NEAT1-mediated regulation of proteostasis and mRNA localization impacts autophagy dysregulation in Rett syndrome

Abstract

Rett syndrome (RTT) is a severe neurodevelopmental disorder primarily caused by loss-of-function mutations in the MECP2 gene, resulting in diverse cellular dysfunctions. Here, we investigated the role of the long noncoding RNA (lncRNA) NEAT1 in the context of MeCP2 deficiency using human neural cells and RTT patient samples. Through single-cell RNA sequencing and molecular analyses, we found that NEAT1 is markedly downregulated in MECP2 knockout (KO) cells at various stages of neural differentiation. NEAT1 downregulation correlated with aberrant activation of the mTOR pathway, abnormal protein metabolism, and dysregulated autophagy, contributing to the accumulation of protein aggregates and impaired mitochondrial function. Reactivation of NEAT1 in MECP2-KO cells rescued these phenotypes, indicating its critical role downstream of MECP2. Furthermore, direct RNA–RNA interaction was revealed as the key process for NEAT1 influence on autophagy genes, leading to altered subcellular localization of specific autophagy-related messenger RNAs and impaired biogenesis of autophagic complexes. Importantly, NEAT1 restoration rescued the morphological defects observed in MECP2-KO neurons, highlighting its crucial role in neuronal maturation. Overall, our findings elucidate lncRNA NEAT1 as a key mediator of MeCP2 function, regulating essential pathways involved in protein metabolism, autophagy, and neuronal morphology.

Graphical Abstract

Introduction

The methyl-CpG binding protein 2 (MeCP2) is an abundant nuclear protein with global regulatory roles primarily associated with its function as a key epigenetic regulator involved in chromatin remodeling and transcriptional control [1–3]. Initially identified as a protein factor binding to methylated CpG dinucleotides [4], MeCP2 exhibits a complex interactome and participates in gene regulation at multiple levels [3], emerging as a master regulator of various cellular programs. MeCP2 is prominently expressed in the developing brain [5], where it plays critical roles in neural differentiation, maturation, and migration [6, 7]. Loss-of-function mutations in the MECP2 gene represent the most common cause of Rett syndrome (RTT, OMIM #312750) [8], a severe and progressive neurodevelopmental disorder affecting ∼1 in 10 000 female births [9, 10]. RTT is characterized by sudden regression and subsequent loss of acquired language and motor skills, leading to profound cognitive impairment. Notably, the developmental regression seen in RTT patients arises from altered neural function rather than neurodegeneration. Despite significant progress in understanding MeCP2 function and the pathological consequences of its loss-of-function mutations over the past two decades, a comprehensive understanding linking the nuclear function of MeCP2 with alterations in key signaling pathways and cellular defects in RTT remains elusive. Such in-depth analysis could critically enhance therapeutic interventions, especially as new treatment avenues continue to advance and approach clinical application [11–13].

The study of MeCP2 function has heavily relied on mutant mouse models, which faithfully replicate pathological traits of the disorder, providing comprehensive insights into MeCP2 function across various biological scales, from the cellular to the organismal level [14]. These models can accurately mimic the dendritic and neuronal morphological abnormalities, synaptic imbalances [15, 16], and associated respiratory, motor and cognitive deficits observed in RTT patients [17, 18]. Recent advancements in cell culture technologies [19], coupled with the revolutionary genome editing capabilities of the CRISPR-Cas system [20, 21], have opened new avenues for modeling human disorders in vitro with unprecedented precision. Stem cell-based models of RTT have emerged as valuable tools for studying the early stages of the disease, allowing researchers to probe the cellular and molecular underpinnings of MeCP2 loss-of-function in human neurons [22, 23].

Furthermore, single-cell RNA sequencing (scRNA-seq) has emerged as a transformative tool, offering unparalleled insights into the intricate cellular composition, heterogeneity, and dynamics of the developing brain. In contrast to traditional bulk RNA-seq methods, which provide averaged gene expression profiles across cell populations, scRNA-seq offers the resolution needed to dissect heterogeneous cell populations at the individual cell level. This approach unveils a rich transcriptional landscape, unraveling novel cell types, lineage relationships, and regulatory networks pivotal in neurodevelopmental processes, such as neurogenesis, neuronal migration, synaptogenesis, and circuit formation [24]. Additionally, scRNA-seq enables the characterization of cell-to-cell variability, revealing the molecular mechanisms driving cellular diversification and specialization during brain development. For example, recent studies utilizing scRNA-seq have reported high variability in the expression of MECP2 within inhibitory neurons of the human brain, highlighting both intra- and inter-individual variability [25]. In the context of RTT, an X-linked disorder predominantly affecting females, innovative approaches have been developed to distinguish between cells expressing wild-type (WT) or mutant MECP2 alleles, enabling the study of mosaic transcriptomic profiles and offering new insights into the pathogenesis of the disorder [26].

In this study, we applied scRNA-seq to dissect the transcriptomic changes in a human cellular model of RTT derived from immortalized neural progenitors [27]. These cells can be differentiated in vitro towards different specific subtypes and have been used to model neurodegenerative disorders [28, 29], as well as to investigate MeCP2 function and the impact of its loss-of-function mutations [30, 31]. Our previous research revealed the role of certain types of noncoding RNAs (ncRNAs) in RTT pathophysiology, demonstrating the involvement of specific circular RNAs (circRNAs) and RNAs derived from ultraconserved regions (T-UCRs) in the biogenesis of the ionotropic glutamate receptor of the AMPA type downstream of MeCP2 function [30]. In the present study, we find through scRNA-seq analysis marked downregulation of the evolutionarily conserved, long noncoding RNA (lncRNA) NEAT1 in MECP2 knockout (MECP2-KO) human neural progenitor and differentiating cells. NEAT1 plays pivotal roles in nuclear organization, RNA metabolism, and cellular stress responses, with dysregulation of NEAT1 implicated through unclear mechanisms in various human diseases, including cancer, neurodegenerative disorders, and autoimmune conditions [32–37]. Our findings indicate that NEAT1 directly regulates the nuclear/cytoplasmic distribution of key autophagy-related messenger RNAs (mRNAs), thereby ensuring proper autophagic flux in response to stress cues. Dysregulation of NEAT1 is associated with the protein accumulation and autophagic defects observed in MECP2 mutant cells. Furthermore, reactivation of NEAT1 in MECP2-KO cells rescues the characteristic morphological defects found in RTT neurons, underscoring the contribution of this abundant lncRNA to MeCP2 function and the disease phenotype.

Materials and methods

Post-mortem samples

Post-mortem brain tissue samples were acquired from both control subjects and individuals with RTT. These samples were sourced from the National Institutes of Health NeuroBioBank at the University of Maryland in Baltimore, MD, USA, and the Human Brain and Spinal Fluid Resource Center at the VA West Los Angeles Healthcare Center in Los Angeles, CA, USA. These centers receive sponsorship from the National Institute of Neurological Disorders and Stroke, the National Institute of Mental Health, the National Multiple Sclerosis Society, and the Department of Veterans Affairs.

ReNCells VM cell culture and differentiation methods

Immortalized human neural progenitors [neural progenitor cell (NPC)] from 10-week-old ventral mesencephalon of fetal brain were purchased (ReNCells VM, SCC008, Merck Millipore, Burlington, MA, USA). The cells were cultured in laminin-coated flasks (20 μg/ml; Sigma–Aldrich, St. Louis, MO, USA), with Dulbecco’s modified Eagle’s medium nutrient mixture (DMEM-F12, L0093-500 Biowest, Nuaillé, France) supplemented with 0.2% heparin (STEMCELL Technologies, Vancouver, BC, Canada), B-27 complete vitamins (17504-044, Invitrogen, Carlsbad, CA, USA), and antibiotic-antimycotic (L0010-100, Biowest), and kept at 37°C with maximum humidity in an atmosphere of 5% CO₂; media was changed every other day. To maintain pluripotency, basic fibroblast growth factor (bFGF-2) (20 ng/ml, SRP4037) and epidermal growth factor (EGF) (20 ng/ml; SRP3027, Sigma–Aldrich) were added to the feeding media. Cells were split every 3–6 days when 90% confluence was reached using Accutase (SCR005, Sigma–Aldrich), and centrifuged for 5 min at 300 × g. For spontaneous differentiation, cells were plated on laminin-coated dishes and allowed to reach 80% confluency, then culture media was administrated without EGF and bFGF-2. Three days after growth factors are withdrawn the cells undergo morphological changes, forming a mixed population of neurons and glia [30]. Glutamatergic differentiation protocol was performed as described previously [30]. Briefly, cells were seeded on laminin-coated plates 24 h before infection. Cells were double infected with TetO-hNGN2-P2A-eGFP-T2A-PuroR and CMV-rtTA. Forty-eight hours after infection, doxycycline (1 μg/ml) was administered to the cells, followed by puromycin (0.5 μg/ml) the next day. After 24 h of puromycin selection, EGF and FGF-2 were withdrawn from the media, and cells were allowed to differentiate for the indicated number of days.

Induced pluripotent stem cells culture and NPC differentiation

RTT patient-derived induced pluripotent stem cells (iPSCs) were kindly donated by the Rett Syndrome Research Trust along with Coriell Institute for Medical Research (Camden, NJ, USA). The cells were sourced from a female RTT patient carrying the T158M mutation on MECP2 gene. Reprogramming and karyotyping of the cells were performed by the Harvard Stem Cell Institute iPS Core Facility (Cambridge, MA, USA). Two iPSC lines were generated, characterized by random X-inactivation: one is an isogenic control expressing the WT MECP2, while the other expresses the T158M mutation. The iPSC colonies were plated on CULTREX™-coated six-well plates (3434-010-02, Bio-Techne R&D Systems) and maintained in mTeSR™1 medium (#85850, STEMCELL Technologies) supplemented with antibiotic-antimycotic (L0010-100, Biowest). They were incubated at 37°C with maximum humidity in a 5% CO₂ atmosphere for 3 days, with daily media changes. Passaging was performed when the cultures reached 80% confluency, using Gentle Cell Dissociation Reagent (STEMCELL Technologies). On day 4, iPSC colonies were dissociated into single cells and plated for NPC differentiation on CULTREX™-coated six-well plates, following the monolayer method and the dual Suppressor of Mothers Against Decapentaplegic (SMAD) inhibition-mediated neural induction protocol (#08581, STEMdiff™ SMADi Neural Induction Kit, STEMCELL Technologies), as per the manufacturer’s instructions. Media were changed daily, and cells were passaged three times during the 21-day protocol. Cell pellets were collected at the undifferentiated stage (as stem cells expressing NANOG and OCT4) and at various stages of differentiation (days 5, 8, and 11). Differentiation into NPCs was confirmed by the expression of SOX1 and PAX6 (see the ‘Immunofluorescence’ section for details).

Single-cell RNA sequencing

Two scRNA-seq experiments were performed. In experiment 1, WT and MECP2-KO (clone A33) cells were cultured and differentiated in a time-course for the experiment: progenitors, 3, 7, 14, 21, and 30 days after differentiation. At each time point, cells were pooled and separated by fluorescence-activated cell sorting (FACS). FACS conditions and cellular selection were optimized beforehand to ensure neuron viability post-sorting. scRNA-seq was carried out at CNAG (CRG, Barcelona, Spain), using the following protocol. The cellular concentration and viability were verified by counting with a TC20™ Automated Cell Counter (Bio-Rad Laboratories, S.A.). Individual cells were partitioned into Gel Bead‐In‐Emulsions with a Target Cell Recovery of 5000 total cells, by using the Chromium Controller system (10X Genomics). Complementary DNA (cDNA) sequencing libraries were prepared using the Chromium Single-cell 3′ mRNA kit (V3; 10X Genomics) following manufacturer’s instructions. Briefly, after GEM-RT clean up, cDNA was amplified during 13 cycles and cDNA QC and quantification were performed on an Agilent Bioanalyzer High Sensitivity chip (Agilent Technologies). Libraries were indexed via polymerase chain reaction (PCR) with the PN-220103 Chromium i7 Sample Index Plate. Final cDNA library size and concentration were confirmed using the same Bioanalyzer system. Sequencing was carried out on a NovaSeq 6000 platform (Illumina), producing ∼40 000 paired-end 50 bp reads per cell.

In experiment 2, WT, MECP2-KO A33 and MECP2-KO 3F cells were cultured and collected as progenitors (pg) and after 7 days of differentiation (7d). Cells were pooled and separated by FACS. Sample preparation and scRNA-seq were performed by the Single Cell Unit (Josep Carreras Research Institute, Badalona, Spain) and Macrogen, Inc. (Seoul, Republic of Korea), following the same protocol described for experiment 1.

