MeCP2 ubiquitination and sumoylation, in search of a function^†

Abstract

MeCP2 (Methyl CpG binding protein 2) is an intrinsically disordered protein that binds to methylated genome regions. The protein is a critical transcriptional regulator of the brain, and its mutations account for 95% of Rett syndrome (RTT) cases. Early studies of this neurodevelopmental disorder revealed a close connection with dysregulations of the ubiquitin system (UbS), notably as related to UBE3A, a ubiquitin ligase involved in the proteasome-mediated degradation of proteins. MeCP2 undergoes numerous post-translational modifications (PTMs), including ubiquitination and sumoylation, which, in addition to the potential functional outcomes of their monomeric forms in gene regulation and synaptic plasticity, in their polymeric organization, these modifications play a critical role in proteasomal degradation. UbS-mediated proteasomal degradation is crucial in maintaining MeCP2 homeostasis for proper function and is involved in decreasing MeCP2 in some RTT-causing mutations. However, regardless of all these connections to UbS, the molecular details involved in the signaling of MeCP2 for its targeting by the ubiquitin-proteasome system (UPS) and the functional roles of monomeric MeCP2 ubiquitination and sumoylation remain largely unexplored and are the focus of this review.

mecp2 homeostasis, rett, ubiquitination, SUMOylation, PEST domains

Introduction

In eukaryote organisms, DNA is the repository of genetic information and is present in cells as a nucleoprotein complex known as chromatin [1]. In addition to DNA, histones are the main protein component of such a complex. Both DNA and histones are subject to post-synthetic chemical modifications that play a critical role in the epigenetic information of the complex. Methylation is one such modification, which has been studied extensively in DNA and histones [2, 3]. The chemical modifications of chromatin are regulatory and constitute one of the main contributors to the epigenetic chromatin component [4]. Their downstream outcome is mediated by the dynamics with which the interacting proteins specifically recognize and bind to them. Hence, these proteins are generically referred to as “readers”. Ubiquitination and sumoylation can target these protein effectors and, in so doing can contribute to the modulation of their function. One such reader is MeCP2 (Methyl CpG binding protein 2) [5], which binds to DNA preferentially methylated at CpG and CpH (H = A or T or C [6–8]). It also binds to histone (H3 K27me³) [9, 10]. MeCP2 is a target of ubiquitination and sumoylation, which can contribute to its functional complexity. Nevertheless, little is known about the molecular mechanisms involved, and their role on MeCP2 function. In this article, we summarize our current knowledge of this important yet relatively unexplored topic.

MeCP2 is an intrinsically disordered protein with two isoforms

From structural and functional perspectives [11], the complexity of MeCP2 is enhanced by the presence of two isoforms (E1 and E2), the product of alternative splicing [12]. The two proteins differ in their first N-terminal amino acids at the N-terminal domain (NTD), with human MeCP2-E1 extending 12 amino acids longer than E2 (Fig. 1A), but they are identical otherwise. MeCP2-E2 seems to have diversified from a vertebrate precursor E1-isoform in mammals [13]. Despite the minor differences in size and amino acid sequence, the two variants bind to chromatin with different affinities [11], and although they can share some functional roles and have controversial functional redundancy [14, 15], they possess distinct functional attributes [11] and different half-lives [13].

Figure 1

Two structural views of MeCP2. (A) Schematic representation of the two human MeCP2 E1 and E2 isoforms indicating their structural/functional domains: NTD, N-terminal domain; MBD, methyl-binding domain; ID, intervening domain; TRD, transcription repression domain and CTD, C-terminal domain. (B) An AlphaFold [131] view of MeCP2 and (C) structure obtained using a combination of all-atom and coarse-grained simulations [19]. The blue margin square highlights the predicted condensed organization of the molecule around the TRD. The red indicates the MBD [132].

