Skip to main content
NCBI home page
eLife logoLink to eLife
. 2021 Aug 6;10:e67085. doi: 10.7554/eLife.67085

Expanding the MECP2 network using comparative genomics reveals potential therapeutic targets for Rett syndrome

Irene Unterman 1, Idit Bloch 1, Simona Cazacu 2, Gila Kazimirsky 3, Bruria Ben-Zeev 4, Benjamin P Berman 1,✉, Chaya Brodie 3,✉, Yuval Tabach 1,✉
Editors: George H Perry5, George H Perry6
PMCID: PMC8346285  PMID: 34355696

Abstract

Inactivating mutations in the Methyl-CpG Binding Protein 2 (MECP2) gene are the main cause of Rett syndrome (RTT). Despite extensive research into MECP2 function, no treatments for RTT are currently available. Here, we used an evolutionary genomics approach to construct an unbiased MECP2 gene network, using 1028 eukaryotic genomes to prioritize proteins with strong co-evolutionary signatures with MECP2. Focusing on proteins targeted by FDA-approved drugs led to three promising targets, two of which were previously linked to MECP2 function (IRAK, KEAP1) and one that was not (EPOR). The drugs targeting these three proteins (Pacritinib, DMF, and EPO) were able to rescue different phenotypes of MECP2 inactivation in cultured human neural cell types, and appeared to converge on Nuclear Factor Kappa B (NF-κB) signaling in inflammation. This study highlights the potential of comparative genomics to accelerate drug discovery, and yields potential new avenues for the treatment of RTT.

Research organism: Human

Introduction

Rett syndrome (RTT [MIM: #312750]) is a rare genetic disorder caused by mutations in the methyl-CpG binding protein 2 (MECP2) gene (Amir et al., 1999). It is an X-linked dominant disorder, almost exclusively affecting females. RTT is characterized by normal early development followed by regression, loss of purposeful hand movements, and intellectual disability (Liyanage and Rastegar, 2014). Characteristics of the disease such as age of onset, range of symptoms, and their severity vary between patients (Neul et al., 2008). The MECP2 protein has a role in transcriptional regulation but its exact mechanism of action, co-factors and downstream targets, are not fully characterized (Picard and Fagiolini, 2019).

The MECP2 protein, as a transcriptional regulator, has been linked to both gene silencing and activation (Horvath and Monteggia, 2018). It binds methylated CG dinucleotides through its methyl-binding domain and has a transcriptional repression domain that mediates binding to co-repressor proteins (Lyst et al., 2013). These two domains are hotspots for missense, nonsense, and frameshift mutations for RTT disease (Ragione et al., 2016; Krishnaraj et al., 2017). Transcriptomic analyses of RTT patient postmortem brain tissues, blood samples, cell lines, and murine models (Shovlin and Tropea, 2018) have revealed hundreds of affected genes but with a low degree of overlap between studies (Picard and Fagiolini, 2019).

Therapies aiming to restore MECP2 loss of function are in early development stages and are not approved for patients. These include attempts at read-through inducing drugs to overcome nonsense mutations (Brendel et al., 2011; Merritt et al., 2020), X reactivating therapies to allow expression of silenced endogenous MECP2 (12), and gene therapy treatments to introduce wild-type MECP2 copies into the brain (Guy et al., 2007; Garg et al., 2013). Furthermore, current gene therapy methods show limited transduction efficiency and are unable to modify all cells in the target tissue (Gadalla et al., 2013). Targeting factors downstream of MECP2 offers a promising avenue which could allow for partial amelioration of Rett symptoms, without the severe neurological phenotypes associates with MECP2 overexpression (Chao and Zoghbi, 2012; Ramocki et al., 2009).

One major downstream target of MECP2 is brain-derived neurotrophic factor (BDNF) (Chen et al., 2003), a key modulator of neuronal development and function. BDNF levels are reduced in symptomatic MECP2 KO mice (Kline et al., 2010) and experimental interventions that increase BDNF levels improve RTT-like phenotypes (Chang et al., 2006). The IGF-1 protein potentiates BDNF activity, and treatments with recombinant human IGF-1 and its analogs have led to improvement in several clinical measurements (Khwaja et al., 2014; Glaze et al., 2019), with a trial of an IGF-1 synthetic analog currently in Phase 3 (NCT04181723). Additional trials include the administration of triheptanoin supplementation which is being investigated in a Phase two clinical trial (NCT02696044) to target mitochondrial dysfunction in RTT (Park et al., 2014). Other interventions, such as ketamine (NCT03633058), targeting N-methyl-D-aspartate receptor (NMDAR) dysfunction, and Anavex 2–73 (NCT03758924), aimed at restoring mitochondrial dysfunction (Kaufmann et al., 2019), are currently being studied. Histone deacetylases (HDACs) mediate MECP2 gene repression, and have also been suggested as therapeutic targets for Rett syndrome (Shukla and Tekwani, 2020), which may act on BDNF trafficking (Xu et al., 2014). The lack of a successful therapy to date emphasizes the need for accurate and comprehensive mapping of the MECP2 network with a focus on targetable proteins that could be used for interventions (Vashi and Justice, 2019). Here, we use a genomics approach that has thus far not been applied to MECP2 (phylogenetic profiling), combined with data from drug target databases, to construct such a map.

A well-tested approach for gene functional prediction and identification of functional gene networks is phylogenetic profiling (PP), which analyzes gene conservation patterns across an evolutionary related set of organisms (Pellegrini et al., 1999). Genes with similar evolutionary patterns are functionally coupled, irrespective of sequence homology (Date and Marcotte, 2003). PP was used successfully to predict protein function (Enault et al., 2004; Eisen and Wu, 2002; Jiang, 2008; Merchant et al., 2007), protein interactions (Kim and Subramaniam, 2006; Sun et al., 2005) and protein localization within the cell (Marcotte et al., 2000; Pagliarini et al., 2008; Avidor-Reiss et al., 2004; Hodges et al., 2012). When a pathway becomes non-functional in a specific lineage (often due to loss of a key gene in the pathway), then other genes involved in the pathway may lose the fitness advantage they provide to the organism if they are not involved in other important pathways/functions. PP captures these pathway-level loss events, providing a quantitative measurement of functional relatedness and prioritizing those genes minimally involved in the same pathway over pleiotropic genes required in multiple pathways. This property is especially attractive for drug development, where the goal to is target the most specific proteins possible. Because PP is based on evolutionary dynamics rather than protein expression or sequence/structure features, it produces results that are highly complementary to traditional methods (Dey et al., 2015; Arkadir et al., 2019).

In our recent work, we developed an improved version of PP that normalizes signals across hundreds of eukaryotic genomes to identify important new members of cellular and disease pathways (Arkadir et al., 2019; Sherill-Rofe et al., 2019; Tabach et al., 2013a; Tabach et al., 2013b; Schwartz et al., 2013). As protein function and evolution are very complex (Bloch et al., 2020), co-evolution within a local lineage (i.e. mammals, vertebrates, fungi, etc.) can provide complementary evidence, and is especially informative for genes with multiple functions that show complex or recent evolution (Sherill-Rofe et al., 2019). We recently incorporated a clade-specific analysis into our Normalized Phylogenetic Profile (NPP) software pipeline (Tsaban et al., 2021), which is used for all the analyses in this study. We have used several functional databases including CORUM (Ruepp et al., 2008), KEGG (Kanehisa and Goto, 2000), and REACTOME (Jassal et al., 2020) to validate the NPP method and show that it recovers known protein complexes and pathways with higher specificity and sensitivity than other phylogenetic methods (Bloch et al., 2020; Tsaban et al., 2021).

We combined our PP-derived interaction map of MECP2 with drug targeting databases to select new candidate drugs for RTT. There are thousands of known bioactive compounds collected in databases such as Drug-Gene interaction database (DGIdb) (Cotto et al., 2018) and Open Targets (Oxford Academic, 2019), that contain information about validated direct or indirect effect on specific proteins or pathways. Network-based drug discovery methods are emerging as important tools to identify novel drug targets and predict the drug’s mode of action and potential side-effects (Iorio et al., 2013), and have been associated with higher success rates in clinical trials (Nelson et al., 2015; Pushpakom et al., 2019). From our PP network, we selected three proteins that showed robust co-evolution with MECP2 across multiple evolutionary scales, had related roles in inflammation, and could be targeted by pharmacologically-validated compounds. These three compounds – pacritinib, EPO and DMF – were then functionally validated in MECP2 Knock-Down (KD) human neural cells, including microglia, astrocytes and neural stem cells. The ability of all these drugs to reverse some aspects of the MECP2-KD phenotypes demonstrates how comparative genome analysis can be used to understand and target this disease network.

Results

Phylogenetic profile analysis of MECP2 in eukaryotes and mammals identifies MECP2 co-evolved genes

To trace the evolutionary history of MECP2, we used BLASTP (Camacho et al., 2009) to compare the MECP2 protein to the proteomes of 1028 eukaryotic species. We repeated the process for each of 20,192 human transcripts to generate a profile of conservation for each human gene. The scores were normalized according to sequence length and the evolutionary distance between humans and the queried species as described previously (Tabach et al., 2013b; Bloch et al., 2020). This Normalized Phylogenetic Profile (NPP) describes how conserved a transcript is in each species across the tree of life, compared to its expected conservation. Genes with similar NPPs (i.e. genes that are either conserved or lost as a group) are functionally related, often belonging to the same pathway (Sherill-Rofe et al., 2019; Tabach et al., 2013a; Tabach et al., 2013b; Schwartz et al., 2013). We have shown that both across-clade and within-clade measures of co-evolution can be informative, and thus we computed NPPs both across all eukaryotes (Figure 1A and Figure 1—figure supplement 1A) and within mammals (Figure 1B and Figure 1—figure supplement 1B) as described previously (Sherill-Rofe et al., 2019; Bloch et al., 2020; Tsaban et al., 2021; Braun et al., 2020).

Figure 1. The phylogenetic profile of MECP2.

(A) The Normalized Phylogenetic Profile (NPP) of MECP2 in eukaryotes along with the top 5000 co-evolved proteins. Each row represents one human protein, ordered by Pearson correlation to MECP2, with the top 200 proteins labeled as E200. Darker color indicates higher normalized conservation of the protein in each species. Organisms are grouped by phylogenetic clade, and clustered within each clade. (B) The NPP of MECP2 in 51 mammals along with the 5000 top co-evolved proteins, ordered by Pearson correlation to MECP within mammals. The top 200 proteins are labeled as M200. (C) Gene overlap of the top 200 most co-evolved genes in eukaryotes and mammals, listing the 10 gene names listed in both.

Figure 1.

Figure 1—figure supplement 1. Scaled phylogenetic profiles of all human genes by similarity to MECP2.

Figure 1—figure supplement 1.

(A) NPP profiles of human genes in 1028 eukaryote species, ordered by the Pearson correlation to the profile of MECP2. (B) Phylogenetic profiles of human genes in 51 mammals, scaled to highlight the conservation patterns within the clade. (C) GeneAnalytics enrichment for the 390 union set of E200 and M200. Within each organ, individual anatomical compartment scores are displayed in dots. (D) Overlap between the 390 union set of E200 and M200 with all STRING MECP2 coexpression, experimental and textmining categories. (E) Overlap between the 200 genes most correlated to MECP2 in mammals (M200) and eukaryotes (E200) with the 390 genes most correlated to MECP2 through STRING.
Open in a new tab

We first used NPP scores to sort all human genes by their phylogenetic similarity to MECP2 across eukaryotes (Figure 1A and Figure 1—figure supplement 1A). We selected the top 200 genes (the ‘E200’ set), which showed high correlation (R>0.47) with MECP2. Similarly, we used NPP scores within the mammalian clade (Figure 1B and Figure 1—figure supplement 1B) to select the top 200 mammal-centric genes (the ‘M200’ set), which also showed high MECP2 correlations (R>0.49) (Supplementary file 1) [code is available on GitHub (Unterman et al., 2021; copy archived at swh:1:rev:14063b44a40688a8024a06347b63cfdac74b96ad)]. These two lists find correlated evolution at different scales, billions of years for eukaryotes and hundreds of millions for mammals. Therefore, the M200 set may show genes more recently co-evolved with MECP2. E200 and M200 had very different functional properties, based on STRING enrichment to Gene Ontology and other annotation sets (Supplementary file 2) The E200 list was enriched in a number of different functional categories, with some of the most significant related to innate immunity and the immune response. These included MHC Class II receptor activity and peptidoglycan receptor activity in GO, and the associated autoimmunity Asthma, Allograft rejection, and Type I in KEGG. In contrast, the M200 list had almost no functional enrichment, except for the general ‘Disease’ category within Uniprot annotated keywords.

