Systematic review and transcriptomic meta-analysis of environmental enrichment reveal core molecular programs of brain plasticity

Enriched Environment, meta-analysis, transcriptomics, cognition, synapse 34 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 2

35 RATIONALE. Environmental enrichment (EE) paradigms in rodents have long demonstrated 36 that enhanced sensory, cognitive, social, and motor stimulation positively impacts brain 37 function, improving learning, memory, and neuroplasticity. These effects have significant 38 implications for understanding cognitive development and mitigating cognitive decline and 39 brain aging. While numerous transcriptomic studies have explored EE -induced molecular 40 changes, a unified view of the genes and pathways consistently modulated remains lacking. 41 METHODS. To address this gap, we performed a systematic review and meta-analysis. We 42 conducted a comprehensive PubMed search for all studies published up to February 2025 43 that matched all the following inclusion criteria: (1) employed EE paradigms; (2) were 44 conducted on rodents; (3) utilized genome-wide transcriptomic methods; (4) examined brain 45 regions or neuronal populations. The 323 retrieved articles were manually screened for 46 relevance to the study aims and data availability. Datasets from 20 eligible RNA-seq reports 47 were reprocessed using a unified analysis pipeline and subjected to a meta -analysis with 48 three complementary statistical methods. 49 RESULTS. Despite considerable heterogeneity across studies, our integrative analysis 50 identified consistent gene expression signatures linked to synaptic function, plasticity and 51 their transcriptional regulation. These molecular insights advance our understanding of how 52 EE impacts on neuronal and behavioural outcomes, and may inform therapeutic strategies 53 aimed at replicating or enhancing EE benefits. To promote open science and foster further 54 research, we developed an accessible web application, mEEtaBrain, that enables the 55 neuroscience community to navigate and interrogate our meta-analysis results. 56 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 3

57 The development of the mammalian brain is guided by complex genetic and epigenetic 58 programs that establish most of its structural framework before birth. Yet, the continuous 59 refinement of neural circuits essential for normal cognitive and behavioural function depends 60 heavily on inputs from the environment during childhood and adulthood 1. There is broad 61 consensus that a stimulating lifestyle, combining physical activity, rich social interactions, 62 and intellectual stimulation, can significantly enhance cognitive development and reduce the 63 risk of cognitive decline in elderly humans 2,3. 64 Enriched environments (EE) have been widely used in rodent models to investigate 65 how external stimulation influences brain function and to uncover the underlying 66 neurobiological mechanisms. In these paradigms, rodents are housed in settings that offer 67 enhanced sensory, cognitive, social and motor stimulation relative to standard conditions. 68 Typically, EE involves keeping animals in large groups within an ample space, with 69 opportunities for voluntary exercise and exploration of novel objects, where toys and running 70 wheels are regularly replaced to maintain novelty. Decades of research have demonstrated 71 that EE can enhance learning and memory performances, both in wild type animals and in 72 a range of neuronal dysfunction models 4–7, including Huntington’s Disease 8,9, Alzheimer’s 73 Disease 10–13, dementia 14, brain injury 15,16, and intellectual disability disorders 17,18. 74 Investigation of the associated neuronal processes revealed that EE stimulates 75 hippocampal neurogenesis 19–21, increases spine density, dendritic length and dendritic 76 complexity in hippocampus and cortex regions 17,22–24, improves neurotransmitter function 77 25,26 and neurotrophic factors expression 27–29. 78 Building on this evidence from rodent studies, clinical occupational therapies and 79 related approaches have gained support as potential interventions for a broad range of 80 human conditions 30,31, from neurodevelopmental disorders in children 32 to cognitive 81 deterioration with aging 33. Moreover, these studies sparked great interest in uncovering the 82 precise molecular mechanisms by which EE improves brain function, aiming to identify novel 83 therapeutic targets which could mimic or potentiate EE benefits. Transcriptional and 84 epigenetic processes are well -established mediators between environmental stimuli and 85 genomic regulation 34,35, and play a pivotal role in orchestrating synaptic plasticity processes 86 that underlie neuronal function and memory 36–40. These mechanisms are thus appealing 87 candidates for translating EE lasting effect into behavioural and cognitive improvements, 88 and their understanding could be key to developing novel therapeutic interventions. 89 However, despite the wealth of genome-wide studies on animals housed in EE over the past 90 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 4 two decades, little effort has been made to integrate their findings, and no clear consensus 91 has emerged on the genes and pathways regulated by EE in the brain. 92 To uncover consistent gene expression signatures associated with EE across multiple 93 studies, we performed a systematic literature screening followed by meta -analysis of 20 94 independent brain transcriptomic studies from rodents exposed to EE. Despite substantial 95 heterogeneity in gene expression profiles across studies, our integrative analysis reliably 96 identified a set of reproducible transcriptional features of EE. These findings offer molecular 97 insights into EE synaptic and behavioural effects, and may guide the development of 98 targeted strategies aimed at mimicking or enhancing its cognitive benefits. To facilitate 99 broader exploration of our meta -analysis by the neuroscience community, we also 100 developed mEEtaBrain (http://regulomics.mimuw.edu.pl/mEEtaBrain/), a freely accessible, 101 user-friendly web application. 102 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 5

103 Compilation of brain transcriptomic datasets from EE studies 104 To conduct our meta-analysis, we aimed to collect all transcriptomic studies which met the 105 following four criteria: (1) employed EE paradigms; (2) used rodents as the animal model, 106 given the extensive EE literature for these species; (3) utilized genome -wide methods to 107 profile gene expression, to ensure an unbiased, discovery -driven approach rather than an 108 hypothesis-driven one limited to preselected genes; (4) examined brain regions or purified 109 neuronal populations as tissue/cell-type of interest. To this end, we queried PubMed using 110 a search string designed to capture all four aspects (see Methods for the full query and 111 dataset compilation details). This search yielded a total of 323 peer-reviewed studies. 112 We then manually screened the abstracts to identify only those which fully satisfied 113 the four inclusion criteria mentioned above. After this initial filtering, we retained 36 RNA -114 seq studies, including one using SAGE -seq, and 26 microarray studies. A second screen 115 was performed based on data availability. At the end of our screening workflow ( Fig. 1A), 116 we retained 20 RNA-seq 41–60 and 13 microarray studies 13,18,26,61–70. More details about the 117 selecting criteria can be found in the methods section. 118 The majority of these datasets were generated from mouse experiments (14 out of 119 20 for RNA-seq, 8 out of 13 for microarrays). These studies displayed large heterogeneity 120 in both the EE paradigms used and the biological materials profiled, which included 121 hippocampus or its subregions (17 studies), various cortical areas (9), striatum or its 122 subregions (4), amygdala (2), brain stem (1) or specific neuronal populations purified from 123 these regions. It should be noted that some papers applied EE in the context of brain 124 pathologies or stressors; in such studies we focused on EE-associated changes rather than 125 in the disease-related changes. Species, genetic background, age, sex, EE paradigm, CNS 126 region, method of transcriptomic profiling, bias assessment, and literature reference for 127 these datasets can be found in Supp. Table 1 and 2. 