For both experiments, data analysis was conducted using the Seurat package version 5.0.1 in R version 4.3.2. Cells were filtered based on the number of unique genes detected, the total number of molecules detected, and the mitochondrial content. The number of cells analyzed per condition for each experiment was: experiment 1 – WT: pg (10 169), 7d (11 216); MECP2-KO A33: pg (9936), 7d (10 604); MECP2-KO 3F: pg (8956), 7d (10 678); experiment 2 – WT: pg (10 169), 7d (11 216); MECP2-KO A33: pg (9936), 7d (10 604); MECP2-KO 3F: pg (8956), 7d (10 678). Expression data were normalized using the LogNormalize function and scaled with ScaleData. Highly variable genes were identified with the FindVariableFeatures function. Dimensionality reduction was performed using principal component analysis, retaining the first 20 principal components. Clustering was performed using FindNeighbors and FindClusters, and clusters were visualized with Uniform Manifold Approximation and Projection (UMAP). Cell cycle phases were scored using the CellCycleScoring function, and cells were classified into NEAT1 high and low expression groups based on the 60th percentile. Differential gene expression analyses were performed with the FindMarkers function (Wilcoxon Rank Sum test, min.pct = 0.1. An astrocyte gene signature was computed using the AddModuleScore function, based on the expression of the following genes: AQP4, ALDOC, S100B, CD44, NF1A, SLC1A3, SLC1A2, CX43, VIMENTIN, and ALDH1L1.

Total RNA sequencing

Total RNA was extracted with Promega’s Maxwell RSC microRNA (miRNA) Tissue kit (AS1460, Promega, Madison, WI, USA). RNA-seq libraries were generated and sequenced by BGI (Warsaw), in 150-bp paired-end, with the DNBSEQ platform, using three biological replicates for each group, with an average yield of 13.27G data per sample. Fastq files were aligned to the hg38 transcriptome using HISAT, with standard options. Clean reads were aligned to the reference genes using Bowtie2. Differentially expressed genes were detected with DESeq2. Only genes with an absolute log₂FC > 1 and a q value < 0.05 were selected as differentially expressed.

Transmission electron microscopy

Plastic cover slips (Thermanox, Thermo Fisher Scientific, Waltham, MA, USA) were coated with laminin and placed in 12-well plates. Progenitor cells were plated the day before the experiment. The next day the cells were fixed with 1.8% glutaraldehyde solution (#16216, Electron Microscopy Sciences) in 0.1 M cacodylate buffer (pH 7.4; #11650, Electron Microscopy Sciences), and then the wells were washed three times with 0.1 M cacodylate buffer for 5 min each. Next, the covers were incubated with 1% osmium tetroxide/1.5% potassium ferricyanide solution for 1 h on ice. Covers were incubated with 1% tannic acid in 0.05 M sodium cacodylate at room temperature (RT) for 45 min, and then washed and dehydrated in an ethanol series. The samples were then embedded in Epon resin mixed with propilene oxide (1:1) for 30 min, followed by fresh 100% Epon for 1 h. The cover slips were placed on solid Epon Beem capsules and baked at 60°C overnight to polymerize the resin. Ultrathin (60–70 nm) sections (Ultracut Reichert S, Leica) were collected on 200 mesh copper grids and contrasted with 2% uranyl acetate in methanol and 0.5% lead citrate in aqueous solution. Ultrastructure images were captured with a transmission electron microscope (JEOL 1010, operated at 60 kV). The number of endosomes and mitophagy events was measured in micrographs obtained at 8000–20 000 magnifications (n > 30 micrographs per sample).

Generation of knockout and mutant cell lines [MECP2-KO (clones A33 and 3F), MECP2-R133C, and NEAT1-KO]

CRISPR/Cas9 technology was used for engineering MECP2 and NEAT1 knockouts (KOs) and for introducing the MECP2-R133C point mutation in human ReNCells VM. For details on the generation of the MECP2-KO A33 clone, refer to Siqueira et al. [30]. The MECP2-KO 3F clone was created using a guide RNA targeting exon 4 (5′-AAAAGCCTTTCGCTCTAAAG-3′), designed with the CRISPR Design Tool (http://crispr.mit.edu/), and cloned into the pSpCas9 (BB)-2A-GFP vector (PX458, Addgene^® #48138, Watertown, MA, USA).

The MECP2-R133C cell line was generated by introducing a cytosine-to-thymine substitution at position g.105697C > T on chr X of the genomic DNA using a donor sequence (5′-CTCTGACATTGCTATGGAGAGCCTCTAATTGTTCCTTGTGTCTTTCTGTTTGTCCCCACAGTCCCCAGGGAAAAGCCTTTTGCTCTAAAGTGGAGTTGATTGCGTACTTCGAAAAGGTAGGCGACACATCCCTGGACCCTAATGATTTTGACTTCACGGTAAC-3′). The guide RNA mentioned above was used to create a double-strand DNA break, while the donor sequence provided the template to introduce the R133C missense mutation.

NEAT1 KO was achieved using a construct described by Li et al. [38]. The pX330-NEAT1pr_v1 plasmid [gift from Archa Fox (Addgene plasmid #97082; http://n2t.net/addgene:97082; RRID: Addgene_97082)] encodes Cas9 protein and a single guide RNA targeting the human NEAT1 promoter region. For all the above cell line generation, ∼2 million cells were transfected by nucleofection (A33 voltage, VPI-1003, Primary Neurons Kit, Amaxa^®, Lonza Group Ltd, Basel, Switzerland) with their corresponding plasmids. MECP2-KO clones A33 and 3F were isolated by FACS. NEAT1-KO cells correspond to the pool of transfected cells. MECP2 depletion was confirmed by western blotting (WB) and Sanger sequencing, while NEAT1 depletion was validated by reverse transcription quantitative polymerase chain reaction (RT-qPCR) and RNA fluorescence in situ hybridization (RNA-FISH). The missense mutation was confirmed by Sanger sequencing.

NEAT1-shRNA depletion

Short-hairpin RNAs (shRNAs) targeting the NEAT1 gene (isoforms 1 and 2) were selected based on published sequences [39, 40] and screened for potential off-target effects. Two distinct shRNAs (shRNA1 and shRNA2, Supplementary Table S2) were cloned into the BamHI and EcoRI restriction sites of the pLVX-shRNA2 vector (Cat# 632179, Clontech). Lentivirus production was performed in HEK293T cells, maintained in DMEM with GlutaMAX (Cat# 31966-021, Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Cat# 10270, Thermo Fisher Scientific), at 37°C in a 5% CO₂ atmosphere. HEK293T cells were co-transfected with the shRNA vectors and packaging plasmids using jetPRIME^® transfection reagent (Cat# 114-15, Polyplus-Transfection) following the manufacturer’s protocol. Forty-eight hours post-transfection, the culture supernatant containing lentiviral particles was collected, filtered through a 0.45 μm low-protein-binding filter, and concentrated using Lenti-X Concentrator (Cat# PT4421-2, Clontech). The concentrated lentiviral particles were stored at −80°C until use. For viral transduction, target cells were infected with the lentivirus, and a 2-day recovery period was allowed before initiating downstream experiments.

NEAT1 reactivation/overexpression

NEAT1 expression was reinstated in human MECP2-KO and NEAT1-KO cells using two distinct strategies: CRISPR activation technology and the introduction of the cloned long isoform of NEAT1. The CRISPR activation method employed a catalytically inactive Cas9, as outlined by Yamazaki et al. [41]. For this approach, we used the following three plasmids: single-guide RNA (sgRNA) [MS2; a guide cloning backbone (Plasmid #61424)], dCAS9-VP64_GFP (Plasmid #61422), and MS2-P65-HSF1_GFP (Plasmid #61423), which were acquired from Addgene (Watertown, MA, USA). Lentiviruses for dCAS9-VP64_GFP and MS2-P65-HSF1_GFP were produced as described above. For the sgRNAs delivery, three sgRNAs targeting the promoter region of NEAT1 (listed in Supplementary Table S2) were synthesized and cloned into sgRNA (MS2). MECP2-KO and WT cells that had been previously infected were seeded in 10-cm dishes and transfected with Lipofectamine™ Stem Transfection Reagent (Thermo Fisher Scientific). Overexpression of NEAT1 was verified by RT-qPCR. As control, WT and MECP2-KO cells were transfected with the empty vector.

MALAT1 and NEAT1 long isoform cloning

Isoform 1 of human MALAT1 (NR_002819.4, 8779 bp) was cloned into the pcDNA4/TO vector (Clontech, Mountain View, CA, USA) using BsiWI and NotI restriction sites. The 8779 bp sequence, along with 100 bp of downstream genomic DNA, was amplified from ReNCells VM DNA using high-fidelity Phusion DNA Polymerase. Primers containing BsiWI and NotI sites were designed (Supplementary Table S2). The entire sequence was confirmed by Sanger sequencing to ensure fidelity. The long isoform of human NEAT1 (NEAT1_2, NR_131012.1, 22 kb) was successfully cloned into the pcDNA4/TO vector (Clontech, Mountain View, CA, USA). Initially, the 5′ region of NEAT1, spanning ∼3 kb, was amplified using the plasmid pCRII_TOPO_hNEAT1 [a gift from Archa Fox [42] (Addgene plasmid #61518; http://n2t.net/addgene:61518; RRID: Addgene_61518)] as a template. Subsequently, the remaining 19 kb segment of the NEAT1 sequence was amplified from ReNCells VM cDNA in two separate PCRs, employing high-fidelity Phusion DNA Polymerase for enhanced accuracy. These PCR products were then assembled and cloned into the pcDNA4/TO vector at the AscI and BamHI restriction sites, creating a comprehensive 22 kb NEAT1 insert. To ensure fidelity of cloning processes, the entire PCR-amplified fragment underwent sequencing, confirming the integrity of the sequence and the absence of PCR-induced mutations. However, for NEAT1_2, sequence analysis revealed a 903 bp deletion within the 3′ region of NEAT1 [chr11:65 436 667–65 437 571 (hg38)]. This deletion was consistently observed in the cDNA from multiple cell lines, suggesting a biological origin rather than a cloning artefact. Moreover, attempts to amplify this genomic region were unsuccessful, indicating potential annotation inacuracy, or technical challenges associated with its sequence complexity or structure.

For the introduction of the constructed vector into cells, Neon NxT Electroporation (Thermo Fisher Scientific) was utilized. Cells were first detached using Accutase^® and subsequently counted. Two million cells were then suspended in 100 μl of electroporation buffer, together with 5 μg of the plasmid, and electroporated under the following conditions: 1400 V for 20 ms, delivered in two pulses. Post-electroporation, cells were cultured in media supplemented with RevitaCell^® (Thermo Fisher Scientific) to enhance recovery, with a media change performed the following day. The successful overexpression of NEAT1 was ascertained using ViewRNA assays, confirming the effectiveness of the cloning and delivery strategy. The empty vector and the vector containing cloned MALAT1 were used as experiment controls.

RNA in situ hybridization

ReNCells VM were seeded onto laminin pre-treated eight-well chambers one day before the experiments or allowed to differentiate for 7 days. Cells were probed for single-molecule sensitivity using the ViewRNA Cell Plus Assay kit (Ref. 88-19000-99, Thermo Fisher Scientific) following the manufacturer’s instructions. In brief, cells were fixed and immunostained overnight with Nestin antibody (for cytoplasmic determination) and secondary antibody Alexa FluorTM 488 chicken anti-mouse (1:1000, #A-21200, Thermo Fisher Scientific). RNA molecules, including ATG16L1 type 6 (VA6-3181792-01VX-01), ATG16L2 type 6 (VA6-3178559-01VX-01), MALAT type 6 (VA6-14180-VCP), ATG5 type 6 (VA1-12125-VC VX-06), NEAT1 type 6 (VA6-14476-VCP VX-06), or type 1 (VA1-12621-VCP), were probed, followed by pre-amplification, amplification, and probe labeling in a HybEZ™ II Oven (Advanced Cell Diagnostic, San Francisco, CA, USA) as per the manufacturer’s recommendations. Nuclei were stained with 4′,6-Diamidino-2-Phenylindole (DAPI). Slides were mounted using FluorSave™ Reagent (345 789, Merck Millipore), dried overnight at room temperature in the dark, and finally stored at 4°C. Images of hybridized cells were captured the following day using a Zeiss Axio Observer Z1 fluorescent microscope with ZEN blue 2012 software (Zeiss, Jena, Germany) and analyzed with ImageJ v1.51h. ImageJ software (v1.51h) was used to quantify the signal intensity of MALAT1 and the number of NEAT1+ foci per nucleus. For signal intensity analysis, DAPI staining was used to define the nuclear area (region of interest, (ROI)), and the MALAT1 fluorescence intensity was measured in multiple nuclei per image (n > 9) across several fields (n = 5 images). The MALAT1 signal intensity ratio was calculated by normalizing fluorescence intensity to the nuclear area. For NEAT1+ foci quantification, the counting tool in ImageJ was employed, analyzing all visible nuclei (n > 9 per image) across at least seven images per cell type. The distribution of NEAT1+ foci was presented as a percentage of the total, or as a distribution plot.