The E200 and M200 lists had 10 genes in common (Figure 1C). As mammals are included within the eukaryote analysis, the two lists are not independent and thus the expected degree of overlap is hard to determine. Nevertheless, in two independent gene lists of this size we would only expect two overlapping genes. We thus considered the set of 10 overlapping genes, as well as the full union set of 390 genes, as two tiers of candidates for further exploration. A GeneAnalytics query of the 390 gene union set revealed it was enriched with genes expressed in the brain (p-value < 0.00024), with 79 genes linked to the cerebral cortex and 54 to the cerebellum (Figure 1—figure supplement 1C; Ben-Ari Fuchs et al., 2016). Only 24 of the 390 gene list were overlapping with a length-matched list of MECP2 interactions from STRING (Figure 1—figure supplement 1D,E), highlighting the complementary role phylogenetic analysis to other measurements and predictions of protein interaction.

MECP2 druggable protein network identifies several drug targets located in evolutionarily conserved chromosomal clusters

We chose to focus on the subset of MECP2 co-evolved genes that could be targeted by readily available compounds. This approach allows for straightforward functional testing in both in vitro and animal models of MECP2 inactivation and moves us closer to the ultimate goal of identifying a candidate therapeutic for RTT. Using DGIdb (Cotto et al., 2018) and Open Targets (Oxford Academic, 2019), we searched specific drug-gene interaction types (e.g. inhibitory, activating) to identify direct effects on the 390 candidate genes. Open Targets identified such protein-compound interactions for 11 proteins from our list, and DGIdb identified an additional 22, resulting in 33 druggable proteins in the MECP2 phylogenetic network. MECP2-linked proteins were neither more or less druggable than other proteins in the genome, since the number expected by chance was 39.9±6.0.

Two of the 33 druggable NPP-linked proteins (IRAK1 and EPOR) were found in both the eukaryotic E200 and the mammalian M200 lists, and were targeted by compounds with clinical efficacy and acceptable safety profiles (Pacritinib and EPI), making these immediately attractive for further study. In a STRING (Szklarczyk et al., 2019) functional interaction map, which is based on independent data types, IRAK1 was one of the six proteins functionally linked to MECP2, whereas EPOR had no known functional links (Figure 2A). To identify additional targets among the 31 druggable NPP-linked proteins found in only one of the E200 and M200 lists, we investigated chromosomal clusters of genes with MECP2-linked genes, since clustered genes often indicate a shared evolutionary function. Indeed, IRAK1 is directly adjacent to MECP2 on an evolutionary conserved chromosomal band on the X chromosome (Figure 2B and Figure 2—figure supplement 1). In a chromosomal clustering analysis of all MECP2-linked genes, the chr19p13.2 band contained more MECP2 co-evolved genes than any other in the genome (Figure 2C). Unexpectedly, this 5.6 Mb domain contained 5 of the 33 druggable MECP2 coevolved genes - EPOR, DNMT1, ICAM1, ICAM3, and KEAP1 (Figure 2D), making this the most strongly enriched band in the genome (p<1E-5, Figure 2D and Figure 2—figure supplement 2). Topologically Associated Domains (TADs) define the 3D organization of functionally co-regulated gene clusters (Fritz et al., 2019), and the entire chr19p13.2 gene cluster defines a single TAD-based high-resolution Hi-C mapping (Rao et al., 2014; Figure 2E). Additionally, this region has been shown to be bound by MECP2 (Yasui et al., 2007).

Figure 2. The MECP2 druggable genes network.

(A) The 390 genes co-evolved with MECP2 in either eukaryotes or mammals contain 33 genes that are targets of known drugs/compounds. These genes are shown in a STRING interaction graph, with known gene-gene interactions shown as colored edges, and those with very strong co-evolution (Pearson correlation > 0.7) shown as gray edges. The two proteins co-evolved in both eukaryotes and mammals (EPOR and IRAK1) are colored yellow and the four other genes found at chr19p13.2 are colored purple. (B) Genomic location of MECP2 interacting genes along chromosome X in humans and mice. (C) Karyotype band locations of MECP2 and the 390 co-evolved genes. * p<0.05, **p<0.01, *** p<0.001 (D) Genomic location of MECP2 interacting genes along chr19p13.2 in humans and chromosome nine in mice. (E) Intra-chromosomal Hi-C contact heatmap for the chr19p13.2 locus in GM12878 cells, adapted from the 3D Genome Browser (Wang et al., 2018). Genes in the MECP2 network are shown, with other genes hidden for clarity. TADs called by 3D Genome Browser shown in blue/yellow track, with a super-TAD containing MECP2 genes outlined in red.

Figure 2.

Figure 2—figure supplement 1. Synteny in the Chr19p13.2 locus.

Figure 2—figure supplement 1.

(A) NCBI Genome Data Viewer view of the genomic locations of DNMT1, ICAM1, ICAM3, KEAP1, and EPOR (top panel, genes highlighted). The gene orthologs are shown in Mus musculus chromosome 9 (bottom panel). (B) Inference of the synteny in ancestral species from Genomicus. Each row represents the same genomic locus at an ancestral state. Each gene is represented by a different color, and selected genes are annotated below. EPOR, ICAM1, ICAM3, KEAP1, and DNMT1 maintain their relative positions as far as the mammalian ancestral genome.
Figure 2—figure supplement 2. Synteny between MECP2 and IRAK1.

Figure 2—figure supplement 2.

(A) NCBI Genome Data Viewer view of the genomic locations of MECP2 and IRAK1 (top panel). The gene orthologs are shown in Mus musculus chromosome X (bottom panel). (B) Inference of the synteny in ancestral species from Genomicus.
Open in a new tab

For the two druggable proteins found in both NPP lists (IRAK1, EPO1) and the other four proteins located in the chr19p13.2 TAD (DNMT1, ICAM1, ICAM3, KEAP1), we further investigated the associated drug’s mode of action and efficacy and safety profile in other contexts. IRAK1 has been identified as a key intermediary in NF-κB signaling in RTT, with MECP2 directly binding the IRAK1 promoter to regulate its expression in MECP2-null mouse brains and NFΚB1 dramatically increasing lifespan in MECP2 null mice (Kishi et al., 2016). Pacritinib is a JAK2/FLT3 inhibitor with IRAK1 inhibiting capabilities (Jensen et al., 2017), presently under clinical investigation for myelofibrosis and glioblastoma. As it is currently the only clinical stage IRAK1 inhibitor with known clinical efficacy and acceptable safety even after prolonged administration (Singer et al., 2018), we chose this compound for our functional studies to target IRAK1.

EPOR forms a heterodimer with β common receptor (βCR), also known as the Tissue-Protective Receptor (TPR) (Brines et al., 2004). The endogenous hormone EPO binds the TPR activating signaling cascades with roles in neuroprotection (Rey et al., 2019). EPO has a very well understood pharmacological profile, and several variants of erythropoietin lacking hematopoietic effects have been developed and shown to preserve the neuroprotective effects of EPO administration (Chen et al., 2015; Patel et al., 2011).

KEAP1 is the key repressor of the transcription factor nuclear factor erythroid 2-related factor 2 (NFE2L2), which is involved in response to antioxidants through its transcriptional regulation of inflammatory pathways. Dimethyl fumarate (DMF) is an approved treatment for both multiple sclerosis and psoriasis (Torkildsen et al., 2016; Blair, 2018). Its administration decreases KEAP1 levels, increases NFE2L2 levels and causes dissociation and nuclear transport of NFE2L2. DMF treatment has proven to be effective and safe in conferring neuroprotection (Montes Diaz et al., 2018).

Since the three candidate proteins targeted by these drug products (IRAK1, EPOR, and KEAP1) all have possible roles mediating inflammation, we chose to proceed with these for functional studies in cells modeling neural inflammatory phenotypes. The other three proteins identified (DNMT1, ICAM1, and ICAM3) may also be interesting targets for future study.

Human neural cell cultures exhibit Rett-like phenotypes when MECP2 is silenced

To analyze the effects of the three selected compounds on a RTT model system, we employed human immortalized primary neural cells silenced for MECP2. We chose to use a human in vitro model system for preliminary validation due to the significant differences in the phenotypes and functions of mouse and human neural cells (Xu et al., 2018). Indeed, recent studies reported the use of iPSCs and immortalized glial cells as reliable human models for analyzing specific therapeutic targets for various diseases including autism (Tsilioni et al., 2020), neuroinflammation (Chiavari et al., 2019; Timmerman et al., 2018) and for analyzing the crosstalk of glioma with glial cells (Henrik Heiland et al., 2019; Zeng et al., 2020; Bier et al., 2020).

Since all three selected compounds were associated with changes in neuroinflammation, we first focused on microglia and astrocytes. Loss or impaired function of MECP2 has been reported to affect the functions of glial cells and these changes have been implicated in the pathogenesis of RTT (Kahanovitch et al., 2019; Jin et al., 2017). We first characterized the effects of MECP2 silencing on the phenotypes of microglia (Figure 3). For these experiments, we transduced the cells with lentivirus vectors expressing control or MECP2 shRNAs. The expression of MECP2 was silenced by over 80% in these cells (Figure 3—figure supplement 1). MECP2 silencing induced a relative increase in the expression of M1 markers (IL1, and CD86) and a relative decrease of M2 markers (CD206 and IL-13) (Figure 3A), suggesting an increase in neuroinflammation. We also found that MECP2 silencing decreased the phagocytosis of the microglia cells (Figure 3B).

Figure 3. Effects of MECP2 knockdown on neural cell phenotypes.

Human microglia (A,B), NSCs (C), astrocytes (D,E), and neurons (E) were silenced for MECP2 using lentivirus vector. The relative expression of M1 and M2 markers was analyzed in microglia cells using RT-PCR (A) and degree of phagocytosis using the pHrodo assay (B). NSCs were transduced with lentivirus vectors expressing the differentiation reporters GFAP and MAP2 and were differentiated as described in the methods. Ten days later, luciferase activity was determined (C). Astrocytes silenced for MECP2 were analyzed for the expression of EAAT2 using western blot analysis (D) and the expression of BDNF mRNA in both MECP2 silenced astrocytes and neurons was determined using RT-PCR (E). The results are a representative experiment of three separate tests analyzed in quadruplet. **p<0.01, ***p<0.001,****p<0.0001.

Figure 3.

Figure 3—figure supplement 1. Silencing of MECP2 in neural cells.

Figure 3—figure supplement 1.

Neurons, astrocytes, and microglia cells were transduced with lentivirus vectors expressing a control or MECP2 shRNAs. Following 3 days, the expression of MECP2 and actin were determined using western blot analysis. The results are representative of three experiments with similar results.
Figure 3—figure supplement 2. Cytotoxic effects of tested compounds.

Figure 3—figure supplement 2.

Human microglia (A), astrocytes (B), neurons (C), and neuronal stem cells (D) were treated with different concentrations of DMF (1, 10, and 50 µM), EPO (10, 30, and 50 ng/ml), and pacrinitib (1 and 10 µM). Following 3 days of treatment, the cells were analyzed for cell death using LDH assay. The results are the means ± SD of six samples for each compound.
Open in a new tab

We next analyzed the effect of MECP2 silencing on human astrocyte differentiation and functions. For these experiments, neural stem cells (NSCs) were transduced with lentivirus vectors expressing the differentiation reporters, GFP/fLuc GFAP or the mCherry/Luc MAP2. The cells were plated on laminin coated plates and luciferase activity was determined 10 days later. In accordance with previous reports (Andoh-Noda et al., 2015; Mok et al., 2020), we found that MECP2 silenced NSCs displayed preferential expression of the glial differentiation marker (GFAP) with a concomitant decrease in expression of the neuronal differentiation marker (MAP2) (Figure 3C). The MECP2 silenced astrocytes expressed lower levels of Excitatory Amino Acid Transporter 2 (EAAT2), indicating impaired function of these cells, as was suggested for mouse astrocytes (Okabe et al., 2012; Figure 3D).

We finally examined the effects of MECP2 silencing on the expression of BDNF in human neurons and astrocytes and found that, as reported for mouse cells (Chang et al., 2006; Li et al., 2012), MECP2 silencing decreased the expression of BDNF mRNA in both astrocyte and neuronal cultures (Figure 3E). Altogether, these finding indicate that the immortalized human cells employed in this study represent a reliable model for studying MECP2-related pathways and potential treatments.

DMF, EPO, and pacritinib differentially abrogate the effects of MECP2 silencing on microglia polarization

We then analyzed the effects of the three selected compounds, DMF, EPO and pacritinib, on MECP2 silenced cells. We first demonstrated that in the concentrations range we used, none of these compounds exerted a toxic effect on neither microglia cells, astrocytes, primary neuronal cells nor NSCs (Figure 3—figure supplement 2).