128 129 Meta-analysis of EE transcriptomic data identifies genes consistently modulated by EE 130 To ensure the robustness of the meta -analysis, we retrieved raw data from public 131 repositories and re-analyzed all RNA-seq datasets using a unified pipeline (Fig. 1B). After 132 quality control, filtering, and trimming, reads were quantified, aggregated to the gene level, 133 visually assessed via PCA and clustering, and finally analysed for differential expression. 134 Overall, this standardized workflow allowed for consistent processing across datasets, 135 providing a reliable foundation for downstream meta-analytic integration. 136 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 6 Regarding RNA-seq data, most mouse studies detected more than 14,000 genes and 137 shared over 10,000, providing a solid basis for data integration, with only one dataset 41 138 containing substantially fewer genes (~ 5,000) (Fig. 2A). Similarly, all rat studies shared over 139 11,000 detected genes ( Supp. Fig. S1A ). Pairwise intersection of differentially expressed 140 genes (DEGs) between EE and Standard Environment (SE) revealed modest overlap across 141 studies (Fig. 2B and Supp. Fig. S1B). In agreement, principal component analysis (PCA) 142 showed no obvious clustering of EE and SE samples in either mouse or rat datasets ( Fig. 143 2C and Supp. Fig. S1C), suggesting that EE-induced transcriptional changes are moderate 144 in magnitude and number of genes affected and likely masked by inter-study variability. 145 Under these conditions, simply intersecting DEGs from individual datasets is likely to 146 yield poor results, as it relies on arbitrary thresholds to include or exclude genes in each 147 study. To overcome this limitation and capture subtle but consistent expression changes 148 shared among datasets, we implemented a meta -analysis. Specifically, we used two 149 complementary approaches 71 (Fig. 1B): 150 i. p-value combination methods , including Fisher’s and weighted Stouffer’s 151 methods, which integrate p-values of individual analysis into a single combined 152 p-value per gene, to assess overall statistical significance across studies 153 ii. the Random Effect Model (REM) , which combines effect sizes to derive a 154 combined fold change per gene, to account for the magnitude and direction of 155 gene expression changes. 156 This dual approach increased our ability to detect robust gene expression patterns 157 associated with EE. Both methods generated meta-analysis statistics for over 10,000 genes, 158 which were detected in at least 16 of the 18 mouse studies and in all rat datasets ( Fig. 2D 159 and Supp. Fig. S1D). Depending on the approach and parameters used, our meta-analysis 160 revealed hundreds to thousands of significantly affected genes across mouse datasets (Fig. 161 2E and Supp. Table 3-5), as illustrated by the volcano plot for REM method (Fig. 2F). One-162 study-out sensitivity analysis confirmed the robustness of our meta -analysis approaches 163 (Supp. Fig. S2A -C). Notably, irrespective of the method used, upregulated genes largely 164 outnumbered downregulated ones, suggesting that EE primarily enhances rather than 165 represses gene expression. In contrast, far fewer significant genes were detected in rat 166 studies, and inter-species overlap was minimal (Fig. 2E). This was likely due to the limited 167 number of available rat datasets, and substantial confounding factors introduced by the use 168 of addiction or brain disease models in the majority of rat studies ( Supp. Table 2 ). To 169 facilitate exploration of the data, all resulting lists of EE -dependent genes can be 170 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 7 interactively accessed through mEEtaBrain , a dedicated app which allows users to 171 dynamically adjust meta-analysis methods and settings, explore integrated results, perform 172 custom gene queries, and generate and download interactive plots and GO analysis. 173 Regarding microarray studies, raw data were unavailable for most datasets. Only lists 174 of DEGs were provided, occasionally with associated fold changes. For genes not included 175 in these lists, it was unclear whether they could no t be detected or simply were not 176 differentially expressed, making it impossible to assess their relative expression between 177 EE and SE. Due to these limitations, a formal meta-analysis was not feasible. To still extract 178 meaningful insight from these datasets, we identified DEGs consistently reported across 179 combined mouse and rat microarray studies. Pairwise comparisons of DEG lists revealed 180 limited overlap (Supp. Fig. S3A), with only 12 genes shared by at least three studies (Supp. 181 Fig. S3B and Supp. Table 6 ). The intersection of the gene lists obtained with the three 182 different RNA -seq meta -analysis methods and from microarray datasets revealed 183 substantial overlaps (Supp. Fig. S3C). As expected, Fisher’s method identified a larger 184 number of DEGs, consistent with its lower stringency compared to Stouffer’s and REM 72. 185 186 EE modulates key genes and complexes operating at the synapse 187 To gain an unbiased insight into the biological processes modulated by EE, we performed 188 gene ontology (GO) enrichment analysis on the genes significantly up - and downregulated 189 by EE. The results illustrated in Fig. 3 are based on Fisher’s method meta -analysis (adj p-190 value < 0.01), which yielded a broader set of significant genes well suited for enrichment 191 analysis. Alternative statistical thresholds and methods can be explored by users through 192 the web app. GO Cellular Compartment analysis revealed a very strong enrichment for 193 neuronal and, specifically, synaptic structures, as well as for terms related to extracellular 194 space and membranes, both across up - and downregulated genes ( Fig. 3A and Supp. 195 Table 7 ). Consistently, Biological Processes GO categories pointed to synaptic 196 transmission, signalling and plasticity (Fig. 3A and Supp. Table 7), while Molecular Function 197 terms were enriched for transmembrane transporters and ion channel activity ( Supp. Fig. 198 S4A and Supp. Table 7 ). Analysis with the manually -curated, synapse -focused SynGO 199 resource confirmed high enrichment of synaptic genes across both presynaptic and 200 postsynaptic compartments (Fig. 3C) and spanning multiple synaptic processes ( Fig. 3D), 201 including 188 upregulated and 120 downregulated genes. In particular, genes related to 202 synapse organization and synaptic signalling were especially enriched among EE -203 upregulated genes, including the postsynaptic scaffolds Dlg4, Ppp1r9b, Homer1, Shank1, 204 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 8 Dlgap3, the ion channels Grin2a, Gabra5, Kcnj6, the synaptic adhesion molecules Nectin1, 205 Nlgn2, Nrxn2, Pcdh7, Pcdh8, Igsf9, Robo2, and multiple actors of the synaptic vesicle cycle 206 such as Snap25, Unc13a, Stx3, Doc2b, Prrt2, Stxbp5l, Syp and Rab3a (Supp. Table 8). 207 Together, these analyses indicate that EE influences neuronal transcriptomes and broadly 208 modulates genes involved in synaptic structure, function and signalling. 