Immunofluorescence

For NeuronStudio analysis and immunofluorescence (IF), ReNCells VM cells were plated on laminin-coated eight-well chambers (Merck Millipore) and iPSC were plated on CULTREX™-coated 18-mm glass slides. The differentiation procedure was the same as previously described. On the day of the experiment, cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences) in washing solution [1 mM NaCl, 1 mM MgCl₂ in phosphate-buffered saline (PBS), pH 7.4] for 20 min at RT. Each chamber was blocked with 5% bovine serum albumin and 5% goat serum (Sigma–Aldrich) in 0.3% Triton^® in washing solution for 1 h at RT. Primary antibodies (Supplementary Table S1) were diluted in blocking solution and incubated overnight (12–16 h). Secondary antibodies (dilution 1:1000) were incubated for 1 h at room temperature in the dark (Supplementary Table S1). Chambers were washed between steps with washing solution three times for 5 min each. DAPI (5 mg/ml) was used for nuclei staining. Chambers were then removed from support, slides were mounted in FluorSave™ reagent (Merck Millipore), and dried overnight at room temperature, protected from light. Images were acquired with a Zeiss Axio Observer Z1 fluorescent microscope with ZEN blue 2012 software (Zeiss, Jena, Germany), and were analyzed in ImageJ software (v1.53k). Signal Intensity was measured using ImageJ software (v1.51h).

Autophagy flux assay

FUW mCherry-GFP-LC3 was a gift from Anne Brunet [43] (Addgene plasmid #110060; http://n2t.net/addgene:110060; RRID: Addgene_110060). LC3-GFP-mCherry lentiviral particles were produced in HEK293T cells as mentioned previously. For infection, ReNCells VM (WT, MECP2-KO, and NEAT1-KO cells) were seeded in laminin-coated flasks until they reached 70% confluency, at which point the virus concentration for infections was optimized in order to obtain optimal fluorescence signals. Following confirmation of Green Fluorescent Protein (GFP) and mCherry expression via microscopy, the cells were transferred to laminin pre-coated eight-well chambers and treated with Bafilomycin A1 (Selleckchem, Catalog No. S1413) for 16 h. The cells were then fixed with 4% Paraformaldehyde (PFA) in PBS for 10 min and washed three times with PBS. Nuclei were stained with DAPI. The cells were mounted in FluorSave™ reagent (Merck Millipore) and left to dry overnight at room temperature, shielded from light. Images were taken using a Leica Stellaris 8 microscope. Quantification was performed by counting the number of green and red foci in the images (n > 10 cells). Overlapping signals were identified by merging the corresponding channels. The analysis was conducted using the ‘Cell Counter’ plugin in ImageJ software (version 1.51 h).

Rapamycin and autophagy inhibitors treatment

For autophagy modulation, cells were seeded one day prior to treatment and cultures under optimal growth conditions, with strict adherence to prevent confluency from surpassing 80%. Cells were treated with Rapamycin (an mTOR inhibitor) at 50 μM, and Bafilomycin A [an inhibitor of vacuolar-type H (+)-ATPase] at a concentration of 10nM for 16h. Parallel control groups were subjected to treatment with the corresponding vehicle control (DMSO) ensuring that Dimethyl Sulfoxide (DMSO) levels did not exceed 0.5% of the total culture volume. Evaluation of treatment efficacy was conducted with WB through analysis of LC3 or S6 phosphorylation levels. Further details can be found in the WB section.

RNA extraction and RT-qPCR

Total RNA was extracted with Promega’s Maxwell RSC miRNA Tissue kit (AS1460, Promega, Madison, WI, USA). A negative control minus reverse transcriptase was run in parallel to control for genomic contamination. Real-time PCR reactions were performed in triplicate with a QuantStudio™ 5 Real-Time PCR System (TermoFisher Scientific, Waltham, MA, USA), using 10–30 ng of cDNA, 6 μl SYBR^® Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA), and 416 nM primers in a final volume of 12 μl for 384-well plates. All data were acquired and analyzed with the QuantStudio Design & Analysis Software v1.3.1 and normalized with L13 as endogenous control. Relative RNA levels were calculated using the comparative Ct method (ΔΔCt). A list of the primers used can be found in Supplementary Table S2.

Western blotting

Fresh cellular pellets from progenitors and 7 day-differentiated cells (>2 × 10⁶ cells) were scrapped and proteins extracted with Laemmli buffer [2% sodium dodecyl sulphate (SDS), 10% glycerol, 60 mM Tris–Cl, pH 6.8, 0.01% bromophenol blue], sonicated, and boiled for 5 min at 95°C after β-mercaptoethanol addition (2% final concentration). Samples’ concentration was determined quantifying the absorbance at 260 nm with the NanoDrop™ One/OneC Microvolume UV Spectrophotometer and using the equivalency between DNA and histones [6 units A_260nm (DNA) = 1 μg/μl of protein]. Equal amounts of protein (20 μg up to 40 μg) were loaded in 6%–15% sodium dodecyl sulphate–polyacrylamide gel electrophoresis gels and separated by electrophoresis. Next, samples were transferred to nitrocellulose membranes (0.2 μm), for 1 h with <400 mA. Blocking was performed for 1 h at RT with 5% skimmed milk with 0.1% Tween 20^® in PBS buffer. Primary antibodies (Supplementary Table S1) were diluted in blocking solution and incubated overnight at 4°C in a shaker. Membranes were washed between incubations with 0.1% Tween 20^® in PBS buffer three times for 7 min each. Secondary antibodies conjugated to horseradish peroxidase (HRP) anti-rabbit IgG (1:10 000; A0545; Sigma–Aldrich) or anti-mouse IgG (1:5000; NA9310, GE HealthCare, Chicago, IL, USA) were incubated for 1 h at RT in slow shaking. Lastly, bands were visualized using an ECL detection kit (GE Healthcare Life Sciences). ß-Actin-HRP (45 kDa, Sigma, #A5441) and α-TUBULIN-HRP (55 kDa, Abcam, Cambridge, UK, ab40742) were used as loading control. Protein abundance in each blot was quantified and is displayed below each western blot panel as the mean relative percentage compared to the control condition, normalized to the loading control. Ponceau staining was also performed to verify the accuracy and consistency of sample loading. Chemiluminescence signal intensity was quantified automatically using iBright software, expressed as signal intensity density for each membrane (n = 3). Supplementary Table S3 presents the mean values ± standard deviation (SD) from two biological replicates and 1–2 technical replicates for all WB panels.

Proteostat

Protein aggregates were quantified using the ProteoStat^® Aggresome Detection kit, adhering to the manufacturer’s protocol (Enzo Life Sciences, Farmingdale, NY). Cells grown under standard conditions were detached with Accutase and centrifuged at 300 × g for 5 min. Following a PBS wash, the cells were resuspended in 1 ml of PBS. They were then fixed with cold 4% paraformaldehyde for 30 min, centrifuged at 400 × g for 10 min, and washed with PBS to eliminate any paraformaldehyde residue. Subsequently, cells were permeabilized with 0.5% Triton X-100 for 15 min and stained with ProteoStat aggresome dye at a 1:2 500 dilution. Aggresomes were analyzed using a BD FACSCanto™ II Flow Cytometer (BD Biosciences) equipped with a 488-nm laser, detecting fluorescence in the PE channel. Data collection utilized gating criteria to isolate viable, single cells, using forward scatter (FSC) and side scatter (SSC) for selection.

Mitochondrial function analysis

Cells were seeded 24 h prior to the experiment on laminin pre-coated 12-well plates and grown under standard conditions. Live cells were treated with Tetramethylrhodamine, Ethyl Ester (TMRE) (T669, Invitrogen, Carlsbad, CA, USA) at a concentration of 50 nM for 30 min at room temperature, Bio Tracker ATP (SCT045, Merck Millipore) at a concentration of 7.5 μM for 15 min at 37°C, or MitoSOX^TM (M36008 Thermo Fisher Scientific) at a concentration of 5 μM for 10 min at 37°C. Simultaneously, mitochondria were dyed with green MitoTracker^TM (M7514, Thermo Fisher Scientific) at a concentration of 100 nM for 30 min RT in accordance with the manufacturer’s recommendations. Following a 1-h exposure to MitoTracker, cells were detached using Versene (15040-033, Gibco^®, Thermo Fisher Scientific). The detached cells were collected with PBS and subsequently centrifuged at 300 × g for 5 min. The resulting pellet was subjected to a single wash in PBS, followed by centrifugation under the aforementioned conditions. The cell population was resuspended in 200 μl of PBS and analyzed by flow cytometry using a FACSCantoII Flow Cytometer (BD Biosciences, CA, USA) equipped with a 488-nm laser and fluorescence detection in the Phycoerythrin (PE) and Allophycocyanin (APC) channels. Data acquisition was performed, employing gating parameters to select viable, single cells based on FSC and SSC. Fluorescence signals for MitoTracker™ and TMRE were acquired, with compensation adjusted as necessary to correct for spectral overlap. Only MitoTracker™-positive cells were considered in the subsequent analytical assessments.

Nuclear and cytoplasmic fractionation

Subcellular fractionation was carried out utilizing a PARIS™ kit (#AM1921, Thermo Fisher Scientific) in accordance with the manufacturer’s instructions. Equivalent quantities of cDNA from each fraction were subjected to RT-qPCR, and the results were standardized by considering the total RNA quantity obtained from each fraction. Endogenous control primers for MALAT1 and GAPDH were used to validate the segregation of mRNA into nuclear and cytoplasmic fractions, respectively. To validate the separation at the protein level, a western blot was conducted using antibodies against HISTONE H3 (1:5000, ab1791, Abcam) and α-TUBULIN-HRP (1:5000; ab40742, Abcam).

RNA pull-down and RIP-sequencing

NEAT1 RNA pull-down was performed as described previously [44, 45], with some modifications. Cells were plated on laminin-coated 150 mm dishes and allowed to grow until 70%–80% confluency. Cells were washed with cold PBS and fixed with freshly prepared 4% PFA for 10 min at room temperature with gentle rocking. Crosslinking was quenched by adding 1/10 volume of 1.25 M Glycine for 5 min under the same conditions. After two PBS washes, cells collected and pelleted by centrifugation at 510 × g for 5 min at 4°C. The cell pellets were lysed in a buffer containing 50 mM Tris–HCl, pH 7.0, 10 mM ethylenediaminetetraacetic acid (EDTA), 1% SDS, 200 U/ml of RNase inhibitor solution, and 5 μl/ml of proteases inhibitor cocktail. Lysates were sonicated using a Bioruptor (four pulses of 30 s ON and 30 s OFF) to assure complete solubilization. The lysate was centrifuged at 12 000 × g for 5 min at 4°C, and the supernatant was collected. An aliquot of the lysate was set aside as the input control. For the pull-down, two 3′-biotinylated NEAT1 probes (AS1: 5′ TGTATCTCTAACCAACCCTCTCCCC/3BioTEG/3′ and AS2: 5′ ACCCTCCCAGCGTTTAGCACACAAT/3BioTEG/3′) were used along with a nonrelated probe (5′ ATAATTTCAAACATCAAATGGTATTTTA/3BioTEG/3′) as a control. Hybridization was carried out using 100 pmol of probe in hybridization buffer (50 mM Tris–HCl, pH 7.0, 750 mM NaCl, 1 mM EDTA, 1% SDS, and 15% formamide) for 4 h on a rotating wheel. Hybridized RNA fragments were captured overnight using streptavidin-coated magnetic beads under rotation. The following day, the beads were washed five times with a 2 × SSC buffer containing 0.5% SDS. RNA–protein complexes were treated with 100 μg of Proteinase K for 45 min at 50°C, followed by incubation at 95°C for 10 min. RNA was separated from the beads using a magnetic rack and purified with 1 ml TRIzol^® Reagent (Thermo Fisher Scientific). Purification included phenol extraction, isopropanol precipitation, and DNase treatment, and the RNA was finally resuspended in 20 μl of nuclease-free water. A portion of the extracted RNA was used for cDNA synthesis and RT-qPCR to confirm NEAT1 enrichment. Input samples processed in parallel were used to evaluate pull-down efficiency.

Approximately 100 ng of total RNA from each sample were used for RIP-sequencing. Ribosomal RNA was removed using the Illumina Stranded Total RNA Prep with Ribo-Zero Plus (Illumina). RNA quality and integrity were assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). Library preparation, sequencing, and bioinformatic analysis were performed by Novogene Co., Ltd. (Cambridge, UK). Libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs) following the manufacturer’s protocol. Sequencing was performed on an Illumina NovaSeq X Plus platform, generating ∼12 gigabases of raw data per sample as paired-end 150 bp reads. Sequence quality was assessed with FASTQC followed by data trimming. Reads were then aligned to the reference genome using BWA. Duplicates were labeled using SAMBLAST, and a mapping quality score threshold of 13 was applied, corresponding to a non-unique region mapping probability of 0.05. Peak calling was performed using MACS2 software (threshold q value = 0.05). Functional enrichment analysis was conducted on all genes associated with NEAT1-AS2 probe peaks identified in the RIP-seq data. The analysis was performed using the clusterProfiler package in R [46–48].

