Original Article Analysis of the circRNA and T-UCR populations identifies convergent pathways in mouse and human models of Rett syndrome Edilene Siqueira,1,2,3,16 Aida Obiols-Guardia,1,2,16 Olga C. Jorge-Torres,1,2,16 Cristina Oliveira-Mateos,1 Marta Soler,1 Deepthi Ramesh-Kumar,1 Fernando Setién,1 Daniëlle van Rossum,4 Ainhoa Pascual-Alonso,5,6 Clara Xiol,5,6 Cristina Ivan,7,9 Masayoshi Shimizu,8,9 Judith Armstrong,6,10,11 George A. Calin,8,9 R. Jeroen Pasterkamp,4 Manel Esteller,1,12,13,14 and Sonia Guil1,2,15 1Josep Carreras Leukaemia Research Institute (IJC), Badalona, Barcelona, 08916 Catalonia, Spain; 2Cancer Epigenetics and Biology Program (PEBC), Bellvitge Biomedical Research Institute (IDIBELL), L’Hospitalet de Llobregat, Barcelona, 08908 Catalonia, Spain; 3National Council for Scientific and Technological Development (CNPq), Brasilia, 71605-001 Federal District, Brazil; 4Department of Translational Neuroscience, University Medical Center Utrecht Brain Center, Utrecht University, 3584 CG Utrecht, the Netherlands; 5Fundación San Juan de Dios, Barcelona, 08950 Catalonia, Spain; 6Institut de Recerca Pediàtrica, Hospital Sant Joan de Déu, Barcelona, 08950 Catalonia, Spain; 7Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA; 8Department of Translational Molecular Pathology, The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA; 9Center for RNA Interference and Non-coding RNAs, The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA; 10Servei de Medicina Genètica i Molecular, Hospital Sant Joan de Déu, Barcelona, 08950 Catalonia, Spain; 11CIBER-ER (Biomedical Network Research Center for Rare Diseases), Instituto de Salud Carlos III, 28029 Madrid, Spain; 12Centro de Investigacion Biomedica en Red Cancer (CIBERONC), Madrid, Spain; 13Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, 08010 Catalonia, Spain; 14Physiological Sciences Department, School of Medicine and Health Sciences, University of Barcelona (UB), Barcelona, 08907 Catalonia, Spain; 15Germans Trias i Pujol Health Science Research Institute, Badalona, Barcelona, 08916 Catalonia, Spain Noncoding RNAs play regulatory roles in physiopathology, but INTRODUCTION their involvement in neurodevelopmental diseases is poorly un- Most of the human genome is transcribed as noncoding RNAs derstood. Rett syndrome is a severe, progressive neurodevelop- (ncRNAs),1 some of which have crucial roles in control of gene regu- mental disorder linked to loss-of-function mutations of the lation, especially in the brain and in neurological diseases.2 Compared MeCP2 gene for which no cure is yet available. Analysis of with other tissues, the brain displays a higher diversity of expressed the noncoding RNA profile corresponding to the brain-abun- ncRNAs and is the organ with the most tissue-specific long noncod- dant circular RNA (circRNA) and transcribed-ultraconserved ing RNAs (lncRNAs), with regulated temporal and spatial expression region (T-UCR) populations in a mouse model of the disease patterns.3,4 lncRNAs are thought to play crucial roles in brain matu- reveals widespread dysregulation and enrichment in glutama- ration and function, and, as a consequence, their dysregulation is tergic excitatory signaling and microtubule cytoskeleton path- likely to be important during onset or progression of a number of ways of the corresponding host genes. Proteomic analysis of neurological diseases.5,6 Although most ncRNAs are linear molecules, hippocampal samples from affected individuals confirms a subgroup of them form a circle through covalent binding; circular abnormal levels of several cytoskeleton-related proteins RNAs (circRNAs) are expressed in a tissue- and organ-specific together with key alterations in neurotransmission. Impor- manner and derive mainly from non-canonical splicing of protein- tantly, the glutamate receptor GRIA3 gene displays altered coding pre-mRNAs.7 Their functions remain largely unexplored, biogenesis in affected individuals and in vitro human cells although one proposed mode of action is based on their ability to and is influenced by expression of two ultraconserved RNAs. titrate out the pool of functional microRNAs (miRNAs) (and also We also describe post-transcriptional regulation of SIRT2 by RNA-binding proteins or other small RNAs) by acting as molecular circRNAs, which modulates acetylation and total protein levels “sponges.”8–11 In the brain, circRNAs are highly abundant in synap- of GluR-1. As a consequence, both regulatory mechanisms ses and more conserved than other ncRNAs, and their expression converge on the biogenesis of AMPA receptors, with an effect on neuronal differentiation. In both cases, the noncoding Received 21 October 2021; accepted 17 December 2021; RNAs antagonize MeCP2-directed regulation. Our findings https://doi.org/10.1016/j.omtn.2021.12.030. indicate that noncoding transcripts may contribute to key al- 16 These authors contributed equally terations in Rett syndrome and are not only useful tools for Correspondence: Sonia Guil, Regulatory RNA and Chromatin Group, Josep Car- revealing dysregulated processes but also molecules of reras Leukaemia Research Institute, Ctra de Can Ruti, Camí de les Escoles s/n, biomarker value. Badalona, Barcelona 08916, Spain. E-mail: sguil@carrerasresearch.org Molecular Therapy: Nucleic Acids Vol. 27 March 2022 ª 2021 The Author(s). 621 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Molecular Therapy: Nucleic Acids (legend on next page) 622 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org seems to increase during development of the CNS.12 In addition, their and motor dysfunction until they fully manifest the RTT-like pheno- presence in the bloodstream and their abundance in exosomes as type at around 6–8 weeks of age. This leads to rapid weight loss and circulating molecules may make them valuable biomarkers for non- death at approximately 10 weeks of age.25 This and other models have invasive diagnosis in a variety of human diseases, including disorders been used to identify the molecular mechanisms and pathways with of the CNS and degenerative diseases.13 key roles in RTT etiology and have provided an essential platform for drug testing in pre-clinical assays. Nevertheless, the lack of effec- Rett syndrome (RTT; MIM: 312750) is a postnatal progressive neuro- tive therapies and biomarkers available warrants further examination developmental pathology that affects 1 in 10,000 female births14,15 of the cellular programs underlying the disorder. and in which loss of function mutations in the X chromosome-linked MeCP2 gene are the genetic basis of most cases.16 MeCP2 is a nuclear One insufficiently understood aspect of RTT is the potential role of protein, expressed widely in a range of tissues, that is most abundant noncoding transcripts whose levels are impaired in the context of in neurons of the mature nervous system. It was first identified by MeCP2 mutations. Although efforts have been made to characterize Lewis et al.17 as a new protein that binds to methylated CpG dinucle- the RTT miRNome,26,27 and specific miRNAs have been investigated otides. RTT is a progressive disorder with a normal prenatal and peri- in the context of RTT,28–35 other classes of ncRNAs, including long natal period and proper brain development during the first few species, have remained largely unexplored. In this study, we describe months of life, associated with normal child neurodevelopment, mo- alterations in two types of ncRNA that are especially enriched in the tor function, and communication skills. However, developmental brain: circRNA and transcribed-ultraconserved region (T-UCR) tran- regression appears between 8 and 24 months of age.18 Thereafter, scripts. Host genes for both classes of altered ncRNAs are particularly the disorder may become evident when affected individuals lose their highly enriched in key processes in RTT (namely, neurotransmission verbal ability and show severe cognitive impairment, stereotypic be- and cytoskeleton dynamics), demonstrating the potential of ncRNAs haviors, as well as cardiac and respiratory abnormalities.19,20 RTT is to reveal crucial abnormalities in neurodevelopmental disorders. considered a synaptopathy, displaying aberrant expression of neuro- transmitters, neuromodulators, transporters, and receptors, which RESULTS are the basis of unbalanced excitatory/inhibitory neurotransmis- circRNAs are dysregulated in Mecp2 KO mouse models sion.21 The lack of an effective cure or even of treatment options to circRNAs are important for many neurodevelopmental processes, ameliorate the most disabling symptoms of the disorder highlights including neurogenesis and synaptic plasticity.36,37 To identify the need for a better understanding of its physiopathology.22–24 Cur- circRNA molecules with altered expression in RTT, we used a rent efforts to tackle RTT include pharmacological and molecular ge- circRNA microarray to accurately quantify their expression profiles netics-based strategies, but the therapeutic approaches developed so in the hippocampus and frontal cortex of symptomatic, 8-week-old far have had limited success (see Leonard et al.24 for a review). Mecp2 KO mice and compared their levels with those of wild-type lit- termates of the same age. circRNAs with a fold change of1.5 or greater Implementation of several mouse models that mimic the disease has and values of p % 0.05 were selected as significantly differentially ex- been crucial to progress in research on RTT. For example, Mecp2 pressed (DE) transcripts (Figures 1A and S1A). There were fewer DE knockout (KO) mice have a range of physiological and neurological circRNAs in the mouse frontal cortex (FC) than in the hippocampus abnormalities that imitate the human syndrome. KO mice are normal (HIP) (47 upregulated and 103 downregulated versus 141 upregulated overall until 4 weeks of age, after which they begin to suffer cognitive and 247 downregulated, respectively). In accordance with the Figure 1. Analysis of the altered circRNA population in MeCP2-deficient mice and cell lines (A) Heatmap and hierarchical clustering analysis of circular RNAs (circRNAs) that are DE in the HIP and FC of 8-week-old Mecp2 KO and WT mice (two-tailed unpaired t test, fold change > 1.5, p < 0.05). (B and C) Enriched Gene Ontology terms for the host genes of altered circRNA (fold change > 2), as identified by functional clustering (Enrichr). The y axis shows the Gene Ontology (GO) terms, and the x axis shows the statistical significance (two-tailed Fisher’s exact test). Categories corresponding to glutamatergic trafficking and/or signaling are highlighted. (D) Top panel: intronic/exonic organization of linear (black) and circular (blue) transcripts in the Sirt2 locus. Coordinates correspond to those of the UCSC Genome Browser (GRCh38/mm10 release). Only the potentially circularized exons and 5ʹ and -30 regions of the Sirt2 transcripts are included. Bottom panel: the expression levels of Sirt2 (left) and circSirt2 (right) were analyzed by qRT-PCR in FC from WT and KO pre-symptomatic (4-week-old) and symptomatic (8-week-old) mice. Graphs show the median and interquartile range of 6–8 replicates from different animals per condition. Two-tailed Mann-Whitney U tests were used (**p < 0.01; ns, not significant). (E) Western blot analysis of MeCP2, SIRT2, acetyl-a-tubulin, and a-tubulin in FC from WT and KO pre-symptomatic and symptomatic mice. Samples from two animals per condition are shown. Graphs on the right represent quantitation of band intensity (unpaired t test, *p < 0.05, **p < 0.01). (F) Western blot analysis of MeCP2 and SIRT2 in N2a cells (2 controls and 2 Mecp2-KO clones). Two sgRNA sequences targeting the Mecp2 gene were used (sg1 and sg2). The graph on the right represents quantitation of band intensity of two representative experiments. (G and H) Expression levels of Sirt2 mRNA and circSirt2 were analyzed by qRT-PCR in the same samples as in (F). Graphs show the median and interquartile range (two-tailed Mann-Whitney U test, *p < 0.05). (I) Western blot analysis of SIRT2, b-actin, and MeCP2 in N2a cells overexpressed with an empty or circSirt2 plasmid. The graph on the right represents quantitation of band intensity of two representative experiments (two-tailed unpaired t test, *p < 0.05). (J) Expression levels of Sirt2 mRNA were analyzed by qRT-PCR in the same samples as in (I). The graph shows the median and interquartile range (two-tailed Mann-Whitney U test). (K) Western blot analysis of MeCP2 and SIRT2 in N2a KO cells (sg2 clone) upon overexpression of circSirt2 and/or MeCP2, as indicated. The graph on the right represents quantitation of band intensity of two representative experiments (two-tailed unpaired t test, *p < 0.05). (L) Expression levels of Sirt2 mRNA were analyzed by qRT-PCR in the same samples as in (K). The graph shows the median and interquartile range (two-tailed Mann-Whitney U test). See also Figure S1. Molecular Therapy: Nucleic Acids Vol. 27 March 2022 623 Molecular Therapy: Nucleic Acids (legend on next page) 624 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org previously described upregulation of circRNAs during neuronal dif- levels of SIRT2 protein by western blot and detected a marked in- ferentiation and maturation12 and the fact that individuals with crease in the FC (but not the HIP) of 8-week-old symptomatic mice RTT have reduced dendritic complexity and spine density,38 we and a reduction in acetylation of the SIRT2 main target, a-tubulin found more downregulated circRNA species in the HIP of Mecp2 (Figures 1E and S1C), which confirms previous observations of KO mice (Figure S1A). circRNAs most commonly derive from reduced levels of acetylated a-tubulin in MeCP2-depleted cells.43 non-canonical splicing of the pre-mRNA of host protein-coding genes, and although in several examples there was little correlation To confirm these observations, we used two cellular models, the between the variation in the abundance of a circRNA and its cognate mouse neuroblastoma N2a cell line and the human neural progenitor linear RNA,39 one major factor that controls the expression of a cell line ReNCell VM, which were engineered to suppress MeCP2 particular circRNA is transcription of its host gene. Therefore, valu- expression through genome editing with CRISPR-Cas9. On one able insights can be gained from analyzing the gene ontologies of hand, and in accordance with the data from mice, decreased levels host genes. Enrichment analysis shows that DE circRNAs with a of MeCP2 protein in N2a cells correlated with an increase in SIRT2 fold change of 2 or greater and p % 0.05 derive from host genes protein levels (Figure 1F), whereas Sirt2 linear transcript levels belonging to categories related to synaptic transmission, specifically remained unaltered (Figure 1G). Notably, circSIRT2 expression was AMPA-type ionotropic glutamatergic