CTNNB1 gene, encoding β-catenin, was located in the region of chromosome 3p21–22. It regulates cell proliferation and differentiation by binding to various proteins and plays a key role in embryonic development and tumorigenesis [13, 14]. Zurawel et al. [15] first discovered point mutations in the CTNNB1 gene in MB, which was later confirmed to be mainly present in the WNT subtype [6, 16, 17]. CTNNB1 exon 3 has four phosphorylation sites. Phosphorylated β-catenin is degraded through the ubiquitin-proteasome pathway, whereas mutations of these sites cause β-catenin accumulation in the cytoplasm, eventually migrating to the nucleus. There it binds and activates the T-cell factor (TCF)/lymphoid enhancer factor (LEF), leading to the upregulation of particular target genes [18, 19]. Combining β-catenin immunohistochemistry and CTNNB1 exon 3 sequencing is a feasible, economical, and effective approach to identifying the WNT subtype of MB, and patients of this subtype have a relatively good prognosis [19].

Activation mutations in the SHH pathway can be found in almost all SHH MBs. The most frequently mutated genes are PTCH1, SMO, and SUFU, and their expressions are mutually exclusive in MB [20–22]. PTCH1 is mainly expressed in mesenchymal cells and is involved in embryonic structure formation and tumorigenesis. SMO proteins are important signal converters in the SHH pathway, and their activity is negatively regulated by PTCH1. SUFU is a major inhibitory factor in the SHH pathway. PTCH1 mutation is the most common mutation in the SHH MB, occurring in all age groups [3, 16], although SMO mutation almost always occurs in adults [23]. A subset of pediatric patients with SHH MBs (aged 3 years to 16 years) showed GLI2 and MYCN amplification to be mutually exclusive with PTCH, but 30% of which harbored phylogenetic (Li-Fraumeni syndrome) or TP53 mutations [24]. Germline SUFU mutation has recently been identified as a genetic background that could cause MB in infants under 3 years old; no SUFU mutation has been found in the adult SHH subtype [23, 25–27]. SHH subtype patients with SUFU germline mutation have a worse prognosis than other SHH MB patients [28].

Human TP53 gene, located on the short arm of chromosome 17, is a tumor-suppressor gene that encodes the P53 protein. It is involved in a number of important biological processes, e.g., cell cycle progression, DNA repair, cell differentiation, and apoptosis [29]. The presence of diffuse, strong P53 immunoreactivity in MB usually indicates a potential TP53 mutation [30]. In a cohort study of 108 cases of MB, mutations in the TP53 gene were reported as an independent predictor of poor prognosis [31]. Zhukova et al. [32]. showed that the prognostic value of somatic TP53 mutation was subtype-dependent in a larger cohort that included 553 cases of MB. Specifically, patients with WNT-subtype tumors who carry somatic TP53 mutation have a good prognosis, whereas patients with SHH subtype tumors who carry the same mutation have a worse prognosis [3, 29, 31]. Moreover, the SHH MB could be further divided into TP53 wild-type and TP53 mutant; TP53 status is the most significant risk factor in SHH MB [13, 29], especially in SHH α subtype [11]. The 5-year overall survival rates of SHH subtype patients with or without TP53 mutation are 41% and 81%, respectively [29]. Therefore, the new WHO classification of the CNS tumors divides SHH subtype MB into TP53 wild-type and TP53 mutant subgroups.

Human genome encodes two functional DDX3 genes: DDX3X and its homologous gene DDX3Y [33]. DDX3X gene is located on the X chromosome and could regulate different steps of RNA metabolism, such as RNA splicing, transcription, and translation initiation. In addition, DDX3X is involved in stress response, cell apoptosis, cell cycle progression, and viral infection [33–35]. The role of DDX3X in tumorigenesis and progression is quite complex, and it plays a dual role in multiple tumors [36]. Downregulation of DDX3X promotes stem cell-like properties and tumorigenesis in hepatocellular carcinoma cells [37], while the upregulation of DDX3 was observed in distant breast cancer metastases and correlated with poor prognosis [38, 39]. DDX3X functions as a tumor-suppressor gene in MB [13, 40]; its functional deletion mutation increased the incidence and severity of tumor formation in mouse models of WNT and SHH MBs [41]. DDX3X mutation is common in adult SHH MB patients, but it is reported to be very rare in the pediatric SHH subtype [23, 25, 26]. DDX3Y gene is located on the Y chromosome and plays a significant role in male fertility [33]. There have currently been no studies connecting DDX3Y with the development of MB.