Neurosphere formation assay

Neurosphere formation was carried out as explained in [30], with some modifications. Briefly, 3000 cells were plated in non-coated flasks with complete medium (EGF and FGF-2) to induce the formation of neurospheres. After 7 days of plating, photos were taken at Leics DM IL LED Fluo microscope (Wetzlar, Germany), and the relative diameter of the spheres was obtained in ImageJ software (v1.53k).

Filopodia density analysis

Immunofluorescence of MAP2 was used to label neurons. Confocal images were taken in Zeiss Confocal Microscope (z = 0.2 μm). The photos were analyzed with ImageJ software, and neurons were reconstructed in NeuronStudio Software (version 0.9.92), as described before [30]. In brief, a three-dimensional dendritic network was constructed using a Sholl analysis through automated segmentation. This network was visualized using spherical representations. Starting from the neuronal soma, concentric circles were drawn outward (covering around ± 100 μm with each circle having a 1 μm interval) along the neuronal projections. This highlighted various neuronal components, with neurites displayed in green, branching indicated by yellow spheres, and spines colored in orange and pink. The soma is shown in red. An assessment of spine density was performed, and the count of spines was determined, based on data obtained from at least 15 neurons for each experimental condition.

Statistical analysis

All experiments were conducted with a minimum of three independent replicates. Graphs and statistical analyses were generated using Graphpad Prism 9. Unless otherwise specified, RT-qPCR data are presented as mean with standard deviation. Comparative analyses to assess differences between experimental groups utilized Mann–Whitney’s U or unpaired samples t-tests, along with one-way Analysis of Variance (ANOVA) complemented by a Tukey or Bonferroni post hoc test for intergroup comparisons. Statistical significance was determined for values of P < .05 (*), P < .01 (**), P < .001 (***), or P < .0001 (****). Gene enrichment analyses from significantly altered genes were provided by BGI or conducted using Enrichr v2.0 software (http://amp.pharm.mssm.edu/Enrichr/) [49] and https://reactome.org/ [50].

Results

NEAT1 is highly dysregulated in mutant MeCP2 models, including human neural cells and samples from RTT patients

Previously, we examined the altered noncoding profile of a human RTT cellular model through bulk analysis of NPCs derived from the ventral mesencephalon (RenCells™), focusing specifically on two types of ncRNAs: circRNAs and T-UCRs [30]. We generated two distinct MECP2-KO cell clones, labeled A33 and 3F, both of which disrupt the reading frame and abolish MeCP2 protein production (Supplementary Fig. S1A and see below description of Fig. 3B). To comprehensively investigate the impact of MECP2 loss during the early stages of neural differentiation with single-cell resolution, we first performed scRNA-seq in the MECP2-KO A33 clone to assess the transcriptomic profile of neural progenitors across a time-course of spontaneous (free) differentiation. As anticipated, scRNA-seq analysis of WT and MECP2-KO NPCs revealed distinct transcriptomic trajectories when comparing KO and WT cells (Fig. 1A). Confirmation of cell cycle arrest upon initiation of differentiation was observed, with some cells still cycling at days 3 and 7 of in vitro differentiation, but the majority of cells transitioning to G1 phase beyond that point (Supplementary Fig. S1B). Dysregulation of transcriptional programs in MECP2-KO cells was evidenced by an upregulation of astrocytic signatures (Fig. 1B and Supplementary Fig. S1C), consistent with observed dysregulation of glial function in both RTT patients and models [51].

Figure 1.

NEAT1 is highly dysregulated in mutant MeCP2 models, including human neural cells and RTT patient samples. (A) UMAP plot of NPCs [WT and MECP2-KO (clone A33)] across all timepoints, labeled according to sample. (B) Astrocyte signature score across samples. (C) Heatmap of the top differentially expressed genes between WT and MECP2-KO (clone A33) cells, across all time points. (D) (Left) UMAP plot of NPCs (WT and MECP2-KO) across all timepoints, labeled according to MECP2 genotype status; (Right) UMAP plot showing the normalized NEAT1 expression in all NPCs (WT and MECP2-KO). (E) Expression levels of NEAT1 measured by RT-qPCR in WT and MECP2-KO progenitor cells (pg) or after 7 days of free differentiation (7 d). Two different MECP2-KO clones (A33 and 3F) were analyzed. Primers used are depicted in the upper diagram. NEAT1_2 signal corresponds to the long isoform, and NEAT1_1 + 2 signal detects both the short (NEAT1_1) and the long (NEAT1_2) isoforms. Graphs show the mean ± SD of three replicates (****P < .0001, one-way ANOVA). (F) Expression levels of MECP2 and NEAT1 detected by RT-qPCR in WT, MECP2-KO (clone A33) or MECP2-KO cells transfected with MECP2_e1 construct. Graphs show the mean ± SD of three (MECP2 and NEAT1_2 levels) or four (NEAT1_1 + 2) replicates (*P < .05, **P < .01, ****P < .0001, one-way ANOVA). (G) Expression levels of NEAT1 in RTT patient-derived iPS cells (T158M and the isogenic WT control) and in the same cells differentiated into neural progenitor cells (NPCs) for the indicated times. Graphs show the mean ± SD of three (NEAT1_2) or four (NEAT1_1 + 2) replicates (**P < .01, ****P < .0001, ns = not significant, one-way ANOVA). (H) Expression levels of total NEAT1 (both isoforms) in the hippocampus (n = 3), cerebellum (n = 4) and frontal cortex (n = 4) of post-mortem RTT patients or healthy control samples. Graphs show the mean ± SEM of replicates (unpaired t-test, *P < .05, ns = not significant). (I) Expression levels of total NEAT1in the peripheral blood of n = 12 RTT patients. The graph shows the mean ± SEM of replicates (unpaired t-test, *P < .05). See also Supplementary Fig. S1.

We next ranked the transcripts exhibiting the most significant dysregulation upon differentiation and identified the lncRNA NEAT1 as the top downregulated gene when considering all time points (Fig. 1C). This finding was confirmed by an additional scRNA-seq experiment, in which the MECP2-KO 3F clone was analyzed alongside the WT and MECP2-KO A33 clone during a 7-day differentiation experiment (Supplementary Fig. S1D). Single-cell analysis of differentiating neural progenitors revealed a virtual absence of NEAT1 in MECP2-KO cells throughout the differentiation trajectories (Fig. 1D and Supplementary Fig. S1E). NEAT1 is a very abundant lncRNA existing as two isoforms within the cell: a shorter 3.7 kb isoform (NEAT1_1) and a much longer 23 kb long isoform (NEAT1_2), both originating from the same promoter and produced through alternative transcription termination sites (see diagram in Fig. 1E). Downregulation of both NEAT1 isoforms in MECP2-KO clones was confirmed by RT-qPCR in human neural progenitor and differentiated cells (Fig. 1E). Moreover, restoration of MeCP2 levels via ectopic transfection also led to the upregulation of NEAT1 levels, confirming the causal relationship between MeCP2 loss and NEAT1 downregulation (Fig. 1F). Additionally, analysis of RTT patient-derived iPS cells harboring the point mutation MECP2-T158M and its isogenic control revealed comparable levels of NEAT1 expression, which were significantly reduced in the mutant cells during differentiation into neural progenitors (Fig. 1G and Supplementary Fig. S1F). These findings suggest dynamic regulation of NEAT1 by MeCP2 in distinct neural contexts. Given reports of NEAT1 dysregulation in neurological disorders, particularly neurodegenerative diseases such as Alzheimer’s disease and Parkinson’s diseases (PD) [52–54], we further validated altered NEAT1 levels in the hippocampus of post-mortem samples and peripheral blood from patients (Fig. 1H and I). Interestingly, in these mature tissues, NEAT1 levels were upregulated, contrasting with the downregulation observed in neural progenitors and during early stages of differentiation. This underscores NEAT1 impairment as a common feature across various RTT models and tissues, with specific manifestations likely influenced by cell type and developmental stage.

NEAT1 expression exhibits significant variability among cells, and its presence correlates with the negative regulation of mTORC1

Analysis of expression during the differentiation of WT cells revealed a decrease in NEAT1 levels as cells transitioned into neurons (Fig. 2A), consistent with previous findings showing a decline in NEAT1 expression with neuronal activity [55]. Further, scRNA-seq analysis unveiled considerable heterogeneity in NEAT1 expression among progenitor cells and within each time point across the differentiation trajectory when compared for example with MECP2 expression (Fig. 2A). Variance analysis further demonstrated that NEAT1 ranks 45th among the top 1000 most highly variable genes (highlighted in red) in WT NPCs (Supplementary Fig. S2A, standardized variance = 2.9981). When cells from all differentiation timepoints are considered, NEAT1 ranks 237th (Supplementary Fig. S2B, standardized variance = 2.3847). In contrast, MECP2 is not among the most highly variable genes (Supplementary Fig. S2A and B).

Figure 2.

NEAT1 expression is highly variable between cells and its presence correlates with negative regulation of mTORC1. (A) Normalized expression levels for NEAT1 (left) and MECP2 (right) across timepoints in single-cell analysis. (B) (Left) RNA-FISH showing localization of NEAT1 in WT NPCs. Scale bar = 20 μm. (Right) Frequency distribution of NEAT1 RNA-FISH foci per cell. The x-axis represents the number of RNA-FISH foci per cell, and the y-axis indicates the percentage of cells with the corresponding number of foci. Data are represented as the mean ± SD of n = 50 WT NPCs in three independent experiments. (C) Volcano plot showing differentially expressed genes between progenitor cells with high (>60th percentile of normalized NEAT1 expression) and low (<60th percentile) NEAT1 expression. The top differentially expressed genes are highlighted. Dashed lines indicate 0.15 as average log₂ fold change threshold (x axis) and 0.01 as P adjusted value threshold (y axis). For clarity, NEAT1 (log₂FC = 2.4338, P adj. value = 0) and MALAT1 [log₂FC = 0.7312, −log₁₀ (P adj. value = 206.68)] are not included in the plot. (D) (Above) Enriched Gene Ontology terms for the top 20 upregulated genes in cells with high NEAT1 expression as identified by functional clustering (Enrichr). The y axis shows the GO terms and the x axis shows the statistical significance (two-tailed Fisher’s exact test). (Below) Same as above, for the top 20 genes upregulated in cells with low NEAT1 expression. (E) (Left) Volcano plot showing gene expression changes (−10 < log₂FC < 10) in MECP2-R133C point mutant cells relative to WT progenitor cells. Upregulated genes are shown in red, downregulated genes in blue. The dotted vertical line indicates log₂FC = 0, and the dotted horizontal line represents a P-value = .05. (Right) enriched Gene Ontology terms for all differentially expressed genes in R133C cells (P-value < .05) as identified by functional clustering (Enrichr). (F) Enriched Gene Ontology terms for all upregulated genes in R133C cells (FC > 1, P-value < .05) as identified by functional clustering (Enrichr). (G) Expression levels of NEAT1 measured by RT-qPCR in WT and MECP2-R133C progenitor cells (pg) or after 7 days of free differentiation (7 d). NEAT1_2 signal corresponds to the long isoform, and NEAT1_1 + 2 signal detects both the short (NEAT1_1) and the long (NEAT1_2) isoforms. Graphs show the mean ± SD of n = 5 (pg) or n = 3 (7 d) replicates (****P < .0001, one-way ANOVA). See also Supplementary Fig. S2.

RNA-FISH confirmed the high variability of NEAT1 expression among cells in the WT line (Fig. 2B). Prompted by these observations and aiming to elucidate NEAT1’s function in neural progenitors at the cellular level, we categorised cells based on NEAT1 expression levels and identified transcripts exhibiting the highest positive and negative correlation with NEAT1 (as depicted in the Volcano plot in Fig. 2C). Cells expressing high levels of NEAT1 displayed elevated expression of genes involved in catabolic processes and the negative regulation of TOR signaling, including SESN2, SESN3 (which modulate mTORC1 signaling via the GATOR complex), and DDIT4 (which primarily inhibits mTORC1 signaling by targeting the mTORC1 activator, Rheb) (Fig. 2D). In contrast, cells with low NEAT1 expression displayed enrichment in transcripts associated with DNA replication and the cell cycle (Fig. 2D). Given the observed reduction in NEAT1 levels in MECP2 mutant cells, we hypothesized that similar processes might be disrupted in cellular models of RTT. Intriguingly, bulk transcriptomic analysis of human NPCs bearing a common MeCP2 mutation in RTT (R133C) found enrichment in pathways related to translation (Fig. 2E). Specifically, upregulated genes in cells with the R133C mutation were associated with the positive regulation of TOR signaling (Fig. 2F), contrasting with the findings in cells expressing high levels of NEAT1. Notably, these R133C mutant NPCs also exhibited very low levels of NEAT1 (Fig. 2G). Taken together, our observations suggest functional similarities between NEAT1 and MECP2 mutant cells. Subsequently, we aimed to investigate the potential functional involvement of NEAT1 in the MECP2 phenotype.

