Quantitative proteomic analysis shows alterations in patient Rett syndrome iPSC cultures at early neuronal progenitor stages

Rett syndrome (RTT) is a progressive neurodevelopmental disease that is characterized by abnormalities in cognitive, social and motor skills. RTT is often caused by mutations in the X-linked gene encoding methyl-CpG binding protein 2 (MeCP2). The mechanism by which impaired MeCP2 induces the pathological abnormalities in the brain is not understood. Both patients and mouse models have shown abnormalities at molecular and cellular level before typical RTT-associated symptoms appear. This implies that underlying mechanisms are already affected during neurodevelopmental stages. To understand the molecular mechanisms involved in disease onset, we used an RTT patient induced pluripotent stem cell (iPSC)-based model with isogenic controls and performed time-series of proteomic analysis using in-depth high-resolution quantitative mass spectrometry during early stages of neuronal development. We provide mass spectrometry-based quantitative proteomic data, depth of about 7000 proteins, at neuronal progenitor developmental stages of RTT patient cells and isogenic controls. Our data gives evidence of proteomic alteration at early neurodevelopmental stages, suggesting alterations long before the phase that symptoms of RTT syndrome become apparent. We found changes in proteins involved in pathway associated with RTT phenotypes, including dendrite morphology and synaptogenesis. Differential expression increased from early to late neural stem cell phases, although proteins involved in immunity, metabolic processes and calcium signaling were affected throughout all stages analyzed.


Background
Rett syndrome (RTT) is a progressive neurodevelopmental disease that is characterized by abnormalities in cognitive, social and motor skills. RTT is often caused by mutations in the X-linked gene encoding methyl-CpG binding protein 2 (MeCP2). The mechanism by which impaired MeCP2 induces the pathological abnormalities in the brain is not understood. Both patients and mouse models have shown abnormalities at molecular and cellular level before typical RTT-associated symptoms appear. This implies that underlying mechanisms are already affected during neurodevelopmental stages.

Methods
To understand the molecular mechanisms involved in disease onset, we used an RTT patient induced pluripotent stem cell (iPSC)-based model with isogenic controls and performed time-series of proteomic analysis using in-depth high-resolution quantitative mass spectrometry during early stages of neuronal development.

Results
We provide mass spectrometry-based quantitative proteomic data, depth of about 7000 proteins, at neuronal progenitor developmental stages of RTT patient cells and isogenic controls. Our data gives evidence of proteomic alteration at early neurodevelopmental stages, suggesting alterations long before the phase that symptoms of RTT syndrome become apparent. We found changes in proteins involved in pathway associated with RTT phenotypes, including dendrite morphology and synaptogenesis. Differential expression increased from early to late neural stem cell phases, although proteins involved in immunity, metabolic processes and calcium signaling were affected throughout all stages analyzed.

Limitations
The limitation of our study is the number of biological replicates. As the aim of our study was to investigate a large number of proteins, only a limited amount of biological replicates were suitable for inclusions without reducing the number of target proteins. Therefore, larger sample sizes derived from RTT patients will be needed to validate results.

Conclusions
Our results provide a valuable resource of proteins to study potential targets for early treatment of RTT symptoms. We found consistent and time-point specific alterations during early neuronal differentiation in RTT cultures. Insight into altered protein levels can help development of new biomarkers and therapeutic approaches in RTT syndrome. Therefore, we hope that our results give awareness of the early pre-natal onset of RTT, providing new insights to explore early diagnosis and treatment.