CREBBP gene is involved in the transcriptional co-activation of many transcription factors, and its expression product is a nuclear protein that binds to the CREB. CREBBP plays a key role in embryonic development, cell growth control, and homeostasis maintenance through chromatin remodeling and transcription factor recognition [42]. CREBBP has been found to be almost completely mutated in adult SHH subtype MB [40]. Similar to the DDX3X mutation, CREBBP mutation is rare in pediatric SHH patients [11, 26, 43]. The loss of CREBBP acts synergistically with SHH signals to enhance SHH pathway output and drive tumor growth in MB [42]. This provides a new direction for targeted therapy of SHH MB.

MLL2 and MLL3 are genes encoding histone-lysine N-methyltransferases involved in the methylation of histone 3 lysine 4 (H3K4) [44]. Histone methyltransferases affect heterochromatin formation, gene imprinting, and transcriptional regulation. MLL2 and MLL3 are tumor-suppressor genes that are inactivated by mutation [45]. In an early study, their mutations were mainly found in WNT or SHH subtype MB [44, 45], but later, Robinson et al. [40] reported a low incidence of MLL2 mutations in group 3 MB. The discrepancy may be caused by a small sample size and a lack of subtype-specific analysis. Therefore, further studies using larger numbers of MB cases will be needed to reveal the specific relationship between the dysfunction of MLL2/3 signaling and the sub-classification and prognosis of MB.

Telomerase is an RNA-dependent DNA polymerase that can prolong the telomere DNA to maintain telomere homeostasis. Maintaining telomere length is a key step for cancer cells to overcome telomere shortening and induce cell senescence [46]. TERT is the rate-limiting catalytic subunit of telomerase. TERT-promoter mutation leads to the upregulation of TERT transcription, which enables cancer cells to avoid cell senescence and increase their replication potential. TERT-promoter mutation is the most common recurrent somatic point mutation in MB [46], occurring mainly in adult patients with SHH and WNT MB. It is interesting that TERT-promoter mutation in the SHH subtype was associated with a higher overall survival rate and lower incidence of tumor metastasis [47], while group 4 MB patients with TERT-promoter mutation had lower overall survival rates than those with TERT-promoter wild-type [48]. In WNT and group 3 subtypes of MB, TERT-promoter mutations appear to have no effect on overall survival [46]. Though the molecular basis for these differences in survival is unclear, the status of the TERT-promoter may provide a new biomarker for subtype classification and targeted therapy in MB.

MYC and MYCN induce cell proliferation and malignant transformation together with other oncogenes or tumor suppressors [49], and they are the two most frequently amplified oncogenes in MB. MYC and MYCN amplifications account for 5–10% of sporadic MB and have a high incidence in large cell subtypes [50]. MYC amplification is a hallmark alteration almost exclusively found in group 3 MB [51] and predicts an extremely poor prognosis [52, 53]. MYCN amplification is enriched in SHH and group 4 MB [54]. Combined ectopic expression of MYCN and SHH promotes the formation of cerebellar MB in mice after birth [54, 55]. MYCN amplification is associated with poor prognosis in SHH MB [23, 52, 55, 56]. However, neither MYCN gain nor amplification was associated with poor survival in group 4 MB [57].

OTX2 gene is composed of 5 exons, of which the first two are non-coding, and the last three encode OTX2. OTX2 was previously identified as a potential oncogene for some malignancies, but recently it has been identified as a driver gene in MB [58]. Due to gene amplification [59], OTX2 is highly expressed in WNT and non-WNT/non-SHH MB, although it is low or absent in the SHH subtype [60, 61]. The downregulation of OTX2 expression can inhibit the growth of MB cells in vitro [62]. Overexpression of OTX2 directly drives MB cell proliferation by targeting cell-cycle genes [63, 64]. Screening for OTX2 overexpression is becoming an integral part of establishing a molecular classification scheme in MB [62], although the correlation between the expression of OTX2 and patient prognosis has not been investigated yet.

CDK6, a serine/threonine kinase, has strong effects on cell cycle progression [65]. CDK6 activation promotes cell cycle progression through the phosphorylation of substrates, including retinoblastoma protein (pRb) and transcription factors with roles in proliferation and differentiation [66]. Recently, some studies have shown that amplification of CDK6 gene is a vital feature of group 4 MB [13, 48]. A genome-wide analysis of DNA copy number in 47 cases of MB showed that CDK6 amplification was significantly associated with poor prognosis in MB [67]. Therefore, CDK6 amplification/overexpression may be used as a biological marker for molecular stratification and therapeutic interventions in MB patients.