With about 60% unstructured regions, MeCP2 represents a prototypical intrinsically disordered protein (IDP) [16–18]. Thus, while the methyl binding domain (MBD) targets the protein to methylated regions of the genome, its overall IDP nature accounts for the large variety of interactors [17], which further impart the protein with manifold downstream functions. Of interest, while the MBD has a well-defined tertiary structure organization, all-atom and coarse-grained simulations [19] predict a less constrained form, but a significant 3D folded organization encompasses the transcriptional repression domain (TRD) (Fig. 1B); the folded domain overlaps with the short trypsin-resistant peptide around the TRD [18].

MeCP2 in health and disease

Because of its potential for many partner interactions, a process that, in the case of protein interactors, is usually mediated by PTMs [20], it is not surprising that MeCP2 can exhibit contrasting functions as a repressor and an activator of gene expression depending on the partner’s nature [21, 22]. As an essential transcriptional regulator [23] in vertebrates, the protein is involved in a plethora of cellular events with diverse functions [24]. MeCP2 phosphorylation at S433 (421 in E2) has been shown to release it from the promoter of Bdnf (brain-derived neurotrophic factor) to induce its transcription, which is critical for neuronal survival and growth. In its phosphorylated form, MeCP2 contributes to regulating spine maturation and dendritic growth [25], which are essential for developing neuron circuits. Moreover, the protein has been involved in ciliogenesis by associating with the centrosome [26]. In the adult brain, the protein is involved in long-term memory [27] and in the circadian cycle-dependent regulation of gene expression [11, 28] in mice. An essential role in maintaining the blood-brain barrier has been recently reported [29]. Moreover, MeCP2 has been shown to link DNA methylation to brain metabolism, where it affects glucose and cholesterol metabolism [30]. It has also been shown to participate in the differential sensitivity of the protein translation initiation machinery and mTOR signaling [31, 32]. Importantly, MeCP2 is a global chromatin organizer of the brain both during neurodevelopment and in the mature brain [11, 33]. Hence, its involvement in neurodevelopmental and neurodegenerative diseases and many adult brain disorders and conditions is no wonder [34].

While the functionality of MeCP2 is reasonably understood in the brain because of its significant presence in this tissue, it is crucial to recognize that the protein is also expressed in numerous non-neural tissues, including the lung. Although the levels of MeCP2 expression in the lung are still controversial [35–37], MeCP2 has been shown to associate with alpha-smooth muscle actin and promote fibroblast differentiation in this tissue [38]. Furthermore, Vashi et al. [39] found that MeCP2 regulates many lipid metabolism genes in the lung and is required to maintain a normal lipid composition of pulmonary surfactant, which reduces the surface tension in the lungs’ alveoli. The intrinsic ability in transcriptional regulation of many genes and indirectly participating in their downstream effects expands the spectrum of its non-neural functionality. A recent publication describing its regulation of Homeobox A9 (HOXA9) that affects vascular smooth muscle cells differentiation [40] provides an excellent example.

Although MeCP2 was first described in 1992 [5], studies on the relevance of the protein did not become fully embraced by the scientific community until 1999 [41] when Ruth Amir and Huda Zoghbi established its connection with Rett syndrome (RTT), showing that mutations of this protein are the genetic cause of 95% of RTT cases [42]. RTT is a monogenic X chromosome-linked rare neurodevelopmental disease mainly affecting girls [43]. All these mutations affect brain functions to different extents described with phenotypes whose symptoms’ severity depends on the particular mutation involved [44] and in cellular mosaicism, which is prevalent in females [45] and somatic mosaicism in males [46, 47].

Another MeCP2-driven rare neurodevelopmental disease with symptoms overlapping with RTT [48] is MeCP2 duplication syndrome (MDS) [49–51] which is caused by an extra copy of MeCP2, predominantly affecting males. Under-expression of MeCP2 in RTT and over-expression in MDS brought awareness to the critical tightly regulated homeostasis for the proper function of this protein in the brain [52, 53]. Moreover, MeCP2 gain- and loss-of-function involves its isoform-specific homeostasis in this tissue [31]. Such controlled protein balance is widespread in neurons and involves proteostasis regulation [54], which requires tight coordination of protein synthesis, ubiquitination and proteasomal degradation [55].