Background
Rett syndrome (RTT) is a severe neurodevelopmental disorder that mainly affects females with a frequency of ~ 1:10,000 [1]. Clinical features of RTT start to present around 6-18 months of age, and include deceleration of head growth, abnormalities in cognitive, social and motor skill development and seizures [2,3]. Postmortem studies showed increased density of neurons in combination with reduced soma sizes in RTT patient compared to healthy control brains [4,5]. RTT neurons show a decrease in dendritic branching, and a reduced number of dendritic spines and synapses [6,7]. While studies suggest affected neurodevelopment starting at early stages, the molecular mechanisms underlying neuropathology in RTT is not understood.
In 90-95% of the RTT cases, the disease is caused by dominant loss-of-function mutations in the Xlinked gene encoding methyl-CpG binding protein 2 (MeCP2) [8]. Random X chromosome inactivation in females results in somatic mosaics with normal and mutant MECP2 [9]. Males carrying a MECP2 mutation are not viable or suffer from severe symptoms and die early in life [10]. MeCP2 is described as a nuclear protein modulating gene expression, via binding to methylated DNA and hundreds of target genes. These modulations take place through direct repression or activation of genes, or by means of DNA modulation and secondary gene regulation. Consequently, mutations in MECP2 lead to miss-regulation of hundreds of genes, including those influencing brain development and neuronal maturation [11][12][13][14]. So far research in RTT focused on genomic and transcriptomic studies [15][16][17] and less so on proteome changes [18,19], although as molecular effectors of cellular processes, these are better predictors of pathological states. Recent advances in mass spectrometry-based proteomics now facilitate the study of global protein expression and quantification [20]. Considering the broad and complex regulating functions of MeCP2, modulating multiple cellular processes, we need insight into the final molecular effectors reflected by perturbation at the protein level to understand pathological states.
Here we used an iPSC-based RTT model and performed proteome analysis on iPSC-derived neuronal stem cells (NES cells) [21]. Earlier studies proved that iPSCs from RTT patients reflect disease-specific characteristics, including changes in neuronal differentiation at early stages of development [22,23].
However, we lack knowledge on the precise molecular mechanisms at the onset of disease. To study early alterations in the proteome of RTT cells compared to isogenic controls (iCTR), we performed a high-resolution mass spectrometry-based quantitative proteomics at different time points during neuronal stem cell development (Fig. 1). We show that the difference between RTT and iCTR, in terms of the number of differentially expressed proteins, begins at early stages and increases at later progenitor stages. Interestingly, a large group of these proteins are involved in cellular processes, implicated in classical features of typical RTT phenotypes, such as dendrite formation and axonal growth. Proteins involved in immunity and metabolic processes are consistently changed between RTT and iCTR at all time points studied. Here we provide evidence of target proteins that could be explored as potential targets for early treatments to reduce progression of RTT symptoms.

Methods
Cell culture and isogenic controls RTT patient fibroblasts were derived from the Cell lines and DNA bank of Rett syndrome, X-linked mental retardation and other genetic diseases at the University Siena in Italy via the Network of Genetic Biobanks Telethon. We used fibroblast lines carrying MECP2 mutation showing a deletion in Exon 3 and 4 of the MECP2 gene (RTT Ex3-4), (RTT#2282C2). Fibroblasts were derived frozen, thawed and expanded in fibroblast medium (DMEM-F12, 20% FBS, 1%NEAA, 1%Pen/Strep, 50 µM β-Mercaptoethanol). To generate pure RTT, i.e. cells expressing affected X-chromosome, and isogenic control, i.e. cells expressing the healthy X-chromosome, fibroblasts were detached from cell culture plate and single fibroblasts were seeded in a 96-well plate. Cells were further expanded and characterized for their MeCP2 state by immunocytochemistry and PCR [24]. All of our experiments were exempt from the approval of the institutional review board.

Differentiation of neuronal stem cells
The 6 iPSC lines were differentiated towards neuronal stem cells as described before [21,25]. As  Next day cells were washed and secondary antibody Alexa Fluor® 488 (ThermoFisher, 1:1000, mouse or rabbit) and Alexa Fluor® 594 (ThermoFisher, 1:1000, mouse or rabbit) were applied in blocking buffer for 1 h at room temperature. To identify cell nuclei DAPI was used for 5 min before cells were mounted with Fluoromount™ (Sigma-Aldrich).

RNA collection, Sequencing and PCR analysis
To isolate RNA samples, standard TRIzol®-Chloroform isolation was done. RNA was stored at -80 °C until further processing. For PCR analysis RT-PCR was performed. cDNA was synthesized by using SuperScriptIV-Kit (ThermoFisher) following manufacturer's recommendations and could be stored until further processing at -20 °C. To perform PCR different primer sets were used (Table 1) and PCR was executed with Phire Hot Start II DNA Polymerase (ThermoFisher).

