of APC/c -Cdc20 to APC/c-Cdh1 which is required for the transition to anaphase in the cell cycle [14][15][16]. Increased expression of UBE2C (F = 9.92, p = 1.28 × 10 −10 ) observed in ST_EPN_RELA of the Pfister dataset ( Figure 3) has been associated with aggressive progression and poor outcome of malignant glioma [103]. Over-expression of UBE2C protein stimulates cell proliferation [104,105], may act as an important proto-oncogene [105,106] and is found over-expressed in several cancers [105][106][107][108][109][110][111][112][113]. Donato and colleagues proposed suppression of UBE2C (UBCH10) as a potential therapy for astrocytic brain tumors [114]. UBE2C may also be a therapeutic target in breast cancer [115] and ovarian cancer [112]. Increased expression of UBE2C in PF-EPN-RELA ( Figure 3) suggests that this supratentorial EPN may also benefit from UBE2C targeted therapies. ST-EPN_RELA also showed increased expression of UBE2S (F = 16.00, p = 1.19 × 10 −16 ) (Figure 3). The UBE2S protein was recently described as an oncogene with potential as a therapeutic target in various cancers [116]. The elevated expression of UBE2S may qualify as a genetic marker for ST-EPN-RELA and its gene product may present a promising therapeutic target in EPN. Additionally, differentially expressed in EPN subtypes were the ubiquitin ligase adaptors of the APC/c ubiquitin ligase complex, ANAPC1, ANAPC5, and ANAPC7, with functions in mitotic exit ( Table 7). The gene coding for a protein of ANAPC1 was up-regulated (F = 22.95, p = 1.0 × 10 −22 ) in the ST-EPN-RELA subtype relative to the other EPN subtypes, while expression of ANAPC5 was depressed relative to the other EPN subtypes (F = 11.26, p = 5.22 × 10 −12 ). The data suggest dysregulation of the APC/c complex in ST_EPN_RELA and possibly in one or more additional EPN subtypes.

Neddylation
A main Reactome Pathway over-represented by ubiquitin ligase adaptor genes in EPN subtypes was the Neddylation pathway ( Figure 15; Table 6). The Neddylation cycle ( Figure 16) includes various steps and contributes to oncogenesis by affecting cell cycle regulation, autophagy, DNA repair, and genomic stability [42]. Our analysis suggests that transcription of all three levels of the UPS pathway, E1 activation, E2 conjugation, and E3 ubiquitination, are modulated in a subtype specific manner. In EPN, transcriptional regulation of genes associated with the Neddylation pathway appeared to contribute to EPN oncogenesis in a subtype-specific manner and these genes may serve as genetic markers for specific EPN subtypes. The use of specific inhibitors for Neddylation components, such as UBE2M inhibitors [42] or MLN4924, which induced cell cycle arrest in human osteosarcoma [117], may be attractive new avenues for the treatment of the ST_EPN_RELA subtype but the development of more specific inhibitors of CRLs is warranted to benefit EPN patients. The Reactome reactions of the E3 ligase adaptor genes in Table 7 provided further information on differential regulation of neddylation in the EPN subtypes.

Linking UPS with Genomic Stability and DNA Repair in EPN Subtypes
Ubiquitin E2 conjugase genes were expressed greastest in ST_EPN_RELA and included UBE2T, UBE2J2, UBE4B, UBE2Q1, UBE2V2 and UBE2I. Their role in EPN is currently largely unknown. UBE2T is part of the Fanconi Anemia (FA) core complex and required for the activation of the FA pathway in DNA repair [118]. We could not find information in the literature regarding UBE2J2 expression in EPN. However, Chen et al. [119] reported higher expression of UBE2J2 protein in hepatocellular carcinoma compared to normal liver tissue, which was associated with increased epithelial to mesenchymal transition (EMT) in vitro, whereas UBE2J2 gene knockdown promoted EGFR expression [119]. UBE4B regulates the degradation of the tumor suppressor p53 in breast cancer [120] and UBE2Q1 has been proposed as a marker for high-grade serous ovarian cancer [121]. Ubiquitin E2 conjugases affecting ubiquitination and SUMOylation of Proliferating Cell Nuclear Antigen (PCNA) were associated with ST-EPN-RELA subtype and affect the complex functions of PCNA at DNA replication forks and in DNA repair [122,123]. The ubiquitin E2 conjugases UBE2N and UBE2V2 contribute to ubiquitination of PCNA to allow the continuation of the DNA replication process across unrestored damaged DNA sites [124,125]. Expression of both UBE2V2 and UBE2I (aka SUMO Conjugating Enzyme, UBC9) were elevated in the ST_EPN_RELA compared to the other subtypes. UBE2I contributes to SUMOyation of PCNA and is a proposed therapeutic target in colon cancer [124,126,127] and the SUMO conjugating activity of UBE2I also contributes to DNA repair [128]. RNF8 expression was elevated in SP_EPN ( Figure 14). The E3 ligase RNF8 interacts with the E2 conjugase UBE2N during double stranded DNA repair [38,39]. Our analysis suggests a possible misregulation of double stranded DNA damage and repair in SPN_EPN.