Transcriptomic alterations in NEAT1-KO neural cells are similar to those observed in MECP2-KO cells

To develop a cellular model suitable for studying NEAT1’s role in early neural differentiation, we employed a CRISPR/Cas9 strategy to generate NEAT1-KO cells. NEAT1-KO NPCs displayed depletion levels of NEAT1 comparable to those observed in MECP2-KO cells (Fig. 3A). Additionally, we established stable cell lines expressing two different shRNAs, which achieved a moderate but significant reduction in NEAT1 levels (Fig. 3A). Interestingly, NEAT1 depletion did not result in marked alterations in MECP2 expression, at either the RNA or protein level (Fig. 3B and Supplementary Fig. S3A), indicating that NEAT1 acts downstream of MECP2 and does not regulate MECP2 levels in human NPCs, contrary to findings in some cancer models [56]. RNA-FISH confirmed the depletion of NEAT1 expression in NEAT1-KO cells, reducing levels to those observed in MECP2-KO cells (Fig. 3C). The functional impact of NEAT1 depletion was further validated through the analysis of α-synuclein levels (SNCA), a known target of NEAT1 [57, 58]. SNCA levels were diminished in MECP2-KO, NEAT1-KO, and NEAT1-shRNA cells (Supplementary Fig. S3B) and were upregulated upon MeCP2 overexpression, demonstrating a positive correlation with NEAT1 levels (Supplementary Fig. S3C). In contrast, the expression of MALAT1, another highly abundant ncRNA that localizes to nuclear speckles and is located 50 kb downstream of the NEAT1 locus, remained unaltered following genomic edition of the NEAT1 promoter (Supplementary Fig. S3D and S3E, as shown by both RNA-FISH and RT-qPCR approaches).

Figure 3.

NEAT1-KO neural cells display transcriptomic alterations similar to MECP2-KO cells. (A) NEAT1 depletion levels in MECP2-KO (clones A33 and 3F), NEAT1-KO cells edited by CRISPR/Cas9, and in NEAT1 shRNA-depleted cells were assessed by RT-qPCR. Graphs show the mean ± SD of three replicates (**P < .01, ****P < .0001, one-way ANOVA). Diagram was created in BioRender. Soler (2025) https://BioRender.com/o10b971. (B) (Left) MECP2 mRNA levels measured by RT-qPCR. Graph shows the mean ± SD of three replicates (*P < .05, ***P < .001, ****P < .0001, ns = not significant, one-way ANOVA). (Right) Western blot analysis of MeCP2 in WT or KO progenitor cells. Quantification of band intensity is shown below each lane. (C) (Left) RNA-FISH showing abundance of NEAT1 in WT, MECP2-KO (clones A33 and 3F) or NEAT1-KO NPCs. Scale bar = 10 μm. (Right) Quantification of NEAT1 presence per cell nucleus. Graph represents the mean ± SEM of the number of nuclear NEAT1+ foci in n = 9 cells (*P < .05, **P < .01, ****P < .0001, one-way ANOVA). (D) Heatmaps of differentially expressed transcripts (DETs) (|log₂FC| ≥ 1, q value ≤ 0.05) in NEAT1-KO and MECP2-KO (clone A33) NPCs (top) and 7-day differentiated cells (bottom) relative to WT cells. The number of condition-specific or shared DETs is shown to the left of the heatmaps together with the P-value from the hypergeometric test for the overlap between the conditions. (E) Enriched KEGG pathways for all differentially expressed genes (q value ≤ 0.05) in NEAT1-KO NPCs relative to WT cells, as identified by functional clustering (Enrichr). See also Supplementary Fig. S3.

Remarkably, bulk RNA-seq analysis of WT, MECP2-KO, and NEAT1-KO cells revealed a significant overlap in the dysregulated genes when comparing MECP2-KO and NEAT1-KO neural progenitor or freely differentiated cells to the WT condition (Fig. 3D). Specifically, 53% of DETs in MECP2-KO progenitor cells (1770 out of 3332) and 47% in MECP2-KO 7 days differentiated cells (2307 out of 4944) were also dysregulated in NEAT1-KO cells, indicating that the absence of NEAT1 contributes significantly to the MECP2-KO phenotype, particularly in this cellular context (Supplementary Fig. S3F). These percentages remained consistent when analyzing upregulated and downregulated transcripts separately (Supplementary Fig. S3G and S3H). KEGG enrichment analysis of the dysregulated genes in NEAT1-KO cells highlights the impact of NEAT1 loss on neurological diseases, including PD, Huntington’s disease, and other neurodegenerative disorders, underscoring the crucial role of NEAT1 in neural function (Fig. 3E), as previously reported [59, 60].

The downregulation of NEAT1 in MECP2-KO cells impacts protein metabolism, vesicle trafficking, and mitochondrial function in human neural cells

To gain insight into the potential effects of NEAT1 deficiency on the mTOR pathway and protein metabolism, as suggested by transcriptomic and enrichment analysis, we examined the levels of key members of the pathway. DDIT4, which positively correlated with NEAT1 expression in our scRNA-seq analysis (Fig. 2C), serves as a crucial mediator of the cellular stress response by inhibiting mTORC1 activity. Additionally, in neurons, DDIT4 is required for normal neuronal migration during embryonic brain development [61]. Western blot analysis revealed aberrantly low levels of DDIT4 protein in both MECP2-KO and NEAT1-KO NPCs (Fig. 4A). Simultaneously, activation of ribosomal protein S6, which triggers protein translation initiation and whose phosphorylation serves as a readout for mTORC1 activity and protein synthesis, was significantly increased in both mutant cell types, despite a mild decrease in total mTOR levels (Fig. 4B). An abnormal increase of mTOR-S6K signaling has previously been observed in RTT patients [62], although the molecular mechanisms underlying this phenomenon remain unclear. Importantly, ectopic expression of the full-length NEAT1_2 isoform (22 kb) in MECP2-KO cells restored S6 inhibition (Fig. 4B and Supplementary Fig. S4A–C), indicating that NEAT1 plays a pivotal role in mTOR pathway inhibition and that NEAT1 downregulation may lead to an aberrant increase in protein synthesis. Elevated levels of protein synthesis can sometimes result in the presence of aberrant protein aggregates within the cell and were evidenced by ProteoStat^® staining in both MECP2-KO and NEAT1-KO or NEAT1-shRNA cells (Fig. 4C). To further corroborate the importance of NEAT1 absence in driving protein aggregation in MECP2-KO cells, we employed the CRISPR/SAM approach to reactivate endogenous gene expression [41, 63]. The CRISPR/SAM system utilizes a catalytically inactive Cas9 (dCas9) fused to a transcriptional activator domain (such as VP64) to precisely target specific genomic loci and upregulate gene expression through recruitment of transcriptional machinery. To this end, we tested several NEAT1-specific sgRNAs (Supplementary Fig. S4D and E). Using three different sgRNAs directed at the NEAT1 promoter, we observed a decrease in protein aggregation levels in MECP2-KO cells, confirming NEAT1 defects as the underlying cause of aberrant protein aggregation in these cells (Fig. 4D).

Figure 4.

NEAT1 downregulation in MECP2-KO cells impacts protein metabolism, vesicle trafficking and mitochondrial function of human neural cells. (A) Western blot analysis of DDIT4 levels in MECP2-KO (clone A33) and NEAT1-KO progenitor cells. Diagram was created in BioRender. Soler (2025) https://BioRender.com/i12q588. (B) Western blot analysis of mTOR, total S6 and phosphorylated-S6 levels in NEAT1-KO, MECP2-KO and MECP2-KO overexpressing NEAT1 progenitor cells. Quantification of band intensity is shown below each lane. (C) (Left) Assessment of protein aggregation in WT or MECP2-KO (clones A33 and 3F) NPCs with ProteoStat^® dye. Graph shows the mean ± SD of three replicates (****P < .0001, one-way ANOVA). (Right) Same assay in NEAT1-KO or NEAT1-shRNA depleted cells. Graph shows the mean ± SD of four replicates (**P < .01, ****P < .0001, one-way ANOVA). (D) Protein aggregation quantified with ProteoStat^® dye in WT, MECP2-KO (clone A33) cells (empty) or MECP2-KO cells with NEAT1 reactivation with three different sgRNAs. The graph shows the mean ± SD of three replicates (****P < .0001, one-way ANOVA). Diagram was created in BioRender. Soler (2025) https://BioRender.com/r99k484. (E) (Left) Representative transmission electron micrographs of WT, MECP2-KO (clones A33 and 3F), NEAT1-KO or MECP2-KO (A33) with NEAT1_2 overexpression or reactivation by means of CRISPRa (sgRNA8) progenitor cells. Mitophagy is indicated by arrows. Scale bars = 1 μm. (Right) Quantification of the number of endosomes in the same conditions. n = 30 micrographs, (**P < .01, ****P < .0001, ns = not significant, one-way ANOVA). (F) Histogram overlay of MitoTracker™ intensity (FITC detection) to compare mitochondria presence in WT, MECP2-KO (clones A33 and 3F) and NEAT1-KO NPCs. (G) Graphs of the mean fluorescence intensity of TMRE (membrane potential), ATP Tracker (ATP production), and MitoSOX (oxidative stress) measured by the PE-A filter, normalized with FITC-A detection of MitoTracker™. Graphs show the mean ± SD of three replicates (ns = not significant, *P < .05, **P < .01, ***P < .001, ****P < .0001, one-way ANOVA). (H) Expression levels of NEAT1, SNCA and MECP2 mRNAs were analyzed by RT-qPCR in NPCs treated with the indicated compounds. Graphs represent the mean ± SD of four independent replicas (ns = not significant, *P < .05, **P < .01, one-way ANOVA). See also Supplementary Figs S4 and S5.

Using these tools, we compared the ultrastructural features of WT NPCs with those of MECP2-KO and NEAT1-KO cells. All KO cells displayed multiple defects, including mitochondria with high electron density matrices, indicative of hyperactivity, as well as an abnormally high number of vesicles, including endosomal and autophagosomal structures. Notably, autophagosomes with multilamellar membranes were observed forming and fusing with mitochondria organelles (yellow arrows), ultimately leading to mitophagy (Fig. 4E and Supplementary Fig S5A). This suggests a potential overload of mitochondria and/or a slow or defective autophagic process. Consistent with this, a previous study in blood cells from RTT patients had proposed that mitochondria are not properly eliminated in RTT due to autophagy defects [64]. Abnormal mitochondrial function was further confirmed by a Mito-SOX-based assay and flow cytometric analysis, which detects mitochondrial reactive oxygen species (ROS). RTT cells are known for increased ROS levels [65, 66], and interestingly, while MECP2-KO and NEAT1-KO cells showed a similar mitochondrial mass staining (MitoTracker, Fig. 4F, and AIF measurement, Supplementary Fig. S5B and C), we detected increased ROS levels compared to control cells (Fig. 4G). Instead, mitochondrial activity appeared impaired, as indicated by similar alterations in mitochondrial membrane potential and ATP production observed in MECP2-KO, NEAT1-KO, and NEAT1-shRNA cells (Fig. 4G). This finding corroborates the hyperactivity suggested by ultrastructure imaging (Fig. 4E). Taken together, these results suggest that NEAT1 downregulation following MeCP2 loss of function may contribute to mitochondrial abnormalities and the disruption of mitochondrial homeostasis in RTT.

The presence of abnormal protein accumulation, a higher number of vesicles, and mitochondrial impairment strongly suggest a significant involvement of NEAT1 in protein metabolism and related pathways. Therefore, we next investigated whether commonly used inhibitors of the mTOR and autophagy processes affected NEAT1 levels in WT cells. Treatment with the mTOR inhibitor Rapamycin, at levels sufficient to induce S6 dephosphorylation (Supplementary Fig. S5D), did not significantly alter lysosomal or autophagic gene expression and only had a marginal impact on NEAT1 levels (Supplementary Fig. S5E and Fig. 4H). In contrast, treatment with BafilomycinA1, which impedes lysosomal acidification and blocks autophagic flux, leading to the accumulation of autophagosomes in the cell, resulted in enhanced LC3-II protein levels as expected (Supplementary Fig. S5D). Notably, it also induced a sharp increase in NEAT1 levels, along with elevated levels of its downstream target SNCA (Fig. 4H). Conversely, MECP2 expression levels remained unaltered in response to both compounds. These findings suggest that NEAT1 functions in response to altered autophagic dynamics and further confirm the role of this lncRNA in essential cellular processes in NPCs.