Given the functional relevance and abundance of MeCP2 in the brain [9, 56], it is not surprising that alterations of this protein not only affect RTT but are pleiotropic, involved in many psychological disorders such as depression, autism [50], schizophrenia [57], and epilepsy [58]. All the conditions involve MeCP2 destabilization. Outside the brain, the regulation of HOXA9 described earlier was observed in MeCP2’s role in atherosclerosis [40], and MeCP2 has been described to be a bona fide oncogene [59], a role not yet particularly attributed to any of its two different isoforms. The oncogene role has been described in many cancer types (see the recent review [60]). Nevertheless, in cases such as breast cancer where MeCP2 has been described to as an oncogene [61], it exhibits a role to suppress cell proliferation [62]. Thus, the involvement of MeCP2 in cancer seems complex and its role may vary in a tissue-specific context.

Rett-causing mutations impair the overall ubiquitination state of the cell. MeCP2 deficiency has been shown to dysregulate genes encoding E3 ubiquitin ligases [63–67]. However, the direct effects on MeCP2 exerted by ubiquitination and sumoylation are only beginning to be elucidated.

Rett syndrome, ubiquitination and MeCP2 degradation

The RTT pathology has provided us with much information relevant to MeCP2 functionality. The early studies established a relationship between RTT and UBE3A, implicating MeCP2 in UBE3A regulation [63]. UBE3A, also known as E6AP, is a ubiquitin ligase that affects proteasome function and is also associated with the neurodevelopmental diseases Prader-Willi syndrome, Angelman syndrome, and Dup15q syndrome [68]. UBE3A is involved in neural activity-regulated synaptic processes mediated by the protein degradation pathway [69]. It was initially believed that the MeCP2 deficiency associated with RTT triggered histone-PTM epigenetic aberrations that ultimately reduced UBE3A production [70]; the observation was challenged in MeCP2 mutant mice, which did not exhibit reduced UBE3A [70, 71]. Subsequently, UBE3A was identified as an important MeCP2 cofactor, and co-immunoprecipitation analyses demonstrated its physical association with MeCP2 [72]. A study of gene expression profiles modulated in RTT using peripheral blood lymphocytes from patients followed by functional clustering identified 146 specific genes involved in pathways, some related to mitochondrial function, protein ubiquitination and proteasome degradation [73]. Also, fibroblasts from RTT patients with non-sense MeCP2 mutations have impaired and reduced proteasome subunits due to the downregulation of chaperone proteins PAC1 and PAC2 [74].

Although much evidence has been accumulated on the relationship between RTT and the proteasome-mediated protein degradation system, less is known about the effects of MeCP2 mutations on protein stability. An early seminal paper by the late Alan Wolffe’s lab expressing the RNAs of different mutants in Xenopus oocytes followed by cycloheximide pulse-chase of the protein provided evidence for differential stability of the RTT-related MeCP2 mutants [75]. Using a RTT mouse model developed in Rudolph Jaeinsch’s lab expressing MeCP2 missing an essential part of its MBD [76], the amount of the truncated protein expressed in the brain decreased to <50% [77]. While this represents an extreme situation, it underscores the relevance of the whole MeCP2 protein, particularly that of its MBD, to retain its stability. Table 1 summarizes the information to date on the stability of several of the RTT-MeCP2 mutants; mutations affecting the MBD consistently appear to result in lower expression of the protein in agreement with the results observed with Jaenisch mouse model [76]. However, to date, only the T158M mutant decrease in MeCP2 expression has been studied in detail and shown to involve proteasomal degradation [78]. Interestingly, this RTT phenotype was ameliorated, and the average life span in male mice reportedly increased from a median of 90 days to 150 days in transgenic mice overexpressing the mutated T158M. It was further shown that these improvements resulted from the increased binding of MeCP2 T158M to DNA. These observations suggest that the instability of the protein introduced by MBD mutations and their altered binding affinity for methylated cytosine [79, 80] are responsible for their propensity to degrade. Another example of the relevance of the ubiquitin-mediated degradation on MeCP2 with Rett-causing mutations is the missense A2V mutation [13] affecting the MeCP2-E1 isoform. This mutation causes a higher proteasomal degradation rate than the wild type. This degradation involves the removal of N-terminal methionine residues followed by acetylation of the penultimate alanine or valine [81], a process that requires ubiquitination [82].