Mass spectrometry analysis
We used nanoflow LC-MS/MS using Orbitrap Lumos (Thermo Fisher Scientific) coupled to an Agilent

Data processing
Mass spectra were processed using Proteome Discover (version 2.1, Thermo Scientific). Peak list was searched using Swissprot database (version 2014_08) with the search engine Sequest HT. The following parameters were used. Trypsin was specified as enzyme and up to two missed cleavages were allowed. Taxonomy was set for Homo sapiens and precursor mass tolerance was set to 50 p.p.m.
with 0.05 Da fragment ion tolerance. TMT tags on lysine residues and peptide N termini and oxidation of methionine residues were set as dynamic modifications, and carbamidomethylation on cysteine residues was set as static modification. For the reporter ion quantification, integration tolerance was set to 20 ppm with the most confident centroid method. Results were filtered to a false discovery rate (FDR) below 1%. Finally, peptides lower than 6 amino-acid residues were discarded. Within each TMT

Results
Generation of iPSC-derived neuronal progenitors from RTT and isogenic controls RTT patient and iCTR fibroblasts were reprogrammed into iPSCs via electroporation of reprogramming plasmids [24]. Pluripotency was confirmed using classic assays, including immunocytochemistry (Additional file 1: Figure S1a) and RNA expression (Additional file 1: Figure S1b). Expression of mutated and healthy MeCP2 in the RTT and iCRT lines was confirmed by immunocytochemistry and PCR for MECP2 (Fig. 2a, b). Additionally, the protein expression level of MeCP2 detected by mass spectrometry showed a higher expression of MeCP2 in iCTR relative to RTT samples (Fig. 2c, Additional file 1: Figure S1c). A detectable low expression of mutant alleles, here MeCP2 expression in mutant RTT lines, is a common effect caused by the so-called co-isolation issue in TMT-experiments [27]. Generation of NES cells was performed as described before [21]. Neuronal induction into neuronal rosette structures was monitored by visual inspection, and appeared after approximately 12 days of neuronal development initiation (Fig. 1). Time points for sample collection were chosen along the differentiation towards neuronal progenitor cells.

MS-based quantitative proteomics during neuronal development
To study proteomic changes between RTT and iCTR during neuronal development, cell lysates at indicated time points were subjected to tryptic digestion, high-pH fractionation followed by highresolution tandem mass spectrometry (LC-MS/MS) analysis and TMT-10plex quantification (Fig. 1). In total we identified up to 7702 proteins in total, of which 3658 proteins were identified in all samples (Additional file 1: Figure S2, Table S1). Next, to determine protein expression changes over time- and 'forebrain development' were down regulated in RTT. Terms such as brain development were down regulated in D15 as well as D22. Gene set enrichment analysis (GSEA) of all up and down regulated proteins further revealed that RTT associated proteins were strongly enriched in gene sets such as apoptosis, DNA repair and in metabolism (Additional file 1: Figure S4, Table S3). To further verify the results of the mass spectrometry, we performed western blot analysis for proteins SOX2 and SOX9, transcription factors with pivotal role in development and differentiation [28,29], which showed significant differences in expression levels between iCTR and RTT lines at D22 (Fig. 3a, c). In line with mass spectrometry data, western blot analysis showed a significant increase in SOX9 expression levels in RTT lines when compared to iCTR (p = 0.0057, unpaired t-test), and a decrease in SOX2 expression in RTT lines at D22, although this did not reach statistical significance (p = 0.07, unpaired t-test). Together both approaches demonstrate that SOX2 and SOX9 were differentially expressed between RTT and iCTR, thereby validating our findings that RTT samples show proteome changes at early neurodevelopmental changes A previous study identified perturbed astrocyte differentiation of RTT-iPSCs, suggesting skewed differentiation of neural progenitor cells into neuronal cell lineage [30]. Here we studied whether we could find similar changes, i.e. increased neuronal marker MAP2 and decreased glia marker (e.g. ATP1A2, CLU and SLC1A3) expression. While these astrocyte and neuronal markers are expressed in our samples, we found no significant differences between iCTR and RTT samples (Additional file 1: Figure S3). Furthermore, while the authors showed higher expression of LIN28 in RTT samples, we identified two isotopes (LIN28A and B) with no differences between RTT and iCTR samples. In addition, we compared our proteomic altered data to the Allen Brain atlas, which is showing individual gene expression in the different brain areas (Supplementary Table S4). This revealed high variability of expression between transcriptomics and proteomics. The majority of the proteins have measurable expression levels in human brain in vivo. Overall, we show that proteins associated with neuronal development are differentially expressed in RTT at early stages of neuronal differentiation.