receptors (Figures 1B, 1C, upregulated under these conditions (Figure 1H). Further, overexpres- and S1B). These and other genes of special importance in brain devel- sion of circSIRT2 in wild-type N2a cells produced a 2-fold increase in opment were selected for further validation (see below). SIRT2 protein levels without altering Sirt2 mRNA (Figures 1I and 1J). Finally, circSIRT2 counteracts the effect of MeCP2 overexpression in SIRT2 and related circRNAs are altered in mouse and human N2a KO cells, restoring the levels of SIRT2 protein, again without any models of RTT change in Sirt2 mRNA levels (Figures 1K and 1L). On the other hand, The microtubule cytoskeleton plays an essential role in neuronal we observed a similar regulation taking place in human neural pro- function, migration, and differentiation and influences trafficking to genitor cells. As shown in Figures 2A and 2B, levels of SIRT2 protein dendritic spines of AMPA-type receptors.40 Importantly, the acetyla- (but not of its mRNA) were much higher in MeCP2-KO cells, espe- tion state of microtubules regulates molecular motor trafficking of cially in spontaneously differentiating cells, concomitant with a mRNA and of vesicles containing brain-derived neurotrophic factor decrease in the levels of acetylated a-tubulin. Overexpression of (BDNF), which is one of the major downstream targets of MeCP2 MeCP2 in MeCP2 KO cells downregulated SIRT2 protein (Figure 2C) in RTT pathogenesis.41 In addition, a reduction in acetylated but not its mRNA (Figure 2D), again pointing to post-transcriptional a-tubulin has been reported in induced pluripotent stem cell regulation. To ascertain the potential role of circRNAs therein, we (iPSC)-derived neurons and mouse models of the syndrome.42,43 investigated the expression of human circSIRT2 in our neural progen- The HDAC6 and SIRT2 lysine deacetylases are thought to play a itor cells. There are several annotated circRNA molecules from the role in a-tubulin deacetylation, and their deregulation is considered SIRT2 locus, including hsa_circ_0050951, which is highly similar to to be important in certain neurodegenerative and neurodevelopmen- mouse circSIRT2. We were able to detect four of these species, three tal diseases.44,45 In particular, we detected a decrease in circSirt2 in the of which were upregulated in MeCP2 KO progenitor cells (Figure 2E). FC of pre-symptomatic mice, which was not accompanied by changes Cellular fractionation and RNA in situ hybridization (ISH) experi- in host gene mRNA levels (Figure 1D). Because some circRNAs ments indicated that these circRNAs accumulated in the cytoplasm modulate translation of their linear counterparts, we assayed the (Figures 2F and S2A) and were enriched in the cellular soma (Figures Figure 2. Analysis of SIRT2 and circSIRT2 expression in human neural progenitor and differentiating cells (A) Western blot analysis of MeCP2, SIRT2, acetyl-a-tubulin, and a-tubulin in WT or MeCP2 KO neural progenitors and differentiating cells (30 days). (B) Expression levels of SIRT2 mRNA in the same samples as in (A) were analyzed by qRT-PCR. Graphs show the median and interquartile range (two-tailed Mann-Whitney U test, *p < 0.05). (C) Western blot analysis of MeCP2 and SIRT2 in MeCP2 KO neural progenitor cells upon ectopic expression of MeCP2. WT cells are also shown for comparison. The graph on the right represents quantitation of band intensity of two representative experiments (one-way ANOVA was used; *p < 0.05, **p < 0.01). (D) Expression levels of SIRT2 mRNA were analyzed by qRT-PCR in MeCP2 KO control or MeCP2-overexpressing progenitor cells. The graph shows the median and interquartile range (two-tailed Mann-Whitney U test). (E) Left: intronic/exonic organization of linear (black) and circular (blue) transcripts in the human SIRT2 locus. Coordinates correspond to those of the UCSC Genome Browser (GRCh37/hg19 release). Only the detected circRNAs and 5ʹ and -30 regions of SIRT2 transcripts are included. Right: expression levels of the described circRNAs in WT or MeCP2 KO progenitor or differentiating cells (30 days) were analyzed by qRT-PCR. Graphs represent the median with interquartile range (two-tailed Mann-Whitney U test, **p < 0.01). (F) Nuclear/cytoplasmic fractionation of WT or MeCP2 KO differentiating neural progenitor cells (day 14), analyzed by qRT-PCR and western blot to assess fraction purity. (G) RNA ISH showing localization of human hsa_circ_0050946 (derived from the SIRT2 gene) in WT or MeCP2 KO differentiating neural progenitor cells (14 days). Colocalization with the neuronal marker TUJ1 is shown. Scale bar size is indicated in each panel. (H) Western blot analysis of SIRT2 in WT neural progenitor cells upon overexpression of the circular transcripts from the human SIRT2 locus. The graph on the right represents quantitation of band intensity of two representative ex- periments (one-way ANOVA was used; *p < 0.05). (I) Expression levels of SIRT2 mRNA were analyzed by qRT-PCR in the same samples as in (H). One-way ANOVA was used. (J) Chromatin immunoprecipitation (ChIP) coupled to qPCR to determine MeCP2 binding to the SIRT2 promoter in WT progenitor cells. The graph represents the mean ± SD of three independent experiments. The enrichments corresponding to BDNF and GAPDH promoters are shown for comparison. Two-tailed Mann-Whitney U test was used (*p < 0.05). (K) qRT-PCR analysis of retrieved RNA following MeCP2 IP in WT progenitor cells. SIRT2 locus-derived transcripts were analyzed. circSCMH1 was used as a positive control. The graph represents the mean ± SEM of at least 3 independent experiments. Two-tailed Mann-Whitney U test was used (*p < 0.05, **p < 0.01). See also Figure S2. Molecular Therapy: Nucleic Acids Vol. 27 March 2022 625 Molecular Therapy: Nucleic Acids (legend on next page) 626 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org 2G and S2B), and only hsa_circ_0050951 was quantitatively present in the levels of SIRT2 protein, we next investigated whether changes the nucleus (Figure S2A). To test their functional relevance, we over- in circSIRT2 levels could alter the expression of AMPA receptor sub- expressed them in wild-type (WT) progenitor cells and observed that units. As shown in Figures S2F and S2G, overexpression of individual all of them caused an increase in SIRT2 protein levels (Figure 2H) circSIRT2 RNAs in WT neural progenitor cells had little effect on without affecting SIRT2 mRNA (Figure 2I). Of note, this increase GluR-1/2/3 mRNA or protein levels. However, their enforced expres- was attenuated when we combined overexpression of the circRNAs sion in MeCP2 KO cells resulted in a marked decrease in GluR-1 and with overexpression of MeCP2 in MeCP2 KO neural progenitor cells -2 protein levels and, to a minor extent, also in GluR-3 levels, and this (Figures S2C and S2D). We finally wanted to determine how MeCP2 could be reversed by overexpression of MeCP2 (Figure 3A), indi- could influence circSIRT2 production. Chromatin immunoprecipita- cating that varying levels of circRNAs in the background of MeCP2 tion (ChIP) data do not indicate an enrichment of MeCP2 on the pro- deficiency may contribute greatly to AMPAR biogenesis. This effect moter of the SIRT2 gene and on circularized exons (Figure 2J); in on protein levels was not mirrored by major changes in GRIA1– contrast, we could quantitatively retrieve all circSIRT2 species (but GRIA3 mRNA levels (Figure 3B), pointing to post-transcriptional not linear SIRT2 mRNA) in MeCP2 immunoprecipitation assays regulation. To assess circRNA roles in the reported control of (Figure 2K). This is reminiscent of the recently described binding GluR-1 acetylation by SIRT2, we next co-transfected all circSIRT2 of MeCP2 to circSCMH1 in ischemic stroke models.46 Of note, RNAs together and observed a decrease in GluR-1 protein levels circ_0050951, the most nuclear of all circSIRT2 molecules (Fig- (but not its mRNA levels) that was dependent on intact SIRT2 activity ure S2A), displayed the highest enrichment in MeCP2 because use of the specific SIRT2 inhibitor AGK2 blocked the effect. immunoprecipitations(Figure 2K), suggesting that nuclear retention Again, this was independent of changes in mRNA levels of either pro- might be directed by its binding to MeCP2. Use of the catRAPID al- tein (Figures 3C and 3D). In view of this, we next immunoprecipitated gorithm confirmed the high capacity of the NID and C-t domains GluR-1 and confirmed that its acetylation levels were downregulated within MeCP2 to interact with all circSIRT2 RNAs (Figure S2E), coin- upon overexpression of circSIRT2 RNAs to a degree similar to the ef- ciding with previous observations of MeCP2 RNA binding ability fect of overexpressing SIRT2. This effect could be abolished using a (recently reviewed by Good et al.47). These results indicate that linear SIRT2 inhibitor, suggesting that the circRNA molecules control and circular SIRT2 molecules are subject to differential regulation GluR biogenesis by modulating SIRT2 levels (Figure 3E). by MeCP2 and point to a post-transcriptional regulatory role of circSIRT2 that might affect SIRT2 protein levels and, potentially, Other circRNAs dysregulated in RTT are potentially associated microtubule cytoskeleton dynamics. with cytoskeleton and GluR proteins Other circRNA species whose host genes play a role in cytoskeleton circSIRT2 modulates GluR-1 levels through regulation of SIRT2 dynamics include circDnah2 and circMap1a, and it is worth noting Another potentially relevant target of SIRT2 in the context of RTT is that truncating variants of the latter have been associated with autism GluR-1, a subunit of AMPAR complexes. Glutamate receptors have a spectrum disorder.52 Most of Dnah2 and Map1a circular and linear key role in excitatory synaptic transmission in the CNS, and alter- species were downregulated in the FC of pre-symptomatic and symp- ations in MeCP2 levels in glutamatergic neurons contribute to the tomatic mice (Figures S1D and S1E). Another circRNA species of RTT phenotype,48 possibly to varying degrees, depending on the potential relevance to RTT is circFoxp1, which we found to be down- brain region. We have previously observed smaller miniature excit- regulated in the pre-symptomatic FC and upregulated in the symp- atory postsynaptic currents in cultured neurons from Mecp2 KO tomatic HIP (Figures S1F and S1G). The host gene, Foxp1, which mouse cerebellum.49 In hippocampal glutamatergic synapses, most encodes a transcription factor with an important role in brain devel- postsynaptic receptors are heterotetramers that contain GluR-1/ opment and whose loss causes autistic behavior and intellectual GluR-2 and GluR-2/GluR-3 subunits (encoded by the genes disability in mice,53 is dysregulated extensively in several different GRIA1–GRIA3). SIRT2 has been reported to promote deacetylation murine models of the disorder, as seen in recent meta-analyses of of GluR-1, downregulating protein levels and impairing AMPAR traf- transcriptomic changes in RTT.54 Accordingly, we observed a signif- ficking and synaptic plasticity.50,51 Given the effect of circSIRT2 on icant decrease of the linear form in the symptomatic FC of Mecp2 KO Figure 3. Analysis of regulation by circRNAs from the SIRT2 locus on GluR-1–GluR-3 proteins (A) Western blot analysis of GluR-1, -2, and -3 proteins in MeCP2 KO neural progenitor cells upon overexpression of the different circSIRT2 RNAs, individually or in com- bination with MeCP2 ectopic expression, as indicated. The graph on the right represents quantitation of band intensity of three representative experiments (one-way ANOVA was used). (B) Expression levels of GRIA1, GRIA2, or GRIA3 mRNAs were analyzed by qRT-PCR in the same samples as in (A). Graphs represent the median with interquartile range of at least 3 independent replicas (one-way ANOVA was used; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). (C) Western blot analysis of GluR-1 and SIRT2 proteins in KO progenitor cells upon overexpression of all four circSIRT2 RNAs (circ_0050945, circ_0050946, circ_0050948 and circ_0050951). Cells treated with the SIRT2 inhibitor AGK2 were compared with control cells (DMSO treated). The graph on the right represents quantitation of band intensity (mean ± SD) of three representative experiments (unpaired t test, **p < 0.01, ****p < 0.0001). (D) Expression levels of SIRT2 and GRIA1 mRNAs were analyzed by qRT-PCR in the same samples as in (C). Graphs represent the median with interquartile range of at least 3 independent replicas (two-tailed Mann-Whitney U test). (E) GluR-1 IP from MeCP2 KO neural progenitor cells was analyzed by Western blot with antibodies against GluR-1 and acetyl-lysine. Cells were transfected with a control vector (empty) or the four circSIRT2 or SIRT2-HA plasmids and treated with the AGK2 inhibitor or mock treated (DMSO), as indicated. The graph on the right represents quantitation of band intensity (mean ± SD) of three representative experiments (one-way ANOVA, ***p < 0.001, ****p < 0.0001). Molecular Therapy: Nucleic Acids Vol. 27 March 2022 627 Molecular Therapy: Nucleic Acids (legend on next page) 628 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org animals (Figure S1F). Similarly, circBirc6, which was deregulated in sion and suggesting that T-UCR expression has a negative effect on the HIP and the FC (Figures S1H and S1I), originates from a candi- the host genes. date gene in autism spectrum disorders.55 In addition to the regula- tion by circSIRT2 shown above, an enrichment analysis indicates The proteome from postmortem human HIP reveals changes in that many genes corresponding to subunits of AMPA receptors microtubule-cytoskeleton dynamics and AMPAR trafficking contain circRNA species that are altered in KO mice. In the FC, we To compare key pathway alterations in the mouse model with those did indeed observe a tendency (but not a statistically significant found in individuals with RTT, we next analyzed the whole proteome one) toward downregulation of the circGria4 RNA, similar to the of postmortem hippocampal regions from four control individuals linear form (Figure S1J). In the case of Gria2, the circular form ap- and four individuals with RTT. Sequencing the MeCP2 gene detected peared to be downregulated, in contrast to the linear form, but again alterations in all four clinically diagnosed individuals with RTT, the difference was not statistically significant (Figure S1K). It is clear whereas no mutations were found in the control samples. Mutations that there is important dysregulation of the circRNA transcriptome were not expected to produce truncated proteins, and western blot that does not always coincide with changes in the corresponding analysis revealed similar levels of protein (Figure S4A). Proteins host genes, highlighting their putative independent roles. significantly altered (false discovery