209 To better understand the functional relationship among significant EE -upregulated 210 genes, we conducted a protein-protein interactions analysis of their gene products using the 211 STRING database. The resulting network displayed significantly more interactions than 212 expected by chance, indicating strong enrichment in physical connectivity (p -value = 10-16) 213 (Fig. 3E and Supp. Table 9 ). Among the top three hubs of the network, two were well -214 established scaffold proteins of the postsynaptic density (PSD): PSD -95 and Spinophilin, 215 encoded by the genes Dlg4 and Ppp1r9b, respectively. PSD -95 is a master organizer of 216 excitatory synapse architecture. By anchoring NMDA and AMPA receptors and linking them 217 to signalling complexes and the actin cytoskeleton, PSD-95 plays a central role in synapse 218 stabilization, receptor clustering, and activity-dependent synaptic plasticity 73,74. Spinophilin 219 targets protein phosphatase 1 (PP1), a key regulator of synaptic plasticity, to specific 220 substrates, particularly glutamate receptors, thereby modulating AMPAR and NMDAR 221 function and regulating synaptic efficacy and spine morphology 75–77. The identification of 222 Dlg4 and Ppp1r9b and their associated protein network among EE -modulated genes (Fig. 223 3E, zoomed panel 1) highlights PSD and dendritic spines as critical neuronal compartments 224 affected by EE. 225 We also found multiple core components ( Snap25, Stx3) and modulators ( Cplx2, 226 Stxbp5l, Doc2b, Prrt2, Rab3a) of the SNARE complex (Soluble NSF Attachment Protein 227 Receptor complex) ( Fig. 3E , zoomed panel 2). The SNARE complex is essential for 228 neurotransmitter release, driving synaptic vesicle fusion with the presynaptic membrane 229 78,79. It also mediates postsynaptic exocytosis and plasticity, including long-term potentiation 230 (LTP) and spine growth 80–82. The enrichment of SNARE complex genes in our meta-analysis 231 further supports the notion that EE modulates genes involved in the dynamic regulation of 232 synaptic function. 233 234 EE enhances the activity-dependent transcriptional program 235 The diverse motor, social and cognitive stimuli provided by EE are thought to activate 236 neuronal circuits responsive to such inputs across multiple brain areas 6. We reasoned that, 237 if our meta-analysis reliably captures transcriptional features of EE, it should reveal a tonic 238 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 9 increase in the expression of Immediate Early Genes (IEGs). These genes are rapidly 239 induced in response to neuronal activation, and many of them encode transcription factors 240 believed to play key roles in the transcriptional regulation of experience-dependent synaptic 241 plasticity 83. Indeed, among the non-synaptic proteins identified in our network, we found the 242 AP-1 transcriptional complex composed of activity -induced FOS and JUN family members 243 (Fig. 3E, zoomed panel 3). AP-1 is a master regulator of activity-dependent gene expression 244 83 and directly controls multiple synaptic genes 54,84. Therefore, alongside other activity -245 induced transcription factors identified in our meta -analysis such as EGR 1 -3, NPAS4 and 246 NR4A1, AP-1 might likely contribute to the widespread transcriptional modulation of synaptic 247 genes revealed by our GO analysis. 248 To test our prediction that EE increases IEG expression in a more systematic manner, 249 we crossed EE -dependent gene lists from our mouse meta -analysis with a previously 250 published set of activity-induced genes in the mouse hippocampus 37. For both Fisher’s and 251 weighted Stouffer’s methods , we observed a highly significant overlap between IEGs and 252 genes up-regulated, but not downregulated, under EE conditions (Fig. 4A). As combined p-253 value methods primarily assess statistical significance rather than the direction of gene 254 expression changes, we focused on REM results to evaluate the effect of EE on IEG 255 regulation. Analysis of gene expression fold changes ( Fig. 4B) and Gene Set Enrichment 256 Analysis (GSEA) for the IEG gene set ( Fig. 4C ) confirmed a consistent and significant 257 upregulation of activity-induced genes under EE across studies, as illustrated by canonical 258 IEGs such as Fosb, Dusp5 and Nptx2 and (Fig. 4D). Remarkably, 10 out of the 12 DEGs 259 consistently identified across microarray datasets were IEGs ( Supp. Fig. S3B), further 260 supporting our conclusion. Together, these results support the notion that EE promotes the 261 activity-dependent transcriptional program, in agreement with its believed role in stimulating 262 neuronal activity. This underscores the reliability of our meta -analysis strategy in capturing 263 biologically meaningful molecular hallmarks of EE. 264 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 10

265 Literature search 266 The systematic literature review and meta-analysis were undertaken according to Preferred 267 Reporting Items for Systematic Reviews and Meta -Analyses (PRISMA) guide lines 85, and 268 PRISMA checklist is shown in Supp. Table 10. PRISMA flow chart in Fig. 1A outlines our 269 dataset collection and curation pipeline. We aimed to include transcriptomic studies 270 published up to 10 February 2025 which met all the following four inclusion criteria: (1) 271 employed EE paradigms, (2) used rodents as the animal model, (3) utilized genome -wide 272

to measure gene expression, and (4) focused on brain regions or purified neuronal 273 populations as tissue/cell-type of interest. To this aim, we queried the NCBI databases with 274 the following search: ("environmental enrichment" OR "enriched environment" OR "EE 275 model" OR “enriched environments”) AND (mouse OR mice OR murine OR rodent OR rat 276 OR rats) AND (brain OR hippocampus OR cortex OR "central nervous system" OR CNS OR 277 “nucleus accumbens” OR hippocampi OR hippocampal) AND ("RNA -seq" OR "RNA 278 sequencing" OR "transcriptome analysis" OR "transcriptomic profiling" OR "RNA-Seq data" 279 OR "gene expression" OR "differential expression" OR transcriptomics OR microarray OR 280 microarrays OR “molecular atlas” OR omics OR “multi -omics”). This search yielded a total 281 of 322 peer -reviewed studies. In addition, we included a study from Angel Barco’s team, 282 which has been recently accepted for publication 54. 283 We then manually screened the 323 studies to retain only those which fully satisfied 284 the four inclusion criteria mentioned above. This selection process was conducted 285 independently by three researchers through abstract review and, when necessary, access 286 to the full text. Review articles were excluded. Concerning EE protocols (1), studies involving 287 only a single exposure to an EE were excluded, because we considered this approach to 288 represents a novelty exploration paradigm rather than an EE paradigm. Studies employing 289 voluntary exercise without the object component were retained. We included papers which 290 applied EE in the context of genetic, pharmacological or injury models of various brain 291 pathologies; in such studies we focused on EE -associated changes rather than in the 292 disease-related changes. Regarding gene expression analysis methods (3), we included the 293 studies that used RNA sequencing, both bulk and single -cell, or microarrays. One article 294 employing SAGE -seq was also retained at this step, since the data generated with this 295 technique are consistent with RNA-seq analysis pipelines, although the study was ultimately 296 excluded due to its read data being in an incompatible format with our analytical framework. 297 In contrast, studies using non -genome-wide methods restricted to predefined gene sets, 298 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 11 such as qPCR array and TaqMan arrays, were excluded. Four papers focused specifically 299 on short RNA (sRNA) but data were available only for two of them, preventing the analysis 300 of sRNA studies as a separate group. Consequently, we excluded those papers. Regarding 301 brain regions and cell-types of interest (4), studies focused on the peripheral nervous system 302 or on non-neuronal cell-types residing in the CNS were excluded. After this initial filtering, 303 we retained 36 RNA-seq studies, including one using SAGE-seq, and 26 microarray studies. 304 Upon reviewing the full texts of the selected articles, four RNA-seq studies and seven 305 microarray studies were excluded as they did not meet the predefined inclusion criteria. 