Coordinated proteome alteration during neuronal development in RTT syndrome
To gain insight into how the differentially expressed proteins in RTT behave across time points, we further analysed all the significantly up or down regulated proteins at D3, D9, D15 and D22. This Collectively, our data reveals a changing, stage-specific pattern of the differentially expressed proteins during neuronal development in RTT.

MeCP2 network analysis
To further investigate the proteins that are targets by the MeCP2 protein, we drew a protein interaction network (Cytoscape, Genemania plugin) using MeCP2 protein as input (Fig. 4c). The data  (Fig. 5a). We identified 27 proteins being up and 12 proteins being down regulated in RTT compared to iCTR. As expected, MeCP2 was one of the most strongly down-regulated proteins in RTT. GO analysis revealed biological processes such as 'cell-cell adhesion' and 'acyl-CoA metabolic processes' to be up regulated (Fig. 5b), which are also up regulated at individual time points D15 and D22 (Fig. 4a). In contrast, several processes such as 'response to cadmium ion', 'response to drug' and 'behavioral fear response' were down regulated in RTT (Fig. 5b). Analysis of the differentially regulated proteins using Reactome pathway analysis revealed among others, 'JAK/STAT signaling after Interleukin-12 stimulation' and 'regulation of MeCP2 expression and activity' to be differentially expressed in RTT versus iCTR (Fig. 5c). To further visualize the connectivity among these significant proteins, we analyzed their protein networks in the Cytoscape tool (Genemania plugin). A high degree of connectivity, such as being co-expressed and having shared genetic interactions, around these proteins was identified. Interestingly, the majority of the proteins are involved in immunity, actin cytoskeleton organization and calcium binding (Fig. 5d).
Together, we show that proteins associated with immunity and metabolic processes are differentially expressed in RTT in a time-point independent manner during differentiation towards neuronal progenitors.

Discussion
RTT samples present protein expression changes that became more apparent from early to late neuronal progenitor stages.
Already at D3 of neural induction, we found decreased protein levels associated with axon regeneration and filopodium assembly, and increased protein levels in neuronal apoptotic processes.
Furthermore we show a down regulation of proteins associated with dendrite morphogenesis and excitatory postsynaptic potential at D9, and down regulation of axon guidance and brain development at D15 of neural induction. By D22 a clear set of proteins was altered in RTT with up regulated proteins associated with 'cell adhesion', 'cytoskeleton organization' and 'translation initiation'.
Proteins that are down regulated in RTT are involved in 'nervous system development', 'forebrain development' as well as 'histone methylation', which has classical roles with MeCP2 in its organization [32]. Previous study in MeCP2 deficient mice also found altered proteins associated to development and morphology of neurons as well as metabolism [33]. While the proteomic alterations found are relatively small (1.3 fold-change), they are in line with previous studies showing dysregulation of dendrites and axons in RTT patients/mouse models [34][35][36][37] as well as in the juvenile RTT brain [38][39][40]. Although neuronal progenitors did not develop dendrites or axons at these time points yet, our findings indicate that proteins involved in these processes are already expressed and altered in and TUBB2A. These proteins are identified to be dysregulated in other brain disorders such as epilepsy, Niemann-Pick disease, mental retardation, cortical dysplasia, lissencephaly, neurotonia and axonal neuropathy. These candidate proteins need to be confirmed in other RTT cases as well before treatment therapy. Overall, we show that dysregulation of MeCP2 affects protein expression changes associated with neurodevelopmental functions at early stages of neuronal differentiation.
Expression changes of MeCP2-interacting proteins in RTT and iCTR Network analysis using GeneMANIA [41] revealed part of these proteins to be interacting with MeCP2.
Interestingly, of these interacting proteins, MBD4 is a member of the MBD family of proteins together with MeCP2. Mutations in these functional important domains tend to cause RTT-associated phenotypes [42][43][44]. Furthermore, HMGB1, which was down regulated in D9 and D15 in our data, was previously shown to be lower expressed in hippocampal granule neurons of Mecp2 KO mice [45].
Robust protein profile changes along the course of neuronal differentiation When all RTT versus iCTR samples from different time points were pooled, data revealed differentially expressed proteins in RTT involved in immunity, calcium binding and metabolism. This analysis allowed us to compensate for limited amount of biological replicates in such high-throughput technology. Several proteins associated in metabolic processes were differentially expressed in RTT, including ACSF2, ACOT2, ACOT9, LDHA, MIF, NPC1, CAT and GSTO1. Current evidence in perturbed lipid metabolism in the brain and the peripheral tissue of RTT patients and mouse models now also supports the metabolic dysfunctions as a component of RTT [46]. Also mRNA GSTO1 was up regulated in RTT patients' lymphocytes together with several other mitochondrial related genes [47].
Furthermore, our results indicate dysregulated proteins associated with Interleuking-12 signalling.
Interestingly, children with MECP2 duplication show immunological abnormalities and suppressed IFN-ϒ [48]. Although RTT is more often classified as a neurodevelopmental disorder, recent studies in cytokine release also suggest involvement of the immune system [49]. A previous study demonstrated that during early brain formation disturbances in the metabolism produces changes in the morphological and biochemical development of the brains [50]. Another study further observed that models with synaptic defects during development fail to couple to metabolic pathways [51].
These observation might indicate that the metabolic pathways could regulate the alterations in neuronal development. We further identified several proteins associated with calcium signalling to be altered. A disturbance in calcium homeostasis during early postnatal development was reported in Mecp2 knockout model and altered calcium signalling in RTT-iPSC-derived cells [52,53]. Altogether, this indicates that next to dysregulation in neurodevelopmental processes, disease mechanisms underlying RTT phenotypes could also involve immunity, calcium signaling and metabolism.