rate [FDR] < 0.05) in individuals with RTT are listed in Figure 4C. An enrichment analysis revealed Transcripts derived from UCRs are altered in Mecp2 KO mice that categories related to microtubule dynamics were overrepresented The T-UCR transcriptome is another type of ncRNA that has never and also pointed to alterations in microglia/macrophage activation been analyzed previously in RTT models. T-UCRs are derived from (Figure 4D). Deficits in microtubule polymerization have been genomic regions that are 100% conserved between humans, mice, increasingly recognized in animal models of RTT63,64 as well as in and rats. Despite growing evidence of their potential as disease bio- MeCP2-deficient fibroblasts,65 and lower levels of expression of markers,56–58 their extraordinary conservation and potential role in tubulin genes have been observed in the brain tissue of individuals gene regulation are still insufficiently understood. In our mouse with RTT.66 We confirmed the changes in several microtubule cyto- model of RTT, we found a number of T-UCRs that were altered in skeleton-related proteins in the HIP from affected individuals, Mecp2 KO mice. Interestingly, although results from the FC reflected including upregulation of the microtubule-associated proteins the expected enrichment in T-UCRs on host genes involved in mRNA TAU, MAP6, and STMN1, which might contribute to microtubule biogenesis (Figure 4A),59 the results in the HIP again indicate an destabilization (Figure 4E). Unexpectedly, the most strongly downre- enrichment in categories related to AMPA receptor trafficking (Fig- gulated protein in the proteome of HIP from affected individuals was ure 4B; see below). Among the dysregulated T-UCRs, we confirmed the glial-specific protein S100B, an observation that is quite the oppo- the negatively correlated expression of the uc.309 transcript, which site of the enrichment of glial cells seen in RTT models.67,68 However, is antisense to an intron within Btrc. Btrc expression was downregu- early studies of RTT also detected a tendency toward downregulation lated in pre-symptomatic mice, whereas uc.309 was upregulated (Fig- of this astroglial protein in the cerebrospinal fluid of individuals with ure S3A). It is worth bearing in mind that Btrc is an E3 ubiquitin ligase RTT,69 and more recent studies have reported no alterations in S100B required for correct neural differentiation through its ability to expression in the hypothalamus of Mecp2 KO mice.70 This could degrade REST, a master repressor of neuronal programs in non- reflect region-specific disruptions because we observed a decrease in neuronal lineages.60 The transcription factor Zeb2 also has an essen- S100B mRNA expression and protein levels in the HIP of Mecp2 tial role in CNS development (reviewed by Epifanova et al.61). It was KO mice (Figures S4B and S4C) but an increase in the FC of the found to be altered in the HIP and FC of Mecp2 KO mice and con- same mice (Figure S4D). In addition, levels of S100B mRNA and pro- tained several T-UCRs that had distinct patterns of expression (Fig- tein in differentiating human neural progenitors were also higher in ure S3B). Additionally, Hnrnph1, a splicing regulator with pathogenic MeCP2 KO cells (Figures S4E and S4F). It is worth noting that variants associated with intellectual disability,62 was also dysregulated S100B is very important for maturation of a variety of neurons in in the HIP and FC, displaying negatively correlated associations with different CNS regions, including the HIP, as a consequence of its abil- uc.186 (Figure S3C). In all of these examples, expression of the ity to promote and stabilize microtubule assembly.71,72 S100B is also T-UCRs and their respective host genes was not positively correlated, important for glutamate clearance and has been shown to protect hip- reflecting independent regulatory control and/or antagonistic expres- pocampal neurons against toxic concentrations of glutamate.73 Given Figure 4. Analysis of the altered T-UCR population in Mecp2 KO mice and proteomics analysis of postmortem samples from individuals with RTT (A and B) Enriched GO terms for the host genes of altered T-UCRs in the FC and HIP of Mecp2 KO mice (p < 0.05), as identified by functional clustering (reactome.org). The y axis shows GO terms, and the x axis shows statistical significance. (C) List of dysregulated proteins (adjusted p < 0.05) in the HIP of individuals with RTT. The ID of the corresponding gene and the expression change are indicated. Highlighted proteins were validated by western blot in this work. (D) Enriched GO terms for the altered proteins in the HIP of individuals with RTT, as identified by functional clustering (p < 0.05, Enrichr). The y axis shows GO terms, and the x axis shows the degree of statistical sig- nificance. (E) Western blot analysis of TAU, MAP6, and STMN1 in the HIP of brains from individuals with RTT. Samples from three control individuals and three individuals with RTT are shown. Graphs on the right represent quantitation of band intensity (unpaired t test, *p < 0.05). (F) Western blot analysis of GluR-1, -2, and -3 proteins in the FC of WT or Mecp2 KO symptomatic (8-week-old) mice. (G) Western blot analysis of GluR-1 and -2 in the HIP of brains from individuals with RTT. Samples from four control individuals and four individuals with RTT are shown. See also Figures S3 and S4. Molecular Therapy: Nucleic Acids Vol. 27 March 2022 629 Molecular Therapy: Nucleic Acids (legend on next page) 630 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org our findings linking circSIRT2 ncRNAs with AMPAR subunit regula- of the 30 end of GRIA3 transcripts and then cloning and sequencing tion in cell models and the fact that our proteomics analysis showed the amplified products. Even though the flip exon was the one most that AMPA glutamate receptor 3 (GluR-3) was markedly downregu- frequently included in all samples, we noted a 3-fold difference in lated in the HIP of individuals with RTT (Figure 4C), we next assessed the ratio of flop exon inclusion (Figure 5E; from 10% in the control GluR protein levels in mouse and human RTT samples. Indeed, to 30% in RTT). In addition to this, we observed (1) a marked dif- GluR-1, -2, and -3 were downregulated in the FC of symptomatic ference in editing levels (50% in control individuals compared with MeCP2 KO mice (Figure 4F), and GluR-1 and -2 were also found 90% in individuals with RTT) and (2) a significant non-random asso- to be decreased in the HIP of individuals with RTT (Figure 4G). As ciation of alternative exon selection and editing with RTT samples for GluR-3, we looked at its expression in more detail, taking into ac- (Figure 5E). In comparison, the HIP of Mecp2 KO mice was highly count that the gene hosts two ultraconserved transcripts. edited, similar to WT animals (90%), and showed no significant dif- ferences in flip/flop selection or Gria3 and uc.478/uc.479 expression Changes in mRNA editing and protein levels of GluR-3 are (Figures 5F and S5H–S5K), suggesting that these features are not pre- correlated with T-UCR expression in human models of RTT sent in this region of the Mecp2y/– animal model. GRIA3 pre-mRNA undergoes alternative splicing to produce two splice variants containing one of the two mutually exclusive alterna- Editing levels of GRIA receptor subunits were generally correlated tive exons (flip and flop variants), which confer distinct channel with levels of the editing enzyme ADAR2.77 However, we did not opening/closing kinetics to the receptor, with a suggested faster detect any significant changes in ADAR2 protein levels in our RTT desensitization time for the flop variant.74–76 In addition, the diversity samples (Figure 5B), suggesting that other mechanisms determine of GRIA3 transcripts is enhanced by adenosine-to-inosine editing of a the alterations in GRIA3 editing. To investigate this further, we position directly upstream of the alternative exons, which can result explored the transcripts from the GRIA3 locus in human neural pro- in an Arg-to-Gly change in the amino acid sequence of GluR-3 and genitor cells knocked out for MeCP2. When forced to differentiate additional fine-tuning of its receptor features. Importantly, flop and (freely or toward glutamatergic neurons), levels of GluR-3 protein flip exons overlap with UCRs from where two transcripts are gener- in KO cells also decreased (Figures 5G, S5L, and S5M) without any ated: uc.478 and uc.479 (Figure 5A). We first confirmed that GluR-3 significant change in mRNA levels (Figure 5H). Importantly, levels protein levels were lower in the HIP of individuals with RTT (Fig- of uc.478 and uc.479 increased in KO progenitor and differentiating ure 5B). This was not accompanied by a decrease in the total cells (Figure 5H), which is concomitant with an increase (although mRNA level, which was unaltered (Figure 5C) but, rather, by upregu- not a statistically different one) in editing in progenitor cells from lation of expression of uc.478 and, to a lesser extent, of uc.479 (Fig- 40% to 50% and a decrease in differentiating cells from 70% to 50% ure 5D). This seems to be region specific because no noticeable (Fisher’s exact test, p = 0.0152) (Figures 5I and 5J). Under the same changes in GRIA3 mRNA or protein levels or in T-UCR expression conditions, flip/flop exon selection was not altered in MeCP2 KO cells were found in the cerebellum or Brodmann area 10 of the same indi- (Figures 5I and 5J). To investigate the potential role of T-UCR expres- viduals (Figures S5A–S5F). We looked in detail at the frequency of ed- sion in GRIA3 regulation in the context of RTT, we first carried out iting and alternative exon selection in the HIP by performing RT-PCR ChIP experiments and observed an enrichment of MeCP2 on the Figure 5. Characterization of GRIA3 and T-UCR expression levels and editing in HIP from postmortem RTT samples, the mouse model, and human neural progenitor cells (A) Schematic illustrating the GRIA3 locus encompassing the 3ʹ exons, including the alternatively spliced flip and flop exons and the A/I RNA editing site, where the arginine is recoded to a glycine. (B) Western blot analysis of GluR-3 and ADAR2 in the HIP of three postmortem samples from individuals with RTT and three samples from healthy control individuals. (C and D) Expression levels of GRIA3 mRNA and the T-UCRs uc.478 and uc.479 were analyzed in the same samples as in (B) by qRT-PCR. Graphs represent the median with interquartile range of the data from the three subjects per group (two-tailed Mann-Whitney U test, **p < 0.01). (E) Analysis of editing and alternative splicing choice of GRIA3 mRNA in the HIP from three postmortem samples of individuals with RTT and samples from three healthy control individuals. Percentages correspond to data pooled from sequences of at least 60 clones per subject. The contingency graph displays flip/flop selection and the associated edited/unedited status. Fisher’s exact test (two-tailed) was used to measure the statistical significance of the editing and alternative splicing analyses. (F) Analysis of editing and alternative splicing choice of Gria3 mRNA in the HIP of control and Mecp2 KO mice. Percentages correspond to pooled data from sequences of at least two animals per condition. The contingency graph displays flip/flop selection and the associated edited/unedited status. Fisher’s exact test (two-tailed) was used to estimate the statistical significance of the editing and alternative splicing analyses. (G) Western blot analysis of GluR-3 in WT and MeCP2 KO neural progenitor or freely differentiated cells (30 days). Two independent replicates were loaded per condition. (H) Expression levels of GRIA3 mRNA and the T-UCRs uc.478 and uc.479 were analyzed in the same samples as in (G) by qRT-PCR. Graphs represent the median with 95% confidence interval (CI) of the data (two-tailed Mann-Whitney U [GRIA3] or Wilcoxon tests [T-UCRs], *p < 0.05, **p < 0.01). (I) Analysis of editing and alternative splicing choice of GRIA3 mRNA in WT or MeCP2 KO neural progenitor cells. Percentages correspond to pooled data from sequences of at least two replicates per condition. The contingency graph displays flip/flop selection and the associated edited/unedited status. Fisher’s exact test (two-tailed) was used to estimate the statistical significance of the editing and alternative splicing analyses. (J) As in (I) but with cells freely differentiating for 30 days. (K) Western blot analysis of GluR-3 in KO progenitor cells (control or overexpressing MeCP2). The graph on the right represents quantitation of band intensity of two representative experiments (a two-tailed t test was used; ***p < 0.001). (L) Expression levels of GRIA3 mRNA were analyzed in the same samples as in (K) by qRT-PCR. The graph represents the median with 95% CI of the data (two-tailed Mann-Whitney U test). (M) Western blot analysis of GluR-3 in WT neural progenitor cells upon overexpression of uc.478 or uc.479. The graph on the right represents quantitation of band intensity of two representative experiments (one-way ANOVA was used; *p < 0.05). (N) Expression levels of GRIA3 mRNA were analyzed in the same samples as in (M) by qRT-PCR. The graph represents the median with 95% CI of the data (one-way ANOVA was used; *p < 0.05). See also Figure S5. Molecular Therapy: Nucleic Acids Vol. 27 March 2022 631 Molecular Therapy: Nucleic Acids (legend on next page) 632 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org GRIA3 promoter (Figure S5N) but not direct binding of MeCP2 to tion of transcriptional and post-transcriptional mechanisms that GRIA3 mRNA or uc.478 or uc.479 RNAs (Figure S5O), pointing to converge toward alterations in trafficking and glutamate receptor direct transcriptional regulation of the loci by MeCP2. Regulation biogenesis. of GRIA3 by MeCP2 was further confirmed by overexpressing MeCP2 in KO cells, where both protein and mRNA levels were upre- DISCUSSION gulated (Figures 5K and 5L). We next observed that uc.478 and uc.479 The alterations affecting ncRNA species in RTT are largely unknown. are nuclear poly(A) transcripts (Figures S5P and S5Q) and proceeded Several studies have addressed dysregulation of miRNAs acting to overexpress them in WT progenitor cells (where they are present at downstream of MeCP2 or as direct regulators of MeCP2 expression, low levels) from the pLVX-shRNA2 plasmid to generate transcripts and it has been suggested that some of them are potential biomarkers with features that would mimic the endogenous T-UCRs. Overex- or of potential therapeutic value.79,80 However, other types of pression of uc.478 and uc.479 resulted in downregulation of GRIA3 ncRNAs have not been explored. Here we used four experimental mRNA and protein levels (Figures 5M and 5N) but had little effect models (the mouse Mecp2y/ [Bird] model, the mouse