Control and MECP2 silenced microglia were treated with the DMF, EPO, and pacritinib, and their effects on the relative expression M1 (IL1, TNFα and CD86) and M2 (CD206 and IL-13) markers were determined using RT-PCR. While MECP2 wild-type cells were unchanged by the three compounds, all three abrogated the M1 shift induced by MECP2 silencing, albeit to a different degree (Figure 4A–C). EPO exerted the most significant effect, abrogating the effects of MECP2 silencing in all five marker genes (Figure 4A). In contrast, both DMF and pacritinib exerted only partial effects, abrogating the effects of MECP2 silencing in CD206, CD86, and IL-1, but failing to abrogate the effects on IL-13 and TNF-α (Figure 4B–C). We also analyzed the effects on phagocytosis in microglia cells. Treatment of MECP2 silenced cells with DMF exerted a significant increase in the phagocytosis of the silenced cells, whereas EPO and pacritinib did not (Figure 4D).

Figure 4. Effects of tested compounds on the polarization of microglial cells.

Figure 4.

Open in a new tab

Human microglia cells were silenced for MECP2 using lentivirus vectors expressing MECP2 shRNA. Control cells were transduced with lentivirus vectors expressing a control shRNA. After 5 days, the expression of MECP2 was determined by western blot analysis (Figure 3—figure supplement 1). Control and silenced cells were treated with EPO 10 ng/ml (A), DMF 10 µM (B) and pacritinib 10 µM (C), and the expression of M1 and M2-associated markers were determined after 72 hr using RT-PCR. MECP2 silenced microglia cells treated with EPO, DMF, or pacritinib were also analyzed for phagocytosis using the pHrodo assay (D). MECP2-silenced microglia cells were transduced with lentivirus vectors expressing the NF-kB reporter followed by treatment with EPO, DMF, and pacritinib for 24 hr. Luciferase activity was determined (E). The results demonstrate the means ± SD of a representative experiment of three separate tests analyzed in quadruplet. *p<0.05, **p<0.01, ***p<0.001,****p<0.0001.

EPO and DMF inhibit NF-κB activation in MECP2 silenced microglia cells

Knockdown of MECP2 has been reported to lead to increased NF-κB activity in neuronal and myeloid cells (O'Driscoll et al., 2015) and this deregulated activity has been associated with neuroinflammation and decreased dendritic arborization and spine density in a mouse model of RTT (Kishi et al., 2016). We therefore examined the ability of our three candidate compounds to abrogate the increased NF-κB activity in MECP2 KD microglia cells. MECP2-silenced microglia cells were transfected with lentivirus vectors expressing the NF-κB luciferase reporter and a constitutively active Renilla luciferase construct (as a transfection control). The cells were treated with EPO, DMF and pacritinib for 24 hr and luciferase activity was determined thereafter. A significant increase in NF-κB-dependent luciferase activation was observed for EPO and DMF in the MECP2-silenced cells (Figure 4E), indicating that EPO and DMF were able to downregulate the increased NF-κB activation in MECP2 KD cells.

EPO, but not DMF or pacritinib, abrogate improper astrocytic differentiation and astrocyte function in MECP2-silenced neural cells

Various studies support a non-cell-autonomous effect of astrocytes on neuronal cell functions and contribution of astrocytes to various aspects of Rett syndrome pathogenesis via regulation of glutamate levels, homeostasis, and neuroinflammation (Kahanovitch et al., 2019; Maezawa et al., 2009). As presented in Figure 3C, MECP2 silencing increased astrocytic differentiation of NSCs at the expense of neuronal differentiation. Treatment of the silenced NSCs with pacritinib or DMF did not have significant effects on markers of neuronal (βIII tubulin) or astrocytic (GFAP) differentiation (Figure 5A). In contrast, EPO abrogated the impaired NSC differentiation as reflected by an increase of βIII tubulin and a decrease of GFAP (Figure 5A). In human astrocytes, MECP2 silencing decreased the expression of the glutamate transporter EAAT2 (Figure 3D). Treatment of these cells with EPO inhibited the decreased EAAT2 expression, whereas pacritinib and DMF did not have a significant effect (Figure 5B).

Figure 5. Effects of tested compounds on the differentiation of NSCs and EAAT2 expression in astrocytes.

Figure 5.

Open in a new tab

Human NSCs (A) and astrocytes (B) were silenced for MECP2 using lentivirus vectors expressing MECP2 shRNA. Control cells were transduced with lentivirus vectors expressing a control shRNA. After 5 days, the expression of MECP2 was determined by western blot analysis (Figure 3—figure suppplement 1). (A) Control and silenced NSCs were allowed to differentiate for 10 days and the expression of differentiation markers GFAP and βIII tubulin were determined using RT-PCR. (B) Silenced astrocytes treated with the different compounds or with medium were analyzed for the expression of EAAT2 using RT-PCR. The results are of a representative experiment of three separate tests analyzed in quadruplet. *p<0.05, **p<0.01, ***p<0.001,****p<0.0001.

DMF and EPO upregulate BDNF expression in MECP2-silenced cells

BDNF plays important roles in neuronal growth and development and represents a well-recognized transcriptional target of MECP2 (Chen et al., 2003). Silencing of MECP2 decreased BDNF expression in both neurons and astrocytes (Figure 6A–B). DMF and EPO induced an increase in BDNF mRNA in the silenced neurons (Figure 6A) and in the silenced astrocytes (Figure 6B), whereas pacritinib did not exert a significant effect in either. We replicated this experiment in astrocytes, testing DMF and EPI both individually and in combination (Figure 6C). This showed a modest increase in the level of BDNF in the in the combined treatment, suggesting potentially independent modes of action.

Figure 6. BDNF expression in neural cells.

Figure 6.

Open in a new tab

MECP2 expression was silenced in human neurons (A) and astrocytes (B, C) and BDNF expression was analyzed in cells treated with a medium or the specific tested compounds using both RT-PCR (A) and ELISA (B, C). The combinatorial effect of EPO and DMF was assessed on BDNF expression in astrocytes (C). *p<0.05, **p<0.01, ***p<0.001,****p<0.0001.

Discussion

Rett syndrome is a neurodevelopmental disease without evidence of degeneration and thus rescuing MECP2 downstream activity might improve disease pathophysiology even in adulthood (Faundez et al., 2019). In murine models, reintroduction of MECP2 to adult animals led to vast phenotypic improvements (Guy et al., 2007). In parallel with approaches aimed at increasing MECP2 levels directly (Brendel et al., 2011; Merritt et al., 2020; Przanowski et al., 2018), accurate mapping of the MECP2 gene network is important to provide additional avenues for intervention, and avoid the toxicity associated with overexpression of MECP2 (Ramocki et al., 2009). Here, we have presented a framework for identifying potential drug-gene interactions based on comparative genomics, and validate these interactions in human immortalized primary neural cells. We analyzed phylogenetic profiling across mammals and hundreds of eukaryotes to infer functional links and generated a network of co-evolved genes, identifying known targets such as IRAK1 as well as new candidates such as EPOR. We then mined this network to identify actionable drug targets, prioritizing a list of associated compounds using a combination of strong co-evolutionary evidence of the target proteins and a proven safety profile of approved pharmaceutical compounds (Pacritinib, DMF, and EPO). While we prioritized based on the potential for drug repurposing, our evolutionary analysis could also be used to identify additional targets for drug design. Furthermore, we only performed functional validation 3 of the 33 co-evolved genes associated with approved drugs, leaving others for potential future study.

We reproduced a number of RTT phenotypes by reducing expression of MECP2 in human neural cell cultures. When treated with the three selected compounds, these phenotypes could all be reversed to different degrees in a cell-dependent manner. While all the compounds restored the polarization of microglia and reduced the hyperactivity of NF-κB signaling in microglia, only DMF abrogated decreased phagocytosis. EPO alone reversed the impaired neuronal/astrocyte differentiation of NSCs. In neurons and astrocytes, DMF and EPO abrogated the inhibitory effect of MECP2 silencing on BDNF expression and secretion, whereas pacritinib did not have a significant effect.

The activity of these compounds on microglia polarization and NF-κB activity is mechanistically interesting. Activation of microglia cells has been reported in MECP2 KO mice (Zhao et al., 2017) and has been associated with pathological process in RTT (Kahanovitch et al., 2019). An MECP2 KO mouse model also implicated NF-κB signaling and IRAK1 as important mediators of the neuroinflammatory response in RTT (Kishi et al., 2016). While the mechanisms involved in the reversing effects of these three compounds remains to be determined, NF-κB signaling downstream of MECP2 is an attractive candidate (Figure 7). EPOR and KEAP1 have also been implicated as NF-κB modulators involved in the inflammatory response. The endogenous hormone EPO binds the TPR to suppresses inflammatory cytokines in immune cells, a process which involves inhibition of the NF-κB pathway through GSK3β (Patel et al., 2011). NFE2L2, usually bound by KEAP1, also has a direct cytoplasmic role in modulating NF-κB signaling through degradation of IκB and nuclear translocation of NF-κB (Ganesh Yerra et al., 2013). The three compounds tested all inhibited NF-κB-dependent luciferase activation in microglia cells to a certain extent, validating their involvement in this context.

Figure 7. A model of MECP2 network genes converging on NF-κB signaling.

Figure 7.

Open in a new tab

(A) We propose that reduced MECP2 levels could lead to increase in ICAM1, ICAM3, IRAK1, and KEAP1 levels, and a decrease of EPOR levels within the tissue-protective receptor. (B) IRAK-1 plays a role in NF-κB activation through IKK activation, which leads to IκB phosphorylation and subsequent degradation, NF-κB nuclear translocation and the expression of its pro-inflammatory target genes. KEAP1 binds NFE2L2 and prevents its nuclear localization. NFE2L2 target genes inhibits NF-κB signaling through a non-transcriptional mechanism involving degradation of IκBα. EPO binds EPOR and enhances the activation of Akt, resulting in inhibition of GSK-3β and inhibition of NF-κB nuclear transport.

In addition to the inflammatory effect, we found that MECP2 silencing decreased the neuronal while increasing the astrocytic differentiation of NSCs. However, human astrocyte function was impaired in the MECP2 silenced cells and these cells expressed lower levels of the glutamate transporter EAAT2. The signaling pathways involved in these effects of EPO are not fully characterized; however, NF-κB has been recently reported to regulate EAAT2 expression in astrocytes (Wei et al., 2019).

Finally, we found that both DMF and EPO rescued the decreased BDNF expression in MECP2-silenced microglia and astrocytes. Indeed, both NF-κB and NFE2L2 pathways have been reported to be associated with the regulation of BDNF expression (Caviedes et al., 2017; Nair and Wong-Riley, 2016). Thus, DMF can increase BDNF levels through an increase in NFE2L2, which binds the BDNF promotor and acts as a transcriptional activator (Nair and Wong-Riley, 2016). EPO administration activates the STAT3 signaling pathway (Rey et al., 2019), which can in turn elevate BDNF expression (Chen et al., 2016).

While the strong evolutionary linkage of our target proteins to MECP2 strongly suggest that they are mediating the reversal of MECP2 silencing phenotypes by their associated drugs, we have not ruled out the possibility for alternative action of these drugs or joint action upon our involving other protein targets. A good example of this are the HDAC chromatin modifiers, a group of genes which did not come up as specifically linked to MECP2 in our evolutionary analysis but are well-known interactors. HDACs play a crucial role in neuronal development, and MECP2-mediated gene repression occurs mainly through HDAC recruitment via the Sin3a and NCoR complexes (Lyst et al., 2013; Ebert et al., 2013; Nan et al., 1998). HDACs have been suggested as therapeutic targets for Rett Syndrome (Shukla and Tekwani, 2020) and HDAC6 inhibition has been linked to improved BDNF trafficking in a Rett mouse model (Xu et al., 2014). Importantly, one of the compounds we examined, DMF, can reduce HDAC1/2/4 expression in rat astrocytes (Kalinin et al., 2013), and EPO can promote phosphorylation of HDAC5 in rat hippocampal neurons (Jo et al., 2016). It is thus plausible that HDAC inhibition mediates the effects we observe for these drugs, at least in part. However, the effects on HDAC activity are not the principal mechanisms of action of either of the proposed compounds (Kalinin et al., 2013; Jo et al., 2016), nor do they have established interactions with the HDAC most implicated in RTT biology (HDAC6). Future studies should investigate if any of the MECP2-rescuing activities of these drugs act through HDACs in concert with or independent of the target proteins identified here.

In summary, EPO, DMF and pacritinib act differentially on MECP2-silenced NSCs, microglia and astrocytes, providing a preliminary validation of our approach. The investigation into specific drug effects and mechanisms of actions are important areas for future studies, as are drug combination assays and in-vivo studies of EPO, DMF and pacritinib using MECP2 KO mice. These will be to assess whether any of the identified drugs can be further be explored for clinical use.