PTEN is the main inhibitor of the PI3K signaling pathway, and PI3K activation is the main driver of most human cancers [68]. Frequent allele loss of PTEN in MB results in low expression of PTEN, which was associated with a low survival rate in a transgenic mouse model of MB [69]. Homozygous deletions of PTEN have been described in SHH MB [51]. Low expression of PTEN could identify high-risk patients with adverse outcomes in the SHH subtype, but not in the remaining MB subgroups [70]. In contrast, PTEN is highly expressed in group 4 MB and could be used to differentiate group 3 and group 4 subtypes [70], but the underlying mechanism has not yet been elucidated.

KDM6A [also known as ubiquitously transcribed tetratricopeptide repeat on chromosome X (UTX)], a tumor-suppressor gene encoding histone 3 lysine 27 (H3K27) demethylase, plays a vital part in determining cell fate and cell differentiation during development [13, 40 , 44, 71]. Robinson et al. [40] first reported the high-frequence mutation of KDM6A in MB. KDM6A mutations are enriched in group 4 MB and are identified with lower frequencies in SHH and group 3 MB [13, 23, 72]. In addition, KDM6A copy-number loss was often found in female non-WNT/non-SHH MB patients [40]. Although the exact mechanism is not clear, KDM6A gene mutation promotes tumorigenesis in a mouse model of MB [73], providing novel insights into the function of KDM6A in MB.

KBTBD4 gene encodes a Kelch protein belonging to a family of ubiquitin-ligase adapters that facilitates the ubiquitination of target substrates. KBTBD4 gene insertions are located at a hotspot region (codons 308–313) and have been reported exclusively in non-WNT/non-SHH MBs [25, 48, 49]. However, researchers from Brazil analyzed a series of 111 MBs, including 48 cases from the non-WNT/non-SHH subtype; none of the 48 harbored any KBTBD4 mutations at the hotspot region [74]. This may have been the result of population differences or small sample size. Therefore, future studies are warranted to assess the frequency and role of KBTBD4 mutations in MB.

GLI is the end effector of hedgehog (HH) signaling and promotes transcription of HH-target genes, which could regulate cell survival, invasion, and angiogenesis, as well as stem cell self-renewal and epithelial-mesenchymal transition [75–79]. It has been found that GLI2 amplification exists in SHH MB [23, 25] and frequently co-occurs with TP53 loss (defined as SHH α subtype), predicting a worse prognosis in patients with this subtype of MB [80]. Additionally, GLI2 is positively regulated by the PI3K/AKT pathway [81], which is also mutated in a subset of SHH MB patients [23].

Chromosome abnormalities are often observed in MBs, particularly those classified as group 3 and group 4 subtypes [9]. Isochromosome (iso) 17q is the commonest cytogenetic change in group 4 MB, although it is also seen in group 3 MB [54, 60]. Shih et al. [57] reported that iso 17q was a statistically significant predictor of poor outcomes in group 3 but not in group 4 MB. In addition, chromosome 17 gain and chromosome 11 loss were found to be good prognostic factors in group 4 MB [56, 82]. These findings indicate that chromosome 17 aberration is a subtype-specific molecular biomarker in MB.

TNRC6 proteins, including TNRC6A, TNRC6B, and TNRC6C, serve as scaffolding proteins within miRNA-induced silencing complex and therefore play an important role in miRNA-mediated gene silencing [83]. Whole-genome methylation sequencing of circulating tumor DNA (ctDNA) in cerebrospinal fluid showed that DNA methylation in the 3’-untranslated region (UTR) of TNRC6C was significantly increased in all subgroups of MB, and can be used as a potential prognostic marker to predict clinical outcomes in patients with this tumor [84].

MXI1 is a negative regulator of the MYC family of proteins [85], and IL8 has potential involvement in chemokine signaling and angiogenic processes in tumor development [86]. Recently the methylation of MXI1 and IL8 was identified as a novel independent high-risk biomarker in the survival models of SHH and non-WNT/non-SHH MB patients [87]. Incorporation of DNA methylation events into current risk-stratification schemes significantly improved the accuracy of survival prediction, which has important implications for future risk-adapted clinical disease management in MB.