N2a neuro- on editing levels and alternative exon selection (Figure S5R). This blastoma cell line knocked out for Mecp2, neural progenitor cells confirms the ability of ultraconserved transcripts to regulate gene knocked out for MeCP2, and postmortem brain regions from affected expression and specifically shows the complex regulation of the individuals) to investigate two specific types of ncRNAs (circRNAs GRIA3 locus. and T-UCRs) that are highly conserved in mouse and human. Alterations in these ncRNA populations have revealed dysregulated The ultraconserved uc.478 and uc.479 RNAs influence key pathways that are common to the different contexts, specifically iden- aspects of RTT neural progenitor cell physiology tifying two fundamental processes: the dynamics of the microtubule Given the observed influence on GluR-3 levels, we went on to inves- cytoskeleton and glutamatergic trafficking and signaling. It is remark- tigate the potential effect of the ultraconserved uc.478 and uc.479 on able that analysis of distinct types of ncRNAs have revealed more general features of RTT cellular models. Previous reports have converging altered pathways in RTT, and our molecular dissection described that neurosphere formation is defective in MeCP2 mutant provides a working model of regulation that integrates gene expres- cells,78 and we observed a decreased size of spheres in our MeCP2-KO sion regulation by T-UCRs and circRNAs to ultimately modulate neural progenitor cells (Figure 6A). Interestingly, overexpression of GluR subunit biogenesis, possibly in coordination with altered cyto- uc.478 or uc.479 on their own was able to restore sphere diameter skeleton trafficking (Figure 7). Arguably, the different experimental in KO cells while not altering WT cells, suggesting that the levels of settings and the intrinsic nature of neurodevelopmental dynamics these ultraconserved ncRNAs play roles in refining the pluripotent have resulted in some cases exhibiting an unexpected direction or differentiating features in the MeCP2 mutant background. In addi- and/or magnitude of difference (e.g., those in uc.479 levels in post- tion, structural analysis of cells driven toward glutamatergic differen- mortem HIP to those in differentiating neural progenitors; Figures tiation indicated that overexpression of uc.478 or uc.479 resulted in a 5D and 5H). However, the coincidence in the altered pathway as a marked decrease in neuronal complexity, as measured by the number whole (GRIA3 biogenesis) highlights the potential for ncRNAs to un- of branchpoints and spines in WT (Figures 6B–6D) and KO cells (Fig- cover hotspots of dysregulation as well as their potential involvement ures 6E–6G). In accordance with this, as the time course of differen- as targets or regulators. tiation progressed, we detected downregulated levels of the glutamate transporters VGlut1 and VGlut2 (at the mRNA and protein levels; As a proxy for ncRNA function, our enrichment analysis of circRNA Figures 6H and 6I), which are associated with synaptic vesicles and and T-UCR host genes revealed Gene Ontology categories related to represent canonical markers of glutamatergic neurons. This suggests glutamatergic signaling, specifically in the HIP, but not in the FC, of a broad effect of ultraconserved transcripts on neural stem cell phys- Mecp2 KO mice (Figures 1B, 1C, 4A, and 4B). Dysfunction in brain iology in the context of RTT. circuits and in the excitatory/inhibitory balance is a manifestation of altered molecular mechanisms downstream of MeCP2 loss but We detected MeCP2-dependent dysregulation of noncoding tran- may differ from region to region. Glutamate signaling has an impor- scripts of the circRNA and T-UCR type that reveal complex regula- tant role in learning and memory through the plasticity, or Figure 6. Phenotypic analysis of neural progenitors and differentiating cells upon overexpression of uc.478 and uc.479 (A) The diameter of neurospheres was measured for WT or MeCP2 KO cells upon overexpression of uc.478 or uc.479. Graphs represent the median and interquartile range (a two-tailed t test was used; *p < 0.05, **p < 0.01, ***p < 0.001). Images correspond to representative spheres of each condition. Scale bar in the sphere, 100 mm. (B) Control WT neural progenitors or cells overexpressing uc.478 or uc.479, as indicated, were driven toward glutamatergic differentiation for 7 days, stained with MAP2, and re- constructed in silico from confocal images with NeuronStudio software. One representative picture for each condition is shown. (C) Automatic analysis of the cells in (B) allowed total branchpoints and spine count per condition (n = 20 neurons, one-way ANOVA was used; *p < 0.05, **p < 0.01). (D) For the same cells as in (B), the abundance of each spine subtype was determined with NeuronStudio software (n = 20 neurons, one-way ANOVA was used; *p < 0.05, **p < 0.01). (E–G) the same analysis as in (B)–(D) was performed in MeCP2 KO cells. (H) Expression levels of VGlut1 and VGlut2 mRNA were analyzed by qRT-PCR in WT or MeCP2 KO cells overexpressing uc.478 or uc.479 and driven toward glutamatergic differentiation for 7 or 14 days, as indicated. The graph represents the mean ± SD of three independent replicas (unpaired t test was used; *p < 0.05, **p < 0.01, ***p < 0.001). (I) Western blot analysis of VGlut1 and VGlut2 protein levels from the same MeCP2 KO cells as in (H). Molecular Therapy: Nucleic Acids Vol. 27 March 2022 633 Molecular Therapy: Nucleic Acids Figure 7. Schematic summarizing the findings in this work Analysis of two types of ncRNAs in mouse and human models of RTT has revealed converging pathways. In WT cells, binding of MeCP2 to circSIRT2 species may prevent upregulation of SIRT2 and excessive deacetylation of cytoskeleton proteins and AMPAR subunits, whereas low levels of T-UCRs originating from the GRIA3 locus contribute to normal expression of GluR proteins. In MeCP2-mutant cells, high levels of circSIRT2s in the cytoplasm promote post-transcriptional upregulation of SIRT2 protein, which results in tubulin and GluR1 destabilization. In addition, upregulated uc.478 and uc.479 levels further downregulate GluR subunits with a global effect on the complexity of glutamatergic neurons. Dashed arrows indicate still unclear mechanisms of regulation. This figure was created with BioRender. modification, of channel properties and synaptic anatomy, most with alternative splicing selection and editing of the gene, strongly in- notably in the HIP of the mammalian central nervous system.81 In fluence the function of the receptor. Editing and splicing have been this regard, glutamate-induced excitotoxicity, which is prominent linked to transport of AMPA receptors, and unedited transcripts in the HIP, has been associated with decreased neuronal regeneration are more efficiently targeted to dendrites,87 with a subsequent effect and dendritic branching, leading to impaired spatial learning,82 and on protein stability. Thus, the hyperediting we observed in RTT sam- recent studies have reported a link between GluR receptors and ples could account for the lower levels of GluR3 protein noted in the RTT. Remarkably, de novo variants in GRIA2 cause neurodevelop- absence of changes in mRNA levels. The great complexity of the mental disorders, including RTT-like features.83 The biogenesis of glutamate system in the developing brain makes it a challenging GRIA transcripts could be under direct control of MeCP2 because target, but understanding its regulatory networks is essential for the protein directly regulates alternative splicing of many genes,84,85 informed design of effective therapies. T-UCR and circRNA function including the flip/flop exons of the Gria1–Gria3 genes (without may form the basis of this biogenesis, as indicated by this study and a changing total mRNA) in the cortex of Mecp2 KO male and Mecp2/+ recent report on circGRIA1, which is involved in age-related loss of female mice, influencing AMPAR-gated current and synaptic trans- synaptic plasticity by controlling GRIA1 transcription.37 The close mission.86 We found no significant differences in exon selection in relationship of these AMPA-type receptors with the PSD95 pathway, the HIP of KO males, suggesting that this fine-tuning of alternative a major excitatory pathway downstream of BDNF and IGF1 splicing is region specific. In contrast, we detected robust changes signaling, deficient in RTT and the target of current therapeutic in editing rates in the postmortem HIP and an engineered neural pro- strategies,24,88,89 is further proof of the importance of understanding genitor cell line depleted in MeCP2 (Figure 5). Our data are consistent their biogenesis. with direct regulation of T-UCRs by MeCP2 in a hypothetical sce- nario where MeCP2 recruitment to intragenic sites promotes or re- The other major altered pathway revealed by the ncRNA analysis in presses overlapping noncoding transcripts. The opposite roles of mice and also observed in the proteomics analysis of samples from MeCP2, uc.478, and uc.479 in GRIA3 expression and protein levels affected individuals was that regulating trafficking and microtubule (Figure 5) suggest complex regulatory dynamics that, in conjunction dynamics. Several studies have highlighted the occurrence of 634 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org microtubule instability and microtubule-dependent deficits in RTT, mediating the regulation; namely, miRNAs, the most studied targets which may underlie the well-documented abnormal neuronal for circRNAs. Future studies will help clarify the detailed mechanism morphology and synaptic plasticity of MeCP2-deficient brains and that explains how the family of circSIRT2 influences SIRT2 protein contribute substantially to the neuropathology of RTT.38,90,91 This in- levels. cludes alterations in microtubule polymerization dynamics in mouse models and MeCP2-mutant fibroblasts63,64 as well as impairment of Despite major advances over the last two decades, restoration of MeCP2 mitotic spindle organization.92 MeCP2 has been coeluted with tubu- levels to the physiological window of activity remains a major technical lins and other cytoskeleton components such as MAP6 (which we challenge in tackling RTT. No drug has so far been approved by the US identify as upregulated in our proteomics analysis of postmortem Food and Drug Administration for treating RTT, and current treat- RTT HIP) in co-immunoprecipitation studies,93 suggesting a direct ments with other medications only partially address some of the effect of MeCP2 loss on cytoskeleton organization. Also related to symptoms. In the absence of an effective therapeutic agent, a better un- cytoskeleton abnormalities, and similar to what is seen in mouse derstanding of the contribution of the noncoding transcriptome to key models of MeCP2 triplication syndrome94 (which has phenotypic dysregulated pathways is essential if we are to gain insights into two ma- similarities to RTT), we also detected an increase in the microtu- jor aspects: (1) identification of molecules that are useful as biomarkers bule-associated tau protein in postmortem HIP. or therapeutic entry targets at all stages of the disease and (2) identifi- cation of new mechanisms and regulatory agents in gene expression Our data highlight the ability of ncRNA dysregulation to reveal key control, metabolic pathways, and signaling processes downstream of pathway abnormalities in the physiopathology of RTT. One general MeCP2 function. Modulation of the expression of specific lncRNAs caveat of most transcriptomic studies in the field is the fact that has shown promising results in Angelman and Dravet syndromes expression profiles of specific genes differ greatly between studies, (whose symptoms overlap with those of RTT),98,99 and we anticipate as meta-analysis have reported.54,95 This is probably a result of the that ncRNAs may become very useful tools in clinical research of RTT. heterogeneous models of study and the high dependency of MeCP2-directed gene expression regulation on varying neurodeve- MATERIALS AND METHODS lopmental conditions. However, the most severely changed biological Animals and RNA extraction for microarray analysis pathways and processes tend to be the same in the different ap- Total RNA from the B6.129P2(c)-Mecp2tm1+1Bird mouse model of proaches, and our analysis of the most conserved subset of the non- RTT25 was analyzed. The animals were purchased from The Jackson coding transcriptome suggests that dysregulated ncRNAs are Laboratory (Bar Harbor, ME, USA; stock number 003890) and main- involved in the same key pathways. Further research on the mecha- tained on a C57BL/6J background. All animals were kept under spe- nism of action of candidate species will provide insights into their cific pathogen-free conditions with a 12-h light-dark cycle, with value as potential therapeutic targets or as biomarkers of the disease. drinking water and food available ad libitum. 20 Mecp2/y (KO- We anticipate that, in most cases, the concordant expression between null male) and 20 WT mice were used. All procedures and host genes and circRNAs or T-UCRs confers the features of a putative experiments were approved by the Ethics Committee for Animal biomarker. This might be particularly relevant for circular transcripts Experiments of the Bellvitge Biomedical Research Institute (IDI- because they are stable molecules present in exosomes that are able to BELL) under the guidelines of Spanish animal welfare laws. Mice cross the blood-brain barrier, becoming accessible in most human were euthanized following the biomedical research endpoint guide- biofluids and circumventing the problematic availability of brain bi- lines. Each brain region was isolated, weighed, and frozen on dry opsies.96 Further research is needed to ensure the maintenance of ice immediately after extraction and stored at 80 C until use. origination features, but exosomal circRNAs that could be specifically associated with early stages of neurodevelopmental disorders may be Total RNA was extracted using the miRCURY RNA Isolation Kit – a promising tool for improving current diagnostic limitations. Cell & Plant (300110, Exiqon, Vedbaek, Denmark) according to the manufacturer’s recommendations. 