Developmental Factors in EPN Subtypes
We identified over-expression of several homebox genes specifically in SP_MPE. The expression of HOXB9, HOXB13, and PRAC1 and PRAC3 genes are all located in the cytogenetic band 17q21.32 as part of the human HOX gene region ( Figure 11). The overexpression of HOX genes in SP-MPE has been previously reported, along with data showing evidence for the over-expression of the HOXB13 protein by immunohistochemistry [30]. The gene for the ubiquitin conjugase UBE2Z is also located at 17q21.32, a finding not previously noted. Since HOX genes outside of chromosome 17 are also over-expressed in SP_MPE, the data suggest that a single factor stimulates selected HOX gene expression on different chromosomes. UBE2I is likely the ubiquitin conjugase transferring SUMO to PIAS1 [129][130][131]. UBE2I and the E3 ligases RNF20 and RNF40 have been implicated in the regulation of homeobox gene expression and this involves histones H2B residue lysine 120 mono-ubiquitination and subsequent methylation at H3 at lysines 4 and 79 [29]. Genetic markers correlated with the increased HOX genes in SP_MPE included the NEFL gene whose expression is regulated by the UBE2D1 (UBCH5a) dependent ubiquitin E3 ligase TRIM2 [31]. The expression of GATA4 was elevated in ST_SE ( Figure 13). In cardiac tissues, GATA4 activity was demonstrated to be stimulated upon SUMOylation by the E3 ligase PIAS1 [37]. GATA4 has been found to be a tumor suppressor in astrocytomas [132]. Our analysis suggest GATA4 could be a marker for ST_SE. PIAS1 was elevated in PF_EPN_B (F = 20.17, p = 2.14 × 10 −20 ).

Summary Table
A summary table of the major over-expressed pathways associated with the UPS components, E1 activators, E2 conjugases, and E2 ligases and adaptors are shown in Table 8. The table shows that more than one UPS component are associated with antigen processing, neddylation, and cell cycle pathways. Table 8. Major over-expressed pathways associated with differential expression of UPS genes in subtypes of ependymoma.

Current Targeting Approach for the UPS Pathway
Small molecular inhibitors have been reported that target ubiquitin E1 activators, the E2 conjugase UBE2N, the E3 ligases CRBN and MDM2, the E3 ligase adaptors SKP2 and BTRC, and the deubiquitinases USP2, USP7, and USP14 [134]. These molecules are in various stages of preclinical or clinical trials.
A neddylation inhibitor MLN4924 is undergoing clinical trials as an anti-cancer drug for solid tumors and hematologic malignancies [135].
PROTAC (Proteolyis TArgeting Chimera) technology is a way of engineering small molecular inhibitors which use the cells' own E3 ligases to target substrates of interest [136]. This technology has great potential for targeting tumor proteins for degradation by the proteasome. A small molecule PROTAC was reported to inhibit MDM2-p53 interaction [137]. At least 12 PROTACs that target cancer have entered clinical trials since 2019 [138]. A PROTAC designed to target the CDC20-APC/c complex and inhibit mitotic progression of breast cancer cells in vitro was reported in 2019 [139]. PROTAC may also be a promising therapeutic strategy to target ependymoma with elevated expression of CDC20, including the ST_EPN_RELA and PF_EPN_A subgroups. Furthermore, our work has identified several other over-expressed proteins in selected subtypes of ependymoma which may provide attractive targets for future therapeutic applications using PROTAC technology. 5.

Conclusions
Our analysis of the Pfister EPN dataset identified significant differential transcription of specific genes for ubiquitin E1, E2, and E3 components of the UPS system. High expression of the E2 conjugase UBE2C is associated with several aggressive cancers, including glioblastoma [103,106] and we can now include the ST_EPN_RELA subtype as well. UBE2C and UBE2S ubiquitinate substrates sequentially and may serve as suitable new therapeutic targets in ST_EPN_RELA. Another E2 conjugase, UBE2Z, may be serve as EPN subtype-specific therapeutic target for SP_MPE. The E2 conjugase UBE2M, a Nedd8 conjugating enzyme, emerged as a potential target for the inhibition of neddylation of E3 cullin ligases, which depend on this neddylating enzyme. The SUMOylation conjugating enzyme, UBE2I, which has been proposed as a therapeutic target in other cancers, may also be considered a valuable therapeutic target in ST-EPN-RELA. Reactome analysis of differentially expressed genes encoding ubiquitin E3 ligases and their adaptors in EPN identified proteins regulating MHC class 1 antigen presentation. SUMOylation may emerge as an important process regulating differential expression of E3 ligase proteins in EPN subtypes. Reactome analysis of ubiquitin E3 substrate adaptors implicated the Neddylation pathway and the APC/c complex as regulators of cell cycle progression in the pathology of EPN. Viral interference with ubiquitin ligase complexes associated with the nuclear pore may contribute to EPN pathogenesis at least in some EPN subtypes but the precise role of viruses, including HCMV, in EPN pathogenesis has yet to be elucidated. A great deal of research is still needed to validate the role of UPS genes in EPN pathogenesis. Our analysis provides a first unique insight into the proteins encoded by these genes and their substrates that promise a new attractive repertoire of biomarkers and potential therapeutic targets for EPN subtype-specific treatment of ependymoma patients.

Informed Consent Statement: Not applicable.
Data Availability Statement: The data referred to in this manuscript are publicly available at the R2 Genomics Analysis and Visualization Platform (http://r2.amc.nl, accessed on 10 January through 21 August 2022) and are available on reasonable request from the first author.