306 Then, we performed a second screen based on data availability. We excluded six RNA-seq 307 articles due to the unavailability of FASTQ files, either publicly or upon request from the 308 authors. Additionally, five RNA-seq papers were excluded because they did not include new 309 experimental data but instead re -analysed datasets already examined in other selected 310 studies. We also excluded five microarray datasets since no DEG tables were published 311 from these studies or no significant DEGs were detected. Furthermore, we removed another 312 microarray study during data extraction since no fold change data were available. Following 313 this second filtering step, based on data availability, we retained 20 RNA -seq and 13 314 microarray studies. The whole screening workflow is illustrated in Fig. 1A. A full list of the 315 source papers included in the meta-analysis can be found in Supp. Table 1 and 2. 316 317 Bias assessment 318 The risk of bias within individual RNA -seq studies was evaluated across multiple domains, 319 including subjective exclusion of outlier samples, sample pooling, absence of untreated or 320 wild-type controls, and technical inconsistencies across samples. Each study was then 321 assigned a risk -of-bias rating of “low”, “some concerns”, “moderate”, “serious”, or “high”. 322 Potential sources of bias and the overall rating for each study are reported in Supp. Table 323 1. 324 325 RNA-seq dataset meta-analysis 326 Raw sequencing data in FASTQ format were obtained from public repositories using 327 accession identifiers provided in the respective publications and downloaded with fasterq -328 dump (v3.2.1). Sample files were renamed according to the metadata available in the 329 corresponding repository. Quality control of the FASTQ files was performed using FastQC 330 (v0.12.1) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and summary 331 reports were generated with MultiQC (v1.28) 86. Based on these results, filtering and 332 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 12 trimming parameters were manually determined, and preprocessing was performed with 333 fastp (v0.23.4) 87 using combinations of the flags -3, -x, -c, -p, and -f 1–3, depending on the 334 QC outcomes. 335 Reads were mapped using RSEM (v1.3.1) 88 with STAR 89 as the aligner. Mouse 336 datasets were aligned to the GRCm39 primary assembly with ENSEMBL GTF annotation 337 v112, and rat datasets to the mRatBN7.2 primary assembly with ENSEMBL GTF annotation 338 v113 90. Strand specificity was assessed using infer_experiment.py (v5.0.4) 91. Mapping 339 quality was evaluated with rsem -plot and by inspecting BAM files with samtools flagstat 340 (v1.3.1) 92. Transcript-level quantifications from .isoforms.results files were summarized to 341 gene-level counts with tximport (v1.30.0) 93 using a tx2gene object generated from the same 342 GTF file used for alignment. In one dataset, ComBat -seq (sva v3.50.0) 94 was applied to 343 correct a strong batch effect, as recommended in correspondence with the study authors 68. 344 Differential expression analysis was performed with DESeq2 (v1.42.0) 95. Variance-345 stabilized counts (vsn v3.70.0) 96 were explored via principal component analysis (PCA) 346 using pcaExplorer (v2.28.0) 97 and hierarchical clustering heatmaps generated with 347 pheatmap (v1.0.13) (https://github.com/raivokolde/pheatmap), applying Euclidean distance 348 to either the top one hundred most variable genes or all genes. These analyses, combined 349 with quality control observations, were used to identify and exclude outlier samples. In some 350 cases, clustering revealed the presence of multiple experimental batches. The two hundred 351 most significant genes associated with the variable of interest, identified via t -tests, were 352 represented in heatmaps, which can be visualized in the app. MA plots were generated using 353 apeglm 98, ashr 99, and normal shrinkage methods and visualized with ggplot2 (v3.5.2) 354 (https://ggplot2.tidyverse.org; H. Wickham. ggplot2: Elegant Graphics for Data Analysis. 355 Springer-Verlag New York, 2016). The DESeq2 design formula was adapted according to 356 the study complexity: datasets with three variables were split into two batches, those with 357 two variables followed a 2×2 design for meta -analysis, and those with a single variable 358 employed a simple design. For downstream meta-analysis, in addition to the two-sided tests 359 for generating results employed in effect size combination methods, one-sided Wald test p-360 values were generated for use in p -value combination methods. Differentially expressed 361 genes were defined as those with an adjusted p-value < 0.05. DEG identification was robust 362 to the manual curation process ( Supp. Fig. S2D, E ). G ene names and biotypes were 363 annotated using EnsDb.Mmusculus.v79 and EnsDb.Rnorvegicus.v79 100. 364 Log₂ fold change combination for meta -analysis was performed with MetaVolcanoR 365 (v1.16.0) (Prada C, Lima D, Nakaya H, 2024), with custom modifications to the plot_rem 366 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 13 and draw_forest functions for visualization. The resulting metaresult object was filtered to 367 retain only genes represented in at least six contributing studies for mouse datasets and at 368 least two for rat datasets, after which p-values were adjusted using the Benjamini–Hochberg 369 (B–H) method. MetaP (v1.12) ( https://cran.r-project.org/web/packages/metap/index.html) 370 was used to combine p -values via Fisher’s method (sumlog) and the weighted Stouffer 371

(sumz). For the latter, an epsilon correction was applied to weights equal to 1 to 372 ensure all values were below 1. Filtering and B–H adjustment were applied as in the random-373 effects model analysis. 374 For data visualization, interactive plots were generated using plotly (v4.10.4) (Sievert 375 C, 2020; https://plotly-r.com), and batch effects were corrected with ComBat from the sva 376 package (v3.50.0) 101. Color schemes in principal component analysis plots were defined 377 using RColorBrewer (v1.1.3) ( Neuwirth E, 2022; https://CRAN.R-378 project.org/package=RColorBrewer). To enable cross -species integration, rat genes were 379 mapped to their mouse orthologs using biomaRt (v2.58.2) 102, querying the 380 rnorvegicus_gene_ensembl dataset for the attributes "ensembl_gene_id" and 381 "mmusculus_homolog_ensembl_gene." Gene identifiers were subsequently converted to 382 gene symbols using org.Mm.eg.db or org.Rn.eg.db (Carlson M, 2023). 383 384 Microarray dataset integration 385 Gene symbols and fold changes were extracted from microarray studies. Gene symbols 386 were compared with the NCBI database and microarray annotation data. Old gene symbols 387 were updated by using the biomaRt - (v 2.65.0) 102, mygene- (v 1.44.0) (Mark, Thompson, 388 Afrasiabi, Wu (2025). https://bioconductor.org/packages/mygene) and annotate (v 1.86.1) 389 (Gentry J (2025). https://bioconductor.org/packages/annotate) R packages. Fold changes 390 were log2-transformed to symmetrize effect sizes around zero, and a combined log2 FC was 391 calculated as the arithmetic mean of individual log2 FC values. 392 393 Gene Ontology (GO) and gene network analysis 394 Enrichment analysis in the Shiny app was performed using clusterProfiler (v4.10.1) together 395 with the annotation packages org.Mm.eg.db or org.Rn.eg.db. Genes annotated with 396 ENSEMBL IDs were first converted to Entrez IDs prior to analysis. All three Gene Ontology 397 domains provided by the package were included. For the universe parameter, all genes 398 present in the result matrix were used, with the final background set defined as the 399 intersection between the database and the genes in the matrix, consistent with the 400 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 14 package’s default behavior. The CellMarker database for mouse was downloaded from bio-401 bigdata.center in January 2025. For Gene Set Enrichment Analysis (GSEA) 103, the complete 402 gene list from REM analysis was ranked by the product of the sign of the summary log2 fold 403 change and the negative logarithm of the summary p-value. A minimum of six datasets with 404 detected values was required for a gene to be included in the ranked list. For GSEA and 405 EE/SE fold change analysis of IEGs, the 200 genes most strongly up -regulated in the 406 hippocampus after kainate treatment were used 37, among which 144 were detected in at 407 least 6 datasets. For SynGO analysis 104, mouse genes were converted to their human 408 homologues with SynGO ID conversion tool before GO analysis. Protein network analysis 409 was performed with STRING 105 on the significant upregulated genes found from mouse 410 studies with Fisher’s method (adj p-value = 0.01). Only the physical subnetwork was taken 411 into account, using textmining, experiments and databases as interaction sources (minimum 412 required interaction score = 0.4). The resulting network was exported to Cytoscape 106 for 413 visualization. 414 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 15