Table 1

Classification of Rett-causing mutations based on their position in MeCP2 domains.

Domain of MeCP2	Mutation	Molecular phenotype	Protein levels compare to WT	% of RTT cases	Median life span in ♂ mice	Reference
N-terminal Domain (NTD)	MeCP2-E1- A2V	Methionine at the P1 position and the penultimate valine (P’1) were acetylated in vitro. N-methionine excision (NME) was absent when compared with MeCP2-E1. MeCP2 colocalization to chromocenters was unaffected.	~1/4	0.2%	Unknown	[13, 106, 136]
Methyl Binding Domain (MBD)	T158M (T170M)	Impaired stability of Asx-ST motif. Impaired binding to chromocenters due to reduced affinity for mCpG.	~1/3	8.79%	12 weeks	[12, 78, 136, 137]
	P152R/A (P164R/A)	Impaired clustering of MeCP2 to chromocenters.	Unknown	1.5%	Unknown	[136, 138]
	R133C (R145C)	Impaired stability of 4th β-sheet and α-helix. Reduced binding to hmC (retained binding to mC). Impaired interaction with TCF20 complex. Reduced MeCP2-driven LLPS.	~1/2 in the hippocampus; indistinguishable in cortex and cerebellum	4.24%	42 weeks	[136, 137, 139–142]
	R111G (R123G)	Abolished binding to mC; disrupted nuclear localization; disrupted binding to TCF20 complex. Reduced MeCP2-driven LLPS.	unknown	>1%	6 weeks	[80, 136, 141–143]
	R106W (R118W)	Impaired stability of 2nd beta-sheet and reduced affinity to mCpG. Decreased MeCP2-driven LLPS.	~1/3	2.76%	10 weeks	[136, 137, 142, 144–146]
Transcription Repression Domain (TRD)	R294X (R306X)	Increased binding to DNA. Clustering to chromocenters was unaffected.	Indistinguishable	6.1%	35 weeks	[106, 147–149]
NCoR/SMRT Interaction Domain (NID)	R306C (R318C)	Abolished interaction between NCoR/SMRT by disrupting TBLR1 binding. Disrupted binding to ATRX complex. Reduced HDAC activity in the forebrain of mice.	Indistinguishable	6.8%	18 weeks	[150–152]

Regardless of the MeCP2-UBE3A-RTT connection and the critical role played by ubiquitin and ubiquitin-like proteins in the equilibrium between synapse physiology and neurodevelopment disorders [83], the role of MeCP2 ubiquitination and sumoylation has been less explored, and its involvement in the mutation phenotypes is less clear. In the last two sections, we will describe how these PTMs impact MeCP2.

MeCP2 ubiquitination

MeCP2 is subject to many PTMs, including phosphorylation, acetylation, methylation, O-glycosylation, ubiquitination, sumoylation [84, 85], lactylation [86] and N-methionine excision (NME) [13]. Among them, phosphorylation and ubiquitination are prominent, as many sites for both have been experimentally identified [87]. Yet, while the phosphorylation of MeCP2 has already been extensively investigated [25, 85, 87], the ubiquitination profile of MeCP2 (Fig. 2) has remained unexplored, and its role remains largely unknown.