Materials and methods

Mapping MECP2 conservation along 1028 eukaryotic species and identifying correlated genes

We used the pipeline we described in Tsaban et al., 2021 to calculate the Normalized Phylogenetic Profile (NPP) of 1028 eukaryotic species and rank genes based on their evolutionary similarity to MECP2. Briefly, 20,192 human proteins (one representative protein sequence per gene) were downloaded from UniProt reference proteomes (June 2018 release) (Bateman and UniProt Consortium, 2019). We used the Uniprot canonical isoform for each gene. For the 29 genes with more than one canonical isoform, we selected the longest isoform to maximize sequence information. The proteins were searched with BLASTP (Camacho et al., 2009) against the proteomes of the 1028 eukaryotes. A bit-score of 20.4, corresponding to a BLAST e-value of 0.05, was set as a minimal similarity threshold. The top scoring protein in each organism was selected. The bit-scores were normalized to protein length and phylogenetic distance from humans. The output is a matrix P of size 20,242 x 1028 where each entry Pab is the best BLASTP bit score between a human protein sequence ‘a’ and the top result in organism ‘b’. The Pearson correlation was calculated for each profile with the MECP2 profile. The top 200 genes in all eukaryotes were selected as the E200 group for further analysis. This procedure was repeated using only the 51 mammal species, and the top 200 genes in were selected as the M200 group for further analysis.

Filtering genes with drug interactions and constructing protein network

Drug gene interactions were collected from DGIdb (Cotto et al., 2018) and from Open Targets (Oxford Academic, 2019). Drugs at any stage of development were included in the search as long as the interaction with the target gene had known directionality (e.g. inhibitory, activating). Drug-gene interactions were validated through a literature review and known gene-gene interactions were collected from STRING (Szklarczyk et al., 2019). Network diagram was constructed using STRING and Cytoscape (Lopes et al., 2011).

Synteny and co-localization analysis

Gene locations were retrieved from BioMart (Kasprzyk, 2011), the NCBI genome data viewer (Figure 2—figure supplement 1A, Figure 2—figure supplement 2A), and Genomicus (Nguyen et al., 2018; Figure 3A). Synteny in ancestral species was obtained from Genomicus (Nguyen et al., 2018). Intra-chromosomal Hi-C contact heatmaps and TAD borders were collected from the 3D Genome Browser (Wang et al., 2018), using GM12878 cell data from the high-resolution Hi-C dataset of Rao et al., 2014.

Neural cell cultures

Immortalized human microglial cells and astrocytes were obtained from Applied Biological Material (Richmond, BC, Canada). Human NSCs (H9, hESC-derived) (ReNcell) were obtained from Invitrogen, Merck (Germany). Human neurons were obtained from ScienCell (Carlsbad, CA, USA). All cells employed in this study were tested for mycoplasma contamination (Mycoplasma PCR Detection Kit) and found negative.

Transduction of neural cells

Lentivirus vectors (System Biosciences, Mountain View, CA, USA) expressing the MECP2 or control shRNAs and the reporters GFP/fLuc GFAP, mCherry/Luc MAP2 and NF-kB luciferase were packaged and used to transduce the cells according to the manufacturer's protocol and as previously described (Giladi et al., 2015).

NSC differentiation

Human NSCs were maintained as spheroids in an NSC maintenance medium containing fibroblast growth factor 2 (FGF-2) and an epidermal growth factor (EGF, 20 ng/ml) on laminin-coated flasks. For differentiation, the cells were maintained in NSC maintenance medium without FGF-2 and EGF and neuronal and glial differentiation were observed.

Microglia polarization

Human microglia cells were silenced for MECP2 using lentivirus vectors expressing MECP2 or control shRNAs. After 5e days, the expression of MECP2 was determined by western blot analysis and cells exhibiting a decrease of at least 80% were employed in further studies. Control and silenced cells were treated with the specific compounds and the expression of M1- and M2-associated markers were determined after 72 hr using RT-PCR. All experiments were done in triplicates and were repeated three times.

Cytotoxicity assay

Cells were treated with different concentrations of DMF (1 10 and 50 mM), EPO (10, 30 and 50 ng/ml), and Pacritinib (1 and 10 mM). Following 3 days of treatment, the cells were analyzed for cell death using LDH assay. The results are presented as the means ± SD of six samples for each compound.

Western blot analysis

Cell pellet preparation and Western blot analyses were performed as previously described (Lomonaco et al., 2009). Equal loading was verified using an anti-β-actin or tubulin antibodies as described (Giladi et al., 2015; Bier et al., 2018).

Real-time PCR

Total RNA was extracted using RNeasy midi kit according to the manufacturer’s instructions (Qiagen, Frederick, MD, USA). Reverse transcription reaction was carried out using 2 μg total RNA (Bier et al., 2018). The primer sequences are described in the supplementary files (Supplementary file 3).

Phagocytosis analysis

Human microglial cells were silenced for MECP2. Phagocytosis was determined using the pHrodo Green zymosan bioparticle assay (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Briefly, microglia were incubated with a solution of pHrodo Green zymosan bioparticles in Live Cell Imaging Solution (0.5 mg/ml) for 2 hr. Phagocytosis was determined using a fluorescence plate reader at Ex/Em 509/533.

Luciferase activity

The firefly luciferase activity of the NF-kB, GFAP and MAP two activities and the control Renilla luciferase activity were analyzed using the Dual-Luciferase Reporter Assay System (Promega Corporation).

Statistical analysis for cell assays

The results are presented as the mean values ± SD. Data were analyzed using a Student's t-test with correction for data sets with unequal variances (Welch's unequal variances t-test).

Acknowledgements

We thank Dr. Yaron Daniely and the Yissum team for being a major driving force for this project. We thank Dr. Liana Patt and the Integra Holdings team for their support. We thank Dr. Hae Kyung Lee and Susan Finniss for their technical support. Figures include vector images designed by Freepik. We thank Dr. Amir Eden and Prof. Yinon Ben Neriah for their insight and comments. Funding: This project was supported by an Israel Science Foundation grant 1591/19 to YT and by Integra Holdings. Additional support to BPB and IU came from a grant from the Beethoven Foundation.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Benjamin P Berman, Email: ben.berman@mail.huji.ac.il.

Chaya Brodie, Email: chaya@brodienet.com.

Yuval Tabach, Email: yuvaltab@ekmd.huji.ac.il.

George H Perry, Pennsylvania State University, United States.

George H Perry, Pennsylvania State University, United States.

Funding Information

This paper was supported by the following grants:

  • Israel Science Foundation 1591/19 to Yuval Tabach.

  • Integra LifeSciences to Yuval Tabach.

  • Beethoven Foundation to Benjamin P Berman.

Additional information

Competing interests

No competing interests declared.

Author contributions

Software, Formal analysis, Investigation, Visualization, Writing - original draft.

Resources, Data curation, Software, Formal analysis.

Validation, Investigation, Visualization, Methodology.

Validation, Investigation, Visualization, Methodology.

Conceptualization, Methodology, Writing - review and editing.

Conceptualization, Resources, Supervision, Writing - original draft, Writing - review and editing.

Conceptualization, Supervision, Investigation, Methodology, Writing - original draft.

Conceptualization, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing.

Additional files

Supplementary file 1. Genes most co-evolved with MECP2.
elife-67085-supp1.csv (6.2KB, csv)
Supplementary file 2. STRING GO-term enrichment of the E200 set.
elife-67085-supp2.txt (6.5KB, txt)
Supplementary file 3. RT-PCR primer sequences.
elife-67085-supp3.docx (13.3KB, docx)
Transparent reporting form

Data availability

The phylogenetic data required to reproduce the analysis are openly available at Zenodo [https://zenodo.org/record/4464120#.YEoFpHmxVPY]. Code used for the analysis is openly available through Github https://github.com/GuIrene/MECP2_phylogeny (copy archived at https://archive.softwareheritage.org/swh:1:rev:14063b44a40688a8024a06347b63cfdac74b96ad).

The following previously published dataset was used:

Stupp D, Tsaban T, Bloch I, Sharon E, Tabach Y. 2021. CladeOScope: elucidating functional interactions via a clade co-evolution prism. Zenodo.