20 mg of HIP and FC from 3 In contrast, we also confirmed several cases of antagonistic expres- symptomatic (8-week-old) and 3 pre-symptomatic (4-week-old) an- sion, whereby production of the ncRNA may cross-talk with expres- imals was used per condition (WT and KO). The isolated RNA was sion of the host gene. One important example is the SIRT2 locus, treated with RQ1 RNase-Free DNase (M6101, Promega, Madison, where early alterations in circRNA expression could confer WI, USA) to remove all DNA traces, and RNA integrity was assessed biomarker features and reveal an extra layer of regulation of the with a 2100 BioAnalyzer (Agilent Technologies, Santa Clara, CA, SIRT2 gene mediated by ncRNAs. In this case, we hypothesize that USA) microcapillary electrophoresis system using the RNA Nano the regulation we have shown by the circRNAs originating from the 6000 Kit (Agilent Technologies, Santa Clara, CA, USA) in accordance same locus in mouse and human models could function at the trans- with the manufacturer’s instructions. The integrity of the RNA of all lational stage or even through interaction with the SIRT2 protein, isolated RNA samples was analyzed using 5 ng RNA/sample input. similar to what has been described for circPABPN1 or circMBNL1,9,97 and underlie the changes in a-tubulin and GluR1 modifications that Postmortem samples contribute to aberrant microtubule dynamics and AMPAR forma- Postmortem brain tissue from control individuals and individuals tion. However, we cannot rule out involvement of third partners with RTT was obtained from the National Institutes of Health Molecular Therapy: Nucleic Acids Vol. 27 March 2022 635 Molecular Therapy: Nucleic Acids (NIH) NeuroBioBank at the University of Maryland (Baltimore, MD, cells were detached and resuspended in PBS for flow cytometry selec- USA) and the Human Brain and Spinal Fluid Resource Center, VA tion. Cells containing green fluorescence (EGFP+) were selected by West Los Angeles Healthcare Care Center (Los Angeles, CA, USA), fluorescence-activated cell sorting (FACS) and then pooled and cloned. which is sponsored by the National Institute of Neurological Disor- Sanger sequencing was performed using a 3730 DNA Analyzer ders (NMSS), and Stroke (NINDS)/National Institute of Mental (Applied Biosystems, Foster City, CA, USA) to confirm insertions or Health (NIMH), the National Multiple Sclerosis Society, and the deletions (indels) within the targeted region after non-homologous Department of Veterans Affairs. end joining, resulting in a premature stop codon at position 133 of the amino acid sequence. For mouse N2a cells, all transfections were Cell culture of progenitor and differentiating cells performed in 6-well plates, with 60%–80% confluence at the time of Immortalized human neural progenitors (ReNCells VM, SCC008) transfection. The gRNAs targeting exons 3 and 4 of the Mecp2 genomic were purchased from Merck Millipore (Burlington, MA, USA). The sequence (5ʹ-GGGCGCTCCATTATCCGTGAC-3ʹ and 5ʹ-GGATTT cells were obtained from 10-week-old ventral mesencephalon of fetal TGACTTCACGGTAAC-3ʹ, respectively) were designed based on brain and immortalized with v-myc transfection, maintaining a CrispR Gold software,103 and forward and reverse oligos were mixed pluripotent and proliferative state in the presence of two growth fac- and phosphorylated individually. Then annealed oligo duplexes were tors: basic fibroblast growth factor (bFGF-2) (20 ng/mL, SRP4037) cloned into the BbsI sites of pSpCas9 (BB)-2A-GFP (PX458) (48138, and epidermal growth factor (EGF) (20 ng/mL; SRP3027, Sigma-Al- Addgene, Watertown, MA, USA). Transfection was performed with drich, St. Louis, MO, USA). ReNCells VM were cultured on laminin- jetPRIME (114-15, Polyplus-transfection, Strasbourg, France) using coated plates (20 mg/mL; Sigma-Aldrich, St. Louis, MO, USA) with 2 mg DNA and 4 mL reagent per well in 2 mL of cell growth medium, DMEM-F12 medium (L0093-500; Biowest, Nuaillé, France) contain- following the manufacturer’s instructions. 48 h after transfection, the ing 0.2% heparin (STEMCELL Technologies, Vancouver, BC, Can- transductant cells (EGFP+) were observed using optical microscopy. ada), B-27 complete vitamins (17504-044, Invitrogen, Carlsbad, Isolation of clonal cells with specific modifications was performed by CA, USA), and antibiotic-antimycotic (L0010-100, Biowest, Nuaillé, serial dilution, followed by an expansion time to establish a new clonal France) at 37 C with maximum humidity in an atmosphere of 5% cell line. Mecp2 KO was confirmed by western blotting. CO2. For subculturing, cells were detached using Accutase (SCR005, Sigma-Aldrich, St. Louis, MO, USA) and centrifuged for Plasmid construction 5 min at 300  g. Pellets were gently resuspended and plated after MeCP2 was restored in mouse and human MeCP2 KO cells. The cell counting. The medium was replaced every other day, and cells mouse Mecp2 sequence was cloned into the pLVX-IRES-ZsGreen1 were sub-cultured every 3–6 days when 90% confluence was reached. (632187, Clontech, Mountain View, CA, USA) vector using the oligos Spontaneous differentiation was performed by withdrawing EGF and mMeCP2-EcoRI Fw and hMeCP2-BamHI Rv, into EcoRI and BamHI bFGF-2 from the culture medium when 90% confluence was reached; cutting sites. For the human MECP2 sequence, exons 4 and 3 were ReNCells VM differentiated into a mixture of populations of neurons amplified from the genome with overlapping sequences using the oli- and glia (as described elsewhere).100 Treatment with AGK2 (A8231, gos hMeCP2e1-Fus Fw and hMeCP2-ex3 Rv (for exon 3) and Sigma-Aldrich, St. Louis, MO, USA) was conducted at 5 mM for 24 hMeCP2-ex4 Fw and hMeCP2-BamHI Rv (for exon 4). Moreover, h; an equivalent concentration of dimethyl sulfoxide vehicle served exon 1 was engineered to replace 14 nucleotides with high content as a control. Mouse Neuro 2a (N2a) cells (American Type Culture of C and G; the triplet substitution was done considering the prefer- Collection [ATCC] CCL-131) were maintained in DMEM supple- ential codon use, and the corresponding 13 amino acids were not mented with 10% fetal bovine serum (10270, Thermo Fisher Scienti- altered. In both strategies, a Kozak sequence was inserted. Human fic, Waltham, MA, USA) at 37 C in a humidified 5% CO2 atmo- circRNAs were cloned into the BstEII-H and SacII sites of the sphere, as described elsewhere.101 pcDNA3.1(+) CircRNA Mini Vector (60648, Addgene, Watertown, MA, USA) with the following oligos: circ_0050945 BstEII Fw (for- MeCP2 editing by CRISPR-Cas9 ward for hsa_circ_0050945, hsa_circ_0050946, and hsa_- CRISPR-Cas9 technology102 was used for engineering MECP2 KO in circ_0050948), circ_0050951 BstEII Fw (forward for hsa_circ_ human ReNCells VM and mice N2a cells. The guide RNA (gRNA) tar- 0050951), circ_0050945 SacII Rv (reverse for hsa_circ_0050945), geting exon 4 of the human MECP2 genomic sequence (5ʹ-AAAA circ_0050946 SacII Rv (reverse for hsa_circ_0050946), circ_ GCCTTTCGCTCTAAAG-3ʹ) was designed using the CRISPR Design 0050948 SacII Rv (reverse for hsa_circ_0050948), and circ_0050951 tool page (http://crispr.mit.edu/), and the best-scored guide was SacII Rv (reverse for hsa_circ_0050951). Mouse circRNA (mmu_ selected and cloned into pSpCas9 (BB)-2A-GFP (PX458) (48138, circRNA_41253) was cloned between EcoRV and BstEII sites with Addgene, Watertown, MA, USA). The plasmid (2 mg) was delivered the following oligos: mmu_circ_41253 EcoRV Fw and mmu_ by nucleofection with A33 voltage (VPI-1003, Primary Neurons circ_41253 BstEII Rv. To improve circularization, a sequence of 35 Kit, Amaxa, Lonza, Basel, Switzerland) in progenitor ReNCells VM nt upstream and downstream of introns was added. uc.478 and (2 million cells per experiment). After nucleofection, cells were coated uc.479 were overexpressed by cloning into the BamHI and EcoRI sites on laminin-coated plates in complete medium (with EGF and bFGF-2) of the vector pLVX-shRNA2 (632179, Clontech, Mountain View, CA, supplemented with Revitacell (A2644501, Thermo Fisher Scientific, USA) using the following oligos: Uc478-BamHI Fw and Uc478-EcoRI Waltham, MA, USA) for the first day. Two days after nucleofection, Rv for uc.478 and Uc479-BamHI Fw and Uc479-EcoRI Rv for uc.479. 636 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org The pcDNA4TO- SIRT2-hemagglutinin (HA) plasmid104 was a gift Analysis of neuronal morphology from Dr. Vaquero’s lab (IJC, Badalona, Spain). WT and MeCP2-KO cells overexpressing T-UCRs and the pLVX- shRNA2 empty vector were differentiated on glass coverslips for Transfection and lentivirus production/infection 7 days following the glutamatergic differentiation protocol Cells were plated on 100-mm dishes 24 h before the experiment, and described above. Neuronal morphology was analyzed after an cloned circRNAs plasmids were transfected with Lipofectamine Stem immunofluorescence assay as follows. the cells were fixed with 4% Reagent (STEM00003, Thermo Fisher Scientific, Waltham, MA, paraformaldehyde in PBS buffer with 1 mM NaCl and 1 mM USA) following the company guidelines. The empty pcDNA3.1(+) MgCl2 for 30 min at room temperature (RT). The coverslips were CircRNA Mini Vector was used as a control. For pLVX-IRES- blocked with 5% BSA and 5% goat serum in 0.3% Triton X-100 ZsGreen1, pLVX-shRNA2 constructs and glutamatergic differentia- (X100, Sigma-Aldrich, St. Louis, MO, USA) in PBS buffer for 1 h tion vectors (TetOhNGN2-P2A-EGFP-T2A-PuroR and CMV- at RT. The primary antibody anti-MAP2 (1:400; 8707, Cell Signaling rtTA), lentiviral particles were produced in HEK293T cells as Technology, Danvers, MA, USA) was diluted in the same blocking described below. HEK293T cells were cultivated in DMEM with solution and incubated overnight. A secondary antibody (1:1,000, GlutaMAX (31966-021, Thermo Fisher Scientific, Waltham, MA, anti-rabbit Alexa Fluor 555; A31572, Thermo Fisher Scientific, Wal- USA) supplemented with 10% fetal bovine serum (10270, Thermo tham, MA, USA) was incubated for 1 h at RT in the dark. Coverslips Fisher Scientific, Waltham, MA, USA) and grown at 37 C in a humid- were washed between steps with PBS three times for 5 min each. ified atmosphere of 5% CO2. HEK293T cells were triple-transfected DAPI (5 mg/mL) was used for nucleus staining. Coverslips were with specific vectors plus packaging plasmids with jetPRIME (114- mounted in Mowiol solution and dried overnight at RT, protected 15, Polyplus-transfection, Strasbourg, France) according to the from the light. The images were acquired under a Leica SP5 confocal manufacturer’s recommendations. Viral particles with pLVX-IRES- microscope (CMRB, L’Hospitalet de Llobregat, Spain) and analyzed ZsGreen1 and pLVX-shRNA2 empty vector were used as a control. using ImageJ software (v.1.53k). Neurons were reconstructed using Lenti-X Concentration (PT4421-2, Clontech, Mountain View, CA, NeuroStudio software (v.0.9.92) as described elsewhere.49 Briefly, a USA) was used for concentrating viral particles in DMEM-F12 me- three-dimensional dendritic network was built with a Sholl analysis dium. Cells were infected with viral particles, followed by a 2-day by automated segmentation, and concentric circles were drawn from recovery period before the experiments started. ZsGreen1 was used the soma (over a length of 20 mm with concentric circles of 10 mm) as a marker to visualize transductants cells by fluorescence throughout the projection’s length. Neuronal components are cate- microscopy for pLVX-IRES-ZsGreen1 plasmids, whereas the TetO- gorized and represented by spheres: the soma is represented in red; hNGN2-P2A-EGFP-T2A-PuroR vector was selected by puromycin neurites are tracked in green, branchpoints are represented in yel- (ant-pr, InvivoGen, San Diego, CA, USA) and visualized with low, and spines are colored in orange, pink, and gray according to EGFP fluorescence. their subtype (stubby, mushroom, and thin). An estimation of spines density was retrieved, and the spines were counted (n = 20 Glutamatergic differentiation protocol neurons per condition). Glutamatergic differentiation was performed as described by Ho et al.105 Briefly, cells were seeded on laminin-coated plates 24 h before circRNA microarray analysis infection. Cells were double-infected with TetO-hNGN2-P2A-EGFP- Total RNA from each sample was treated with RNase R to enrich the T2A-PuroR and CMV-rtTA as described above. 48 h after infection, circRNA. The enriched circRNA was then amplified and transcribed doxycycline (1 mg/mL) was administered to the cells, followed by pu- into fluorescent cRNA using random primers according to the Ar- romycin (0.5 mg/mL) the next day. After 24 h of puromycin selection, raystar Super RNA Labeling protocol (Arraystar, Rockville, MD, EGF and FGF-2 were withdrawn from the medium, and cells were USA). The labeled cRNAs were hybridized onto the Arraystar Mouse allowed to differentiate for 21 days. During the first week of differen- Circular RNA Microarray V2 (8  15 K) and incubated for 17 h at tiation, half of the medium was changed every day, maintaining 65 C in an Agilent hybridization oven. After washing with PBS, slides doxycycline concentration. In the second week of differentiation, were scanned with a G2505C scanner (Agilent Technologies, Santa the medium was changed every other day, reducing doxycycline con- Clara, CA, USA). centration to half with each change until complete removal by the last week of differentiation. qRT-PCR was performed to confirmed over- circRNA data analysis expression of glutamatergic marker VGlut1. Data were extracted using Agilent Feature Extraction software. A se- ries of data processing operations, including quantile normalization, Neurosphere formation assay was performed using the Limma 3.22.7 package in R. After quantile Neurosphere formation was performed as proposed by Donato et al.100 normalization of the raw data, low-intensity filtering was performed, Briefly, 2.5 million cells were cultured in non-coated flasks with com- and the circRNAs for which at least 3 of 12 samples had “P” or “M” plete medium (EGF and FGF-2) to induce neurospheres formation. flags (defined by GeneSpring software) were retained for further an- After 24 h of plating, images were acquired with a 20 objective in un- alyses. Significantly DE circRNAs (t test, p < 0.05) and a greater than der a DM IL LED Fluo microscope (Leica, Wetzlar, Germany). Sphere 1.5-fold difference in expression between the two groups were diameter was measured using ImageJ software (v.1.53k). identified through volcano plot filtering. Hierarchical clustering was Molecular Therapy: Nucleic Acids Vol. 27 March 2022 637 Molecular Therapy: Nucleic Acids performed to distinguish different patterns of circRNA expression pulled down RNA was extracted by adding 1 mL of TRIzol reagent among the samples. (15596-018, Thermo Fisher Scientific, Waltham, MA, USA). After phenol extraction, isopropanol precipitation, and DNase treatment, T-UCR microarray analysis the final pellet was resuspended in 20 mL H2O. Each RNA sample We used a customized single-channel Agilent array (Agilent Technol- was split into 2 tubes; one tube was reverse transcribed, and the other ogies, Santa Clara, CA, USA), which was deposited at the Gene was mock treated (without reverse transcriptase). cDNA was analyzed Expression Omnibus database (accession number GPL22858). It con- by qPCR. Input material processed in parallel was used to estimate tains a collection of sense and antisense probes for various types of pull-down