key: cord-0330374-4squ6r3q
authors: Garay, Patricia M.; Chen, Alex; Tsukahara, Takao; Kohen, Rafi; Christian Althaus, J.; Wallner, Margarete A.; Giger, Roman J.; Sutton, Michael A.; Iwase, Shigeki
title: RAI1 Regulates Activity-Dependent Nascent Transcription and Synaptic Scaling
date: 2019-01-18
journal: bioRxiv
DOI: 10.1101/523456
sha: 15b119b7f92564a6a2aeb5ecf27d463031b5053e
doc_id: 330374
cord_uid: 4squ6r3q

Long-lasting forms of synaptic plasticity such as synaptic scaling are critically dependent on transcription. Activity-dependent transcriptional dynamics in neurons, however, have not been fully characterized, because most previous efforts relied on measurement of steady-state mRNAs. Here, we profiled transcriptional dynamics of primary neuronal cultures undergoing network activity shifts using nascent RNA sequencing. We found pervasive transcriptional changes, in which ~45% of expressed genes respond to network activity shifts. Notably, the majority of these genes respond to increases or decreases of network activity uniquely, rather than reciprocally. We further linked the chromatin regulator Retinoic acid induced 1 (RAI1), the Smith-Magenis Syndrome gene, to the specific transcriptional program driven by reduced network activity. Finally, we show that RAI1 is essential for homeostatic synaptic upscaling but not downscaling. These results demonstrate the utility of bona fide transcription profiling to discover mechanisms of activity-dependent chromatin remodeling that underlie normal and pathological synaptic plasticity.

Proper cognitive development and brain function relies on synaptic plasticity -the ability of 44 synapses to strengthen or weaken in response to sensory or neuromodulatory inputs. Synaptic 45 scaling is one mechanism of plasticity, which buffers potentially destabilizing patterns of network 46 activity (Abbott and Nelson, 2000; Miller and MacKay, 1994; Turrigiano, 2008) . In response to a sustained increase in neuronal firing rate, neurons decrease, or "scale-down", the receptivity of 48 the neuron to excitatory neurotransmitters. Conversely, global decreases in firing rate causes 49 neurons to "scale-up" and increase synaptic efficacy. Synaptic scaling is thought to 50 accommodate other forms of plasticity, such as long-term potentiation (LTP), that impose long-51 lasting increase of individual synaptic efficacy, which if left uncompensated, would result in 52 circuits that are overly active (Turrigiano, 2017) . Synaptic scaling appears to play important 

To understand how homeostatic plasticity contributes to normal and pathological brain 59 development, identifying the molecular mechanisms underlying synaptic scaling is an important 60 first step.

Long-lasting forms of synaptic plasticity, including synaptic scaling, require de novo synthesis of 63 RNAs, which in turn produce the proteins that directly modulate synaptic efficacy (Benito and 

The labeled RNAs were isolated by immunoprecipitation using an anti-BrU antibody and 141 subjected to next-generation sequencing (Fig. 1A ).

We first validated the results of Bru-seq by examining transcription of known activity-dependent 144 genes individually. As shown in Figure 1B (Table S1) 1C ). BIC increased transcription of 2,908 genes, whereas TTX did so for 1,820 genes. The 156 magnitude of transcriptional induction is higher in BIC treatment compared to TTX (Fig. 1D ).

Meanwhile, a similar number of genes were transcriptionally suppressed upon BIC (2,842) and 158 TTX (2,307) treatments. A relatively small fraction of genes, 24% (1,798 of 7,592 activity-159 response genes), displayed reciprocal changes between BIC and TTX treatments, e.g.

upregulation by BIC and downregulation by TTX (Fig.1E ). An even smaller fraction, 6% 161 (487/7,592) of activity-response genes altered their transcription levels in the same direction 7 after BIC and TTX treatments. The remaining 70% of genes (5,307/7,592) responded to BIC or 163 TTX uniquely. Supplemental Table 1 lists genes that displayed greater than 2-fold changes in 164 transcription upon network-activity shifts.