Figure 2

MeCP2 ubiquitination and sumoylation. (A) Tertiary structure of human ubiquitin (1) and SUMO-1 (2). The red circles indicate the lysine sites forming K11 or K48 homotypic/heterotypic linkage polyubiquitin chains [133] that contribute to the signaling for proteasomal degradation. (B) Amino acid sequences of (1) human ubiquitin and (2) inactive human SUMO-1. The red circles are as in (A); the blue squared box is the di-glycine motif involved in the formation of an iso-peptide bond between the alpha-amino group of the C-terminal Gly and the epsilon amino group of the Lys that has been ubiquitinated in the target protein. The orange box in SUMO-1 is the sequence removed by SENP (sentrin/SUMO-specific protease) enzymes during the activation of SUMO previously to its conjugation with the target protein [134]. (C) Schematic representation of human MeCP2-E1 indicating the sites of ubiquitination (light blue pentagons). Empty filled correspond to predicted sites [96], solid/lightly shaded are those experimentally determined [87] the lightly shaded refer to the ones falling within a PEST domain. Sites of sumoylation are fuchsia hexagons. Sumoylation at mouse MeCP2-E1 K429 (mK429)/(mouse MeCP2-E2 K412) [114] corresponds to site K426 in human MeCP2-E1.The intensity of their filling colors is as in ubiquitination [96, 111]. The regions in red correspond to the two PEST domains, and the white numbers within each region correspond to their respective PEST find scores [96].

Ubiquitin is a 76 amino acid long protein expressed in all eukaryotes. It is covalently attached to substrate proteins as a PTM, providing a signaling mechanism involved in versatile functions such as control of stability, cell-signaling, cell proliferation, and immune response, among others [88–90]. The process of attachment involves three enzymes: E1, ubiquitin-activating enzyme; E2, ubiquitin-conjugating enzyme; and E3, ubiquitin ligase [91], resulting in the formation of an isopeptide bond between the C-terminal glycine alpha carboxyl group of ubiquitin (Fig. 2B) and the epsilon amino of lysine in the target protein. Depending on the signaling pathway, a polyubiquitin chain can follow the mono ubiquitin [91]. The diverse functional outcomes of such PTMs are mediated by the nature of their structural diversity, spanning from mono-ubiquitination to a highly complex variety of polyubiquitinylated molecules, which additionally depends on the linkage type which can involve the internal lysine residues (K6, K11, K27, K29, K33, K48, and K63) to be of a homotypic and heterotypic linkage [92]. For example, a polyubiquitin chain attached by internal K11 and K48 heterotypic linkages binds and stimulates proteasomal degradation (Fig. 3) while the homotypic K11 linked polyubiquitin weakly associates with the proteasome machinery [93–95]. The 26S ubiquitin-proteasome system (26S UPS) is responsible for the breakdown of MeCP2 [78, 96]. The ubiquitin linkage type involved in protein degradation is likely the K48-linked homotypic or K48/K11 branched polymer (Figs 2A and3) as these linkages precede the recruitment of the 26S UPS [78, 97].

Figure 3

Structure of the PEST domains in MeCP2 in human and mice. The phosphorylation, ubiquitination and sumoylation of MeCP2 PEST domains are shown by red, green and blue circles, respectively. The yellow boxes display the N and C-terminus PEST amino acid sequences. Predicted (broken circles) [96] and experimentally determined [87] phosphorylation, ubiquitination and sumoylation sites are indicated. The AlphaFold [131] prediction of the helical polyproline type II (PPII) secondary structure arising from sequences containing repeated proline residues [135], which is characteristic of some PEST sequences [100], is displayed underneath the PEST-II domain.