Given that NEAT1 is upregulated in response to deficits in autophagic flux and considering the close relationship between autophagy and mitochondrial function in maintaining cellular homeostasis, energy balance, and overall cell health, we proceeded to compare the responses of cells to these same inhibitors when NEAT1 is absent (MECP2-KO and NEAT1-KO cells) by measuring key readouts of mitochondrial function. Compared to WT cells, treatment with BafilomycinA1 failed to elicit changes in oxidative stress-related processes in either MECP2-KO or NEAT1-KO cells, as indicated by the MitoSOX assay (Supplementary Fig. S5F). Similarly, ATP production increased only in WT cells but not in MECP2-KO or NEAT1-KO cells (Supplementary Fig. S5G). In contrast, the membrane potential (measured using TMRE) remained unchanged under all conditions (Supplementary Fig. S5H). Overall, these findings suggest that the downregulation of NEAT1 as a consequence of MECP2 mutation hinders the cellular response to autophagy inhibitors, underscoring the importance of NEAT1 in maintaining mitochondrial function and cellular homeostasis.

Next, we aimed to explore the molecular mechanisms underlying NEAT1’s involvement in autophagy and how this process is compromised in MECP2 mutant cells.

Reactivation of NEAT1 in MECP2-KO cells restores the levels of autophagy-related proteins

To gain more insight into the status of autophagic flux in the mutant cells, we employed a mCherry-GFP-LC3 reporter to visualize vesicle maturation trafficking, enabling the differentiation between autophagosomes and autolysosomes [43]. Autophagosomes exhibit both GFP and mCherry fluorescence, producing a green and yellow signal, while autolysosomes are labeled red due to the inability of GFP to maintain its native conformation in the acidic lysosomal environment (Fig. 5A). Following 16 h of autophagy induction, both MECP2 and NEAT1-KO cells exhibit a similar increase in the ratio of autolysosomes to autophagosomes compared to WT cells (Fig. 5B), validating abnormal dynamics of the autophagic flux in the mutant cells and implying that the defects in MECP2-KO cells result from NEAT1 downregulation. Analysis of lysosomal markers LAMP1 and LAMP2 (integral membrane glycoproteins primarily localized to the lysosome and late endosome membranes) further confirmed abnormal accumulation in MECP2-KO cells, both at the progenitor stage and during differentiation (Fig. 5C), which could be reversed by ectopic overexpression of MeCP2 protein (Fig. 5D). Interestingly, similar LAMP1 accumulation was observed in NEAT1-KO and NEAT1-shRNA cells, as evidenced by both western blot and immunofluorescence assays (Fig. 5E and F).

Figure 5.

NEAT1 reactivation in MECP2-KO cells rescues the levels of autophagy-related family of proteins. (A) Top diagram: Autophagic vesicle maturation can be monitored in cells expressing the mCherry-GFP-LC3 reporter, which is sensitive to the increased acidic conditions in autolysosomes relative to autophagosomes. Concomitant expression of GFP- and mCherry-fused LC3 in autophagosomes is visualized as yellow–green, whereas only the expression of mCherry is allowed by acidic lysosomes and thus autolysosomes are visualized as red fluorescence. Bottom: representative images of WT, MECP2-KO (clones A33 and 3F) and NEAT1-KO NPCs transfected with the mCherry-GFP-LC3 reporter after autophagy induction (through mTOR inhibition) by administration of Rapamycin (50 μM) for 16 h. Diagram was created in BioRender. Soler (2025) https://BioRender.com/z98e924. (B) Quantification of the number of autophagosomes and autolysosomes in WT, MECP2-KO (clones A33 and 3F) and NEAT1-KO cells (n ≥ 12 cells per condition, **P < .01, ***P < .001, ns = not significant, one-way ANOVA). (C) Western blot analysis of lysosomal markers LAMP1 (left) or LAMP2 (right) levels in WT or MECP2-KO cells (clone A33) at the progenitor stage or upon free differentiation (times are indicated). (D) Western blot analysis of lysosomal markers LAMP1 and LAMP2 in WT or MECP2-KO (clone A33) NPCs, or upon transfection of MeCP2_e1 isoform. MeCP2 levels are also blotted for reference. (E) (Left) LAMP1 and LAMP2 levels assessed by western blot in WT, NEAT1-KO or MECP2-KO (clone A33) cells. (Right) LAMP1 and LAMP2 levels assessed by western blot in control and NEAT1-shRNA1 depleted cells. (F) (Left) LAMP1 levels assessed by immunofluorescence in WT, MECP2-KO (clones A33 and 3F) or NEAT1-KO NPCs. (Right) Quantification of the LAMP1 immunofluorescence signal, where the graph represents the mean ± SD of n = 6 images (***P < .001, one-way ANOVA) (G) Western blot analysis of autophagy-related proteins in WT or MECP2-KO (clones A33, left, and 3F, right) NPCs transfected with empty vector or MeCP2_e1 isoform. MeCP2 levels are also blotted for reference. (H) RT-qPCR analysis of the mRNA of the same genes shown in (G). Graphs represent the mean ± SD of four independent replicas (*P < .05, **P < .01, ****P < .0001, ns = not significant). (I) Western blot analysis of autophagy-related proteins in WT or NEAT1-KO (left), or in NEAT1-shRNA1 depleted (right) NPCs. (J) Western blot analysis of autophagy-related proteins in WT or MECP2-KO (clones A33, left, and 3F, right) NPCs transfected with empty vector or two different sgRNAs to reactivate NEAT1 expression. (K) RT-qPCR analysis of the mRNA of autophagy-related genes shown in (J). Graphs represent the mean ± SD of three independent replicates (**P < .01, ***P < .001, ****P < .0001, ns = not significant). (L) Western blot analysis of autophagy-related proteins in WT or NEAT1-KO NPCs transfected with empty vector, NEAT1_2 or MALAT1 vectors. Quantification of band intensity is shown below each western blot lane. See also Supplementary Fig. S6.

Accumulation of autolysosomes and other undegraded autophagic structures can result from various defects, including deficiencies in cargo sequestration into autophagosomes, impairment in autophagosome formation, and improper fusion with lysosomes. Autophagy-related (ATG) proteins play crucial roles in all these stages of the autophagy process, encompassing autophagosome assembly, cargo recognition, and fusion with lysosomes, all of which may contribute to autolysosome accumulation. Therefore, we next investigated ATG protein levels following MECP2 KO. As shown in Fig. 5G, several ATG proteins, including ATG3, ATG16L1, ATG16L2, ATG12-ATG5, and Beclin-1/ATG6, were decreased in MECP2-KO cells, with a minor reduction observed for ATG7. Overexpression of MeCP2 restored these deficits (Fig. 5G), confirming the role of MeCP2 in regulating ATG family protein levels. However, these protein-level changes did not correlate with changes in the respective mRNAs, which were largely unaffected by either MeCP2 downregulation or overexpression, except for ATG16L1 mRNA, which was significantly upregulated in MECP2-KO clones (Fig. 5H). These findings suggest that MeCP2 regulates ATG proteins through mechanisms beyond transcriptional control.

To evaluate the contribution of NEAT1 in the downregulation of ATG proteins, we examined protein and mRNA levels in NEAT1-depleted cells. Remarkably similar to MECP2-KO cells, a pronounced reduction in protein levels was observed for ATG3, ATG16L1, ATG16L2, ATG12-ATG5, and Beclin-1, with less evident changes for ATG7 (Fig. 5I). Interestingly, mRNA levels remained unaltered (Supplementary Fig. S6A), indicating that similar regulatory mechanisms are at play in both MECP2- and NEAT1-mutant lines. Examination of RNA-seq data from WT, MECP2-KO, and NEAT1-KO cells also revealed no decrease in the mRNA levels of autophagy genes (Supplementary Fig. S6B). Crucially, reactivation of NEAT1 in MECP2-KO cells using CRISPR/SAM with different sgRNAs successfully restored deficits in ATG protein levels, again independently of mRNA levels (Fig. 5J and K). Furthermore, overexpression of NEAT1, but not MALAT1 (used as a control), rescued ATG protein levels in NEAT1-KO cells (Fig. 5L and Supplementary Fig. S6C). Altogether, these findings confirm NEAT1 as the primary mediator of MeCP2 regulation of autophagy-related genes and the principal contributor to impaired autophagic flux in MECP2-KO neural cells.

NEAT1 interacts with mRNAs of autophagy genes and promotes their nuclear retention

Considering the observed impact on the autophagy process following the loss of NEAT1 and the unaltered steady-state levels of ATG gene mRNAs (Fig. 5), we investigated their subcellular localization. Previous studies have suggested that one important paraspeckles/NEAT1 function is to retain certain mRNAs encoding mitochondrial protein components (mito-mRNAs) in the nucleus [67]. In addition, direct RNA targets of NEAT1 have been reported to preferentially bind to the 5′ end of NEAT1 [45]; however, the scope and specific mRNAs subject to such regulation remain unclear. To investigate NEAT1’s potential role in regulating ATG protein expression via mRNA metabolism, we performed a RIP-seq experiment in WT NPCs. Cells were crosslinked, and endogenous NEAT1 was pulled down using antisense probes targeting the 5′ (common) region of NEAT1 isoforms. Associated RNAs were then purified and analyzed by RT-qPCR. The NEAT1-AS2 probe effectively retrieved endogenous NEAT1, recovering ∼7% of the total NEAT1 RNA, while the NEAT1-AS1 probe showed no significant NEAT1 retrieval (Fig. 6A and Supplementary Fig. S7A). Subsequently, RIP-seq analysis of RNAs from NEAT1-AS2 pull-down and control experiments (performed in triplicates) identified enriched transcripts. Gene ontology analysis of the bound transcripts revealed enrichment for categories such as mRNA processing, splicing, and transport, consistent with previous reports [45], along with other processes like ubiquitin-mediated proteolysis, autophagy, and endocytosis (Fig. 6B). Notably, among autophagy-related transcripts, ATG16L2 exhibited a distinct RIP-seq peak signal with the NEAT1-AS2 probe (Fig. 6C), whereas other ATG transcripts showed no specific enrichment (Supplementary Fig. S7B). Further, in silico predictions of direct RNA–RNA interactions using IntaRNA [68–71] (http://rna.informatik.uni-freiburg.de/IntaRNA/) suggested energetically stable base-pairing (−28.8 kcal/mol) between the canonical ATG16L2 transcript (NM_033388.2) and a region within NEAT1 near the NEAT1-AS2 probe (positions 192–242 of ATG16L2 mRNA; Fig. 6D). RT-qPCR confirmed that this region near the 5′UTR of ATG16L2 mRNA was enriched in the NEAT1 pull-down, with ∼3% of input ATG16L2 mRNA recovered. Given that only ∼7% of NEAT1 was retrieved, this suggests a significant interaction between ATG16L2 mRNA and NEAT1. None of the other tested ATG mRNAs were present in the pull-down with comparable enrichments, except for one ATG5 isoform (NM_001286106.2), whose 5′UTR was also enriched in the pull-down (Fig. 6E). Predicted base-pairing interactions between NEAT1 and this specific ATG5 region were also energetically stable (Supplementary Fig. S7C). These findings suggest that NEAT1 selectively interacts with autophagy-related mRNAs, particularly ATG16L2, and, to a lesser extent, a specific ATG5 isoform, likely through direct RNA–RNA interactions.

Figure 6.

NEAT1 directly interacts with mRNAs of autophagy-related genes and promotes their nuclear retention. (A) Diagram summarizing the strategy for the NEAT1 pulldown experiment. Cells were crosslinked and endogenous NEAT1 was retrieved with biotinylated, antisense probes. RNAs bound to NEAT1 were amplified by next-generation sequencing or RT-qPCR analysis. Diagram was created in BioRender. Soler (2025) https://BioRender.com/w62q250. (B) Top 10 enriched functional categories for NEAT1-AS2 probe peak-related genes (threshold q value = 0.05) from RIP-seq analysis. (C) RIP-seq peak signal profiles across ATG16L2 (GRCh38/hg38 chr11:72 814 411–72 829 635). (D) Predicted interaction between ATG16L2 (transcript NM_033388.2) with NEAT1 in the 500 nt region flanking the AS2 probe. The interaction was predicted using IntaRNA (http://rna.informatik.uni-freiburg.de/IntaRNA/) with default parameters, including a minimum seed length of 7 nucleotides. (E) RIP-qPCR analysis of the indicated autophagy-related genes. For each transcript, several amplicons were tested, whose locations are indicated above each graph. The graphs represent the mean ± SD of three independent experiments. The enrichments are normalized by the input material and shown relative to scramble (scr) probe. One-way ANOVA test was used, *P < .05, **P < .01, ****P < .0001, ns = not significant. (F) (Above) RNA-FISH showing localization of NEAT1 and ATG16L2 mRNAs in WT, NEAT1-KO, MECP2-KO (clones A33 and 3F) or MECP2-KO clones overexpressing full-length NEAT1_2 progenitor cells. NESTIN antibody and DAPI staining were used at the same time to visualize the cells. For each cell condition, the white-squared inset is amplified on the right panel. White arrows indicate cytoplasmic signals. Scale bars = 10 μm. (Below) Quantification of ATG16L2 RNA-FISH signal. Total number of foci were counted in each cell and the percentage of cytoplasmic foci per cell is represented. Graphs show the mean ± SEM of n ≥ 18 cells (*P < .05, ***P < .001, ns = not significant, one-way ANOVA). (G) RT-qPCR assessment of the nuclear-cytoplasmic distribution of the indicated autophagy-related and mitochondrial protein-encoding mRNAs. Biochemical fractionation was done in WT, NEAT1-KO and MECP2-KO (clone A33) cells transfected with empty or sgRNAs to reactivate NEAT1. Percentage of the mRNA present in each fraction relative to total mRNA is shown. Graphs represent the mean ± SD of four independent replicas. See also Supplementary Fig. S7.