References

  1. Amir RE, Van den Veyver IB, Wan M, Tran CQ, Francke U, Zoghbi HY. Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl-CpG-binding protein 2. Nature Genetics. 1999;23:185–188. doi: 10.1038/13810. [DOI] [PubMed] [Google Scholar]
  2. Andoh-Noda T, Akamatsu W, Miyake K, Matsumoto T, Yamaguchi R, Sanosaka T, Okada Y, Kobayashi T, Ohyama M, Nakashima K, Kurosawa H, Kubota T, Okano H. Differentiation of multipotent neural stem cells derived from rett syndrome patients is biased toward the astrocytic lineage. Molecular Brain. 2015;8:31. doi: 10.1186/s13041-015-0121-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Arkadir D, Lossos A, Rahat D, Abu Snineh M, Schueler-Furman O, Nitschke S, Minassian BA, Sadaka Y, Lerer I, Tabach Y, Meiner V. MYORG is associated with recessive primary familial brain calcification. Annals of Clinical and Translational Neurology. 2019;6:106–113. doi: 10.1002/acn3.684. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Avidor-Reiss T, Maer AM, Koundakjian E, Polyanovsky A, Keil T, Subramaniam S, Zuker CS. Decoding cilia function: defining specialized genes required for compartmentalized cilia biogenesis. Cell. 2004;117:527–539. doi: 10.1016/s0092-8674(04)00412-x. [DOI] [PubMed] [Google Scholar]
  5. Bateman A, UniProt Consortium UniProt: a worldwide hub of protein knowledge. Nucleic Acids Research. 2019;47:D506–D515. doi: 10.1093/nar/gky1049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Ben-Ari Fuchs S, Lieder I, Stelzer G, Mazor Y, Buzhor E, Kaplan S, Bogoch Y, Plaschkes I, Shitrit A, Rappaport N, Kohn A, Edgar R, Shenhav L, Safran M, Lancet D, Guan-Golan Y, Warshawsky D, Shtrichman R. GeneAnalytics: an integrative gene set analysis tool for next generation sequencing, RNAseq and microarray data. OMICS: A Journal of Integrative Biology. 2016;20:139–151. doi: 10.1089/omi.2015.0168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bier A, Berenstein P, Kronfeld N, Morgoulis D, Ziv-Av A, Goldstein H, Kazimirsky G, Cazacu S, Meir R, Popovtzer R, Dori A, Brodie C. Placenta-derived mesenchymal stromal cells and their exosomes exert therapeutic effects in Duchenne muscular dystrophy. Biomaterials. 2018;174:67–78. doi: 10.1016/j.biomaterials.2018.04.055. [DOI] [PubMed] [Google Scholar]
  8. Bier A, Hong X, Cazacu S, Goldstein H, Rand D, Xiang C, Jiang W, Ben-Asher HW, Attia M, Brodie A, She R, Poisson LM, Brodie C. miR-504 modulates the stemness and mesenchymal transition of glioma stem cells and their interaction with microglia via delivery by extracellular vesicles. Cell Death & Disease. 2020;11:1–14. doi: 10.1038/s41419-020-03088-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Blair HA. Dimethyl fumarate: a review in moderate to severe plaque psoriasis. Drugs. 2018;78:123–130. doi: 10.1007/s40265-017-0854-6. [DOI] [PubMed] [Google Scholar]
  10. Bloch I, Sherill-Rofe D, Stupp D, Unterman I, Beer H, Sharon E, Tabach Y. Optimization of co-evolution analysis through phylogenetic profiling reveals pathway-specific signals. Bioinformatics. 2020;36:4116–4125. doi: 10.1093/bioinformatics/btaa281. [DOI] [PubMed] [Google Scholar]
  11. Braun M, Sharon E, Unterman I, Miller M, Shtern AM, Benenson S, Vainstein A, Tabach Y. ACE2 Co-evolutionary pattern suggests targets for pharmaceutical intervention in the COVID-19 pandemic. iScience. 2020;23:101384. doi: 10.1016/j.isci.2020.101384. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Brendel C, Belakhov V, Werner H, Wegener E, Gärtner J, Nudelman I, Baasov T, Huppke P. Readthrough of nonsense mutations in rett syndrome: evaluation of novel aminoglycosides and generation of a new mouse model. Journal of Molecular Medicine. 2011;89:389–398. doi: 10.1007/s00109-010-0704-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Brines M, Grasso G, Fiordaliso F, Sfacteria A, Ghezzi P, Fratelli M, Latini R, Xie QW, Smart J, Su-Rick CJ, Pobre E, Diaz D, Gomez D, Hand C, Coleman T, Cerami A. Erythropoietin mediates tissue protection through an erythropoietin and common beta-subunit heteroreceptor. PNAS. 2004;101:14907–14912. doi: 10.1073/pnas.0406491101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, Madden TL. BLAST+: architecture and applications. BMC Bioinformatics. 2009;10:421–429. doi: 10.1186/1471-2105-10-421. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Caviedes A, Lafourcade C, Soto C, Wyneken U. BDNF/NF-κB signaling in the neurobiology of depression. Current Pharmaceutical Design. 2017;23:3154–3163. doi: 10.2174/1381612823666170111141915. [DOI] [PubMed] [Google Scholar]
  16. Chang Q, Khare G, Dani V, Nelson S, Jaenisch R. The disease progression of Mecp2 mutant mice is affected by the level of BDNF expression. Neuron. 2006;49:341–348. doi: 10.1016/j.neuron.2005.12.027. [DOI] [PubMed] [Google Scholar]
  17. Chao HT, Zoghbi HY. MeCP2: only 100% will doNat. NeuroSci. 2012;15:176–177. doi: 10.1038/nn.3027. [DOI] [PubMed] [Google Scholar]
  18. Chen WG, Chang Q, Lin Y, Meissner A, West AE, Griffith EC, Jaenisch R, Greenberg ME. Derepression of BDNF transcription involves calcium-dependent phosphorylation of MeCP2. Science. 2003;302:885–889. doi: 10.1126/science.1086446. [DOI] [PubMed] [Google Scholar]
  19. Chen J, Yang Z, Zhang X. Carbamylated erythropoietin: a prospective drug candidate for neuroprotection. Biochemistry Insights. 2015;8:BCI.S30753. doi: 10.4137/BCI.S30753. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Chen B, Liang Y, He Z, An Y, Zhao W, Wu J. Autocrine activity of BDNF induced by the STAT3 signaling pathway causes prolonged TrkB activation and promotes human non-small-cell lung Cancer proliferation. Scientific Reports. 2016;6:1–8. doi: 10.1038/srep30404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Chiavari M, Ciotti GMP, Navarra P, Lisi L. Pro-Inflammatory activation of A new immortalized human microglia cell line. Brain Sciences. 2019;9:111. doi: 10.3390/brainsci9050111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Cotto KC, Wagner AH, Feng YY, Kiwala S, Coffman AC, Spies G, Wollam A, Spies NC, Griffith OL, Griffith M. DGIdb 3.0: a redesign and expansion of the drug-gene interaction database. Nucleic Acids Research. 2018;46:D1068–D1073. doi: 10.1093/nar/gkx1143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Date SV, Marcotte EM. Discovery of uncharacterized cellular systems by genome-wide analysis of functional linkages. Nature Biotechnology. 2003;21:1055–1062. doi: 10.1038/nbt861. [DOI] [PubMed] [Google Scholar]
  24. Dey G, Jaimovich A, Collins SR, Seki A, Meyer T. Systematic discovery of human gene function and principles of modular organization through phylogenetic profiling. Cell Reports. 2015;10:993–1006. doi: 10.1016/j.celrep.2015.01.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Ebert DH, Gabel HW, Robinson ND, Kastan NR, Hu LS, Cohen S, Navarro AJ, Lyst MJ, Ekiert R, Bird AP, Greenberg ME. Activity-dependent phosphorylation of MeCP2 threonine 308 regulates interaction with NCoR. Nature. 2013;499:341–345. doi: 10.1038/nature12348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Eisen JA, Wu M. Phylogenetic analysis and gene functional predictions: phylogenomics in action. Theoretical Population Biology. 2002;61:481–487. doi: 10.1006/tpbi.2002.1594. [DOI] [PubMed] [Google Scholar]
  27. Enault F, Suhre K, Poirot O, Abergel C, Claverie JM. Phydbac2: improved inference of gene function using interactive phylogenomic profiling and chromosomal location analysis. Nucleic Acids Research. 2004;32:W336–W339. doi: 10.1093/nar/gkh365. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Faundez V, Wynne M, Crocker A, Tarquinio D. Molecular systems biology of neurodevelopmental disorders, rett syndrome as an archetype. Frontiers in Integrative Neuroscience. 2019;13:30. doi: 10.3389/fnint.2019.00030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Fritz AJ, Sehgal N, Pliss A, Xu J, Berezney R. Chromosome territories and the global regulation of the genomeGenes Chromosom. Cancer. 2019;58:407–426. doi: 10.1002/gcc.22732. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Gadalla KKE, Bailey MES, Spike RC, Ross PD, Woodard KT, Kalburgi SN, Bachaboina L, Deng JV, West AE, Samulski RJ, Gray SJ, Cobb SR. Improved survival and reduced phenotypic severity following AAV9/MECP2 gene transfer to neonatal and juvenile male Mecp2 knockout mice. Molecular Therapy. 2013;21:18–30. doi: 10.1038/mt.2012.200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Ganesh Yerra V, Negi G, Sharma SS, Kumar A. Potential therapeutic effects of the simultaneous targeting of the Nrf2 and NF-κB pathways in diabetic neuropathy. Redox Biology. 2013;1:394–397. doi: 10.1016/j.redox.2013.07.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Garg SK, Lioy DT, Cheval H, McGann JC, Bissonnette JM, Murtha MJ, Foust KD, Kaspar BK, Bird A, Mandel G. Systemic delivery of MeCP2 rescues behavioral and cellular deficits in female mouse models of rett syndrome. Journal of Neuroscience. 2013;33:13612–13620. doi: 10.1523/JNEUROSCI.1854-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Giladi ND, Ziv-Av A, Lee HK, Finniss S, Cazacu S, Xiang C, Ben-Asher HW, deCarvalho A, Mikkelsen T, Poisson L, Brodie C. RTVP-1 promotes mesenchymal transformation of glioma via a STAT-3/IL-6-dependent positive feedback loop. Oncotarget. 2015;6:22680–22697. doi: 10.18632/oncotarget.4205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Glaze DG, Neul JL, Kaufmann WE, Berry-Kravis E, Condon S, Stoms G, Oosterholt S, Della Pasqua O, Glass L, Jones NE, Percy AK, Rett 002 Study Group Double-blind, randomized, placebo-controlled study of trofinetide in pediatric rett syndrome. Neurology. 2019;92:e1912–e1925. doi: 10.1212/WNL.0000000000007316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Guy J, Gan J, Selfridge J, Cobb S, Bird A. Reversal of neurological defects in a mouse model of rett syndrome. Science. 2007;315:1143–1147. doi: 10.1126/science.1138389. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Henrik Heiland D, Ravi VM, Behringer SP, Frenking JH, Wurm J, Joseph K, Garrelfs NWC, Strähle J, Heynckes S, Grauvogel J, Franco P, Mader I, Schneider M, Potthoff AL, Delev D, Hofmann UG, Fung C, Beck J, Sankowski R, Prinz M, Schnell O. Tumor-associated reactive astrocytes aid the evolution of immunosuppressive environment in glioblastoma. Nature Communications. 2019;10:1–12. doi: 10.1038/s41467-019-10493-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Hodges ME, Wickstead B, Gull K, Langdale JA. The evolution of land plant cilia. New Phytologist. 2012;195:526–540. doi: 10.1111/j.1469-8137.2012.04197.x. [DOI] [PubMed] [Google Scholar]
  38. Horvath PM, Monteggia LM. MeCP2 as an activator of gene ExpressionTrends. NeuroSci. 2018;41:72–74. doi: 10.1016/j.tins.2017.11.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Iorio F, Saez-Rodriguez J, Bernardo Ddi. Network based elucidation of drug response: from modulators to targets. BMC Systems Biology. 2013;7:139. doi: 10.1186/1752-0509-7-139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Jassal B, Matthews L, Viteri G, Gong C, Lorente P, Fabregat A, Sidiropoulos K, Cook J, Gillespie M, Haw R, Loney F, May B, Milacic M, Rothfels K, Sevilla C, Shamovsky V, Shorser S, Varusai T, Weiser J, Wu G, Stein L, Hermjakob H, D'Eustachio P. The reactome pathway knowledgebase. Nucleic Acids Research. 2020;48:D498–D503. doi: 10.1093/nar/gkz1031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Jensen KV, Cseh O, Aman A, Weiss S, Luchman HA. The JAK2/STAT3 inhibitor pacritinib effectively inhibits patient-derived GBM brain tumor initiating cells in vitro and when used in combination with temozolomide increases survival in an orthotopic xenograft model. PLOS ONE. 2017;12:e0189670. doi: 10.1371/journal.pone.0189670. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Jiang Z. Protein function predictions based on the phylogenetic profile method. Critical Reviews in Biotechnology. 2008;28:233–238. doi: 10.1080/07388550802512633. [DOI] [PubMed] [Google Scholar]
  43. Jin X-R, Chen X-S, Xiao L. MeCP2 deficiency in Neuroglia: new progress in the pathogenesis of rett syndrome. Frontiers in Molecular Neuroscience. 2017;10:316. doi: 10.3389/fnmol.2017.00316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Jo H-R, Kim Y-S, Son H. Erythropoietin and carbamylated erythropoietin promote histone deacetylase 5 phosphorylation and nuclear export in rat hippocampal neurons. Biochemical and Biophysical Research Communications. 2016;470:220–225. doi: 10.1016/j.bbrc.2016.01.039. [DOI] [PubMed] [Google Scholar]
  45. Kahanovitch U, Patterson KC, Hernandez R, Olsen ML. Glial dysfunction in MeCP2 deficiency models: implications for rett syndrome. International Journal of Molecular Sciences. 2019;20:3813. doi: 10.3390/ijms20153813. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Kalinin S, Polak PE, Lin SX, Braun D, Guizzetti M, Zhang X, Rubinstein I, Feinstein DL. Dimethyl fumarate regulates histone deacetylase expression in astrocytes. Journal of Neuroimmunology. 2013;263:13–19. doi: 10.1016/j.jneuroim.2013.07.007. [DOI] [PubMed] [Google Scholar]
  47. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Research. 2000;28:27–30. doi: 10.1093/nar/28.1.27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Kasprzyk A. BioMart: driving a paradigm change in biological data management. Database. 2011;2011:bar049. doi: 10.1093/database/bar049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Kaufmann WE, Sprouse J, Rebowe N, Hanania T, Klamer D, Missling CU. ANAVEX2-73 (blarcamesine), a Sigma-1 receptor agonist, ameliorates neurologic impairments in a mouse model of rett syndrome. Pharmacology Biochemistry and Behavior. 2019;187:172796. doi: 10.1016/j.pbb.2019.172796. [DOI] [PubMed] [Google Scholar]
  50. Khwaja OS, Ho E, Barnes KV, O'Leary HM, Pereira LM, Finkelstein Y, Nelson CA, Vogel-Farley V, DeGregorio G, Holm IA, Khatwa U, Kapur K, Alexander ME, Finnegan DM, Cantwell NG, Walco AC, Rappaport L, Gregas M, Fichorova RN, Shannon MW, Sur M, Kaufmann WE. Safety, pharmacokinetics, and preliminary assessment of efficacy of mecasermin (recombinant human IGF-1) for the treatment of rett syndrome. PNAS. 2014;111:4596–4601. doi: 10.1073/pnas.1311141111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Kim Y, Subramaniam S. Locally defined protein phylogenetic profiles reveal previously missed protein interactions and functional relationships, proteins struct. Funct. Genet. 2006;62:1115–1124. doi: 10.1002/prot.20830. [DOI] [PubMed] [Google Scholar]
  52. Kishi N, MacDonald JL, Ye J, Molyneaux BJ, Azim E, Macklis JD. Reduction of aberrant NF-κB signalling ameliorates rett syndrome phenotypes in Mecp2-null mice. Nature Communications. 2016;7:1–13. doi: 10.1038/ncomms10520. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Kline DD, Ogier M, Kunze DL, Katz DM. Exogenous brain-derived neurotrophic factor rescues synaptic dysfunction in Mecp2-null mice. Journal of Neuroscience. 2010;30:5303–5310. doi: 10.1523/JNEUROSCI.5503-09.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Krishnaraj R, Ho G, Christodoulou J. RettBASE: rett syndrome database update. Human Mutation. 2017;38:922–931. doi: 10.1002/humu.23263. [DOI] [PubMed] [Google Scholar]
  55. Li W, Calfa G, Larimore J, Pozzo-Miller L. Activity-dependent BDNF release and TRPC signaling is impaired in hippocampal neurons of Mecp2 mutant mice. PNAS. 2012;109:17087–17092. doi: 10.1073/pnas.1205271109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Liyanage VR, Rastegar M. Rett syndrome and MeCP2. NeuroMolecular Medicine. 2014;16:231–264. doi: 10.1007/s12017-014-8295-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Lomonaco SL, Finniss S, Xiang C, DeCarvalho A, Umansky F, Kalkanis SN, Mikkelsen T, Brodie C. The induction of autophagy by γ-radiation contributes to the radioresistance of glioma stem cells. International Journal of Cancer. 2009;125:717–722. doi: 10.1002/ijc.24402. [DOI] [PubMed] [Google Scholar]