Initial attempts to determine the PTM profile of MeCP2 were made by stable transfection of human neuroblastoma cells SH-SY5Y with mouse MeCP2-E1 isoform. Tandem mass spectrometry (MS/MS) analysis revealed nine putative ubiquitylation sites at K24, K131, K142, K147, K245, K262, K256, K283 and K333 [87] of the human MeCP2-E1 isoform (Fig. 2C). The E3 ubiquitin ligase that ubiquitinates MeCP2 is the RING finger protein 4 (RNF4) (Fig. 4A) [98]. It is yet to be determined via which linkage type, and on which lysine residues within MeCP2, that RNF4 ubiquitinates [98]. However, RNF4 is a known recruiter for protein degradation [99]. It also triggers the removal of MeCP2 from methylated promoters and acts as a transcriptional activator [98] (Fig. 4A).

Figure 4

MeCP2 ubiquitination and PEST-mediated degradation. (A) RNF4 is the ubiquitin ligase of MeCP2 that acts as a transcriptional co-activator in a DNA methylation-dependent way, likely regulating its turnover [98]. (B) MeCP2 (red) contains two PEST domains (orange) [96]. Mutations in MeCP2 MBD (T158M [78]) impair the binding of the protein to methylated DNA, favoring its ubiquitination at the PEST domains and resulting in a decrease of the overall amount of protein as a result of degradation by the 26S UPS. The light brown/blue lollipop structures correspond to methylated CpG/CH (H = A, C or T) [8], respectively.

Indeed, most of what we know about MeCP2 ubiquitination has to do with its potential role in degradation by the 26S UPS [13, 78]. In 2009, we proposed a hypothesis that the MeCP2 turnover in the cell is partially regulated by PEST-mediated degradation. We identified two bona fide PEST (enriched in proline, glutamic acid, serine, and threonine) domains [96] (Fig. 2C) which are highly conserved across vertebrate evolution (Fig. 3 and Supplementary Fig. 1). These motifs are common in proteins with a short half-life. PEST domains have marginal stability due to their secondary structures, such as polyproline type II (PPII) and alpha helices (Fig. 3). The PEST hypothesis of protein degradation itself proposes that upon phosphorylation at a serine or threonine, the conformational stability of PEST domains becomes compromised [78, 100] and this instability is recognized by the E3 ubiquitin ligase, which ubiquitinates the flanking lysine residues via K48 or K11/48 linked polyubiquitination [100, 101] acting as a signal for the 26S UPS degradation (Fig. 4B). A phosphorylation site for each of the PEST domains at S92 [87, 102] and S413 [87] have been experimentally identified (Fig. 3), supporting the PEST mediated degradation hypothesis.

The ubiquitin-proteasome system is involved in recognizing the destabilizing N-terminal residues (N-degrons) following the N-terminal removal of methionine (NME) and N-terminal acetylation of the following alanine or valine as mentioned earlier. However, there is still a need to experimentally confirm if the E3 ubiquitin ligase that binds to N-degrons following NME and N-acetylation is RNF4 or a different E3 ubiquitin ligase.

Besides the putative role of ubiquitination in MeCP2 degradation, very little is known about the experimentally determined ubiquitination sites which fall outside the realm of PEST domains (Fig. 2C). One such residue is K283 within the TRD, which was heavily represented in the number of fragments of the MS/MS analysis [87]. In 2015, Pandey et al. reported that when K283 and a few other ubiquitin-targeted lysines are acetylated in cancerous cells by p300 acetyltransferase upon pharmacological inhibition of the histone deacetylase SIRT1 [103], it decreases the binding of ATP-dependant helicase (ATRX) and histone deacetylase 1 (HDAC1) to methylated gene promoters. Beyond cancer, MeCP2 interacts with ATRX [104], and there is a Rett frameshift mutation at K283 [105, 106]. Indeed, some of the lysines acetylated upon histone deacetylase treatment overlap with previously reported ubiquitinated sites [87]. Given the suggested competition between acetylation and ubiquitination of the same lysine residues and its role in regulating protein stability, such ubiquitination might, in this instance, play a stabilizing role [107]. With the observed overlap, it is possible that ubiquitination operates as a functional switch when presented as a monomer [108]. Altogether, there has been a paucity of investigation of MeCP2 ubiquitination and the importance of this PTM. Despite its high relevance in the critical homeostasis for MeCP2’s proper function, it remains largely unknown.