415 Since the widespread adoption of microarray technology in the 2000s and next -generation 416 sequencing in the 2010s, numerous studies have investigated the impact of EE on the brain 417 transcriptome 7. However, there has been limited effort to integrate these datasets and 418 determine consistent gene expressions signatures associated with EE. To address this gap, 419 we conducted what it is, to our knowledge, the first meta -analysis of brain transcriptomic 420 datasets from rodents exposed to EE. 421 Across our broad dataset collection (33 studies), a direct intersection of EE -induced 422 DEGs from individual datasets revealed only limited overlap. This inconsistency likely 423 reflects both methodological and biological factors. Differences in experimental design, such 424 as sex, age, strain, group size, duration of EE exposure, object characteristics (e.g. shape, 425 texture, size), object replacement frequency, and the presence or absence of running 426 wheels, likely influenced transcriptional outcomes, and argue for standardization of EE 427 protocols to improve reproducibility. In addition, prolonged EE exposure can trigger 428 neuroadaptive gene expression responses which may differ between individuals 21,107, 429 introducing additional variability even across animals housed under identical conditions. 430 Biological sources of variability also include the specific brain region analysed in each study, 431 and the high cell -type heterogeneity of neural tissue, which may mask cell -type-specific 432 responses. 433 To overcome study heterogeneity and uncover gene expressions signatures 434 consistently associated with EE, we implemented three meta-analysis approaches for RNA-435 seq datasets: Fisher’s and weighted Stouffer’s methods, which combine p -values, and the 436 Random Effect Model, which accounts for the magnitude and direction of gene expression 437 changes. These methods differ in their assumptions and statistical properties 71. Indeed, 438 Fisher’s analysis found the largest number of significant genes, consistent with its greater 439 sensitivity to small p -values compared to the more conservative Stouffer’s method 72. 440 Regarding microarray studies, the lack of available raw data restricted their integration to 441 comparisons of DEG lists. 442 Our meta-analysis of mouse RNA-seq datasets identified sets of genes and pathways 443 consistently modulated by EE. A key finding was the enrichment of synaptic genes. Despite 444 most source datasets were obtained from bulk tissue and thus lacked cell -type resolution, 445 the pronounced overrepresentation of synaptic GO terms clearly points to a major effect of 446 EE on neuronal populations. The involvement of structural components of the postsynaptic 447 density (PSD), core subunits and regulators of the synaptic vesicle machinery, ion channels, 448 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 16 and synaptic adhesion proteins supports the growing consensus that EE influences neuronal 449 and spine morphology and synaptic function. Several studies have shown that EE increases 450 spine density, dendritic length, and dendritic complexity in hippocampal and cortical regions 451 17,22–24. Recordings from hippocampal neuron revealed that EE increases cell excitability, 452 specifically enhances excitatory synaptic transmission in DG granule neurons, and 453 facilitates long -term potentiation (LTP) in CA1 pyramidal neurons 5,108–110. While 454 experimental manipulation of the identified genes and pathways is required to infer causality, 455 the transcriptomic changes revealed by our meta-analysis likely act together to support the 456 morphological and functional synaptic adaptation underlying EE beneficial effects. 457 Notably, Gene Ontology (GO) terms related to synapses were strongly 458 overrepresented among both upregulated and downregulated genes. This pattern could, in 459 principle, arise from the properties of the Fisher’s method, which emphasizes small p-values 460 from individual studies even when others show weak or no effects, potentially leading to the 461 same gene being classified as both upregulated and downregulated. However, we found 462 that only 22 genes were modulated in both directions, representing a median of 3.37% of 463 the genes within each of the top enriched GO categories. Thus, the shared enrichment of 464 synaptic GO terms among up- and downregulated genes is unlikely to be an artifact of the 465 meta-analysis method itself, but instead genuinely reflects the complex and widespread 466 remodeling of the synaptic transcriptome induced by environmental enrichment. 467 We also found that EE upregulates the activity-dependent transcriptional program, a 468 finding that was also supported by microarray studies. This evidence supports, at the gene 469 expression level, the general notion that enhanced sensory, cognitive, motor and social 470 stimulation triggers neuronal activation across cortical and hippocampal regions 6. Whereas 471 other studies had shown IEG upregulation under EE 45,46,54, our meta-analysis generalizes 472 this conclusion across tens of datasets. Notably, among the significantly affected IEGs, we 473 identified multiple FOS and JUN family members which form the activity -induced AP-1 474 transcriptional complex. AP-1 is a master regulator of the activity-dependent transcriptional 475 program 83, controls synaptic gene expression 84,111, and has been proposed to mediate the 476 cognitive benefits of EE 54. In light of these findings, AP -1 likely plays a central role in the 477 modulation of synaptic genes by EE revealed by our meta -analysis. Gene regulatory 478 network analysis could potentially provide further insights in the transcriptional regulatory 479 layers underlying the observed gene expression changes, but attempting it would require 480 adapting the approach to each study and is therefore beyond the scope of this work. 481 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 17 Altogether, these results demonstrate the power of our systematic approach in uncovering 482 consistent molecular signatures of EE across diverse experimental conditions. 483 To facilitate biological interpretation of the RNA -seq meta-analysis and enable data 484 interrogation tailored to specific research questions, we developed mEEtaBrain, an 485 interactive online app to explore and compare results across methods. This tool is made 486 publicly available to all of the research community so that any researcher interested in a 487 subset of the studies we have analyzed can get their own lists of genes that would meet 488 their own significance criteria. We hope that this will greatly improve the impact of our 489 analysis on the scientific community. 490 The intersection of EE-dependent genes between mice and rats was very small. We 491 believe that the discrepancy might more likely originate from the quality and quantity of rat 492 datasets rather than from genuine inter -species biological differences. Importantly, in 5 of 493 the 6 rat studies 55–59, the authors examined the effect of EE in the context of severe 494 stressors such as cocaine administration and lead exposure, or using models of brain 495 conditions, thereby introducing a substantial confounding factor in the comparison between 496 SE and EE. Also, the limited number of rat studies likely reduced statistical power, resulting 497 in much fewer significant genes compared to the mouse meta -analysis and contributing to 498 the modest inter-species overlap. Furthermore, the large majority of mouse datasets were 499 obtained from cortex or hippocampus, which predominantly contain glutamatergic neurons, 500 whereas half of rat studies was conducted on striatum, almost entirely composed of 501 GABAergic neurons. This discrepancy in the brain regions examined, together with the other 502 factors mentioned above, likely contributes to the apparent lack of overlap between the two 503 species. Finally, reported differences in cognitive performance and stress resilience between 504 mice and rats 112,113 could also influence how EE affects brain transcriptomes in each 505 species. 