Since NEAT1 is able to dictate target RNA localization, RNA in situ hybridization assays were next conducted to visualize the localization of ATG16L2 mRNA in NPCs (Fig. 6F). In WT cells, this mRNA was abundantly detected in the nucleus, whereas in MECP2-KO or NEAT1-KO cells, it was frequently found in the cytoplasm. Notably, overexpression of the long NEAT1 isoform (NEAT1_2) in MECP2-KO cells restored nuclear retention (Fig. 6F). Intriguingly, although not detected in the pull-down assay, RNA-FISH experiments revealed a similar pattern for ATG16L1: the mRNA was localized to the nucleus in WT cells, often overlapping with NEAT1 signal, while both MECP2-KO and NEAT1-KO cells exhibited a more abundant cytoplasmic signal, which could be partially reverted by overexpression of NEAT1 (Supplementary Fig. S7D).

To quantitatively assess the effect on mRNA localization, we conducted biochemical cellular fractionation, revealing that ATG5, ATG16L1 and 2, and BECLIN/ATG6 mRNAs were abundantly localized in the nucleus of WT NPCs, with ATG16L2 mRNA displaying particularly striking nuclear enrichment (up to 90%). This nuclear localization in WT cells may serve as a rapid response mechanism for increased protein production during physiological stress. Interestingly, both NEAT1-KO and MECP2-KO cells showed significantly lower nuclear levels of these mRNAs. Similar patterns were observed for the mito-mRNAs CPT1A and SMURF1 [67], while the localization of MALAT1 or GAPDH remained unaltered in MECP2-mutant cells (Fig. 6G and Supplementary Fig. S7E). When considering the impact of this relocation relative to the amount of nuclear mRNA in WT cells, we observed that up to 20% of nuclear mRNA for ATG genes was redistributed to the cytoplasm in MECP2-KO cells. For comparison, 25% and 30% of nuclear CPT1A and SMURF1 mito-mRNAs, respectively, were relocated (Supplementary Fig. S7F). Importantly, CRISPR-mediated reactivation of NEAT1 using two different sgRNAs in MECP2-KO cells restored the nuclear levels of all mRNAs to those of WT cells (Fig. 6G and Supplementary Fig. S7F). The impact of NEAT1 downregulation on mRNA localization was specific, as the localization of other ATG mRNAs, such as ATG3, ATG4B, ATG7, ATG12, and ATG13, remained unaltered, as did LAMP1/2 mRNAs and other mito-mRNAs (Supplementary Fig. S7G). Overall, this suggests that the specific nuclear retention of ATG mRNAs may not be limited to transcripts interacting with NEAT1, even though this lncRNA appears to regulate the nuclear-to-cytoplasmic localization of many ATG mRNAs. It is noteworthy that WT cells display higher nuclear levels of ATG mRNAs than MECP2-KO or NEAT1-KO cells, while the corresponding proteins are present at lower levels in both KO cells. This counterintuitive finding illustrates how the correct trafficking trajectory and nuclear history of an mRNA may impact its subsequent translation and protein level.

Overall, these findings indicate that the downregulation of NEAT1 plays a significant role in the dysfunction of the autophagic process observed in MECP2-KO cells. The abnormal accumulation of autolysosomes may result from impaired ATG expression and autophagic complex biogenesis, potentially leading to the accumulation of misfolded proteins and damaged organelles, as observed in our ultrastructural analysis.

Re-expression of NEAT1 reverses the morphological defects of MECP2-KO neurons

Given the fundamental roles of autophagy and mitochondrial function in cellular homeostasis, the impact of NEAT1 on these processes in the context of MECP2 loss of function suggests a broader contribution of NEAT1 to the RTT phenotype. Therefore, we further investigated how NEAT1 contributes to the morphological phenotypes of MECP2-KO cells. We and others have previously described that neurosphere formation is defective in human MeCP2 mutant cells [30, 72], a phenomenon we observed in NPCs depleted of NEAT1. Crucially, the reintroduction of NEAT1 in MECP2-KO cells was able to reverse this defect (Fig. 7A). Furthermore, as our progenitor cells underwent glutamatergic differentiation, structural analysis revealed similar defects in neural complexity for MECP2-KO and NEAT1-KO cells, characterized by reduced branching (Fig. 7B and Supplementary Fig. S8A) and decreased filopodia count (Fig. 7C and Supplementary Fig. S8A). Both mutant cells exhibited shorter neurite projections, with branches and dendritic filopodia closer to the soma compared to WT cells (Fig. 7D), reminiscent of the morphological features observed in RTT neurons [73]. Considering these findings, we further evaluated the potential of NEAT1 recovery to ameliorate the morphological phenotype of RTT neurons. Endogenous reactivation of NEAT1 through CRISPR/SAM did not fully rescue the number of branches but did restore the total filopodia count (Fig. 7E). However, ectopic overexpression of NEAT1 showed improvement in both aspects (Fig. 7F and Supplementary Fig. S8B), underscoring the critical role of NEAT1 in neuronal maturation.

Figure 7.

NEAT1 reexpression restores the morphological defects of MECP2-KO neurons. (A) Diameter length of neurospheres was measured for WT, NEAT1-KO or MeCP2-KO (clones A33 and 3F) cells (left) and upon transfection of an empty vector or a vector carrying NEAT1_2 (right). Graphs represent the mean ± SEM of at least 12 measurements (****P < .0001, one-way ANOVA). Images correspond to representative spheres of each condition. Scale bar = 0.25 mm. (B) (Left) Control WT neural progenitors, MECP2-KO (clone A33) or NEAT1-KO cells were driven towards glutamatergic differentiation for 7 days, stained with MAP2 and reconstructed in silico from confocal images with the NeuronStudio software. One representative picture for each condition is shown. (Right) Automatic analysis of the cells in (B) allowed total branch points count per each condition (graphs represent mean ± SEM, n = 15 neurons, *P < .05, one-way ANOVA). Scale bar = 10 μm. (C) For the same cells as in (B), the total filopodia count and the abundance of each filopodia subtype was determined with the NeuronStudio software (n = 15 neurons, **P < .01, ****P < .0001, one-way ANOVA). (D) The same quantification in panels (B) and (C) was represented in the histograms as distance from the soma. (E) Morphological analysis as in panels (B) and (C) with WT or MECP2-KO (clone A33) differentiated cells expressing empty vector or sgRNAs to reactivate NEAT1. Graphs represent mean ± SEM, n = 10 neurons (*P < .05, **P < .01, ***P < .001, ****P < .0001, ns = not significant, one-way ANOVA). (F) Morphological analysis as in panel (E) with MECP2-KO (clone A33) differentiated cells expressing NEAT1_2 or an empty vector as control. Graphs represent mean ± SEM, n = 15 neurons (*P < .05, ****P < .0001, two-tailed Mann–Whitney U test). See also Supplementary Fig. S8.

Overall, this study reveals the substantial role of the abundant lncRNA NEAT1, significantly downregulated in MECP2 mutant cells, as a crucial downstream mediator of MeCP2 function. NEAT1 regulates pathways controlling protein metabolism, ATG protein biogenesis, and autophagic flux. Consequently, NEAT1 deficiency profoundly impacts ultrastructure and morphological aspects in RTT. Crucially, NEAT1 reactivation restores autophagy and neuronal morphology, highlighting its pivotal role as a mediator of key aspects in RTT pathophysiology (Fig. 8).

Figure 8.

Schematic summarizing the findings in this work. (Left) In control NPCs, the lncRNA NEAT1 regulates protein metabolism by reducing the activation of S6 protein through regulation of mTORC1 and by directly interacting with autophagy-related transcripts, promoting their biogenesis and translation. NEAT1 responds to stress stimuli and positively influences the autophagic response. (Right) In MECP2-KO cells, the marked downregulation of NEAT1 expression causes hyperactivation of S6 and protein accumulation. At the same time, dysregulation of autophagy-related genes results in the blockage of the autophagic flux, loss of protein homeostasis, and mitochondrial dysfunction, which can be observed at the functional and structural level. Altogether, neurons derived from mutant NEAT1 or MECP2 cells display similar morphological defects, which can be rescued by the reactivation of NEAT1 expression. This suggests a central role for NEAT1 in mediating the metabolic dysregulation typical of RTT. The dashed arrow indicates as yet unclear mechanisms of regulation. Diagram was created in BioRender. Soler (2025) https://BioRender.com/e64d138.

Discussion

Despite significant advances in understanding MeCP2’s role in the cell and the consequences of its loss of function in neurodevelopment, particularly in RTT, there is sparse literature detailing the molecular mechanisms underlying dysregulated cellular pathways in RTT. Loss of function in MECP2 mutant cells leads to various metabolic defects, including impaired glucose and cholesterol metabolism, disrupted mitochondrial homeostasis, and redox imbalance, as observed in patients and animal models of the disorder [74–76]. One critical pathway linking MeCP2 function and cellular metabolism is mTORC1 signaling, which modulates synaptic plasticity and learning and is increasingly recognized as a hallmark of RTT across different study models [62, 77, 78]. However, the mechanistic details of how MeCP2 controls metabolic dysfunction are underdefined. In this study, we aimed to address this research gap and integrate scRNA-seq analysis, revealing significant downregulation of NEAT1 in MECP2 mutant cells. While single-cell transcriptomic analysis has been applied to study mosaicism in female mouse models and postmortem human Rett brain tissue [26], our study represents the first investigation of a human cellular model of RTT during early differentiation stages at the single-cell level. This allowed the identification of NEAT1 as an important MECP2-regulated gene, a functional link not previously reported in RTT. Notably, in murine models of Huntington’s disease and spinocerebellar ataxias, elevated Neat1 levels were associated with proposed protective roles against cytotoxicity, with MeCP2 protein suggested to repress Neat1 function through direct protein:RNA binding [79]. This indicates a different mechanistic scenario from the downregulation of NEAT1 observed in MECP2-KO cells and underscores the positive contribution of NEAT1 downstream of MeCP2 function in our RTT cellular model.

The significant decrease in NEAT1 expression in MECP2 mutant cells in both NPCs and throughout all differentiation time points assessed here is an unusual finding in RTT cells, where individual gene expression changes typically exhibit small magnitudes of misregulation [80]. In contrast, NEAT1 upregulation is also prominent in post-mortem tissue, with a fold-change of ∼8 in the hippocampus (Fig. 1H). This may reflect the differences in cellular composition, as well as variations in developmental stages and the pathological environment, suggesting context-dependent regulation of NEAT1. This highlights the importance of studying different conditions and stages of neural development in RTT physiopathology. Altogether, NEAT1 expression may play a crucial role in the neurological phenotype of RTT, supported by the high percentage (53%) of dysregulated transcripts in MECP2-KO progenitor cells also dysregulated in NEAT1-KO cells (Fig. 3D and Supplementary Fig. S3E). Similar dysregulation of NEAT1 has been observed in other neurological diseases such as Alzheimer’s disease, PD, and Huntington’s disease [53, 54, 81, 82], all characterized by abnormal protein aggregation, although the specific role of NEAT1 in these contexts remains incompletely understood. NEAT1 is implicated in promoting the nuclear retention of TDP-43, a protein central to many proteinopathies [83], yet the mechanisms by which elevated NEAT1 levels prevent neurodegeneration are unclear [84]. While NEAT1 overexpression has shown neuroprotective effects in Huntington’s disease [79], its inhibition appears beneficial in Alzheimer’s models [85, 86]. Elucidation of NEAT1’s mechanism of action in the diverse neurological contexts is therefore highly relevant. Along these lines, the observed drastic downregulation of NEAT1 in neural cells contrasts with its upregulation in the mature diseased brain and suggests dynamic changes during neurodevelopment. However, given the central role of NEAT1 in metabolic and autophagic control, and its ability to rescue RTT morphological deficits, this upregulation may serve as a compensatory response to pathological conditions; for example, while typically absent in the healthy brain, NEAT1_2 and associated paraspeckles are often increased in response to cellular stress and inflammation [87]. Given that RTT and neurodegenerative diseases like Alzheimer’s disease or amyotrophic lateral sclerosis are associated with cellular stress, neuroinflammation, and altered RNA metabolism, NEAT1 upregulation could serve as a protective mechanism to mitigate disease-related dysfunction and promote neuronal survival. Additionally, while Neat1-KO mice do not exhibit overt neurological phenotypes [88, 89], under stress conditions they display distinct behavioral defects such as decreased anxiety, hyperlocomotion, and impaired sociability [60], further highlighting the neuroprotective role of NEAT1.