MeCP2 SUMOylation

As mentioned above, experimental evidence already exists for the existence of SUMOylation in MeCP2 [85, 109] (Fig. 2C). Yet, the functional relevance and consequences of this MeCP2 PTM are just starting to come to light. SUMO (Small Ubiquitin-related Modifier) proteins are a family of relatively small proteins of about 100 amino acids, and in humans they comprise four members (SUMO-1/SUMO-4). Except for SUMO-4, after their translation as inactive precursors, several C-terminal amino acids are removed by highly SUMO-specific proteases (SENPs) to expose conserved C-terminal Gly-Gly residues (Fig. 2B-2) that participate in the formation of the iso-peptide bond during their conjugation to the target protein. The resulting PTM usually occurs as a co-modification with phosphorylation [109]. The duality of sumoylation and phosphorylation plays an essential role in the cytoplasmic-nuclear translocation of MeCP2 during neuronal differentiation/maturation, as MeCP2 is also present in the neuronal cytosol before differentiation occurs [110].

The first MeCP2 sumoylation site reported was at K235 or K223 in E2 isoform, and it was found to be required for the recruitment of histone deacetylase complexes 1/2 [111], which together with Sin3a, was one of the first described MeCP2 interacting co-repressor complexes [112]. Moreover, such sumoylation is essential in MeCP2 repressor activity, as mutation of K235 to an arginine abolishes its mediated gene silencing in mouse primary cortical neurons. In addition, sumoylation at K235 is crucial in synaptic development in the central nervous system [111, 113].

Another sumoylation site is that of mouse MeCP2-E2 K412 [114] that corresponds to human MeCP2 K426 (Fig. 3), which falls within one of the PEST regions (Fig. 2C), and is mediated by E3 ligase PIAS. Interestingly, as mentioned earlier, this sumoylation takes place as a co-modification that involves the phosphorylation of MeCP2-E2 mouse S421 and T308 [114]. The E3 SUMO ligase protein inhibitor of activated STAT1 (PIAS1) is an essential component of sumoylation pathways. A model was proposed in which induction of PIASI K412 sumoylation by NMDA, IGF-1 and corticotropic releasing factor (CRF) stimuli in the brain enhances MeCP2 binding to methylated DNA. This, in turn, results in the release of CREB from the Sin3a-HDAC1 co-repressor complex, allowing it, in association with CBP, to interact with the Bdnf CRE promoter element and increase the expression of the gene [114] (Fig. 5). The model fits well with the notion that one of the roles of sumoylation is to regulate transcription factor activity by changing the interaction with DNA and chromatin [115]. Of interest to this review, the authors of the paper found that MeCP2 mutants (residues refer to E2 isoform) R106W, R133C, P152A, T158M, R306C, and P376R (Table 1) significantly decreased MeCP2 SUMOylation [116]. Moreover, transduction to the basolateral amygdala (BLA) of a conditional MeCP2 knockout (cKO) in mice with lenti-mRFP-MeCP2WT-SUMO1 fusion vector rescued the deficits of social interaction, and it improved their memory performance when compared to the cKO. All in all, the paper’s results by Tai et al. revealed an essential functional role of MeCP2 sumoylation in synaptic plasticity that is altered in RTT [114].

Figure 5

MeCP2 sumoylation. (A) MeCP2 binds to methylated DNA under normal physiological conditions. It is associated with the Sin3a-HDAC1 co-repressor complex and represses gene expression [112], thereby suppressing the interaction of CREB with the corticotropic releasing factor (CRE) of Bdnf in neurons and inhibiting its expression. (B) Upon stimulation by different factors and phosphorylation of MeCP2 (sites T308 and S421 of mouse MeCP2-E2), the protein becomes sumoylated at lysine 412 (mouse MeCP2-E2) by protein inhibitor of activated STAT 1 (PIAS1). As a result, the MeCP2 interaction with methylated DNA tightens (red halo), releasing CREB from the repressor complex and dissociation of Sin3a-HDAC1, allowing CREB in association with CBP to bind to the CRE in the promoter region of Bdnf and increasing its expression. [Figure adapted and modified from Tai et al. [114]].