efficiency. ncRNAs (18,009 probes corresponding to 1,271 human pre-miRNAs, 8,660 probes corresponding to 626 mouse pre-miRNAs [miRBase RNA extraction and qRT-PCR 21], 2,745 probes corresponding to 479 ultra-conserved elements, For mRNA, circRNA, and T-UCR expression analysis, total RNA was 16,314 probes corresponding to 1,283 human-specific transcribed py- extracted with the Maxwell RSC miRNA Tissue Kit (AS1460, Prom- knons, and 2,197 probes corresponding to 97 lncRNAs). The hybrid- ega, Madison, WI, USA). The isolated RNA was treated with RQ1 ization was performed as described.106 Data preprocessing steps of RNase-Free DNase (M6101, Promega, Madison, WI, USA) (twice background correction, normalization, and summarization were per- for the T-UCR analysis) and reverse transcribed using the RevertAid formed in R v.3.5.1 (https://www.r-project.org/) using functions in RT Reverse Transcription Kit (K1691, Thermo Fisher Scientific, Wal- the Limma library.107 A threshold for positive spot selection was tham, MA, USA). A negative control without reverse transcriptase calculated as the mean value of all dark corner spots plus twice the was run in parallel to control for genomic contamination. Real-time standard deviation.108 Linear models and empirical Bayes methods PCR reactions were performed in triplicate in the QuantStudio 5 from Limma were used to obtain the statistics and assess differential real-time PCR system (Thermo Fisher Scientific, Waltham, MA, gene expression between two sample types. Statistical significance was USA), using 10–60 ng cDNA, 6 mL SYBR Green PCR Master Mix defined as a p value of less than 0.05, and we imposed a cutoff of func- (Applied Biosystems, Foster City, CA, USA), and 416 nM primers tional relevance on the fold change in absolute value of 1.1. in a final volume of 12 mL for 384-well plates. All data were acquired and analyzed with QuantStudio Design & Analysis Software v.1.3.1 GluR-1 immunoprecipitation (IP) and normalized with respect to B2m as an endogenous control in Cell pellets were lysed in stringent RIPA buffer (50 mM Tris-HCl [pH mouse samples and L13 in human samples. Relative RNA levels 7.4], 50 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 1% SDS, and were calculated using the comparative Ct method (DDCt). A list of 1 protease inhibitor cocktail) supplemented with deacetylase inhib- the primers used can be found in Table S1. itors (100 mM trichostatin A, 50 mM sodium butyrate, and 50 mM nicotinamide), incubating for 30 min in a rocker at 4 C. Lysates Western blot were then cleared by centrifugation, and protein concentration was 20 mg of mouse FC or HIP and human postmortem HIP were resus- measured. 1 mg of total extract was incubated with 5 mg of anti- pended in Laemmli buffer (2% SDS, 10% glycerol, 60 mM Tris-Cl GluR-1 rabbit antibody (ab109450, Abcam, Cambridge, UK) or con- [pH 6.8], and 0.01% bromophenol blue), sonicated, and boiled for trol rabbit immunoglobulin G (IgG) antibody (I5006, Sigma-Aldrich, 5 min at 95 C. The sample concentration was determined by quanti- St. Louis, MO, USA) and 60 mL of Dynabeads M-280 sheep anti-rab- fying the absorbance at 260 nm with a NanoDrop One/OneC Micro- bit IgG (11204D, Thermo Fisher Scientific, Waltham, MA, USA) for volume UV spectrophotometer and using the equivalency between 1 h at 4 C. Beads were then washed three times with lysis buffer and DNA and histone quantity (6 units A260nm [DNA] = 1 mg/mL of pro- boiled in the presence of Laemmli buffer for western blot analysis. tein). Proteins were transferred to a 0.2-mm nitrocellulose membrane (10600001, GE Healthcare, Chicago, IL, USA) and incubated over- RNA IP night at 4 C with a primary antibody diluted in 5% skim milk in To retrieve the RNA species bound by MeCP2, we used the protocol PBS containing 0.1% Tween 20 (663684B, Atlas Chemical, Houston, described by Cheng et al.109 with minor variations. Cell pellets were TX, USA). After three washes with PBS containing 0.1% Tween 20, resuspended in 1 mL of CLIP lysis buffer (50 mM Tris-HCl [pH membranes were incubated for 1 h at RT in a benchtop shaker with 7.4], 100 mM NaCl, 1% Igepal CA-630, 0.1% SDS, 0.5% sodium de- the secondary antibodies conjugated to horseradish peroxidase anti- oxycholate, and 1 protease inhibitor cocktail), sonicated on ice at rabbit IgG (1:10,000, A0545; Sigma-Aldrich, St. Louis, MO, USA) 10%–15% of amplitude for 30 s (5 s on and 10 s off), and treated or anti-mouse IgG (1:5,000; NA9310, GE Healthcare, Chicago, IL, with 14 U of Turbo DNase (AM2238, Thermo Fisher Scientific, Wal- USA). Enhanced chemiluminescence (ECL) reagents (Luminata- tham, MA, USA) at 37 C for 5 min. Lysates were then cleared by HRT; Merck Millipore, Burlington, MA, USA) were used to visualize centrifugation, and protein concentration was measured. mg of pro- the proteins. Films were scanned, and the band analysis tool of the tein was immunoprecipitated for 2 h at RT with 5 mg of MECP2 anti- Quantity One software application was used for background subtrac- body (M9317, Sigma-Aldrich, St. Louis, MO, USA) or control rabbit tion and to determine the density of the bands. The proteins detected IgG antibody (I5006, Sigma-Aldrich, St. Louis, MO, USA) preincu- were MeCP2 (1:1,000; ab2829, Abcam, Cambridge, UK), SIRT2 bated for 30 min at RT with 50 mL Dynabeads M-280 sheep anti-rab- (1:2,000; ab211033, Abcam, Cambridge, UK), acetyl-a-tubulin bit IgG (11204D, Thermo Fisher Scientific, Waltham, MA, USA). The (1:2,000; T6793, Sigma-Aldrich, St. Louis, MO, USA), a-tubulin 638 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org horseradish peroxidase (HRP) (1:5,000; ab40742, Abcam, Cambridge, 4 mg of normal IgG control antibody (I5006, Sigma-Aldrich, St. Louis, UK), b-actin HRP (1:30,000; a3854, Sigma-Aldrich, St. Louis, MO, MO, USA) were coupled for 2 h at 4 C with 20 mL of Dynabeads. Pre- USA), TAU (1:1,000; ab64193, Abcam, Cambridge, UK), MAP6/ cleared extracts were incubated with the antibody (Ab)-bead com- STOP (1:1,000; 4265, Cell Signal, Danvers, MA, USA), STMN1 plexes overnight at 4 C. The recovered material was treated with (1:500; 3352, Cell Signal, Danvers, MA, USA), S100b (1:1,000; RNAse, DNAse-free (11119915001, Merck Millipore, Burlington, Z0311, Dako, Carpinteria, CA, USA), GluR-1 (1:1,000; ab109450, Ab- MA, USA) and reverse cross-linked with 0.2 mg/mL of Proteinase cam, Cambridge, UK), GluR-2 (1:1,000, 13607, Cell Signal, Danvers, K overnight at 65 C, followed by phenol/chloroform extraction and MA, USA), GluR-3 (1:2,000; 4676, Cell Signal, Danvers, MA, USA), ethanol precipitation. Immunoprecipitated DNA samples (MeCP2, acetyl-lysine (1:1,000; ab80178, Abcam, Cambridge, UK), ADAR2 IgG, input, and histone H3, the two former two with 1:10 dilution) (1:400; HPA018277, Atlas Antibodies, Bromma, Sweden), SOX2 were analyzed by qRT-PCR using SYBR Green Master Mix (Applied (1:1,000; 3579, Cell Signal, Danvers, MA, USA), and Ki67 (1:1,000; Biosystems, Foster City, CA, USA) in the QuantStudio 5 real-time ab16667, Abcam, Cambridge, UK). PCR system (Thermo Fisher Scientific, Waltham, MA, USA). The primers used are shown in Table S1. Nuclear and cytoplasmic fractionation and poly(A) selection Subcellular fractionation was performed with a PARIS kit RNA ISH (AM1921, Thermo Fisher Scientific, Waltham, MA, USA) ReNCells VM were seeded on coverslips pre-treated with laminin. following the manufacturer’s instructions. Proportional amounts Following two washes with PBS (5 mM CaCl2 and 5 mM MgCl2), cells of RNA from each fraction were subjected to qRT-PCR, and the were fixed with 4% paraformaldehyde for 10 min and washed twice results were normalized, considering the total quantity of RNA with PBS. RNA was hybridized with the hsa_circ_0050946 probe recovered from each fraction. MALAT1 and GAPDH primers (VPTZ76X, Thermo Fisher Scientific, Waltham, MA, USA) using were used as controls to verify nuclear and cytoplasmic separation the ViewRNA miRNA ISH cell assay (QVCM0001, Thermo Fisher of the mRNA. At the protein level, separation was confirmed Scientific, Waltham, MA, USA) in a HybEZ II oven (Advanced Cell by western blot with histone H3 (1:5,000; ab1791, Abcam, Cam- Diagnostic, San Francisco, CA, USA) according to the manufacturer’s bridge, UK) and a-tubulin HRP (1:5,000; ab40742, Abcam, recommendations. Coverslips were blocked with blocking solution Cambridge, UK). Additionally, MeCP2 (1:1,000; ab2829, Abcam, (5% chicken serum and 0.1% Triton X-100 in PBS) for 1 h at RT. Cov- Cambridge, UK) was detected by western blot to confirm nuclear erslips were then incubated with the neural protein TUJ1 Ab (1:500; localization and absence in human KO cells. A Dynabeads mRNA T8660, Sigma-Aldrich, St. Louis, MO, USA) in blocking solution Purification Kit (61006, Thermo Fisher Scientific, Waltham, MA, overnight at 4 C in the dark and covered with wet tissues. The next USA) was used to separate poly(A)+ and poly(A) RNAs, day, coverslips were washed three times with PBS and incubated following the manufacturer’s instructions and performing three with the secondary Ab Alexa Fluor 488 chicken anti-mouse rounds of selection. RNA enrichment in each sample was analyzed (1:1,000; A-21200, Thermo Fisher Scientific, Waltham, MA, USA) by qRT-PCR, using GAPDH and MALAT1 as controls for in blocking solution for 45 min at RT. After three washes with PBS, poly(A)+ and poly(A) samples, respectively. coverslips were incubated with DAPI (1:10,000) for 5 min at RT before mounting on microscope slides with FluorSave reagent ChIP assay (345789, Merck Millipore, Burlington, MA, USA), dried overnight For MeCP2 ChIP, the commercially available truChIP Chromatin at RT in the dark, and finally stored at 4 C. Images of hybridized cells Shearing Reagent Kit was used (PN 520154, Covaris, Woburn, MA, were collected using an Axio Observer Z1 fluorescence microscope USA). Cross-linking, quenching, nucleus preparation, and chromatin with ZEN blue 2012 software (Zeiss, Jena, Germany) and analyzed shearing are described above. In brief, in 150-mm dishes, 15 million with ImageJ v.1.51h. cells were washed once with cold PBS and cross-linked with fresh 1% methanol-free paraformaldehyde (final concentration) for 10 min at Characterization of mutations of individuals with RTT RT with gentle shaking. Quenching buffer was added to stop cross- Genomic DNA was extracted from frozen cerebral tissue using the sa- linking and incubated with gentle shaking at RT for 5 min. Cell pellets line-urea method as follows. Frozen tissue was washed with saline were lysed for 10 min at 4 C on a rotor. Nuclei pellets were resus- buffer (2 M Tris and MgCl2), and the pellet was rescued after centri- pended in 1 mL shearing buffer containing protease inhibitor and fugation and washed with urea buffer (urea, 5 M NaCl, 0.5 M EDTA, transferred to milliTUBE AFA Fiber (520130, Covaris, Woburn, 2 M Tris, and 10% SDS). Samples were then incubated for 2 h at 56 C MA, USA). Chromatin was sheared during 10 min in an M220 Cova- and subjected to phenol-chloroform extraction and precipitation with ris Focused-ultrasonicator at 10% duty factor, 75 P.I.P., 200 cbp, and 3 M sodium acetate and ethanol. Pellets were resuspended with TE 6.5 W to obtain fragments of 150–400 bp. 30 mg of chromatin was buffer. All DNA samples were quantified using a Qubit 2.0 fluorom- used as the input for each IP. Chromatin extracts were precleared eter (Invitrogen, Carlsbad, CA, USA). DNA purity was quantified by for 2 h at 4 C with 20 mL of Dynabeads M-280 sheep anti-rabbit calculating the absorbance ratio (A260/A280) with a NanoDrop 1000 IgG (11203D, Thermo Fisher Scientific, Waltham, MA, USA). 4 mg (Thermo Fisher Scientific, Waltham, MA, USA). The DNA was of MeCP2 antibody (M9317, Sigma-Aldrich, St. Louis, MO, USA), also Sanger sequenced to confirm MeCP2 mutations in RTT samples 4 mg histone H3 antibody (ab1791, Abcam, Cambridge, UK), and and to exclude mutations in control samples, as described Molecular Therapy: Nucleic Acids Vol. 27 March 2022 639 Molecular Therapy: Nucleic Acids elsewhere.110 A list of the primers used for sequencing can be found in were chosen randomly and sequenced using a 3730 DNA Analyzer Table S1. (Applied Biosystems, Foster City, CA, USA). Results were analyzed with BioEdit v.7.2.5 software, and methylated cytosines were mapped. Whole-proteome analysis Cytosine nucleotides that remained unaltered were transformed into Four biopsies from human healthy HIP and HIP from individuals percentages of CpGs showing methylation. with RTT were processed, digested, and analyzed in an Orbitrap Fusion Lumos (Thermo Fisher Scientific, Waltham, MA, USA) at Statistical analysis the CRG Proteomics Core Facility (Barcelona, Spain) following stan- All experiments were performed at least in triplicate. Graphs and sta- dard procedures. Raw data were analyzed with MaxQuant software tistical analyses were obtained with GraphPad Prism 8.2.0. Unless v.1.6.2.6a. The statistical analysis was performed with the Limma otherwise stated, qRT-PCR data are expressed as the median with package in R, which was used to implement a linear model. Proteins interquartile range. Comparative analyses of differences between quantified in at least three of four samples of each condition (control experimental groups were performed using Mann-Whitney U test and RTT) were analyzed further. 