LncRNAs have gained increased attention in neurodevelopmental diseases due to their diverse roles in maintaining cellular homeostasis [59]. Despite growing evidence linking lncRNAs to neurogenesis, neuronal differentiation, maturation, and synaptic plasticity [90, 91], the precise mechanisms by which these molecules regulate these processes, particularly their interactions with other macromolecules, remain unclear. NEAT1, extensively studied in tumorigenesis, functions as a structural RNA essential for forming membraneless nuclear hubs known as paraspeckles, which play a crucial role in RNA biogenesis by selectively sequestering components of ribonucleoprotein complexes [92]. The nuclear retention of specific mRNAs can be paraspeckle-dependent, cellular stress-dependent [93, 94], and/or follow circadian rhythms [45, 95]. For example, several mRNAs encoding mitochondrial proteins have been found to be retained in the nucleus through direct interaction with NEAT1, highlighting the close relationship between mitochondrial and nuclear stresses [67]. While our data does not exclude the possibility of paraspeckle-dependent retention of ATG transcripts, further research is needed to clarify the specific roles of each NEAT1 isoform and the potential involvement of paraspeckles in the nuclear accumulation of ATG mRNAs.

To our knowledge, this is the first report linking nuclear retention of autophagy-related transcripts to the positive regulation of corresponding protein levels. Nuclear retention of developmental or stress response genes is a known regulatory mechanism, allowing cells to modulate protein expression in response to changing conditions. Typically, when specific cues are present, these mRNAs relocate to the cytoplasm for translation, ensuring precise spatial and temporal control of gene expression [93, 96]. In contrast, cytoplasmic accumulation of ATG mRNAs in NEAT1-KO and MECP2-KO cells leads to downregulation rather than increased protein levels, suggesting that NEAT1-mediated nuclear retention is essential for efficient mRNA biogenesis, and/or export and translation. One intriguing possibility is that RNA modifications, such as A-to-I editing sustained by paraspeckles, or methylation, required for downstream processes in the expression route of these genes, depend on NEAT1-mediated nuclear retention. A limitation of our study is that we have not distinguished between the roles of the short and long NEAT1 isoforms. Although ectopic overexpression of full-length NEAT1_2 (22 kb) can restore neuronal morphology, it remains unclear whether the short NEAT1_1 isoform, also generated from the same construct, also plays a role. Future studies should investigate whether paraspeckle formation per se (dependent on the long isoform but unlikely to be restored with ectopic NEAT1_2 overexpression) is required for transcript localization and autophagy control or if these are NEAT1-dependent functions independent of subnuclear bodies. Equally intriguing are the sequence features on ATG mRNAs that sustain interaction with NEAT1. For example, we could find no evidence of Alu repeats on the nuclear-retained ATG mRNAs, a feature previously linked to retention in paraspeckles [97, 98]. Although we have identified the interaction regions between NEAT1 and ATG16L2 or ATG5 mRNAs, further research is needed to confirm the exact sequence determinants underlying NEAT1–mRNA interactions.

Autophagy stands as a pivotal mechanism for preserving cellular cleanliness in eukaryotic cells, orchestrating the degradation of long-lived cytoplasmic proteins, complexes, and organelles. Significantly, the autophagy–lysosomal pathway is implicated in neurodegenerative disorders, where it functions in clearing accumulated abnormal proteins like huntingtin, tau, and alpha-synuclein from within cells [99, 100]. Moreover, emerging evidence underscores the role of autophagy regulation in processes central to neurodevelopment [101]. Metabolic dysfunction, notably involving glucose and cholesterol, and redox imbalance, are well-established hallmarks of RTT [74–76, 102]. While autophagy plays a crucial role in maintaining cellular metabolic homeostasis, particularly in post-mitotic neurons, its involvement in RTT pathogenesis remains unexplored. Previous studies have indicated an association between MeCP2 and mTORC1/2 complexes and signaling, with a general reduction observed in mTOR signaling pathway [78, 103, 104]. However, some brain regions from RTT patients have shown increased mTOR protein levels and phosphorylation status, linked to ribosomal biogenesis and ribosomal RNA induction [62]. Our findings reveal activation of S6 ribosomal protein and protein accumulation in MECP2-KO and NEAT1-KO neural progenitors despite the reduction in total mTOR levels. These results partially align with previous studies in RTT iPSCs, indicating reduced global translation, compromised mTOR signaling, and impairment in ubiquitination processes leading to protein accumulation [105]. While mechanistic discrepancies exist across various model systems and cellular types, accumulating evidence suggests that dysregulated protein translation and accumulation, leading to cellular resource exhaustion, are significant burdens for MECP2 mutant cells. Proper control of autophagy may hold the key to restoring the RTT phenotype, as the accumulation of abnormally shaped autophagosomes is a hallmark of neurodegenerative diseases and lysosomal pathologies. Despite limited focus to date on autophagy status in neurodevelopmental disorders and RTT models specifically, our study provides compelling evidence that autophagy dysregulation significantly contributes to the mutant MECP2 phenotype. Indeed, a study has reported defects in autophagy activation and autophagosome formation in fibroblasts from RTT patients [64], providing evidence that autophagy impairment is a characteristic feature of RTT pathophysiology. Moreover, our data highlights NEAT1 downregulation in RTT as the primary defect triggering autophagic impairment, a feature potentially applicable to other neurodevelopmental disorders.

The influence of NEAT1 on autophagy has been documented in numerous cancer studies, where NEAT1 is shown to promote autophagy in various tumorigenic contexts. While the precise mechanism remains largely unspecified, it is generally proposed that NEAT1 modulates the activity of several miRNAs, ultimately impacting autophagy-related genes [106–108]. For example, in mouse hepatic stellate cells, NEAT1 has been reported to regulate autophagy by binding directly to miR-29b, thereby regulating ATG9A expression [109]. Moreover, in myeloma, the replacement of miR-29b disrupts the autophagy pathway [110]. Interestingly, miR-29b is also highly expressed in the brain and predicted to interact with ATG16L1 and BECN1 (as assessed by the multiMiR 1.24.0 package [111]). Similarly, in neurodegenerative conditions such as PD and cerebral ischemia-reperfusion injury, NEAT1 has been implicated in promoting autophagy, often via miRNA regulation or other yet unidentified mechanisms [112, 113]. While the literature supports the possibility of NEAT1 regulating autophagy-related genes through miRNA modulation, our findings suggest a distinct role for NEAT1 in the control of ATGs through mRNA relocation, indicating a direct RNA–RNA interaction between NEAT1 and the mRNAs. However, considering the intricate landscape of RTT, various mechanisms may contribute significantly to the observed effects.

Our findings reveal a direct interaction between NEAT1 and specific autophagy-related mRNAs, including ATG16L2 and ATG5 (Fig. 6). Interestingly, while additional autophagy proteins such as ATG12 and ATG3 are affected in NEAT1-deficient cells (Fig. 5), there are no quantitative changes in their mRNA localization. This suggests that NEAT1 may selectively regulate key players to exert global control over ATG complexes. For instance, the E3 ligase complex ATG16L1-ATG5-ATG12 acts as a scaffold for the E2-like enzyme ATG3, facilitating the conjugation of LC3 to phosphatidylethanolamine and the maturation of the nascent phagophore into the autophagosome [114]. Changes in the localization of ATG5 and ATG16L1 mRNAs, and consequently their protein levels, are likely to influence the levels of ATG12 and ATG3 proteins, despite no significant relocalization of ATG12 and ATG3 mRNAs is observed in MECP2-KO cells. While ATG16L2, a paralog of ATG16L1, remains less studied, it has been suggested to play a potential autophagic inhibitory role by interfering with the formation of the ATG12-ATG5-ATG16L1 complex and LC3 lipidation [115]. Notably, alterations in ATG16L2 levels have been observed in several neurodegenerative diseases [116, 117], although its implications in neurodevelopment have yet to be elucidated.

Our findings also reveal NEAT1 as a crucial negative regulator of S6 phosphorylation in NPCs, which contrasts with the typical positive correlation between NEAT1 expression and mTOR activation observed in cancer cells [118–121]. This suggests that the role of NEAT1 may vary depending on cell type and context. This additional layer of regulation exerted by NEAT1 on the autophagic flux leads to impaired autophagosomal/autolysosomal ratio and disrupted autophagic dynamics in NEAT1-KO cells. Consequently, this disruption may result in the accumulation of misfolded or aggregated proteins and a loss of cellular protein quality control mechanisms, ultimately leading to abnormal neuronal morphology.

Our findings underscore the potential therapeutic importance of targeting NEAT1 and autophagy as promising approaches to alleviate the molecular and cellular dysfunctions associated with RTT. This offers hope for the development of novel treatment strategies for this debilitating neurodevelopmental disorder.

Acknowledgements

We thank the CERCA program (Generalitat de Catalunya) and the Josep Carreras Foundation for institutional support and the Guil lab for discussion and help. We are grateful for the technical support provided by the CCiTUB, University of Barcelona, L’Hospitalet de Llobregat, Spain, and the Advanced Microscopy Unit, IJC, co-financed by FEDER ‘A way of making Europe’. We thank the Rett Syndrome Research Trust for the donation of the iPSCs used in this study (ID 2037). We are indebted to the ‘Biobanc de l’Hospital Infantil Sant Joan de Déu per a la Investigació’, integrated in the Spanish Biobank Network of ISCIII for the sample and data procurement. Angelica Garcia (IJC) is gratefully acknowledged for her valuable technical assistance. Graphical Abstract was created in BioRender.com (Soler 2025) and can be accessed in https://BioRender.com/m66m166.

Author contributions: Conceptualization, E.S., A.L., and S.G.; Methodology, E.S., C.D.V., T.S., F.S., and L.M.-V.; Formal analysis, M.C.-P. and L.F.A.; Investigation, E.S., C.D.V., A.T., M.S., and C.O.-M.; Resources, J.A.; Writing—original draft, E.S. and S.G.; Writing—review & editing, E.S., T.S., L.F.A., A.L., and S.G.; Supervision, M.E., A.L., and S.G.; Funding acquisition, M.E., A.L., and S.G.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

The authors declare no competing interests.

Funding

This work was supported by the Spanish Ministerio de Ciencia, Innovación y Universidades, co-funded by European Regional Development Fund, ERDF, [PID2022-142829OB-I00, AEI/10.13039/501100011033 to S.G., PID2021-124536NB-I00 to A.L., PID2021-125282OB-I00, MCIN/AEI/10.13039/501100011033/ to M.E.]; the Departament de Recerca i Universitats/Generalitat de Catalunya [2021 SGR 01309 to S.G., 2021 SGR 01494 to M.E.]; the Catalan and the Spanish Associations for Rett Syndrome [FinRett to S.G.]; the Cellex Foundation [CEL007 to M.E.]; and the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brasilia, Brazil), ‘Science without borders’ scholarship [202074/2015-3 to E.S.]. M.E. is an ICREA Research Professor. Funding to pay the Open Access publication charges for this article was provided by Spanish Ministry of Science and Innovation.

Data availability

The data underlying this article are available in the European Nucleotide Archive (ENA), and can be accessed with the following accession numbers: PRJEB75396 (single-cell transcriptomic analysis of mutant MECP2 human neural cells (MECP2-KO, clone A33) over a time-course of in vitro differentiation, https://www.ebi.ac.uk/ena/browser/view/PRJEB75396); PRJEB82130 (single-cell transcriptomic analysis of progenitor and 7 days differentiated mutant MECP2 human neural cells (clones A33 and 3F), https://www.ebi.ac.uk/ena/browser/view/PRJEB82130); PRJEB75392 (transcriptomic analysis of mutant MECP2 human neural progenitor cells (MECP2-R133C), https://www.ebi.ac.uk/ena/browser/view/PRJEB75392); PRJEB75393 (transcriptomic analysis of mutant MECP2-KO (clone A33) and NEAT1-KO human neural cells, https://www.ebi.ac.uk/ena/browser/view/PRJEB75393); PRJEB81894 (NEAT1 RIP-seq in neural progenitor cells, https://www.ebi.ac.uk/ena/browser/view/PRJEB81894).