506 This study presents some limitations. Substantial heterogeneity in experimental 507 design, EE protocol, and tissue of interest across the source studies likely increased 508 variability in the meta-analysis outcomes. The use of stressors or disease models in some 509 studies introduced a confounding factor and limited reliable interspecies comparison. Raw 510 data were unavailable for most microarray studies, preventing to conduct a proper meta -511 analysis. Overall, the studies included in the meta-analysis exhibited a low to moderate risk 512 of bias. 513 In addition to transcriptomic analysis, several works have examined the impact of EE 514 on chromatin profiles, including histone acetylation, histone and DNA methylation, and 515 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 18 chromatin accessibility 46,49,51,54. Futures research should aim to determine core EE -516 dependent epigenetic modifications across studies using a systematic approach, and relate 517 these changes to the gene expression patterns uncovered in our analysis. Pinpointing 518 critical chromatin marks that regulate EE-responsive genes could open new opportunities to 519 modulate their expression, and support mechanistic studies to identify potential therapeutic 520 targets able to mimic or strengthen the cognitive benefits of EE. 521 In conclusion, our meta -analysis extracted core gene expression signatures of EE 522 across dozens of studies. To make these findings accessible, we created an intuitive web 523 app for navigating genes and pathways influenced by EE. Importantly, the mEEtaBrain app 524 also makes it possible for the community to use this interface to ask their own questions by 525 adjusting the subset of studies that can be taken into account and the required significance 526 level. We believe that this resource will advance understanding of the molecular basis of 527 EE-driven behavioural effects, and foster new hypotheses within the neuroscience 528 community. 529 530 AUTHOR CONTRIBUTIONS 531 These authors contributed equally: Marcelina Kurowska , Federico Miozzo, Robert 532 Schroeder. 533 Correspondence to Bartek Wilczyński ([email protected]) or Federico Miozzo 534 ([email protected]). 535 Conceptualization: B.W. Literature screening: M.K., F.M. and R.S. Statistical analysis and 536 app development: M.K. Microarray analysis: R.S. Drafting of the manuscript: F.M. Editing of 537 the manuscript: M.K., F.M., R.S., M.A.M., R.P.G., K.M., A.F., A.B., A-L.B., B.W. 538 539 DATA AVAILABILITY 540 All data analysed in this study were obtained from publicly available databases, with the 541 exception of the dataset from Utsunomiya et al., which is not publicly available and was 542 provided directly by the authors upon request . Detailed information on each dataset is 543 available in Supplementary Tables 1 and 2, and all publicly accessible data can be retrieved 544 through the respective platforms. Scripts used for preprocessing raw FASTQ files, including 545 Fastp trimming and RSEM quantification, are available in Supplementary File 1. All pre -546 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 19 processed datasets used in the mEEtaBrain application 547 (http://regulomics.mimuw.edu.pl/mEEtaBrain/), representing the results of DESeq2 548 differential expression analyses, are publicly available via the Zenodo repository at 549 https://doi.org/10.5281/zenodo.19680157. A video tutorial demonstrating the use of the 550 mEEtaBrain application is available at https://www.youtube.com/watch?v=neqPSVcyRwA. 551 552 FUNDING 553 R.P.G., K.M., A.F., A.B., A-L.B., and B.W research was supported by the grant JPND2022 -554 115 ( EPI-3E: Defining (sex and age) cell -specific epigenetic mechanisms underlying 555 Environmental Enrichment/Exercise as non -pharmacological intervention for Alzheimer’s 556 and Huntington’s disease and related potential noninvasive biomarkers ) from the EU Joint 557 Programme - Neurodegenerative Disease (JPND) Research. 558 559 COMPETING INTERESTS 560 The authors declare no competing interests. 561 562 DECLARATION OF GENERATIVE AI IN THE WRITING PROCESS 563 During the preparation of this work the authors used ChatGPT to revise English grammar 564 and usage, as well as to assist with software development tasks as a coding copilot. After 565 using this tool, the authors reviewed and edited the content as needed and take full 566 responsibility for the content of the publication. 567 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 20

568 1. Tooley, U. A., Basse8, D. S. & Mackey, A. P . Environmental influences on the pace of brain 569 development. Nat Rev Neurosci 22, 372–384 (2021). 570 2. BeQo, L. E. B., Rajendran, L. & Gil-Mohapel, J. The effects of aging in the hippocampus and 571 cogniZve decline. Neuroscience & Biobehavioral Reviews 79, 66–86 (2017). 572 3. Mandolesi, L. et al. Environmental Factors PromoZng Neural PlasZcity: Insights from Animal 573 and Human Studies. Neural Plas7city 2017, 7219461 (2017). 574 4. van Praag, H., Kempermann, G. & Gage, F. H. Neural consequences of environmental 575 enrichment. Nat Rev Neurosci 1, 191–198 (2000). 576 5. Ohline, S. M. & Abraham, W. C. Environmental enrichment effects on synapZc and cellular 577 physiology of hippocampal neurons. Neuropharmacology 145, 3–12 (2019). 578 6. Nithianantharajah, J. & Hannan, A. J. Enriched environments, experience-dependent 579 plasZcity and disorders of the nervous system. Nat Rev Neurosci 7, 697–709 (2006). 580 7. Singhal, G. & Baune, B. T. A bibliometric analysis of studies on environmental enrichment 581 spanning 1967-2024: pa8erns and trends over the years. Front Behav Neurosci 18, 1501377 582 (2024). 583 8. Spires, T. L. et al. Environmental enrichment rescues protein deficits in a mouse model of 584 HunZngton’s disease, indicaZng a possible disease mechanism. J Neurosci 24, 2270–2276 (2004). 585 9. van Dellen, A., Blakemore, C., Deacon, R., York, D. & Hannan, A. J. Delaying the onset of 586 HunZngton’s in mice. Nature 404, 721–722 (2000). 587 10. Adlard, P . A., Perreau, V. M., Pop, V. & Cotman, C. W. Voluntary exercise decreases amyloid 588 load in a transgenic model of Alzheimer’s disease. J Neurosci 25, 4217–4221 (2005). 589 11. Arendash, G. W. et al. Environmental enrichment improves cogniZon in aged Alzheimer’s 590 transgenic mice despite stable beta-amyloid deposiZon. Neuroreport 15, 1751–1754 (2004). 591 12. Jankowsky, J. L. et al. Environmental Enrichment MiZgates CogniZve Deficits in a Mouse 592 Model of Alzheimer’s Disease. J. Neurosci. 25, 5217–5224 (2005). 593 13. Lazarov, O. et al. Environmental enrichment reduces Abeta levels and amyloid deposiZon in 594 transgenic mice. Cell 120, 701–713 (2005). 595 14. Choi, D.-H., Lee, K.-H. & Lee, J. Effect of exercise-induced neurogenesis on cogniZve funcZon 596 deficit in a rat model of vascular demenZa. Mol Med Rep 13, 2981–2990 (2016). 597 15. Bondi, C. O., Klitsch, K. C., Leary, J. B. & Kline, A. E. Environmental Enrichment as a Viable 598 NeurorehabilitaZon Strategy for Experimental TraumaZc Brain Injury. J Neurotrauma 31, 873–888 599 (2014). 600 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 21 16. Ohlsson, A. L. & Johansson, B. B. Environment influences funcZonal outcome of cerebral 601 infarcZon in rats. Stroke 26, 644–649 (1995). 602 17. Rampon, C. et al. Enrichment induces structural changes and recovery from nonspaZal 603 memory deficits in CA1 NMDAR1-knockout mice. Nat Neurosci 3, 238–244 (2000). 604 18. Lopez-Atalaya, J. P . et al. CBP is required for environmental enrichment-induced 605 neurogenesis and cogniZve enhancement. The EMBO Journal 30, 4287–4298 (2011). 