51 significantly regulated proteins or unpaired-sample t tests and one-way ANOVA with a Tukey or (Benjamini-Hochberg-adjusted p < 0.05) were obtained. Bonferroni post hoc test for intergroup comparisons. Results were considered significant for values of p < 0.05, p < 0.01, or p < 0.001 Editing and alternative splicing analysis (***). Gene enrichment analyses from significantly altered genes To identify the selected exon and the editing levels in GRIA3, the re- were conducted using Enrichr v.2.0 software (http://amp.pharm. gion surrounding the interrogated edited position (including the mssm.edu/Enrichr/)111,112 and https://reactome.org/.113 alternative exon) was amplified with specific primers by using 250 ng of cDNA from each sample and Immolase DNA polymerase SUPPLEMENTAL INFORMATION (BIO21047, Meridian Bioscience, Cincinnati, OH, USA) for 35 cycles. Supplemental information can be found online at https://doi.org/10. The resulting PCR products were gel purified with NucleoSpin Gel 1016/j.omtn.2021.12.030. and PCR Clean-up (740609.250, Macherey-Nagel, Düren, Germany) and cloned into the pGEM-T Easy Vector (A1360, Promega, Madi- ACKNOWLEDGMENTS son, WI, USA) according to the manufacturer’s protocol. The We thank the CERCA program (Generalitat de Catalunya) and the GRIA3 insert of each colony was amplified with Immolase DNA po- Josep Carreras Foundation for institutional support, Dr. Vaquero lymerase (BIO21047, Meridian Bioscience, Cincinnati, OH, USA) for (IJC, Badalona, Spain) for the generous gift of the SIRT2-HA plasmid, 35 cycles using specific pGEM-T primers and precipitated with 87.7% and members of the Pasterkamp lab for discussions and help. This of absolute ethanol and 12.3% of 3 M NaAc. The precipitated PCR work was supported by the Ministerio de Economía y Competitividad product was sequenced using the T7 primer 50 TAATACGAC and the Instituto de Salud Carlos III, co-financed by the European TCACTATAGGG0 3 in the 3730 DNA Analyzer (Applied Biosystems, Development Regional Fund, “A way to achieve Europe” ERDF, un- Foster City, CA, USA) and analyzed with BioEdit v.7.2.5 software. For der grants PI15/01159 and PI20/00389 (to J.A.), the crowd-funding the postmortem samples, results represent the pool of sequences from program PRECIPITA, FondoBiorett and Mi Princesa Rett (to J.A.), three controls and three individuals with RTT. After sequencing anal- the Health and Science Departments of the Catalan Government ysis, the percentages of A or G and the corresponding exons were (Generalitat de Catalunya), the Catalan and Spanish Associations calculated and analyzed with Fisher’s exact test. for Rett Syndrome (FinRett, to S.G.), the Epilepsiefonds (WAR18- 05, to R.J.P.), and the European Union Horizon 2020 Research and DNA methylation analysis Innovation Program under Marie Skłodowska-Curie grant agreement The methylation state of GRIA3 alternative exons was determined by 721890 (cirRTrain ITN, to R.J.P.). We are indebted to the “Biobanc de bisulfite genomic sequencing (BSP). The Methyl Primer Express v.1.0 l’Hospital Infantil Sant Joan de Déu per a la Investigació,” integrated software (Thermo Fisher Scientific, Waltham, MA, USA) was used to in the Spanish Biobank Network of ISCIII, for sample and data pro- design specific primers to amplify the regions of interest (a list of oli- curement. G.A.C. is the Felix L. Haas Endowed Professor in Basic Sci- gos can be found in Table S1). Genomic DNA (1 mg) was subjected to ence. G.A.C.’s laboratory is supported by National Institutes of Health sodium bisulfite treatment using the EZ DNA Methylation-Gold Kit (NIH/NCATS) grant UH3TR00943-01 through the NIH Common (D5006, Zymo Research, Irvine, CA, USA), which induces chemical Fund, Office of Strategic Coordination (OSC); NCI grants 1R01 conversion of unmethylated, but not methylated, cytosine to uracil. CA182905-01 and 1R01CA222007-01A1; NIGMS 1R01GM122775- Fragments of 200- to 300-bp pairs were amplified using 1.2 mL 01; a Team DOD grant (CA160445P1); a Chronic Lymphocytic Leu- of bisulfite-converted DNA with Immolase DNA polymerase kemia Moonshot Flagship project; a CLL Global Research Foundation (BIO21047, Meridian Bioscience, Cincinnati, OH, USA) in a final vol- 2019 grant; a CLL Global Research Foundation 2020 grant; donor ume of 15 mL for 40 cycles. The resulting PCR products were gel support through Dr. Jaffer Ajani; and the Estate of C.G. Johnson, purified with NucleoSpin Gel and PCR Clean-up (740609.250, Ma- Jr. A.O.-G. is a pre-doctoral fellow funded by MINECO (BES-2015- cherey-Nagel, Düren, Germany) and then cloned into the pGEM-T 071452) and the recipient of an EMBO short-term fellowship Easy Vector (A1360, Promega, Madison, WI, USA) according to (8129). E.S. is a pre-doctoral fellow funded by a “Science without bor- the manufacturer’s protocol. For all samples, a minimum of 20 clones ders” scholarship (202074/2015-3) from Conselho Nacional de 640 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org Desenvolvimento Científico e Tecnológico (CNPq, Brasilia, Brazil). 15. Rett, A. (1966). On an unusual brain atrophy syndrome in hyperammonemia in childhood. Wien Med. Wochenschr. 116, 723–726. M.E. is an ICREA Research Professor. 16. Amir, R.E., Van Den Veyver, I.B., Wan, M., Tran, C.Q., Francke, U., and Zoghbi, H.Y. (1999). Rett syndrome is caused by mutations in X-linked MECP2, encoding AUTHOR CONTRIBUTIONS methyl- CpG-binding protein 2. Nat. Genet. 23, 185–188. All experiments were conceived by E.S., A.O.-G., O.C.J.-T., and S.G. 17. Lewis, J.D., Meehan, R.R., Henzel, W.J., Maurer-Fogy, I., Jeppesen, P., Klein, F., and and carried out mainly by E.S. and A.O.-G. O.C.J.-T. carried out Bird, A. (1992). Purification, sequence, and cellular localization of a novel chromo- the proteomics analysis and experiments with the N2a model. M.S. somal protein that binds to Methylated DNA. Cell 69, 905–914. performed the methylation analysis and provided technical support. 18. Hagberg, B. (2002). Clinical manifestations and stages of Rett syndrome. Ment. F.S. carried out plasmid construction. C.O.-M. and D.R.-K. per- Retard. Dev. Disabil. Res. Rev. 8, 61–65. formed circRNA characterization. A.P.-A., C.X., and J.A. genotyped 19. Castro, J., Mellios, N., and Sur, M. (2013). Mechanisms and therapeutic challenges in the RTT samples. G.A.C., M.S., and C.I. performed T-UCR microar- autism spectrum disorders: insights from Rett syndrome. Curr. Opin. Neurol. 26, 154–159. ray hybridization and analysis. D.v.R. and R.J.P. provided training 20. Weese-Mayer, D.E., Lieske, S.P., Boothby, C.M., Kenny, A.S., Bennett, H.L., Silvestri, and assisted with RNA ISH experiments. S.G. wrote the manuscript J.M., and Ramirez, J.M. (2006). Autonomic nervous system dysregulation: breathing with input from all authors. S.G. and M.E. supervised the project. and heart rate perturbation during wakefulness in young girls with rett syndrome. Pediatr. Res. 60, 443–449. DECLARATION OF INTERESTS 21. Calfa, G., Li, W., Rutherford, J.M., and Pozzo-Miller, L. (2015). Excitation/inhibition The authors declare no competing interests. imbalance and impaired synaptic inhibition in hippocampal area CA3 of Mecp2 knockout mice. Hippocampus 25, 159–168. 22. Katz, D.M., Bird, A., Coenraads, M., Gray, S.J., Menon, D.U., Philpot, B.D., and REFERENCES Tarquinio, D.C. (2016). Rett syndrome: crossing the threshold to clinical translation. 1. Flicek, P., Ahmed, I., Amode, M.R., Barrell, D., Beal, K., Brent, S., Carvalho-Silva, D., Trends Neurosci. 39, 100–113. Clapham, P., Coates, G., Fairley, S., et al. (2013). Ensembl 2013. Nucleic Acids Res. 41, D48–D55. 23. Lyst, M.J., and Bird, A. (2015). Rett syndrome: a complex disorder with simple roots. Nat. Rev. Genet. 16, 261–275. 2. Qureshi, I.A., and Mehler, M.F. (2012). Emerging roles of non-coding RNAs in brain evolution, development, plasticity and disease. Nat. Rev. Neurosci. 13, 24. Leonard, H., Cobb, S., and Downs, J. (2017). Clinical and biological progress over 50 528–541. years in Rett syndrome. Nat. Rev. Neurol. 13, 37–51. 3. Derrien, T., Johnson, R., Bussotti, G., Tanzer, A., Djebali, S., Tilgner, H., Guernec, 25. Guy, J., Hendrich, B., Holmes, M., Martin, J.E., and Bird, A. (2001). A mouse Mecp2- G., Martin, D., Merkel, A., Knowles, D.G., et al. (2012). The GENCODE v7 catalog null mutation causes neurological symptoms that mimic rett syndrome. Nat. Genet. of human long noncoding RNAs: analysis of their gene structure, evolution, and 27, 322–326. expression. Genome Res. 22, 1775–1789. 26. Lyu, J.W., Yuan, B., Cheng, T.L., Qiu, Z.L., and Zhou, W.H. (2016). Reciprocal regu- 4. Washietl, S., Kellis, M., and Garber, M. (2014). Evolutionary dynamics and tissue lation of autism-related genes MeCP2 and PTEN via microRNAs. Sci. Rep. 6, 20392. specificity of human long noncoding RNAs in six mammals. Genome Res. 24, 27. Wu, H., Tao, J., Chen, P.J., Shahab, A., Ge, W., Hart, R.P., Ruan, X., Ruan, Y., and 616–628. Sun, Y.E. (2010). Genome-wide analysis reveals methyl-CpG-binding protein 2- 5. Briggs, J.A., Wolvetang, E.J., Mattick, J.S., Rinn, J.L., and Barry, G. (2015). dependent regulation of microRNAs in a mouse model of Rett syndrome. Proc. Mechanisms of long non-coding RNAs in mammalian nervous system develop- Natl. Acad. Sci. USA 107, 18161–18166. ment, plasticity, disease, and evolution. Neuron 88, 861–877. 28. Nomura, T., Kimura, M., Horii, T., Morita, S., Soejima, H., Kudo, S., and Hatada, I. 6. Aprea, J., and Calegari, F. (2015). Long non-coding RNAs in corticogenesis: deci- (2008). MeCP2-dependent repression of an imprinted miR-184 released by depolar- phering the non-coding code of the brain. EMBO J. 34, 2865–2884. ization. Hum. Mol. Genet. 17, 1192–1199. 7. Chen, I., Chen, C.Y., and Chuang, T.J. (2015). Biogenesis, identification, and func- 29. Tsujimura, K., Irie, K., Nakashima, H., Egashira, Y., Fukao, Y., Fujiwara, M., Itoh, tion of exonic circular RNAs. Wiley Interdiscip. Rev. RNA. 6, 563–579. M., Uesaka, M., Imamura, T., Nakahata, Y., et al. (2015). MiR-199a links MeCP2 with mTOR signaling and its dysregulation leads to rett syndrome phenotypes. 8. Memczak, S., Jens, M., Elefsinioti, A., Torti, F., Krueger, J., Rybak, A., Maier, L., Cell Rep. 12, 1887–1901. Mackowiak, S.D., Gregersen, L.H., Munschauer, M., et al. (2013). Circular RNAs are a large class of animal RNAs with regulatory potency. Nature 495, 333–338. 30. Zhang, Y., Chen, M., Qiu, Z., Hu, K., McGee, W., Chen, X., Liu, J., Zhu, L., and Wu, J.Y. (2016). MiR-130a regulates neurite outgrowth and dendritic spine density by 9. Ashwal-Fluss, R., Meyer, M., Pamudurti, N.R., Ivanov, A., Bartok, O., Hanan, M., targeting MeCP2. Protein Cell 7, 489–500. Evantal, N., Memczak, S., Rajewsky, N., and Kadener, S. (2014). CircRNA biogenesis competes with pre-mRNA splicing. Mol. Cell. 56, 55–66. 31. Rodrigues, D.C., Kim, D.S., Yang, G., Zaslavsky, K., Ha, K.C.H., Mok, R.S.F., Ross, P.J., Zhao, M., Piekna, A., Wei, W., et al. (2016). MECP2 is post-transcriptionally 10. Guo, J.U., Agarwal, V., Guo, H., and Bartel, D.P. (2014). Expanded identification regulated during human neurodevelopment by combinatorial action of RNA-bind- and characterization of mammalian circular RNAs. Genome Biol. 15, 409. ing proteins and miRNAs. Cell Rep. 17, 720–734. 11. Dragomir, M., and Calin, G.A. (2018). Circular RNAs in cancer - lessons learned 32. Han, K., Gennarino, V.A., Lee, Y., Pang, K., Hashimoto-Torii, K., Choufani, S., Raju, from microRNAs. Front. Oncol. 8, 179. C.S., Oldham, M.C., Weksberg, R., Rakic, P., et al. (2013). Human-specific regula- 12. Rybak-Wolf, A., Stottmeister, C., Glazar, P., Jens, M., Pino, N., Giusti, S., Hanan, M., tion of MeCP2 levels in fetal brains by microRNA miR-483-5p. Genes Dev. 27, Behm, M., Bartok, O., Ashwal-Fluss, R., et al. (2015). Circular RNAs in the mamma- 485–490. lian brain are highly abundant, conserved, and dynamically expressed. Mol. Cell. 58, 33. Klein, M.E., Lioy, D.T., Ma, L., Impey, S., Mandel, G., and Goodman, R.H. (2007). 870–885. Homeostatic regulation of MeCP2 expression by a CREB-induced microRNA. Nat. 13. Lukiw, W.J. (2013). Circular RNA (circRNA) in Alzheimer’s disease (AD). Front. Neurosci. 10, 1513–1514. Genet. 4, 307. 34. Alvarez-Saavedra, M., Antoun, G., Yanagiya, A., Oliva-Hernandez, R., Cornejo- 14. Hagberg, B., Aicardi, J., Dias, K., and Ramos, O. (1983). A progressive syndrome of Palma, D., Perez-Iratxeta, C., Sonenberg, N., and Cheng, H.Y. (2011). miRNA- autism, dementia, ataxia, and loss of purposeful hand use in girls: rett’s syndrome: 132 orchestrates chromatin remodeling and translational control of the circadian report of 35 cases. Ann. Neurol. 14, 471–479. clock. Hum. Mol. Genet. 20, 731–751. Molecular Therapy: Nucleic Acids Vol. 27 March 2022 641 Molecular Therapy: Nucleic Acids 35. Im, H.I., Hollander, J.A., Bali, P., and Kenny, P.J. (2010). MeCP2 controls BDNF 55. Wu, H., Li, H., Bai, T., Han, L., Ou, J., Xun, G., Zhang, Y., Wang, Y., Duan, G., Zhao, expression and cocaine intake through homeostatic interactions with microRNA- N., et al. (2020). Phenotype-to-genotype approach reveals head-circumference-asso- 212. Nat. Neurosci. 13, 1120–1127. ciated genes in an autism spectrum disorder cohort. Clin. Genet. 97, 338–346. 36. Hu, J.-H., Park, J.M., Park, S., Xiao, B., Dehoff, M.H., Kim, S., Hayashi, T., Schwarz, 56. Fabris, L., and Calin, G.A. (2017). Understanding the genomic ultraconservations: M.K., Huganir, R.L., Seeburg, P.H., et al. (2010). Homeostatic scaling requires group T-UCRs and cancer. Int. Rev. Cell Mol. Biol. 333, 159–172. I mGluR activation mediated by Homer1a. Neuron 68, 1128–1142. 57. Calin, G.A., Liu, C., Ferracin, M., Hyslop, T., Spizzo, R., Sevignani, C., Fabbri, M., 37. Xu, K., Zhang, Y., Xiong, W., Zhang, Z., Wang, Z., Lv, L., Liu, C., Hu, Z., Zheng, Y.T., Cimmino, A., Lee, E.J., Wojcik, S.E., et al. (2007). Ultraconserved regions encoding Lu, L., et al. (2020). CircGRIA1 shows an age-related increase in male macaque brain ncRNAs are altered in human leukemias and carcinomas. Cancer Cell 3, 215–229. and regulates synaptic plasticity and synaptogenesis. Nat. Commun. 11, 3594. 58. Ferdin, J., Nishida, N., Wu, X., Nicoloso, M.S., Shah, M.Y., Devlin, C., Ling, H., 38. Armstrong, D., Dunn, J.K., Antalffy, B., and Trivedi, R. (1995). Selective dendritic Shimizu, M., Kumar, K., Cortez, M.A., et al. (2013). HINCUTs in cancer: hypox- alterations in the cortex of rett syndrome. J. Neuropathol. Exp. Neurol. 54, 195–201. ia-induced noncoding ultraconserved transcripts. Cell Death Differ. 12, 1675–1687. 39. Salzman, J., Chen, R.E., Olsen, M.N., Wang, P.L., and Brown, P.O. (2013). Cell-type 59. Bejerano, G., Pheasant, M., Makunin, I., Stephen, S., Kent, W.J., Mattick, J.S., and specific features of circular RNA expression. PLoS Genet. 9, e1003777. Haussler, D. (2004). Ultraconserved elements in the human genome. Science 304, 40. Goo, M.S., Sancho, L., Slepak, N., Boassa, D., Deerinck, T.J., Ellisman, M.H., 1321–1325. Bloodgood, B.L., and Patrick, G.N. (2017). Activity-dependent trafficking of lyso- 60. Westbrook, T.F., Hu, G., Ang, X.L., Mulligan, P., Pavlova, N.N., Liang, A., Leng, Y., somes in dendrites and dendritic spines. J. Cell Biol. 216, 2499–2513. Maehr, R., Shi, Y., Harper, J.W., et al. (2008). SCFbeta-TRCP controls oncogenic 41. Chang, Q., Khare, G., Dani, V., Nelson, S., and Jaenisch, R. (2006). The disease pro- transformation and neural differentiation through REST degradation. Nature 452, gression of Mecp2 mutant mice is affected by the level of BDNF expression. Neuron 370–374. 49, 341–348. 61. Epifanova, E., Babaev, A., Newman, A.G., and Tarabykin, V. (2019). Role of Zeb2/ 42. Landucci, E., Brindisi, M., Bianciardi, L., Catania, L.M., Daga, S., Croci, S., Frullanti, Sip1 in neuronal development. Brain. Res. 15, 1724–1731. E., Fallerini, C., Butini, S., Brogi, S., et al. (2018). iPSC-derived neurons profiling re- 62. Reichert, S.C., Li, R., Turner, S.A., van Jaarsveld, R.H., Massink, M.P.G., van den veals GABAergic circuit disruption and acetylated a-tubulin defect which improves Boogaard, M.-J.H., Del Toro, M., Rodríguez-Palmero, A., Fourcade, S., Schlüter, after iHDAC6 treatment in Rett syndrome. Exp. Cell Res. 368, 225–235. A., et al. (2020). HNRNPH1-related syndromic intellectual disability: seven addi- 43. Gold, W.A., Lacina, T.A., Cantrill, L.C., and Christodoulou, J. (2014). MeCP2 defi- tional cases suggestive of a distinct syndromic neurodevelopmental syndrome. ciency is associated with reduced levels of tubulin acetylation and can be restored Clin. Genet. 