  58. Lopes CT, Franz M, Kazi F, Donaldson SL, Morris Q, Bader GD, Dopazo J. Bioinformatics. Oxford University Press; 2011. [Google Scholar]
  59. Lyst MJ, Ekiert R, Ebert DH, Merusi C, Nowak J, Selfridge J, Guy J, Kastan NR, Robinson ND, de Lima Alves F, Rappsilber J, Greenberg ME, Bird A. Rett syndrome mutations abolish the interaction of MeCP2 with the NCoR/SMRT co-repressor. Nature Neuroscience. 2013;16:898–902. doi: 10.1038/nn.3434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Maezawa I, Swanberg S, Harvey D, LaSalle JM, Jin LW. Rett syndrome astrocytes are abnormal and spread MeCP2 deficiency through gap junctions. Journal of Neuroscience. 2009;29:5051–5061. doi: 10.1523/JNEUROSCI.0324-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Marcotte EM, Xenarios I, van Der Bliek AM, Eisenberg D. Localizing proteins in the cell from their phylogenetic profiles. PNAS. 2000;97:12115–12120. doi: 10.1073/pnas.220399497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Merchant SS, Prochnik SE, Vallon O, Harris EH, Karpowicz SJ, Witman GB, Terry A, Salamov A, Fritz-Laylin LK, Maréchal-Drouard L, Marshall WF, Qu LH, Nelson DR, Sanderfoot AA, Spalding MH, Kapitonov VV, Ren Q, Ferris P, Lindquist E, Shapiro H, Lucas SM, Grimwood J, Schmutz J, Cardol P, Cerutti H, Chanfreau G, Chen CL, Cognat V, Croft MT, Dent R, Dutcher S, Fernández E, Fukuzawa H, González-Ballester D, González-Halphen D, Hallmann A, Hanikenne M, Hippler M, Inwood W, Jabbari K, Kalanon M, Kuras R, Lefebvre PA, Lemaire SD, Lobanov AV, Lohr M, Manuell A, Meier I, Mets L, Mittag M, Mittelmeier T, Moroney JV, Moseley J, Napoli C, Nedelcu AM, Niyogi K, Novoselov SV, Paulsen IT, Pazour G, Purton S, Ral JP, Riaño-Pachón DM, Riekhof W, Rymarquis L, Schroda M, Stern D, Umen J, Willows R, Wilson N, Zimmer SL, Allmer J, Balk J, Bisova K, Chen CJ, Elias M, Gendler K, Hauser C, Lamb MR, Ledford H, Long JC, Minagawa J, Page MD, Pan J, Pootakham W, Roje S, Rose A, Stahlberg E, Terauchi AM, Yang P, Ball S, Bowler C, Dieckmann CL, Gladyshev VN, Green P, Jorgensen R, Mayfield S, Mueller-Roeber B, Rajamani S, Sayre RT, Brokstein P, Dubchak I, Goodstein D, Hornick L, Huang YW, Jhaveri J, Luo Y, Martínez D, Ngau WC, Otillar B, Poliakov A, Porter A, Szajkowski L, Werner G, Zhou K, Grigoriev IV, Rokhsar DS, Grossman AR. The Chlamydomonas genome reveals the evolution of key animal and plant functions. Science. 2007;318:245–250. doi: 10.1126/science.1143609. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Merritt JK, Collins BE, Erickson KR, Dong H, Neul JL. Pharmacological read-through of R294X Mecp2 in a novel mouse model of rett syndrome. Human Molecular Genetics. 2020;29:2461–2470. doi: 10.1093/hmg/ddaa102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Mok RS, Zhang W, Sheikh T, Fernandes I, DeJong L, Hildebrandt M, Mufteev M, Rodrigues D, Wei W, Piekna A, Liu J, Muotri A, Vincent J, Salter M, Ellis J. Wide phenotypic spectrum of human stem cell-derived excitatory neurons with rett syndrome-associated MECP2 mutations. bioRxiv. 2020 doi: 10.1101/2020.07.12.189621. [DOI] [PMC free article] [PubMed]
  65. Montes Diaz G, Hupperts R, Fraussen J, Somers V. Dimethyl fumarate treatment in multiple sclerosis: recent advances in clinical and immunological studies. Autoimmunity Reviews. 2018;17:1240–1250. doi: 10.1016/j.autrev.2018.07.001. [DOI] [PubMed] [Google Scholar]
  66. Nair B, Wong-Riley MT. Transcriptional regulation of Brain-derived neurotrophic factor coding exon IX: role of nuclear respiratory factor 2. The Journal of Biological Chemistry. 2016;291:22583–22593. doi: 10.1074/jbc.M116.742304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Nan X, Ng HH, Johnson CA, Laherty CD, Turner BM, Eisenman RN, Bird A. Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature. 1998;393:386–389. doi: 10.1038/30764. [DOI] [PubMed] [Google Scholar]
  68. Nelson MR, Tipney H, Painter JL, Shen J, Nicoletti P, Shen Y, Floratos A, Sham PC, Li MJ, Wang J, Cardon LR, Whittaker JC, Sanseau P. The support of human genetic evidence for approved drug indications. Nature Genetics. 2015;47:856–860. doi: 10.1038/ng.3314. [DOI] [PubMed] [Google Scholar]
  69. Neul JL, Fang P, Barrish J, Lane J, Caeg EB, Smith EO, Zoghbi H, Percy A, Glaze DG. Specific mutations in methyl-CpG-binding protein 2 confer different severity in rett syndrome. Neurology. 2008;70:1313–1321. doi: 10.1212/01.wnl.0000291011.54508.aa. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Nguyen NTT, Vincens P, Roest Crollius H, Louis A. Genomicus 2018: karyotype evolutionary trees and on-the-fly synteny computing. Nucleic Acids Research. 2018;46:D816–D822. doi: 10.1093/nar/gkx1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. O'Driscoll CM, Lima MP, Kaufmann WE, Bressler JP. Methyl CpG binding protein 2 deficiency enhances expression of inflammatory cytokines by sustaining NF-κB signaling in myeloid derived cells. Journal of Neuroimmunology. 2015;283:23–29. doi: 10.1016/j.jneuroim.2015.04.005. [DOI] [PubMed] [Google Scholar]
  72. Okabe Y, Takahashi T, Mitsumasu C, Kosai K, Tanaka E, Matsuishi T. Alterations of gene expression and glutamate clearance in astrocytes derived from an MeCP2-null mouse model of rett syndrome. PLOS ONE. 2012;7:e35354. doi: 10.1371/journal.pone.0035354. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Oxford Academic Open targets platform: new developments and updates two years on | nucleic acids research |. 2019. [January 8, 2019]. https://academic.oup.com/nar/article/47/D1/D1056/5193331 [DOI] [PMC free article] [PubMed]
  74. Pagliarini DJ, Calvo SE, Chang B, Sheth SA, Vafai SB, Ong SE, Walford GA, Sugiana C, Boneh A, Chen WK, Hill DE, Vidal M, Evans JG, Thorburn DR, Carr SA, Mootha VK. A mitochondrial protein compendium elucidates complex I disease biology. Cell. 2008;134:112–123. doi: 10.1016/j.cell.2008.06.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Park MJ, Aja S, Li Q, Degano AL, Penati J, Zhuo J, Roe CR, Ronnett GV. Anaplerotic triheptanoin diet enhances mitochondrial substrate use to remodel the metabolome and improve lifespan, motor function, and sociability in MeCP2-Null mice. PLOS ONE. 2014;9:e109527. doi: 10.1371/journal.pone.0109527. [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. Patel NS, Nandra KK, Brines M, Collino M, Wong WF, Kapoor A, Benetti E, Goh FY, Fantozzi R, Cerami A, Thiemermann C. A nonerythropoietic peptide that mimics the 3D structure of erythropoietin reduces organ injury/dysfunction and inflammation in experimental hemorrhagic shock. Molecular Medicine. 2011;17:883–892. doi: 10.2119/molmed.2011.00053. [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Pellegrini M, Marcotte EM, Thompson MJ, Eisenberg D, Yeates TO. Assigning protein functions by comparative genome analysis: protein phylogenetic profiles. PNAS. 1999;96:4285–4288. doi: 10.1073/pnas.96.8.4285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Picard N, Fagiolini M. MeCP2: an epigenetic regulator of critical periods. Current Opinion in Neurobiology. 2019;59:95–101. doi: 10.1016/j.conb.2019.04.004. [DOI] [PubMed] [Google Scholar]
  79. Przanowski P, Wasko U, Zheng Z, Yu J, Sherman R, Zhu LJ, McConnell MJ, Tushir-Singh J, Green MR, Bhatnagar S. Pharmacological reactivation of inactive X-linked Mecp2 in cerebral cortical neurons of living mice. PNAS. 2018;115:7991–7996. doi: 10.1073/pnas.1803792115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Pushpakom S, Iorio F, Eyers PA, Escott KJ, Hopper S, Wells A, Doig A, Guilliams T, Latimer J, McNamee C, Norris A, Sanseau P, Cavalla D, Pirmohamed M. Drug repurposing: progress, challenges and recommendations. Nature Reviews Drug Discovery. 2019;18:41–58. doi: 10.1038/nrd.2018.168. [DOI] [PubMed] [Google Scholar]
  81. Ragione FD, Vacca M, Fioriniello S, Pepe G. MECP2, a multi-talented modulator of chromatin architecture | briefings in functional genomics | oxford academic, br. Funct Genomics. 2016;15:420–431. doi: 10.1093/bfgp/elw023. [DOI] [PubMed] [Google Scholar]
  82. Ramocki MB, Peters SU, Tavyev YJ, Zhang F, Carvalho CM, Schaaf CP, Richman R, Fang P, Glaze DG, Lupski JR, Zoghbi HY. Autism and other neuropsychiatric symptoms are prevalent in individuals with MeCP2 duplication syndrome. Annals of Neurology. 2009;66:771–782. doi: 10.1002/ana.21715. [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Rao SS, Huntley MH, Durand NC, Stamenova EK, Bochkov ID, Robinson JT, Sanborn AL, Machol I, Omer AD, Lander ES, Aiden EL. A 3D map of the human genome at Kilobase resolution reveals principles of chromatin looping. Cell. 2014;159:1665–1680. doi: 10.1016/j.cell.2014.11.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  84. Rey F, Balsari A, Giallongo T, Ottolenghi S, Di Giulio AM, Samaja M, Carelli S. Erythropoietin as a neuroprotective molecule: an overview of its therapeutic potential in neurodegenerative diseases. ASN Neuro. 2019;11:175909141987142. doi: 10.1177/1759091419871420. [DOI] [PMC free article] [PubMed] [Google Scholar]
  85. Ruepp A, Brauner B, Dunger-Kaltenbach I, Frishman G, Montrone C, Stransky M, Waegele B, Schmidt T, Doudieu ON, Stümpflen V, Mewes HW. CORUM: the comprehensive resource of mammalian protein complexes. Nucleic Acids Research. 2008;36:D646–D650. doi: 10.1093/nar/gkm936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Schwartz S, Agarwala SD, Mumbach MR, Jovanovic M, Mertins P, Shishkin A, Tabach Y, Mikkelsen TS, Satija R, Ruvkun G, Carr SA, Lander ES, Fink GR, Regev A. High-resolution mapping reveals a conserved, widespread, dynamic mRNA methylation program in yeast meiosis. Cell. 2013;155:1409–1421. doi: 10.1016/j.cell.2013.10.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  87. Sherill-Rofe D, Rahat D, Findlay S, Mellul A, Guberman I, Braun M, Bloch I, Lalezari A, Samiei A, Sadreyev R, Goldberg M, Orthwein A, Zick A, Tabach Y. Mapping global and local coevolution across 600 species to identify novel homologous recombination repair genes. Genome Research. 2019;29:439–448. doi: 10.1101/gr.241414.118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  88. Shovlin S, Tropea D. Transcriptome level analysis in rett syndrome using human samples from different tissuesOrphanet J. Rare Diseases. 2018;13:113. doi: 10.1186/s13023-018-0857-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Shukla S, Tekwani BL. Histone deacetylases inhibitors in neurodegenerative diseases, neuroprotection and neuronal differentiation. Frontiers in Pharmacology. 2020;11:537. doi: 10.3389/fphar.2020.00537. [DOI] [PMC free article] [PubMed] [Google Scholar]
  90. Singer JW, Fleischman A, Al-Fayoumi S, Mascarenhas JO, Yu Q, Agarwal A. Inhibition of interleukin-1 receptor-associated kinase 1 (IRAK1) as a therapeutic strategy. Oncotarget. 2018;9:33416–33439. doi: 10.18632/oncotarget.26058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  91. Sun J, Xu J, Liu Z, Liu Q, Zhao A, Shi T, Li Y. Refined phylogenetic profiles method for predicting protein-protein interactions. Bioinformatics. 2005;21:3409–3415. doi: 10.1093/bioinformatics/bti532. [DOI] [PubMed] [Google Scholar]
  92. Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, Simonovic M, Doncheva NT, Morris JH, Bork P, Jensen LJ, Mering CV. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Research. 2019;47:D607–D613. doi: 10.1093/nar/gky1131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. Tabach Y, Golan T, Hernández-Hernández A, Messer AR, Fukuda T, Kouznetsova A, Liu JG, Lilienthal I, Levy C, Ruvkun G. Human disease locus discovery and mapping to molecular pathways through phylogenetic profiling. Molecular Systems Biology. 2013a;9:692. doi: 10.1038/msb.2013.50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  94. Tabach Y, Billi AC, Hayes GD, Newman MA, Zuk O, Gabel H, Kamath R, Yacoby K, Chapman B, Garcia SM, Borowsky M, Kim JK, Ruvkun G. Identification of small RNA pathway genes using patterns of phylogenetic conservation and divergence. Nature. 2013b;493:694–698. doi: 10.1038/nature11779. [DOI] [PMC free article] [PubMed] [Google Scholar]
  95. Timmerman R, Burm SM, Bajramovic JJ. An overview of in vitro methods to study microglia. Frontiers in Cellular Neuroscience. 2018;12:242. doi: 10.3389/fncel.2018.00242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  96. Torkildsen Ø, Myhr KM, Bø L. Disease‐modifying treatments for multiple sclerosis – a review of approved medications. European Journal of Neurology. 2016;23:18–27. doi: 10.1111/ene.12883. [DOI] [PMC free article] [PubMed] [Google Scholar]
  97. Tsaban T, Stupp D, Sherill-Rofe D, Bloch I, Sharon E, Schueler-Furman O, Wiener R, Tabach Y. CladeOScope: functional interactions through the prism of clade-wise co-evolution. NAR Genomics and Bioinformatics. 2021;3:lqab024. doi: 10.1093/nargab/lqab024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  98. Tsilioni I, Pantazopoulos H, Conti P, Leeman SE, Theoharides TC. IL-38 inhibits microglial inflammatory mediators and is decreased in amygdala of children with autism spectrum disorder. PNAS. 2020;117:16475–16480. doi: 10.1073/pnas.2004666117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  99. Unterman I, Berman BP, Tabach Y. MECP2_phylogeny. swh:1:rev:14063b44a40688a8024a06347b63cfdac74b96adSoftware Heritage. 2021 https://archive.softwareheritage.org/swh:1:dir:e2c60f2c5038244f7192b1070c80c10bad4daeb0;origin=https://github.com/GuIrene/MECP2_phylogeny;visit=swh:1:snp:0c28d2131db499777078eb129237064f9b0c9be0;anchor=swh:1:rev:14063b44a40688a8024a06347b63cfdac74b96ad
  100. Vashi N, Justice MJ. Treating rett syndrome: from mouse models to human therapies. Mammalian Genome. 2019;30:90–110. doi: 10.1007/s00335-019-09793-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  101. Wang Y, Song F, Zhang B, Zhang L, Xu J, Kuang D, Li D, Choudhary MNK, Li Y, Hu M, Hardison R, Wang T, Yue F. The 3D genome browser: a web-based browser for visualizing 3D genome organization and long-range chromatin interactions. Genome Biology. 2018;19:151. doi: 10.1186/s13059-018-1519-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  102. Wei L, Chen C, Ding L, Mo M, Zou J, Lu Z, Li H, Wu H, Dai Y, Xu P, Lu Z. Wnt1 promotes EAAT2 expression and mediates the protective effects of astrocytes on dopaminergic cells in Parkinson’s Disease. Neural Plasticity. 2019;2019:1–12. doi: 10.1155/2019/1247276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  103. Xu X, Kozikowski AP, Pozzo-Miller L. A selective histone deacetylase-6 inhibitor improves BDNF trafficking in hippocampal neurons from Mecp2 knockout mice: implications for rett syndrome. Frontiers in Cellular Neuroscience. 2014;8:68. doi: 10.3389/fncel.2014.00068. [DOI] [PMC free article] [PubMed] [Google Scholar]
  104. Xu X, Stoyanova EI, Lemiesz AE, Xing J, Mash DC, Heintz N. Species and cell-type properties of classically defined human and rodent neurons and Glia. eLife. 2018;7:e37551. doi: 10.7554/eLife.37551. [DOI] [PMC free article] [PubMed] [Google Scholar]
  105. Yasui DH, Peddada S, Bieda MC, Vallero RO, Hogart A, Nagarajan RP, Thatcher KN, Farnham PJ, Lasalle JM. Integrated epigenomic analyses of neuronal MeCP2 reveal a role for long-range interaction with active genes. PNAS. 2007;104:19416–19421. doi: 10.1073/pnas.0707442104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  106. Zeng A, Wei Z, Rabinovsky R, Jun HJ, El Fatimy R, Deforzh E, Arora R, Yao Y, Yao S, Yan W, Uhlmann EJ, Charest A, You Y, Krichevsky AM. Glioblastoma-Derived extracellular vesicles facilitate transformation of astrocytes via reprogramming oncogenic metabolism. iScience. 2020;23:101420. doi: 10.1016/j.isci.2020.101420. [DOI] [PMC free article] [PubMed] [Google Scholar]
  107. Zhao D, Mokhtari R, Pedrosa E, Birnbaum R, Zheng D, Lachman HM. Transcriptome analysis of microglia in a mouse model of rett syndrome: differential expression of genes associated with microglia/macrophage activation and cellular stress. Molecular Autism. 2017;8:17. doi: 10.1186/s13229-017-0134-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision letter