Of note, the mouse MeCP2-E2 K412 (MeCP2-E1 429) sumoylation site happens within one of the predicted PEST sequences (Figs 2C and3). As indicated by Heaton and Wilson, the ubiquitination and SUMOylation systems are likely to have many points of intersection and that their coordination contributes to regulated fundamental cellular events [117]. They can work in concert to lead to proteasomal degradation [118]. The mono and poly-SUMOylated forms of the centromeric proteins have been shown to play a critical role in centromere homeostasis [119]. It would be interesting to determine whether these PTMs, in conjunction with their ubiquitinated counterparts or independently, play a similar role in the critical homeostasis required for proper MeCP2 function [119].

Recently, HDAC4 has been shown to form a complex with MeCP2. Compared to HDAC1/2, which are members of class 1 of HDACs, HDAC4 is a member of class IIa [120], and its increased nuclear abundance impairs neuronal development and long-term memory [121]. In addition to regulating protein acetylation, HDAC4 functions as a cytoplasmic SUMO E3 ligase [122]. However, the molecular mechanism relating to HDAC4-mediated cytoplasmic MeCP2 sumoylation and the relevance of sumoylated MeCP2 HDAC in the nucleus needs additional investigation.

Future perspectives

As with intellectual disability and autism spectrum disorder, where the ubiquitin system (UbS) is one of the relevant cellular processes affected [123, 124], through this review it should also be evident that dysregulation of this system plays an equally critical role in RTT. To better understand the molecular mechanisms involved, in-depth studies are required. An important reason for the difficulty in addressing impaired UbS and SUMO pathways is that intermediate states of the ubiquitin and sumo are short-lived [125]. In this regard, recently developed powerful techniques, such as the chemical trapping strategies [126], enable site-selective targeting of E2-ubiquitin-conjugates to full-length protein substrates. These techniques are expedient for the study of E3 ligases, as they can trap the transient intermediates of ubiquitin ligase, and thus could prove to be practical for studying the short-lived intermediates of ubiquitin and SUMO pathways.

We have also described how some of the RTT-relevant MeCP2 mutations (Table 1) alter and, in several instances, decrease the levels of MeCP2 in a proteasome-dependent way involving ubiquitination. In this regard, the possibility of using therapeutics hijacking E3 ligases using the PROteolysis Targeting Chimera (PROTAC) approach [127, 128] should provide good therapeutic strategies for treatment in such conditions. Moreover, this technology might also provide essential insights into fundamental biological aspects of MeCP2 ubiquitination and sumoylation.

Despite the functional relevance of MeCP2 ubiquitination and sumoylation described above, a key challenge in advancing our knowledge on this topic resides in the low abundance and highly dynamic nature of these PTMs [129], and the lack of antibodies with enough specificity, particularly for ubiquitin. In this regard, approaches such as that of sequential ChIPseq for sumoylated MeCP2 in neurons [130] and modifications thereof prove to be an informative and invaluable tool to gain information on the occupancy of SUMOylated transcription and chromatin factors at specific genomic loci.

Acknowledgements

We are very thankful to Mikko Karttunen and Cecilia Chavez Garcia at the University of Western Ontario for kindly providing the image shown in Fig.1B. This paper was supported by the Center for Addition and Mental Health Foundation Discovery Fund grant to J.B.V. and J.A.

Conflict of interest statement: None declared.

Footnotes

^†The amino acid positions referred to in this paper are those of human MeCP2-E1 isoform which is the most abundant isoform in brain.