98, 91–98. using HDAC6 inhibitors. J. Mol. Med. 93, 63–72. 63. Nectoux, J., Florian, C., Delepine, C., Bahi-Buisson, N., Khelfaoui, M., Reibel, S., 44. Pellegrini, L., Wetzel, A., Grannó, S., Heaton, G., and Harvey, K. (2017). Back to the Chelly, J., and Bienvenu, T. (2012). Altered microtubule dynamics in Mecp2-defi- tubule: microtubule dynamics in Parkinson’s disease. Cell Mol. Life Sci. 74, 409–434. cient astrocytes. J. Neurosci. Res. 90, 990–998. 45. Li, L., and Yang, X.-J. (2015). Tubulin acetylation: responsible enzymes, biological 64. Delépine, C., Meziane, H., Nectoux, J., Opitz, M., Smith, A.B., Ballatore, C., Saillour, functions and human diseases. Cell. Mol. Life Sci. 72, 4237–4255. Y., Bennaceur-Griscelli, A., Chang, Q., Williams, E.C., et al. (2016). Altered micro- 46. Yang, L., Han, B., Zhang, Z., Wang, S., Bai, Y., Zhang, Y., Tang, Y., Du, L., Xu, L., tubule dynamics and vesicular transport in mouse and human MeCP2-deficient as- Wu, F., et al. (2020). Extracellular vesicle-mediated delivery of circular RNA trocytes. Hum. Mol. Genet. 25, 146–157. SCMH1 promotes functional recovery in rodent and nonhuman primate ischemic 65. Delépine, C., Nectoux, J., Bahi-Buisson, N., Chelly, J., and Bienvenu, T. (2013). stroke models. Circulation 142, 556–574. MeCP2 deficiency is associated with impaired microtubule stability. FEBS Lett. 47. Good, K.V., Vincent, J.B., and Ausió, J. (2021). MeCP2: the genetic driver of Rett 587, 245–253. syndrome epigenetics. Front. Genet. 12, 620859. 66. Abuhatzira, L., Shemer, R., and Razin, A. (2009). MeCP2 involvement in the regu- 48. Meng, X., Wang, W., Lu, H., He, L.-J., Chen, W., Chao, E.S., Fiorotto, M.L., Tang, B., lation of neuronal alpha-tubulin production. Hum. Mol. Genet. 18, 1415–1423. Herrera, J.A., Seymour, M.L., et al. (2016). Manipulations of MeCP2 in glutamater- 67. Andoh-Noda, T., Akamatsu, W., Miyake, K., Matsumoto, T., Yamaguchi, R., gic neurons highlight their contributions to Rett and other neurological disorders. Sanosaka, T., Okada, Y., Kobayashi, T., Ohyama, M., Nakashima, K., et al. (2015). Elife 5, e14199. Differentiation of multipotent neural stem cells derived from Rett syndrome pa- 49. Jorge-Torres, O.C., Szczesna, K., Roa, L., Casal, C., Gonzalez-Somermeyer, L., Soler, tients is biased toward the astrocytic lineage. Mol. Brain 8, 31. M., Velasco, C.D., Martínez-San Segundo, P., Petazzi, P., Sáez, M.A., et al. (2018). 68. Tsujimura, K., Abematsu, M., Kohyama, J., Namihira, M., and Nakashima, K. Inhibition of Gsk3b reduces Nfkb1 signaling and rescues synaptic activity to (2009). Neuronal differentiation of neural precursor cells is promoted by the improve the rett syndrome phenotype in Mecp2-knockout mice. Cell Rep. 23, methyl-CpG-binding protein MeCP2. Exp. Neurol. 219, 104–111. 1665–1677. 69. Ahlsén, G., Rosengren, L., Belfrage, M., Palm, A., Haglid, K., Hamberger, A., and 50. Wang, G., Li, S., Gilbert, J., Gritton, H.J., Wang, Z., Li, Z., Han, X., Selkoe, D.J., and Gillberg, C. (1993). Glial fibrillary acidic protein in the cerebrospinal fluid of chil- Man, H.Y. (2017). Crucial roles for SIRT2 and AMPA receptor acetylation in syn- dren with autism and other neuropsychiatric disorders. Biol. Psychiatry. 33, aptic plasticity and memory. Cell Rep. 20, 1335–1347. 734–743. 51. O’Connor, M.O., Shentu, Y.-P., Wang, G., Hu, W.-T., Xu, Z.-D., Wang, X.C., Liu, R., 70. Forbes-Lorman, R.M., Kurian, J.R., and Auger, A.P. (2014). MeCP2 regulates GFAP and Man, H.Y. (2020). Acetylation of AMPA receptors regulates receptor trafficking expression within the developing brain. Brain Res. 1543, 151–158. and rescues memory deficits in Alzheimer’s disease. iScience 23, 101465. 71. Donato, R., and Giambanco, I. (1989). Interaction between S-100 proteins and 52. Satterstrom, F.K., Walters, R.K., Singh, T., Wigdor, E.M., Lescai, F., Demontis, D., steady-state and taxol-stabilized microtubules in vitro. J. Neurochem. 52, 1010– Kosmicki, J.A., Grove, J., Stevens, C., Bybjerg-Grauholm, J., et al. (2019). Autism 1017. spectrum disorder and attention deficit hyperactivity disorder have a similar burden of rare protein-truncating variants. Nat. Neurosci. 22, 1961–1965. 72. Sorci, G., Agneletti, A.L., and Donato, R. (2000). Effects of S100A1 and S100B on microtubule stability. An in vitro study using triton-cytoskeletons from astrocyte 53. Araujo, D.J., Toriumi, K., Escamilla, C.O., Kulkarni, A., Anderson, A.G., Harper, M., and myoblast cell lines. Neuroscience 99, 773–783. Usui, N., Ellegood, J., Lerch, J.P., Birnbaum, S.G., et al. (2017). Foxp1 in forebrain pyramidal neurons controls gene expression required for spatial learning and syn- 73. Tramontina, F., Tramontina, A.C., Souza, D.F., Leite, M.C., Gottfried, C., Souza, aptic plasticity. J. Neurosci. 37, 10917–10931. D.O., Wofchuk, S.T., and Gonçalves, C.A. (2006). Glutamate uptake is stimulated by extracellular S100B in hippocampal astrocytes. Cell Mol. Neurobiol. 26, 81–86. 54. Sanfeliu, A., Kaufmann, W.E., Gill, M., Guasoni, P., and Tropea, D. (2019). Transcriptomic studies in mouse models of rett syndrome: a review. 74. Pei, W., Huang, Z., and Niu, L. (2007). GluR3 flip and flop: differences in channel Neuroscience 413, 183–205. opening kinetics. Biochemistry 46, 2027–2036. 642 Molecular Therapy: Nucleic Acids Vol. 27 March 2022 www.moleculartherapy.org 75. Mosbacher, J., Schoepfer, R., Monyer, H., Burnashev, N., Seeburg, P.H., and with different protein and DNA interactions compared to MeCP2-E2. Epigenetics Ruppersberg, J.P. (1994). A molecular determinant for submillisecond desensitiza- Chromatin. 12, 63. tion in glutamate receptors. Science 266, 1059–1062. 94. Montgomery, K.R., Louis Sam Titus, A.S.C., Wang, L., and D’Mello, S.R. (2018). 76. Koike, M., Tsukada, S., Tsuzuki, K., Kijima, H., and Ozawa, S. (2000). Regulation of Elevated MeCP2 in mice causes neurodegeneration involving tau dysregulation kinetic properties of GluR2 AMPA receptor channels by alternative splicing. and excitotoxicity: implications for the understanding and treatment of MeCP2 trip- J. Neurosci. 20, 2166–2174. lication syndrome. Mol. Neurobiol. 55, 9057–9074. 77. Balik, A., Penn, A.C., Nemoda, Z., and Greger, I.H. (2013). Activity-regulated RNA 95. Ehrhart, F., Coort, S.L., Eijssen, L., Cirillo, E., Smeets, E.E., Bahram Sangani, N., editing in select neuronal subfields in hippocampus. Nucleic Acids Res. 41, 1124– Evelo, C.T., and Curfs, L.M.G. (2019). Integrated analysis of human transcriptome 1134. data for Rett syndrome finds a network of involved genes. World J. Biol. Psychiatry 25, 1–14. 78. Trujillo, C.A., Adams, J.W., Negraes, P.D., Carromeu, C., Tejwani, L., Acab, A., Tsuda, B., Thomas, C.A., Sodhi, N., Fichter, K.M., et al. (2021). Pharmacological 96. Lu, D., and Xu, A.-D. (2016). Mini review: circular RNAs as potential clinical bio- reversal of synaptic and network pathology in human MECP2-KO neurons and markers for disorders in the central nervous system. Front. Genet. 7, 53. cortical organoids. EMBO Mol. Med. 13, e12523. 97. Abdelmohsen, K., Panda, A.C., Munk, R., Grammatikakis, I., Dudekula, D.B., De, S., 79. Sheinerman, K., Djukic, A., Tsivinsky, V.G., and Umansky, S.R. (2019). Brain-en- Kim, J., Noh, J.H., Kim, K.M., Martindale, J.L., et al. (2017). Identification of HuR riched microRNAs circulating in plasma as novel biomarkers for Rett syndrome. target circular RNAs uncovers suppression of PABPN1 translation by PLoS One 14, e0218623. CircPABPN1. RNA Biol. 14, 361–369. 98. Meng, L., Ward, A.J., Chun, S., Bennett, C.F., Beaudet, A.L., and Rigo, F. (2015). 80. Mellios, N., Feldman, D.A., Sheridan, S.D., Ip, J.P.K., Kwok, S., Amoah, S.K., Rosen, Towards a therapy for Angelman syndrome by targeting a long non-coding RNA. B., Rodriguez, B.A., Crawford, B., Swaminathan, R., et al. (2018). MeCP2-regulated Nature 518, 409–412. miRNAs control early human neurogenesis through differential effects on ERK and AKT signaling. Mol. Psychiatry 23, 1051–1065. 99. Hsiao, J., Yuan, T.Y., Tsai, M.S., Lu, C.Y., Lin, Y.C., Lee, M.L., Lin, S.W., Chang, F.C., Liu Pimentel, H., Olive, C., et al. (2016). Upregulation of haploinsufficient gene 81. Barco, A., Bailey, C.H., and Kandel, E.R. (2006). Common molecular mechanisms in expression in the brain by targeting a long non-coding RNA improves seizure explicit and implicit memory. J. Neurochem. 97, 1520–1533. phenotype in a model of Dravet syndrome. EBioMedicine. 9, 257–277. 82. Cortese, B.M., and Phan, K.L. (2005). The role of glutamate in anxiety and related 100. Donato, R., Miljan, E.A., Hines, S.J., Aouabdi, S., Pollock, K., Patel, S., Edwards, F.A., disorders. CNS Spectr. 10, 820–830. and Sinden, J.D. (2007). Differential development of neuronal physiological respon- 83. Salpietro, V., Dixon, C.L., Guo, H., Bello, O.D., Vandrovcova, J., Efthymiou, S., siveness in two human neural stem cell lines. BMC Neurosci. 8, 36. Maroofian, R., Heimer, G., Burglen, L., Valence, S., et al. (2019). AMPA receptor 101. Tremblay, R.G., Sikorska, M., Sandhu, J.K., Lanthier, P., Ribecco-Lutkiewicz, M., GluA2 subunit defects are a cause of neurodevelopmental disorders. Nat. and Bani-Yaghoub, M. (2010). Differentiation of mouse Neuro 2A cells into dopa- Commun. 10, 3094. mine neurons. J. Neurosci. Methods 186, 60–67. https://doi.org/10.1016/j.jneumeth. 84. Osenberg, S., Karten, A., Sun, J., Li, J., Charkowick, S., Felice, C.A., Kritzer, M., 2009.11.004. Nguyen, M.V.C., Yu, P., and Ballas, N. (2018). Activity-dependent aberrations in 102. Ran, F.A., Hsu, P.D., Wright, J., Agarwala, V., Scott, D.A., and Zhang, F. (2013). gene expression and alternative splicing in a mouse model of Rett syndrome. Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. 8, 2281–2308. Proc. Natl. Acad. Sci. USA 115, E5363–E5372. 103. Chu, V.T., Graf, R., Wirtz, T., Weber, T., Favret, J., Li, X., Petsch, K., Tran, N.T., 85. Cheng, T.-L., Chen, J., Wan, H., Tang, B., Tian, W., Liao, L., and Qiu, Z. (2017). Sieweke, M.H., Berek, C., et al. (2016). Efficient CRISPR-mediated mutagenesis in Regulation of mRNA splicing by MeCP2 via epigenetic modifications in the brain. primary immune cells using CrispR Gold and a C57BL/6 Cas9 transgenic mouse Sci. Rep. 7, 42790. line. PNAS 113, 12514–12519. 86. Li, R., Dong, Q., Yuan, X., Zeng, X., Gao, Y., Chiao, C., Li, H., Zhao, X., Keles, S., 104. Serrano, L., Martínez-Redondo, P., Marazuela-Duque, A., Vazquez, B.N., Dooley, Wang, Z., et al. (2016). Misregulation of alternative splicing in a mouse model of S.J., Voigt, P., Beck, D.B., Kane-Goldsmith, N., Tong, Q., Rabanal, R.M., et al. rett syndrome. PLoS Genet. 12, e1006129. (2013). The tumor suppressor SirT2 regulates cell cycle progression and genome sta- bility by modulating the mitotic deposition of H4K20 methylation. Genes Dev. 27, 87. La Via, L., Bonini, D., Russo, I., Orlandi, C., Barlati, S., and Barbon, A. (2013). 639–653. Modulation of dendritic AMPA receptor mRNA trafficking by RNA splicing and editing. Nucleic Acids Res. 41, 617–631. 105. Ho, S.M., Hartley, B.J., Tcw, J., Beaumont, M., Stafford, K., Slesinger, P.A., and Brennand, K.J. (2016). Rapid Ngn2-induction of excitatory neurons from hiPSC- 88. Yoshii, A., and Constantine-Paton, M. (2007). BDNF induces transport of PSD-95 derived neural progenitor cells. Methods 15, 113–124. https://doi.org/10.1016/j. to dendrites through PI3K-AKT signaling after NMDA receptor activation. Nat. ymeth.2015.11.019. Neurosci. 10, 702–711. 106. Dragomir, M.P., Tudor, S., Okubo, K., Shimizu, M., Chen, M., Giza, D.E., He, W.R., 89. Zheng, W.-H., and Quirion, R. (2004). Comparative signaling pathways of insulin- Ivan, C., Calin, G.A., and Vasilescu, C. (2019). The non-coding RNome after sple- like growth factor-1 and brain-derived neurotrophic factor in hippocampal neurons nectomy. J. Cell Mol. Med. 23, 7844–7858. and the role of the PI3 kinase pathway in cell survival. J. Neurochem. 89, 844–852. 107. Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., and Smyth, G.K. 90. Asaka, Y., Jugloff, D.G.M., Zhang, L., Eubanks, J.H., and Fitzsimonds, R.M. (2006). (2015). Limma powers differential expression analyses for RNA-sequencing and mi- Hippocampal synaptic plasticity is impaired in the Mecp2-null mouse model of Rett croarray studies. Nucleic Acids Res. 43, e47. syndrome. Neurobiol. Dis. 21, 217–227. 108. Vallée, M., Gravel, C., Palin, M.-F., Reghenas, H., Stothard, P., Wishart, D.S., and 91. Moretti, P., Levenson, J.M., Battaglia, F., Atkinson, R., Teague, R., Antalffy, B., Sirard, M.A. (2005). Identification of novel and known oocyte-specific genes using Armstrong, D., Arancio, O., Sweatt, J.D., and Zoghbi, H.Y. (2006). Learning and complementary DNA subtraction and microarray analysis in three different species. memory and synaptic plasticity are impaired in a mouse model of Rett syndrome. Biol. Reprod. 73, 63–71. J. Neurosci. 26, 319–327. 109. Cheng, C., Spengler, R., Keiser, M.S., Monteys, A.M., Rieders, J.M., Ramachandran, 92. Bergo, A., Strollo, M., Gai, M., Barbiero, I., Stefanelli, G., Sertic, S., Cobolli Gigli, C., S., and Davidson, B.L. (2018). The long non-coding RNA NEAT1 is elevated in pol- Di Cunto, F., Kilstrup-Nielsen, C., and Landsberger, N. (2015). Methyl-CpG binding yglutamine repeat expansion diseases and protects from disease gene-dependent protein 2 (MeCP2) localizes at the centrosome and is required for proper mitotic toxicities. Hum. Mol. Genet. 27, 4303–4314. spindle organization. J. Biol. Chem. 290, 3223–3237. 110. Vidal, S., Brandi, N., Pacheco, P., Gerotina, E., Blasco, L., Trotta, J.-R., Derdak, S., 93. Martínez de Paz, A., Khajavi, L., Martin, H., Claveria-Gimeno, R., Tom Dieck, S., Del Mar O’Callaghan, M., Garcia-Cazorla, À., Pineda, M., et al. (2017). The utility Cheema, M.S., Sanchez-Mut, J.V., Moksa, M.M., Carles, A., Brodie, N.I., et al. of next generation sequencing for molecular diagnostics in Rett syndrome. Sci. (2019). MeCP2-E1 isoform is a dynamically expressed, weakly DNA-bound protein Rep. 7, 12288. Molecular Therapy: Nucleic Acids Vol. 27 March 2022 643 Molecular Therapy: Nucleic Acids 111. Chen, E.Y., Tan, C.M., Kou, Y., Duan, Q., Wang, Z., Meirelles, G.V., Clark, N.R., and comprehensive gene set enrichment analysis web server 2016 update. Nucleic Ma’ayan, A. (2013). Enrichr: interactive and collaborative HTML5 gene list enrich- Acids Res. 44, W90–W97. ment analysis tool. BMC Bioinformatics. 14, 128. 113. Jassal, B., Matthews, L., Viteri, G., Gong, C., Lorente, P., Fabregat, A., Sidiropoulos, 112. Kuleshov, M.V., Jones, M.R., Rouillard, A.D., Fernandez, N.F., Duan, Q., Wang, Z., K., Cook, J., Gillespie, M., Haw, R., et al. (2020). The reactome pathway knowledge- Koplev, S., Jenkins, S.L., Jagodnik, K.M., Lachmann, A., et al. (2016). Enrichr: a base. Nucleic Acids Res. 48, D498–D503. 644 Molecular Therapy: Nucleic Acids Vol. 27 March 2022