Editor: George H Perry1

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Acceptance summary:

This manuscript will interest scientists who are aiming to leverage concepts and tools of evolutionary biology to identify novel gene targets, in this case towards much-needed therapeutic interventions for Rett syndrome. The follow up experiments presented in the paper are detailed, well thought out, and help establish the potential of the identified drugs in alleviating molecular signatures in in vitro disease models.

Decision letter after peer review:

Thank you for submitting your article "Expanding the MECP2 network using comparative genomics reveals potential therapeutic targets for Rett syndrome" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by George Perry as the Senior and Reviewing Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The manuscript has potential but requires major revisions for further consideration by eLife. In consultation, the editor and reviewers considered the level of revision to be possible, but we emphasize that we have high expectations, expecting at least the major comments made by each reviewer (below) to be robustly addressed. These essential revisions include improving the comprehensiveness of the comparative genomics pipeline and the need for more explicit validation experiments.

Reviewer #1 (Recommendations for the authors):

On line 367 – The differential mechanisms of restoration are interesting in the face of MeCP2s broad genomic role. While this is more of a future direction, could these treatments have a combinatorial beneficial effect? The in-vivo experiments that the authors are conducting should be very interesting. I wonder if all 33 identified genes could be screened for their toxicity/efficacy in these KD cells. The chosen 3 candidate proteins are all from the same evolutionarily conserved chromosomal cluster yet have meaningfully different cell specific mechanisms of RTT amelioration. Perhaps the other 33 identified genes with actionable druggable targets might have even more diverse mechanisms.

On line 51 there is a small typo, intro -> into.

Reviewer #2 (Recommendations for the authors):

1. Addressing/answering the following comments would help provide supporting evidence for the comparative evolutionary pipeline used by the authors:

a. Can the authors elaborate on their NPP methodology and how did they avoid the false positives?

b. Is the functional nature of genes different between E200 and M200?

c. Is the overlap of 10 genes between E200 and M200 statistically significant, given that overlapping input data was used? Is it higher or lower than expected?

d. Among the 390 genes, how many well established functional and biophysical partners of MECP2 were identified? What positive controls did the authors use to validate the robustness of their comparative genomic pipeline?

e. HDACs appear to be highly connected functional and biophysical partners of MECP2. Yet the authors do not seem to have identified any of them. Can the authors speculate on the reasons behind that?

f. How many novel MECP2 partners (without any previously known links) do the authors identify?

g. Some of the partners identified by the authors have functional association in Stringdb. Can the authors perform an enrichment analysis, with randomized controls, to demonstrate whether this functional enrichment is significantly more than expected? It will be an important validation for the authors PP approach.

2. The authors detect that 33 out of the 390 genes are druggable targets. Is that higher/lower than expected?

3. The authors show a series of sophisticated experiments to demonstrate the impact of the three drugs. The following clarifications will help strengthen the importance of the evolutionary pipeline to identify these drugs:

a. Are any of the previously known MECP2 functional partners targeted by EPO, DMF and Pacritinib in DGIdb and Open Targets? For example, a quick literature survey shows that all three molecules have direct or indirect effects on HDAC activity, which are functional and biophysical interaction partners of MECP2.

b. If the authors were to use known functional partners of MECP2 from Stringdb, PDB, or biogrid and check for their druggability in DGIdb and Open Targets, will the authors identify EPO, DMF and Pacritinib? If not, it proves that the comparative genomic pipeline was critical for identification of these drugs. If yes, it shows that the authors didn't need the pipeline. Or perhaps the authors mean to demonstrate that EPOR, KEAP1 and IRAK work upstream of HDACs and are therefore contributing to a better understanding of the mechanism of action of MECP2. If so, the authors do not have enough text, analysis or experiments to demonstrate that.

c. Can the authors perform silencing or knockdown experiment to demonstrate that the effects of the three drugs are mediated through EPOR, KEAP1 and IRAK and not through other targets?

eLife. 2021 Aug 6;10:e67085. doi: 10.7554/eLife.67085.sa2

Author response

Reviewer #1 (Recommendations for the authors):

On line 367 – The differential mechanisms of restoration are interesting in the face of MeCP2s broad genomic role. While this is more of a future direction, could these treatments have a combinatorial beneficial effect? The in-vivo experiments that the authors are conducting should be very interesting. I wonder if all 33 identified genes could be screened for their toxicity/efficacy in these KD cells. The chosen 3 candidate proteins are all from the same evolutionarily conserved chromosomal cluster yet have meaningfully different cell specific mechanisms of RTT amelioration. Perhaps the other 33 identified genes with actionable druggable targets might have even more diverse mechanisms.

We tested the combinatorial effect of EPO with DMF on BDNF secretion, since these two both had an individual effect on this phenotype and Pacritinib did not (new Figure 6C). There may be an additive effect to EPO+DMF, compared to EPO or DMF alone, though we did not see a substantial synergistic effect. While the points you make are largely beyond the scope of this initial study, we did incorporate some of them in the improvements to our Discussion section discussing future directions (lines 420-423, 483).

We note that only 2 of the 3 candidate proteins are from the same chromosomal cluster (EPOR and KEAP1), while IRAK1 is from a conserved cluster with MECP2. We have modified the text on lines 202-204 to make this clearer.

On line 51 there is a small typo, intro → into.

Fixed.

Reviewer #2 (Recommendations for the authors):

1. Addressing/answering the following comments would help provide supporting evidence for the comparative evolutionary pipeline used by the authors:

a. Can the authors elaborate on their NPP methodology and how did they avoid the false positives?

We described our two new methodology and benchmarking studies in the new Introduction on lines 104-109. Bloch et al. 2020 explicitly deals with the specificity issue using control datasets from the CORUM and KEGG databases. We have found that the most important factor in reducing false positives is our incorporation of within-clade normalization (discussed in Bloch 2020). In addition to our standard approach, we tried to additionally limit false positives by selecting candidates that were top-ranked in both the mammalian and eukaryotic lists, and that clustered within an evolutionarily conserved and topologically organized chromosomal domain.

b. Is the functional nature of genes different between E200 and M200?

We have done a new analysis of Gene Ontology and other functional annotations using STRING (now added as supplementary file 2). The E200 list is enriched in a number of different functional categories, with some of the most significant related to innate immunity and the immune response. These included MHC Class II receptor activity and peptidoglycan receptor activity. Consistent with this, autoimmunity annotations such as Asthma, Allograft rejection, and Type I diabetes are highly enriched in the KEGG network. In contrast, the M200 list has almost no functional enrichment, except for the general “Disease” category within Uniprot annotated keywords. This indicates that the two lists might be revealing fundamentally different relationships to MECP2, and makes the 10 overlapping genes especially interesting. We have added a description of this to the Results section in lines 152-159.

c. Is the overlap of 10 genes between E200 and M200 statistically significant, given that overlapping input data was used? Is it higher or lower than expected?