Lmx1A is a LIM homeobox transcription factor 1, alpha previously shown to function as a critical regulator of cell-fate determination in cerebellar development [88]. Lin et al. [89] reported that both Lmx1A enhancer activity and expression are important for the identification of group 4 MB. Their chromatin immunoprecipitation (ChIP)-sequencing data supported Lmx1A as a master regulator of the transcription factor in the transcriptional program of group 4 MB.

PRDM6 belongs to the PRDM family of transcriptional repressors, a family that is essential for the growth of smooth muscle cells [90]. It was reported that PRDM6 gene expression was markedly upregulated in a subset of group 4 MB patients, due to “enhancer hijacking” induced DNA rearrangement [25]. Further studies are needed to confirm the function of PRDM6 as an oncogene in this subtype of MB.

GFI1B is a paralog of GFI1. Both genes functioned as sustained natural apophyseal glides (SNAG) domain-containing zinc finger transcriptional repressors essential for a variety of developmental processes [91]. GFI1 and GFI1B were identified as prominent oncogenes specifically activated in non-WNT/non-SHH MB, and somatic genomic rearrangements together with mutually exclusive activation of GFI1 and GFI1B were found in approximately one-third of group 3 MB patients [92]. These oncogenes are now considered the commonest enhancer hijacker in this subtype of MB [13, 25, 92], and they may have a synergistic effect on MYC gene amplification in promoting the malignant progression of MB [92]. Therefore, GFI1 and GFI1B are promising biomarkers for molecular typing and targeted therapy in non-WNT/non-SHH MB.

ERBB4 is the only member of the ERBB receptor family with growth-inhibiting properties. According to the Cancer Cell Line Encyclopedia database, its messenger RNA (mRNA) expression is only present in a small fraction of tumor cell lines, whereas the other ERBB receptors are highly expressed in the majority of tumor cell lines [93]. The controversies around the anti- or pro-oncogenic role of ERBB4 can in part be explained by the multiple ligands that can activate ERBB4, its numerous intracellular phosphorylation sites, the presence of alternative splice variants, the different intracellular signaling pathways affected, and the different downstream responses in different cell types and different disease stages. Using quantitative (phospho)-proteomics in primary human MBs, Forget et al. [94] unraveled distinct post-transcriptional regulation leading to highly divergent oncogenic signaling and kinase activity profiles, e.g., aberrant ERBB4-SRC [a key protein tyrosine kinase to regulate receptor tyrosine kinase (RTK) signaling] signaling in group 4 MB. These findings indicated that ERBB4 promoted MB malignance and could serve as a therapeutic target in group 4 subtype.

MYC amplification is a “hallmark” of MYC-active MB, but not all tumors of this type have MYC amplification [25, 95]. Archer et al. [96] quantitatively profiled global proteomes and phospho-proteomes in 45 MB samples, and found that increased post-translational modifications of MYC, e.g., acetylation and phosphorylation, are associated with poor outcomes in group 3 MB, and correlate with the increased phosphorylation of protein kinase, DNA-activated, catalytic subunit (PRKDC). Inhibiting the activity of PRKDC sensitizes MYC-active MB cells to radiation [96], which offers a new strategy for the treatment of group 3 MB.

miRNAs are a class of endogenous non-coding small RNAs with a length of 18–22 basepair (bp) that regulate the expression of target genes by inducing mRNA degradation or translation inhibition [97]. They control basic cellular processes such as development, differentiation, metabolism, proliferation, and apoptosis. Non-transcriptional expression of miRNAs is associated with the development and progression of a variety of cancers, and such expression changes can be caused by mutations, methylation, deletions, and gains in the miRNA coding region [98]. miR-124 was first reported to be positively associated with the survival of MB patients, and it can inhibit tumor cell growth by targeting CDK6 [99, 100]. Kunder et al. [101] carried out a miRNA expression analysis in different subtypes of MB and identified miR-592 and miR-182 as surrogate markers for non-WNT/non-SHH group 3/group 4 MB. The two mRNAs were also useful for risk stratification of this category of MB. They later found that restoration of miR-193a expression or overexpression of miR-206 suppressed tumor cell growth in MB cells [102, 103]. In addition, high expression of miR-182 and miR-183 was positively associated with the metastasis of non-SHH MB [104]. These studies suggest that miRNA profiling might be a promising marker for risk stratification, molecular typing, and prognosis estimation of MB [105].