606 19. Nilsson, M., Perfilieva, E., Johansson, U., Orwar, O. & Eriksson, P . S. Enriched environment 607 increases neurogenesis in the adult rat dentate gyrus and improves spaZal memory. J Neurobiol 39, 608 569–578 (1999). 609 20. Bruel-Jungerman, E., Laroche, S. & Rampon, C. New neurons in the dentate gyrus are 610 involved in the expression of enhanced long-term memory following environmental enrichment. 611 Eur J Neurosci 21, 513–521 (2005). 612 21. Kempermann, G. Environmental enrichment, new neurons and the neurobiology of 613 individuality. Nat Rev Neurosci 20, 235–245 (2019). 614 22. Pons-Espinal, M., MarZnez de Lagran, M. & Dierssen, M. Environmental enrichment rescues 615 DYRK1A acZvity and hippocampal adult neurogenesis in TgDyrk1A. Neurobiology of Disease 60, 18–616 31 (2013). 617 23. Leggio, M. G. et al. Environmental enrichment promotes improved spaZal abiliZes and 618 enhanced dendriZc growth in the rat. Behav Brain Res 163, 78–90 (2005). 619 24. Rojas, J. J. et al. Effects of daily environmental enrichment on behavior and dendriZc spine 620 density in hippocampus following neonatal hypoxia-ischemia in the rat. Exp Neurol 241, 25–33 621 (2013). 622 25. Darna, M., Beckmann, J. S., Gipson, C. D., Bardo, M. T. & Dwoskin, L. P . Effect of 623 environmental enrichment on dopamine and serotonin transporters and glutamate 624 neurotransmission in medial prefrontal and orbitofrontal cortex. Brain Res 1599, 115–125 (2015). 625 26. Koh, S., Magid, R., Chung, H., SZne, C. D. & Wilson, D. N. Depressive behavior and selecZve 626 down-regulaZon of serotonin receptor expression aler early-life seizures: reversal by 627 environmental enrichment. Epilepsy Behav 10, 26–31 (2007). 628 27. Sun, H. et al. Environmental enrichment influences BDNF and NR1 levels in the 629 hippocampus and restores cogniZve impairment in chronic cerebral hypoperfused rats. Curr 630 Neurovasc Res 7, 268–280 (2010). 631 28. Birch, A. M., McGarry, N. B. & Kelly, A. M. Short-term environmental enrichment, in the 632 absence of exercise, improves memory, and increases NGF concentraZon, early neuronal survival, 633 and synaptogenesis in the dentate gyrus in a Zme-dependent manner. Hippocampus 23, 437–450 634 (2013). 635 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 22 29. Dandi, E. et al. Beneficial effects of environmental enrichment on behavior, stress reacZvity 636 and synaptophysin/BDNF expression in hippocampus following early life stress. Int J Dev Neurosci 637 67, 19–32 (2018). 638 30. Smallfield, S. & Heckenlaible, C. EffecZveness of OccupaZonal Therapy IntervenZons to 639 Enhance OccupaZonal Performance for Adults With Alzheimer’s Disease and Related Major 640 NeurocogniZve Disorders: A SystemaZc Review. The American Journal of Occupa7onal Therapy 71, 641 7105180010p1-7105180010p9 (2017). 642 31. McDonald, M. W., Hayward, K. S., Rosbergen, I. C. M., Jeffers, M. S. & Corbe8, D. Is 643 Environmental Enrichment Ready for Clinical ApplicaZon in Human Post-stroke RehabilitaZon? 644 Front Behav Neurosci 12, 135 (2018). 645 32. Ball, N. J., Mercado, E. & Orduña, I. Enriched Environments as a PotenZal Treatment for 646 Developmental Disorders: A CriZcal Assessment. Front Psychol 10, 466 (2019). 647 33. Leon, M. & Woo, C. Environmental Enrichment and Successful Aging. Front Behav Neurosci 648 12, 155 (2018). 649 34. Feinberg, A. P . & Fallin, M. D. EpigeneZcs at the Crossroads of Genes and the Environment. 650 JAMA 314, 1129–1130 (2015). 651 35. Cavalli, G. & Heard, E. Advances in epigeneZcs link geneZcs to the environment and 652 disease. Nature 571, 489–499 (2019). 653 36. Day, J. J. et al. DNA methylaZon regulates associaZve reward learning. Nat Neurosci 16, 654 1445–1452 (2013). 655 37. Fernandez-Albert, J. et al. Immediate and deferred epigenomic signatures of in vivo 656 neuronal acZvaZon in mouse hippocampus. Nat Neurosci 22, 1718–1730 (2019). 657 38. Fischer, A., Sananbenesi, F., Wang, X., Dobbin, M. & Tsai, L.-H. Recovery of learning and 658 memory is associated with chromaZn remodelling. Nature 447, 178–182 (2007). 659 39. Fuentes-Ramos, M. & Barco, Á. Unveiling TranscripZonal and EpigeneZc Mechanisms 660 Within Engram Cells: Insights into Memory FormaZon and Stability. Adv Neurobiol 38, 111–129 661 (2024). 662 40. Zovkic, I. B., Guzman-Karlsson, M. C. & Swea8, J. D. EpigeneZc regulaZon of memory 663 formaZon and maintenance. Learn Mem 20, 61–74 (2013). 664 41. Barker, S. J. et al. MEF2 is a key regulator of cogniZve potenZal and confers resilience to 665 neurodegeneraZon. Sci Transl Med 13, eabd7695 (2021). 666 42. Grégoire, C.-A. et al. RNA-Sequencing Reveals Unique TranscripZonal Signatures of Running 667 and Running-Independent Environmental Enrichment in the Adult Mouse Dentate Gyrus. Front. 668 Mol. Neurosci. 11, (2018). 669 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 23 43. Morè, L. et al. MSK1 is required for the beneficial synapZc and cogniZve effects of enriched 670 experience across the lifespan. Aging 15, 6031–6072 (2023). 671 44. Privitera, L. et al. Experience Recruits MSK1 to Expand the Dynamic Range of Synapses and 672 Enhance CogniZon. J Neurosci 40, 4644–4660 (2020). 673 45. Wassouf, Z. et al. Environmental Enrichment Prevents TranscripZonal Disturbances Induced 674 by Alpha-Synuclein Overexpression. Front Cell Neurosci 12, 112 (2018). 675 46. Zhang, T.-Y. et al. Environmental enrichment increases transcripZonal and epigeneZc 676 differenZaZon between mouse dorsal and ventral dentate gyrus. Nat Commun 9, 298 (2018). 677 47. Feng, Y ., Fan, J., Cheng, Y ., Dai, Q. & Ma, S. Stress regulates Alzheimer’s disease progression 678 via selecZve enrichment of CD8+ T cells. Cell Rep 42, 113313 (2023). 679 48. Schmidt, S. et al. Restoring Age-Related CogniZve Decline through Environmental 680 Enrichment: A Transcriptomic Approach. Cells 11, 3864 (2022). 681 49. Pérez, R. F. et al. A mulZomic atlas of the aging hippocampus reveals molecular changes in 682 response to environmental enrichment. Nat Commun 15, 5829 (2024). 683 50. Herrera-Rivero, M. et al. TranscripZonal profiles in the mouse amygdala aler a cogniZve 684 judgment bias test largely depend on the genotype. Front Mol Neurosci 15, 1025389 (2022). 685 51. Espeso-Gil, S. et al. Environmental Enrichment Induces Epigenomic and Genome 686 OrganizaZon Changes Relevant for CogniZon. Front Mol Neurosci 14, 664912 (2021). 687 52. Methi, A. et al. A Single-Cell Transcriptomic Analysis of the Mouse Hippocampus Aler 688 Voluntary Exercise. Mol Neurobiol 61, 5628–5645 (2024). 689 53. Caradonna, S. G. et al. Genomic modules and intramodular network concordance in 690 suscepZble and resilient male mice across models of stress. Neuropsychopharmacology 47, 987–691 999 (2022). 692 54. Alaiz-Noya, M. et al. Neuronal type-specific modulaZon of cogniZon and AP-1 signaling by 693 early-life rearing condiZons. Nat Commun 16, 9710 (2025). 694 55. NovaZ, A. et al. Environment-dependent striatal gene expression in the BACHD rat model 695 for HunZngton disease. Sci Rep 8, 5803 (2018). 696 56. Zhang, Y . et al. Transcriptomics of Environmental Enrichment Reveals a Role for ReZnoic 697 Acid Signaling in AddicZon. Front Mol Neurosci 9, 119 (2016). 698 57. Powell, G. L. et al. Environmental enrichment during forced absZnence from cocaine self-699 administraZon opposes gene network expression changes associated with the incubaZon effect. 700 Sci Rep 10, 11291 (2020). 701 58. Singh, G. et al. Altered genome-wide hippocampal gene expression profiles following early 702 life lead exposure and their potenZal for reversal by environmental enrichment. Sci Rep 12, 11937 703 (2022). 704 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 24 59. Utsunomiya, R. et al. Rearing in an Enriched Environment Ameliorates the ADHD-like 705 Behaviors of Lister Hooded Rats While Suppressing Neuronal AcZviZes in the Medial Prefrontal 706 Cortex. Cells 11, 3649 (2022). 707 60. Smail, M. A. et al. Molecular neurobiology of loss: a role for basolateral amygdala 708 extracellular matrix. Mol Psychiatry 28, 4729–4741 (2023). 709 61. Rampon, C. et al. Effects of environmental enrichment on gene expression in the brain. Proc 710 Natl Acad Sci U S A 97, 12880–12884 (2000). 711 62. Thiriet, N. et al. Environmental enrichment during adolescence regulates gene expression in 712 the striatum of mice. Brain Res 1222, 31–41 (2008). 