Early treatment
The finding, that neuronal progenitor cells of RTT patients show altered protein expressions responsible for neuronal development and maturation, indicates that RTT influences the patients much earlier than first symptoms suggest. Therefore, early treatment should be explored. As there is no cure for RTT yet, mutations in MECP2 are not screened for in pre-natal diagnostics. However, early post-natal testing could still provide enough time to start treatment approaches, for example focused on supporting neuronal maturation with treatments, such as IGF-1 or Bumetanide [54][55][56], to extenuate disease symptoms. The progress of RTT is based on different aspects. There is evidence that mosaicism and therefore the ratio between cells expressing healthy compared to mutated MeCP2 plays a major role [57]. Furthermore, the mutation itself has a huge impact on severity of the disease [58]. However, in almost all cases of patients with MECP2 mutation a post-natal phase of compensation can be observed. In terms of diagnostics, mutations in MECP2 are not always associated with RTT and a mutation in MECP2 is not sufficient to make the diagnosis of RTT. RTT diagnosis is now based on clinical criteria rather than molecular findings. A set of altered signaling pathways or proteins, as found in this study, could be used as biomarkers to guide the disease diagnosis. Therefore, we hope that our results give awareness of the early pre-natal onset of RTT and could be further explored as potential targets for treatment.

Limitations
A limitation of our study is the number of biological replicates that was used (n = 3). Because of the incredible work that was needed, this number of replicates was feasible for this study. A possible avenue for future research would test few target proteins of a larger scale of biological replicates.

Conclusion
Before typical RTT-associated symptoms appear, both RTT patients and mouse models of RTT already show abnormalities [59,60]. This implies that underlying mechanisms are already affected during early neurodevelopmental stages. Much of our understanding of how MeCP2 deficiency contributes to RTT disease is derived from genomic and transcriptomic studies. So far, only a few proteomic studies have been performed involving RTT human derived tissue [19,61,62]. The current study provides mass spectrometry-based quantitative proteomic data, depth of 7702 proteins, using an earlier developed iPSC-based models involving RTT patient and isogenic control cells [24]. We showed that changes in dendrite morphology or synaptic defects, previously associated with RTT [22,63], already become apparent at early developmental stages. Proteins involved in immunity and metabolism, also in line with previous studies on RTT pathology [37,46], are consistently differentially expressed at all time points studied. Insight into differentially expressed protein levels could support identification of novel biomarkers as well as therapeutic strategies. In total 24 samples were subjected for tryptic digestion, TMT-based isotope labeling, high-pH fractionation and LC MS/MS analysis. Different bioinformatic approaches were then used to analyze the data.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download. Table S2.xlsx Table S1.xlsx Figure S4.pdf Figures S1.pdf Table S4.xlsx Figure S2 .pdf Figure S3.pdf Table S3.xlsx figure X1.pdf