We next analyzed two published mRNA-seq datasets, which profiled the steady-state 

Conventional mRNA-seq may not be the ideal approach to detect downregulation of 177 transcription, because after transcription ceases, synthesized RNAs persist for certain periods 178 of time. To test whether Bru-seq can detect transcriptional downregulation sensitively, we 179 compared the induction and suppression of known immediate-early genes in our Bru-seq and 180 published mRNA-seq datasets. We found significantly larger suppression of Fos, Arc, Bdnf, and 181 Npas4 (4-to 16-fold) by TTX treatment in Bru-seq compared to conventional mRNA-seq, in 182 which downregulation was less than 2-fold ( Fig. 1F ). In the Bru-seq data, the four genes showed 183 smaller magnitudes of upregulation in response to BIC, likely because the early transcriptional 184 induction is largely complete 4 hours after BIC treatment (Fig. S1C) . These data highlight an 185 advantage of the Bru-seq approach to probe mechanisms underlying highly-dynamic activity-186 dependent transcription.

We next examined the cell-type specificity of activity-dependent genes in our datasets. Recent CD11b, and Olig2, we estimated that our cultures consist of 41% excitatory neurons, 11% 192 inhibitory neurons, 33% astrocytes, 15% of cells within the oligodendrocyte lineage, and no 193 microglia ( Fig. S2A-B) . Indeed, several non-neuronal genes are represented in our dataset,

including Thbs1, a known synaptic regulator specifically expressed in astrocytes (Risher and 195 8 Eroglu, 2012) . Gene ontology analysis of the Bru-seq data detected enrichment of biological 196 processes specific for both neurons and non-neuronal cell types (Table S2) . For example, 197 regulation of axon diameter (p adj = 2.6 x 10 −5 ), axonal transport of mitochondrion (p adj =3.5 x 198 10 −4 ), and glial cell proliferation (p adj : 7.8 x 10 −3 ) represent genes that are transcriptionally 199 induced upon BIC treatment, while transcription of genes involved in astrocyte activation are 

The superior sensitivity of Bru-seq over conventional RNA-seq prompted us to assess the role 210 of RAI1 in activity-dependent gene expression. We first determined RAI1 expression in our 211 culture systems. Publically-available databases indicate ubiquitous Rai1 expression in a broad 212 array of cell types in the brain (Fig. S3 ). Previous studies have demonstrated that Rai1 mRNA 213 expression rises from E13.5 to peak at P7, and its expression continues throughout adulthood were found at promoters of ~80% genes that are expressed at detectable levels, regardless of 227 their activity-dependent transcriptional changes (Fig. S6A) . We did not find any statistically-9 significant enrichment or depletion of RAI1 occupancy of TTX-or BIC-response genes ( Fig.   229 S6A). Thus, these data indicate that neuronal activity does not influence RAI1's expression level 230 or subcellular localization and that steady-state chromatin occupancy by RAI1 is not selective 231 between BIC or TTX-response genes.

To directly test RAI1's role in activity-dependent transcription, we went on to perform Rai1 were assessed by Bru-seq as described above.

We initially sought to establish if Rai1-KD alone was sufficient to alter nascent transcription in Having uncovered that RAI1 is essential to suppress the TTX-associated transcriptional 293 program under baseline activity conditions (Fig. 2) , we next tested if Rai1-KD has any impact on 294 transcriptional induction and suppression upon TTX and BIC treatments. By calculating fold-295 changes of transcription, we found that Rai1-KD led to a significant impairment of transcriptional 296 response to TTX, while transcriptional response to BIC was slightly weakened only for Table S3 ).

We then sought to determine if the strongly-impaired transcriptional response to TTX ( 

Taken together, the Bru-seq results led us to conclude that 1) RAI1-deficiency shifts 318 transcriptional profiles towards an activity-suppressed state in the resting network (Fig. 2) , and 319 2) RAI1 is selectively required for the transcriptional response driven by network activity 320 suppression (Fig. 3 ).

We sought to explore biological implications for such small but pervasive deficits in 322 transcriptional response to TTX. We utilized RNA-Enrich, a gene ontology algorithm, in which 323 the entire output of DESeq2 is analyzed, such that the program takes into account statistically- 

Our Bru-seq method and analyses have provided novel insights into activity-dependent 364 transcription. We found widespread transcriptional responses to network activity shifts and its 365 high sensitivity in detecting transcriptional downregulation. Furthermore, our data indicate that 366 most dynamically regulated genes altered by hyperactivity or suppression are unique, not 367 reciprocal. (Fig. 1) . This is particularly interesting given that gene expression studies have 368 tended to focus on bidirectional regulation of target genes (e.g. Arc, Fos, Homer1, Bdnf) 369 (Okuno, 2011) . Our results agree with a nascent proteome study on rat hippocampal neurons, in 

Our data indicate that RAI1 is as a chromatin regulator that is selectively required for the 379 transcription program of activity-suppression. Loss of RAI1 leads to misregulation of TTX-380 response genes, while leaving the uniquely BIC-genes unaffected (Fig. 2) . In addition to its 381 exclusive impact on TTX-response genes, an intriguing feature of RAI1 is its state-dependent 

(n = 6-6) treated neurons. All bar graphs represent mean ± SEM, and comparisons between sh-523 Ctrl and sh-Rai1 were made with unpaired Student's t-tests. *p < 0.05, **p < 0.01, ***p < 0.001. 

The cortices and hippocampi from E18.5 mouse pups were pooled into biological replicates with 614 identical female-male ratios. Sex of the pups was determined by PCR using primers for the ZFY 615 gene (Table S5 ). Primary culture of neurons was carried out as previously described (Iwase et (Table S5) (Table S5 ). Adaptor duplexes with 5-or 6-base pair random 647 nucleotide overhangs were ligated to the 3' end of the cDNA (Table S5) 

Immunoreactivity was quantified semi-automatedly using a custom ImageJ script after Fisher's LSD test (Fig. 6) . Transcription of activity-dependent genes after drug treatments

Synaptic plasticity: taming the beast

LSD1/KDM1A Maintains Genome-wide 761 Homeostasis of Transcriptional Enhancers. bioRxiv

Chromatin Acetylation, Memory, and LTP Are Impaired in CBP+/− 764 Mice: A Model for the Cognitive Deficit in Rubinstein-Taybi Syndrome and Its 765 Amelioration

Allen Developing Mouse Brain Atlas

Excitatory/inhibitory synaptic imbalance leads to hippocampal 770 hyperexcitability in mouse models of tuberous sclerosis

Histone Methylation by the 773

Kleefstra Syndrome Protein EHMT1 Mediates Homeostatic Synaptic Scaling

The neuronal activity-driven transcriptome

Inactivation of Rai1 in mice recapitulates phenotypes observed in 779 chromosome engineered mouse models for Smith-Magenis syndrome

Mutations of RAI1, a PHD-containing protein, in nondeletion 783 patients with Smith-Magenis syndrome

Rai1 deficiency in mice causes 786 learning impairment and motor dysfunction, whereas Rai1 heterozygous mice display 787 minimal behavioral phenotypes

Abnormal circadian rhythm of melatonin in Smith-Magenis syndrome patients with RAI1 790 point mutations

Neurons and neuronal activity 864 control gene expression in astrocytes to regulate their development and metabolism

Neuronal activity-regulated alternative 867 mRNA splicing

Single-cell 870 analysis of experience-dependent transcriptomic states in the mouse visual cortex

Early adolescent Rai1 reactivation reverses transcriptional and social interaction 874 deficits in a mouse model of Smith-Magenis syndrome

Molecular and Neural 878 Functions of Rai1, the Causal Gene for Smith-Magenis Syndrome

Rapid synaptic scaling induced by 880 changes in postsynaptic firing

Two Time Periods of 882 Hippocampal mRNA Synthesis Are Required for Memory Consolidation of Fear-883 Motivated Learning

Role of Tet proteins in 5mC to 5hmC conversion, ES-cell self-renewal and inner cell 886 mass specification

A Mouse Model of

Intellectual Disability Associated with Impaired Removal of Histone Methylation

KMT2B) Is Required for Memory Formation in Mice. The 894

TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions 897 and gene fusions

LRpath analysis reveals common pathways 900 dysregulated via DNA methylation across cancer types

Sleep recalibrates homeostatic and 903 associative synaptic plasticity in the human cortex

Circadian abnormalities in mouse models 905 of smith-magenis syndrome: Evidence for involvement of RAI1

Fast gapped-read alignment with Bowtie 2

RNA-Enrich: a cut-off free functional 910 enrichment testing method for RNA-seq with improved detection power

featureCounts: an efficient general purpose 913 program for assigning sequence reads to genomic features

PKCα-mediated phosphorylation of LSD1 is required for 916 presynaptic plasticity and hippocampal learning and memory

Moderated estimation of fold change and 918 dispersion for RNA-seq data with DESeq2

Activity-Induced Regulation of 921

Synaptic Strength through the Chromatin Reader L3mbtl1

The Role of Constraints in Hebbian Learning

Abnormal social behaviors and altered gene 926 expression rates in a mouse model for Potocki-Lupski syndrome

Unifying Views of Autism Spectrum 929 Disorders: A Consideration of Autoregulatory Feedback Loops

Brain-penetrant LSD1 inhibitors 932 can block memory consolidation

Classical Rubinstein-Taybi Syndrome Phenotypes in Mice: Implications for a 936 Dominant-Negative Mechanism

Regulation and function of immediate-early genes in the brain: 938 Beyond neuronal activity markers

DNA methylation: a permissive mark in memory formation and 940 maintenance

Use of Bru-Seq and BruChase-Seq for genome-wide 943 assessment of the synthesis and stability of RNA

Coordinated regulation of 946 synthesis and stability of RNA during the acute TNF-induced proinflammatory response

Lupski syndrome (dup(17)(p11.2p11.2)) and delineation of a dosage-sensitive critical 951 interval that can convey an autism phenotype

Circadian rhythm abnormalities of melatonin in Smith-Magenis syndrome

Thrombospondins as key regulators of 956 synaptogenesis in the central nervous system

Tet1 is critical for neuronal activity-regulated gene 959 expression and memory extinction

Nascent Proteome Remodeling following Homeostatic Scaling at 962

An Intrinsic Transcriptional Program Underlying Synaptic Scaling during 965 Activity Suppression

Mutations in RAI1 associated with Smith-Magenis syndrome

Fragile X protein FMRP is required for homeostatic 969 plasticity and regulation of synaptic strength by retinoic acid

REVIGO Summarizes and 972 Visualizes Long Lists of Gene Ontology Terms

Histone methyltransferases G9a 975 and GLP form heteromeric complexes and are both crucial for methylation of 976 euchromatin at H3-K9

Adult mouse cortical cell taxonomy revealed 979 by single cell transcriptomics

The self-tuning neuron: synaptic scaling of excitatory synapses

The dialectic of Hebb and homeostasis

Altered Gene-Regulatory Function of KDM5C by a Novel Mutation Associated 986 With Autism and Intellectual Disability

The neuron-specific 989 chromatin regulatory subunit BAF53b is necessary for synaptic plasticity and memory

LSD1n is an H4K20 demethylase regulating memory formation via 993 transcriptional elongation control

Regulation of transcription 995 factors by neuronal activity

The m6A-epitranscriptomic signature in 997 neurobiology: from neurodevelopment to brain plasticity

Smith-magenis syndrome results in disruption of CLOCK gene transcription and reveals 1000 an integral role for RAI1 in the maintenance of circadian rhythmicity

A metaplasticity view of the interaction 1003 between homeostatic and Hebbian plasticity

Tet3 regulates synaptic transmission and 1007 homeostatic plasticity via DNA oxidation and repair

Bioorthogonal Metabolic Labeling 1010 of Nascent RNA in Neurons Improves the Sensitivity of Transcriptome-Wide Profiling

An RNA-sequencing 1014 transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral 1015 cortex

Model-based analysis of 1018 ChIP-Seq (MACS)

MeCP2 phosphorylation is required for 1020 modulating synaptic scaling through mGluR5

Carmona-Mora, P., Canales, C.P., Cao, L., Perez, I.C., Srivastava, A.K., Young, J.I., 808and Walz, K. (2012). RAI1 Transcription Factor Activity Is Impaired in Mutants 809Associated with Smith-Magenis Syndrome. PLoS One 7. reg. of cyclase activity synaptic vesicle localization neuromuscular process controlling balance long-term memory marginal zone B cell differentiation reg. of receptor activity striated muscle atrophy spinal cord oligodendrocyte cell differentiation reg. of branching in lung morphogenesis autophagic cell death cellular response to copper ion reg. of mesenchymal stem cell differentiation pos. reg. of interleukin-12 biosynthetic process elastic fiber assembly renal absorption reg.of establishment of cell polarity pos.reg. of apoptotic signaling pathway mitotic sister chromatid segregation cellular copper ion homeostasis reg. of establishment/maintenance of cell polarity copper ion transmembrane transport kidney vasculature morphogenesis fatty acid oxidation lung growth cellular response to hydroperoxide DNA packaging protein-DNA complex subunit organization mitotic sister chromatid segregation mitotic sister chromatid segregation noradrenergic neuron differentiation neutral amino acid transport reg. of guanylate cyclase activity long-term memory photoperiodism entrainment of circadian clock by photoperiod