It is statistically significant if you consider the two lists independent (p<3.2E-4 by the hypergeometric test), but as you point out the phylogenetic correlation in Eukaryotes can be affected by the correlation within the mammalian clade. We do not know a way to factor this out from the significance estimate, so we have just edited the description in the text to reflect this fact (lines 160-165).

d. Among the 390 genes, how many well established functional and biophysical partners of MECP2 were identified? What positive controls did the authors use to validate the robustness of their comparative genomic pipeline?

In the original manuscript, we relied on the fact that the 10 proteins in the E200/M200 intersection list included the well-established MECP2-interacting protein IRAK1. Given that evidence, and the fact that we have used this pipeline before and validated it extensively (Bloch et al., 2020; Tsaban et al., 2021), this gave us enough confidence to proceed to functional drug follow-up. However, based on your feedback we have now performed new comparisons to STRING to give a more comprehensive analysis for MECP2. See our response to question 1g, below.

e. HDACs appear to be highly connected functional and biophysical partners of MECP2. Yet the authors do not seem to have identified any of them. Can the authors speculate on the reasons behind that?

First, we investigated only the very most strongly linked of our PP lists, which would necessarily leave out some true interactions. However, we have gone back to our full original lists and found that the relevant HDACs have quite low PP similarity scores to MECP2 in general. We have plotted the phylogenetic profiles for MECP and the two major classes of HDACs in Author response image 1, showing significant differences. MECP2 is distinguished by a relative loss among a large group of vertebrates, and stronger conservation among mammals. Class I HDACS 1/2/3/8 clustered together and were much better conserved in ancient branches. Class II HDACS 4/5/6/7/9/10 clustered together and were characterized by loss within a sub-branch of fungi and all non-fungi single-cell eukaryotes (“other organisms”). While it is hard to interpret these differences at more ancient time points, the split within the non-mammalian vertebrates may be important in understanding the divergence between MECP2 and HDACs. While we did not add this more in depth analysis of HDAC evolutionary conservation, we have added new text describing the HDAC role in MECP2 function and RTT in our Introduction on lines 72-80 and their potential role as a mediator of our drug candidates in the Discussion on lines 462-479.

Author response image 1. Normalized phylogenetic profiles of MECP2 and HDAC proteins.

Author response image 1.

Open in a new tab

f. How many novel MECP2 partners (without any previously known links) do the authors identify?

As shown in Figure 1—figure supplement 1D), there are 1,398 known interactions with MECP2 in the String-db categories “text-mining”, “co-expression”, and “experimental”. 366 of 390 genes in out E200+M200 list were not among these genes, indicating that large majority of our genes are novel relationships. We have added text describing this on lines 168-171.

g. Some of the partners identified by the authors have functional association in Stringdb. Can the authors perform an enrichment analysis, with randomized controls, to demonstrate whether this functional enrichment is significantly more than expected? It will be an important validation for the authors PP approach.

As described in response to your question 1f, only 24 of 390 genes in our E200+M200 gene lists (6.2%) had Stringdb associations with MECP2. Given that 1,398 (7.3%) of 19,257 total genes in Stringdb are associated with MECP2, this is compatible with random chance. However, in the 10 high priority genes included in both the M200 and E200 groups, 3 overlapped Stringdb (Author response image 2). The chance of observing 3 or more overlaps is 0.03 by the hypergeometric test. While this is not a high degree of overlap, it does help to establish the validity of our approach, along with our previous benchmarking of our pipeline using the CORUM, REACTOME, and KEGG databases (Bloch et al., 2020; Tsaban et al., 2021). These recent benchmarking studies are now described in lines 104-109. We also analyzed the 390 gene list using GeneAnalytics, and they were most enriched with genes expressed in the brain compared to all other tissues (see Figure 1—figure supplement 1C ).

Author response image 2. Overlap between the E200 and M200 intersection set (PP) and all STRING MECP2 interactions.

Author response image 2.

Open in a new tab

2. The authors detect that 33 out of the 390 genes are druggable targets. Is that higher/lower than expected?

We performed random sampling of 390 genes from our PP input gene list. This resulted in an average of 39.9 druggable genes (std 6.0). Thus our gene lists was not significantly either enriched or depleted in druggable genes. This does not seem surprising to us, given that druggability depends on priorities of past research efforts and chemical properties of the gene product, rather than anything having to do with evolutionary links. We have added a sentence describing this to the Results section (line 191). If we missed your point here, please let us know.

3. The authors show a series of sophisticated experiments to demonstrate the impact of the three drugs. The following clarifications will help strengthen the importance of the evolutionary pipeline to identify these drugs:

a. Are any of the previously known MECP2 functional partners targeted by EPO, DMF and Pacritinib in DGIdb and Open Targets? For example, a quick literature survey shows that all three molecules have direct or indirect effects on HDAC activity, which are functional and biophysical interaction partners of MECP2.

We addressed HDACs as well as other known associations with MECP2 , by performing an analysis of the top linked MECP2 genes from the STRING database. Neither HDACs nor other highly-ranked MECP2 associated genes were linked to either EPO or DMF. We used STRING as our baseline method for finding known interactions, since it includes text-mining, co-expression, and experimental associations (which includes links between MECP2 and many HDACs). Additionally, we searched for direct HDAC interactions for all class 1 and 2 HDACs in DGIDB, GeneCards and OpenTargets, and found no links to the 3 drugs we identified here. Below is a list of 46 drugs with direct HDAC association we found through this process.

Gene Drug Interaction Gene Drug Interaction
HDAC1 TACEDINALINE inhibitor HDAC
1		 VORINOSTA
T		 inhibitor
HDAC1 PANOBINOSTAT LACTATE inhibitor HDAC4 TASQUINIMOD allosteric modulator
HDAC1 ROMIDEPSIN antagonist|inhibitor HDAC6 BUFEXAMAC inhibitor
HDAC1 FIMEPINOSTAT inhibitor HDAC6 TUBACIN inhibitor
HDAC1 CUDC-101 inhibitor HDAC1 Valproic acid inhibitor
HDAC1 SCRIPTAID inhibitor HDAC1 Fingolimod inhibitor
HDAC1 RESMINOSTAT inhibitor HDAC1 4SC-202 inhibitor
HDAC1 NANATINOSTAT inhibitor HDAC1 sodium phenylbutyrate inhibitor
HDAC1 APICIDIN inhibitor HDAC1 CHR-3996 inhibitor
HDAC1 ABEXINOSTAT inhibitor HDAC1 MGCD-0103 inhibitor
HDAC1 BELINOSTAT inhibitor HDAC1 PCI-24781 inhibitor
HDAC1 DEPAKOTE inhibitor HDAC1 Pivanex inhibitor
HDAC1 MOCETINOSTAT inhibitor HDAC1 SB939 inhibitor
HDAC1 DACINOSTAT inhibitor HDAC2 Theophylline activator
HDAC1 PANOBINOSTAT inhibitor HDAC2 Aminophylline activator
HDAC1 AN-9 inhibitor HDAC2 Atorvastatin inhibitor
HDAC1 TUCIDINOSTAT inhibitor HDAC2 Fluvastatin inhibitor
HDAC1 RICOLINOSTAT inhibitor HDAC2 Oxtriphylline activator
HDAC1 ENTINOSTAT inhibitor HDAC2 Pravastatin inhibitor
HDAC1 GIVINOSTAT inhibitor HDAC2 Simvastatin inhibitor
HDAC1 CITARINOSTAT inhibitor HDAC3 Romidepsin inhibitor
HDAC1 QUISINOSTAT inhibitor HDAC3 RG2833 inhibitor
HDAC1 PRACINOSTAT inhibitor HDAC7 TACEDINALINE inhibitor
Open in a new tab

b. If the authors were to use known functional partners of MECP2 from Stringdb, PDB, or biogrid and check for their druggability in DGIdb and Open Targets, will the authors identify EPO, DMF and Pacritinib? If not, it proves that the comparative genomic pipeline was critical for identification of these drugs. If yes, it shows that the authors didn't need the pipeline. Or perhaps the authors mean to demonstrate that EPOR, KEAP1 and IRAK work upstream of HDACs and are therefore contributing to a better understanding of the mechanism of action of MECP2. If so, the authors do not have enough text, analysis or experiments to demonstrate that.

For STRING, please see our response to question 3a above. Biogrid only contains only 182 proteins linked to MECP2, and none are linked to the 3 drugs tested here through these databases. This reinforces what we have shown previously, that PP is complementary to other functional relationship databases. We were not sure what you had in mind for PDB. Only one protein (TBLR1) has a PDB structure in complex with MECP2, and this gene is not linked to any of our 3 drugs. There are three structures with DMF in complex with a human gene (SMN1, UHRF1, TDRD3), two with Pacritinib (NQO2, SMYD3), and none with EPO. Only UHRF1 has any connection to MECP2 through STRING.

c. Can the authors perform silencing or knockdown experiment to demonstrate that the effects of the three drugs are mediated through EPOR, KEAP1 and IRAK and not through other targets?

This is a completely valid point and definitely an important area for future work, as evidenced by the point you brought up about HDACs. We have now added text to our Discussion regarding this on lines 478-483. Given our resources, these experiments would be beyond the scope of what we could accomplish in a short time frame. We believe that our results here, while not 100% conclusive, are highly compelling given the fact that all three drugs predicted by our pipeline showed reversal of multiple Rett-like phenotypes of MECP2 depletion. It seems highly unlikely that by chance we picked three drugs that could reverse these changes in the MECP2 depleted condition but have no effect on the wildtype cells. We believe it is important to publish our initial findings based on this methodological workflow, as well as our gene and drug lists, so that other labs may follow up on these results in parallel to us.

References:

Bateman, A. (2019). UniProt: A worldwide hub of protein knowledge. Nucleic Acids Research, 47(D1), D506–D515. https://doi.org/10.1093/nar/gky1049

Bloch, I., Sherill-Rofe, D., Stupp, D., Unterman, I., Beer, H., Sharon, E., and Tabach, Y. (2020). Optimization of co-evolution analysis through phylogenetic profiling reveals pathway-specific signals. Bioinformatics, 36(14), 4116–4125. https://doi.org/10.1093/bioinformatics/btaa281

Dunwell, T. L., Paps, J., and Holland, P. W. H. (2017). Novel and divergent genes in the evolution of placental mammals. Proceedings of the Royal Society B: Biological Sciences, 284(1864). https://doi.org/10.1098/rspb.2017.1357

Fuchs, S. B. A., Lieder, I., Stelzer, G., Mazor, Y., Buzhor, E., Kaplan, S., … Shtrichman, R. (2016). GeneAnalytics: An Integrative Gene Set Analysis Tool for Next Generation Sequencing, RNAseq and Microarray Data. OMICS A Journal of Integrative Biology, 20(3), 139–151. https://doi.org/10.1089/omi.2015.0168

Philippon, H., Souvane, A., Brochier-Armanet, C., and Perrière, G. (2017). IsoSel: Protein isoform selector for phylogenetic reconstructions. PLoS ONE, 12(3).

https://doi.org/10.1371/journal.pone.0174250

Tsaban, T., Stupp, D., Sherill-Rofe, D., Bloch, I., Sharon, E., Schueler-Furman, O., … Tabach, Y. (2021).

CladeOScope: functional interactions through the prism of clade-wise co-evolution. NAR Genomics and Bioinformatics, 3(2). https://doi.org/10.1093/nargab/lqab024

Villanueva-Cañas, J. L., Laurie, S., and Albà, M. M. (2013). Improving genome-wide scans of positive selection by using protein isoforms of similar length. Genome Biology and Evolution, 5(2), 457– 467. https://doi.org/10.1093/gbe/evt017

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Stupp D, Tsaban T, Bloch I, Sharon E, Tabach Y. 2021. CladeOScope: elucidating functional interactions via a clade co-evolution prism. Zenodo. [DOI] [PMC free article] [PubMed]

    Supplementary Materials

    Supplementary file 1. Genes most co-evolved with MECP2.
    elife-67085-supp1.csv (6.2KB, csv)
    Supplementary file 2. STRING GO-term enrichment of the E200 set.
    elife-67085-supp2.txt (6.5KB, txt)
    Supplementary file 3. RT-PCR primer sequences.
    elife-67085-supp3.docx (13.3KB, docx)
    Transparent reporting form
    elife-67085-transrepform.bpb.docx (112.3KB, docx)

    Data Availability Statement

    The phylogenetic data required to reproduce the analysis are openly available at Zenodo [https://zenodo.org/record/4464120#.YEoFpHmxVPY]. Code used for the analysis is openly available through Github https://github.com/GuIrene/MECP2_phylogeny (copy archived at https://archive.softwareheritage.org/swh:1:rev:14063b44a40688a8024a06347b63cfdac74b96ad).

    The following previously published dataset was used:

    Stupp D, Tsaban T, Bloch I, Sharon E, Tabach Y. 2021. CladeOScope: elucidating functional interactions via a clade co-evolution prism. Zenodo.

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES