We used SynoR to study the possible relationship between the SARE regulatory region and genes related to the nervous system 
, specifically those involved in synaptic activity and mental processes. We sought sequence regions containing clusters of the consensus binding sites for CREB, MEF2 and SRF () in the human genome, and compared them to the mouse genome to identify conserved sequences. Based on these criteria, we identified 887 genetic regions containing SARE sequences (Table S1
and data deposited in the SynoR tool, ID: s1219104005847). The SARE regions are assigned to the gene(s) of which they form part or to which they are proximal, and are classified as intergenic, intronic, utr (untranslated), cds (coding sequence), or promoter, depending on their position within the gene (Table S1
and ). Control searches for clusters containing combinations of other unrelated TFBS yield significantly less number of regions and were not enriched in neural biological functions ( and Experimental Procedures). The original SARE sequence of the Arc promoter is not identified in our search because it contains only half of the CREB binding site, and its MEF2 binding sequence shows 2 nucleotide mismatches compared to the consensus 
. Binding site predictions for individual TF using matrix analysis can be conducted with the Match tool of TRANSFAC. We have validated the presence of the SARE cluster in more than ten of our candidates manually using this tool. These include ATF3
, FOXP1, FOXP2, HOMER1
, NRG1, NPAS4. NR4A1, PLXNA4 and SEMA6A.
Similarly, there might be additional SARE sequences not identified by this search because of the analysis procedure: it first identifies TFBS clusters in the human genome and subsequently searches for homology to this specific sequence in mice.
SARE sequences are found in intergenic and intronic genetic regions.
The analysis showed that SARE clusters were most abundant in intergenic and intronic regions (), potential areas for gene expression control. The list of SARE-containing genes showed genes with central roles in the nervous system such as NMDA (Grin2a
, essential for excitatory synapses), Robo2
(with major functions in axon guidance) and Cutl/Cux2
(determinant for cerebral cortex layer II-IV) (examples in ). Classification of genes containing SARE sequences at the GO categories using Toppfun application (http://toppgene.cchmc.org/“ToppFun”)
indicated that the processes potentially affected by SARE regulation are clearly related to the nervous system (Table S2
). This analysis yielded several enriched GO categories, out of 112 significantly enriched GO biological processes, 21 (18,75%) of them related to neural functions (). All of these categories are specifically related to nervous system development and maintenance, and many showed significant greater enrichment than other categories ( and Table S2
). In accordance with our hypothesis, this prevalence of neural functions supports potentially important, selective action of SARE-mediated mechanisms in the nervous system. Next, SARE containing genes were grouped into two main categories representing potential distal and proximal regulatory sequences: intergenic; and intragenic, cds, promoter and utr regions (Table S2
) and GO analysis was performed separately with these two groups. Both groups showed similar enrichment in neural functions; therefore it did not favor proximal or distal regulatory regions as more relevant to plasticity.
SARE regulation affects several aspects of neuron function.
Classification of SARE-containing genes at GO indicated significant enrichment of processes related to the nervous system.
The analysis of genes containing the SARE cluster appeared an appropriate approach for identification of mechanisms of homeostasis, plasticity and activity-dependent remodeling in the nervous system. This study disclosed a large number of genes known to participate in plasticity and synaptogenesis (examples in ); Homer1
is an example in this category. Homer
genes encode scaffolding proteins that bind Ca2+
signaling proteins and target them to their correct subcellular localization 
; they are essential for dynamic regulation of the synapse, synaptic plasticity, and spatial learning 
. Coincident expression of experience-triggered Homer and Arc proteins is found in hippocampal and cortical neurons 
, which supports simultaneous activation, as predicted by our analysis. We also identified axonal guidance molecules ( and Table S1
), including PlxnA4
and its ligand Sema6A
as molecules potentially regulated by SARE. The semaphorin and plexin receptor families, together with neuropilins, are crucial during nervous system development and homeostasis, and mark the pathway for axon growth 
. These proteins also control synaptogenesis, axon pruning, the density and maturation of dendritic spines and are implicated in a number of developmental, psychiatric and neurodegenerative disorders 
. As for the axon guidance cues, we found a number of genes that encode cytoskeletal remodeling molecules at the synapse (). For example, ankyrins link integral membrane proteins to the underlying spectrin-actin cytoskeleton; they have key roles in activities such as cell motility, activation, proliferation, contact, and maintenance of specialized membrane domains. They might be involved in bipolar disorder and other mental alterations 
Less anticipated were SARE-containing genes not previously implicated in plasticity or structural maintenance of the synapse; in this category, we found neuronal subtype-specific TF such as Cux1
( and Table S1
). Cux TF expression is restricted to neurons of layers II-III and IV of the cerebral cortex. During development, Cux regulate dendritic branching, spine morphogenesis and synapse maturation 
. Cux expression is maintained through adulthood, but nothing is yet known of their function in mature neurons. Whereas Cux functions could be associated with plasticity at the postsynaptic site 
, Zic2 might act on the presynaptic terminal, as it is associated with axon development in retinal ganglion cells; Sox6, in turn, is described as essential for neuronal differentiation 
. These observations suggest that activity-dependent mechanisms act on pathways specific to neuronal subtypes.
To test the relevance of our findings and the predictive capability of our gene set to identify genes up-regulated upon neuronal activation we searched for experimental confirmation. Several studies of gene expression changes induced by neuronal activation have been reported and made useful available contributions. Many of them analyze the effects of the gabaergic inhibitor bicuculline to trigger neuronal excitatory response. We therefore compare the list of SARE containing genes with those of genes which expression was modify in studies analyzing the in vivo
effects of infusion of bicuculline into the accessory olfactory bulb 
; in vitr
o bicuculline treatment of cortical cells 
; and hippocampal neuronal cultures 
. This allowed us to extend our validation to several neuronal types. In all three cases, the comparison revealed a highly significant enrichment between the SARE containing genes and those up-regulated upon neuronal activation, but none or of lower statistical significance, when compared to the list of genes that are down-regulated (). Several of the SARE genes, such as Homer
are common to two or the three studies, and may represent a general pan-neuronal response, while unique ones might represent tissue-specific responses. These significant overlapping validate our results with external independent data. We next took the reverse approach and tested the predictive capability of our study by testing the expression of genes picked from our list upon neuronal activation. Cells from E18 mouse cortex were dissociated, neurons were cultivated and neuronal activity triggered using bicuculline 
. RNA was obtained and transcript expression of twelve SARE containing genes, including Arc, analyzed by using quantitative real time RT-PCR (Q-PCR) (). Up-regulation of Arc gene demonstrated efficient neuronal activation and, also expected, the levels of the S-isoform of Homer1
were increased 
. Six more genes showed up-regulation when neurons were activated. Atf3
, and Npas4
up-regulation in cortical cells was in agreement with our own analysis of the raw data obtained from gene expression arrays reported by other investigators 
, and further confirmed our comparison with external sources (). Interestingly, up-regulation of Cux1
, and PlxnA4,
genes not suspected to be regulated by activity, again confirmed the predictive capacity of our study. Four genes, Lmo4
did not show significant changes. This can be ascribed to the almost certain possibility of a number of false positive in our list, to the fact that other splicing variants might be affected, or to the possibility that subsets of genes may respond differently depending on the stimulus that triggers neuronal response.
The number of SARE containing genes is significantly enriched on genes up-regulated upon bicuculline triggering of neuronal activity.
Up-regulation of the mRNA of SARE containing genes and promoter activation in response to neuronal activity. A)
Next, the sequence corresponding to two of these SARE sequences were cloned upstream of a minimal promoter into vectors containing luciferase reporters to test their ability to activate transcription in response to neuronal activity. Cells from E18 mouse cortex were transfected with reporter constructs, neuronal activity triggered using bicuculline 
, and luciferase activity compared to control tetrodotoxin (TTX) treated neurons. These experiments demonstrated that these novel identified SARE sequences replicate the promoter activity of the SARE sequence corresponding to the Arc gene and significantly increase transcription upon depolarization ().
Our analyses thus point to overlooked pathways that might participate in activity-dependent regulatory mechanisms and, by extension, suggests the identification of genes potentially linked to mental diseases caused by plasticity defects 
. This is the case of genes reported as candidates for autism in which we found SARE sequences, such as, NRXN1
and 2 
and others (http://gene.sfari.org
(see Table S1
Validation of our prediction nonetheless required evaluation of true enrichment of genes involved in cognitive dysfunction. Fragile X syndrome (FXS) is a well-characterized form of autism, caused by loss of function of the Fragile X mental retardation protein (FMRP), which regulates local translation and plasticity at pre- and postsynaptic sites 
. Based on a recent extensive list of genes targeted by FMRP, from which the authors extract a stringent set of 842 reliable targets 
, we hypothesized that the list of SARE-regulated genes will be enriched in FMRP targets. Comparison of SARE-containing genes (including those containing SARE clusters at intergenic locations) with the stringent list of FMRP targets resulted in 70 genes common to both (8.5% overlap), an enrichment of biological relevance (p
4.3909–13; see Methods) (). The relationship between the SARE-containing genes and FMRP targets thus strongly supports SARE involvement in activity-dependent regulation. In addition, it suggests that mutations in SARE or SARE-containing genes and pathways can contribute to mental retardation, autism spectrum disorders and other psychiatric diseases.
FMRP targets containing SARE sequence genes.
Correct function of nervous system networks and subnetworks is possible thanks to the extraordinary spatial and temporal coordination of gene expression that is guided by the TF subset expressed by each neuronal population. Our findings suggest that cooperation between CREB, SRF, and MEF2 transcription factors at the SARE region is one of the precisely regulated mechanisms that govern the transcriptional program of activated neurons. This transcriptional cooperation might also apply to other TF to initiate an appropriate, specific transcriptional response in other biological processes. This study also highlights the value of the development and use of computational tools and databases for the comprehensive analysis of biological events. We identified a subset of genes whose transcription is potentially regulated by the SARE cluster after synaptic activation. Most of these genes are directly related to nervous system development and maintenance; several of them are reported at the synapse, some are mutated in human mental disorders, and many form part of FMRP-regulated mechanisms. The identification and functional analysis of SARE-containing genes provided here is thus a useful for implicating new candidate genes in plasticity, memory, and mental retardation, and suggests new approaches to the study of mental disorders in which synaptic activity might have a central role.