713 63. Costa, D. A. et al. Enrichment improves cogniZon in AD mice by amyloid-related and 714 unrelated mechanisms. Neurobiol Aging 28, 831–844 (2007). 715 64. Huang, G.-J. et al. Neurogenomic Evidence for a Shared Mechanism of the AnZdepressant 716 Effects of Exercise and Chronic FluoxeZne in Mice. PLOS ONE 7, e35901 (2012). 717 65. Dong, S., Li, C., Wu, P ., Tsien, J. Z. & Hu, Y . Environment enrichment rescues the 718 neurodegeneraZve phenotypes in presenilins-deficient mice. Eur J Neurosci 26, 101–112 (2007). 719 66. Li, C., Niu, W., Jiang, C. H. & Hu, Y . Effects of enriched environment on gene expression and 720 signal pathways in cortex of hippocampal CA1 specific NMDAR1 knockout mice. Brain Res Bull 71, 721 568–577 (2007). 722 67. Shono, Y . et al. Gene expression associated with an enriched environment aler transient 723 focal ischemia. Brain Res 1376, 60–65 (2011). 724 68. Keyvani, K., Sachser, N., Wi8e, O. W. & Paulus, W. Gene expression profiling in the intact 725 and injured brain following environmental enrichment. J Neuropathol Exp Neurol 63, 598–609 726 (2004). 727 69. Schneider, A. et al. Forced arm use is superior to voluntary training for motor recovery and 728 brain plasZcity aler corZcal ischemia in rats. Exp Transl Stroke Med 6, 3 (2014). 729 70. Zai, L. et al. Inosine augments the effects of a Nogo receptor blocker and of environmental 730 enrichment to restore skilled forelimb use aler stroke. J Neurosci 31, 5977–5988 (2011). 731 71. Toro-Domínguez, D. et al. A survey of gene expression meta-analysis: methods and 732 applicaZons. Brief Bioinform 22, 1694–1705 (2021). 733 72. Loughin, T. M. A systemaZc comparison of methods for combining p-values from 734 independent tests. Computa7onal Sta7s7cs & Data Analysis 47, 467–485 (2004). 735 73. Ehrlich, I. & Malinow, R. PostsynapZc density 95 controls AMPA receptor incorporaZon 736 during long-term potenZaZon and experience -driven synapZc plasZcity. J Neurosci 24, 916–927 737 (2004). 738 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 25 74. El-Husseini, A. E., Schnell, E., Chetkovich, D. M., Nicoll, R. A. & Bredt, D. S. PSD-95 739 involvement in maturaZon of excitatory synapses. Science 290, 1364–1368 (2000). 740 75. Feng, J. et al. Spinophilin regulates the formaZon and funcZon of dendriZc spines. Proc Natl 741 Acad Sci U S A 97, 9287–9292 (2000). 742 76. Foley, K., McKee, C., Nairn, A. C. & Xia, H. RegulaZon of SynapZc Transmission and PlasZcity 743 by Protein Phosphatase 1. J Neurosci 41, 3040–3050 (2021). 744 77. Yan, Z. et al. Protein phosphatase 1 modulaZon of neostriatal AMPA channels: regulaZon by 745 DARPP–32 and spinophilin. Nat Neurosci 2, 13–17 (1999). 746 78. Rizo, J. & Xu, J. The SynapZc Vesicle Release Machinery. Annu Rev Biophys 44, 339–367 747 (2015). 748 79. Washbourne, P . et al. GeneZc ablaZon of the t-SNARE SNAP-25 disZnguishes mechanisms of 749 neuroexocytosis. Nat Neurosci 5, 19–26 (2002). 750 80. Acuna, C. et al. Microsecond DissecZon of Neurotransmi8er Release: SNARE-Complex 751 Assembly Dictates Speed and Ca2+ SensiZvity. Neuron 82, 1088–1100 (2014). 752 81. Jurado, S. et al. LTP requires a unique postsynapZc SNARE fusion machinery. Neuron 77, 753 542–558 (2013). 754 82. Kennedy, M. J. & Ehlers, M. D. Mechanisms and FuncZon of DendriZc Exocytosis. Neuron 755 69, 856–875 (2011). 756 83. Yap, E.-L. & Greenberg, M. E. AcZvity-Regulated TranscripZon: Bridging the Gap between 757 Neural AcZvity and Behavior. Neuron 100, 330–348 (2018). 758 84. Yap, E.-L. et al. BidirecZonal perisomaZc inhibitory plasZcity of a Fos neuronal network. 759 Nature 590, 115–121 (2021). 760 85. Page, M. J. et al. PRISMA 2020 explanaZon and elaboraZon: updated guidance and 761 exemplars for reporZng systemaZc reviews. BMJ 372, n160 (2021). 762 86. Ewels, P ., Magnusson, M., Lundin, S. & Käller, M. MulZQC: summarize analysis results for 763 mulZple tools and samples in a single report. Bioinforma7cs 32, 3047–3048 (2016). 764 87. Chen, S., Zhou, Y ., Chen, Y . & Gu, J. fastp: an ultra-fast all-in-one FASTQ preprocessor. 765 Bioinforma7cs 34, i884–i890 (2018). 766 88. Li, B. & Dewey, C. N. RSEM: accurate transcript quanZficaZon from RNA-Seq data with or 767 without a reference genome. BMC Bioinforma7cs 12, 323 (2011). 768 89. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinforma7cs 29, 15–21 (2013). 769 90. Dyer, S. C. et al. Ensembl 2025. Nucleic Acids Res 53, D948–D957 (2025). 770 91. Wang, L., Wang, S. & Li, W. RSeQC: quality control of RNA-seq experiments. Bioinforma7cs 771 28, 2184–2185 (2012). 772 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 26 92. Danecek, P . et al. Twelve years of SAMtools and BCFtools. Gigascience 10, giab008 (2021). 773 93. Soneson, C., Love, M. I. & Robinson, M. D. DifferenZal analyses for RNA-seq: transcript-level 774 esZmates improve gene-level inferences. Preprint at 775 h8ps://doi.org/10.12688/f1000research.7563.1 (2016). 776 94. Zhang, Y ., Parmigiani, G. & Johnson, W. E. ComBat-seq: batch effect adjustment for RNA-seq 777 count data. NAR Genom Bioinform 2, lqaa078 (2020). 778 95. Love, M. I., Huber, W. & Anders, S. Moderated esZmaZon of fold change and dispersion for 779 RNA-seq data with DESeq2. Genome Biology 15, 550 (2014). 780 96. Huber, W., von Heydebreck, A., Sültmann, H., Poustka, A. & Vingron, M. Variance 781 stabilizaZon applied to microarray data calibraZon and to the quanZficaZon of differenZal 782 expression. Bioinforma7cs 18 Suppl 1, S96-104 (2002). 783 97. InteracZve and Reproducible Workflows for Exploring and Modeling RNA-seq Data with 784 pcaExplorer, Ideal, and GeneTonic - Ludt - 2022 - Current Protocols - Wiley Online Library. 785 h8ps://currentprotocols.onlinelibrary.wiley.com/doi/10.1002/cpz1.411. 786 98. Zhu, A., Ibrahim, J. G. & Love, M. I. Heavy-tailed prior distribuZons for sequence count data: 787 removing the noise and preserving large differences. Bioinforma7cs 35, 2084–2092 (2019). 788 99. False discovery rates: a new deal | BiostaZsZcs | Oxford Academic. 789 h8ps://academic.oup.com/biostaZsZcs/arZcle/18/2/275/2557030. 790 100. Rainer, J., Ga8o, L. & Weichenberger, C. X. ensembldb: an R package to create and use 791 Ensembl-based annotaZon resources. Bioinforma7cs 35, 3151–3153 (2019). 792 101. Leek, J. T., Johnson, W. E., Parker, H. S., Jaffe, A. E. & Storey, J. D. The sva package for 793 removing batch effects and other unwanted variaZon in high-throughput experiments. 794 Bioinforma7cs 28, 882–883 (2012). 795 102. Durinck, S., Spellman, P . T., Birney, E. & Huber, W. Mapping idenZfiers for the integraZon of 796 genomic datasets with the R/Bioconductor package biomaRt. Nat Protoc 4, 1184–1191 (2009). 797 103. Reimand, J. et al. Pathway enrichment analysis and visualizaZon of omics data using 798 g:Profiler, GSEA, Cytoscape and EnrichmentMap. Nat Protoc 14, 482–517 (2019). 799 104. Koopmans, F. et al. SynGO: An Evidence-Based, Expert-Curated Knowledge Base for the 800 Synapse. Neuron 103, 217-234.e4 (2019). 801 105. Szklarczyk, D. et al. STRING v11: protein-protein associaZon networks with increased 802 coverage, supporZng funcZonal discovery in genome-wide experimental datasets. Nucleic Acids 803 Res. 47, D607–D613 (2019). 804 106. Shannon, P . et al. Cytoscape: a solware environment for integrated models of biomolecular 805 interacZon networks. Genome Res 13, 2498–2504 (2003). 806 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 27 107. Körholz, J. C. et al. SelecZve increases in inter-individual variability in response to 807 environmental enrichment in female mice. eLife 7, e35690 (2018). 808 108. Artola, A. et al. Long-lasZng modulaZon of the inducZon of LTD and LTP in rat hippocampal 809 CA1 by behavioural stress and environmental enrichment. Eur J Neurosci 23, 261–272 (2006). 810 109. Duffy, S. N., Craddock, K. J., Abel, T. & Nguyen, P . V. Environmental enrichment modifies the 811 PKA-dependence of hippocampal LTP and improves hippocampus-dependent memory. Learn Mem 812 8, 26–34 (2001). 813 110. Green, E. J. & Greenough, W. T. Altered synapZc transmission in dentate gyrus of rats reared 814 in complex environments: evidence from hippocampal slices maintained in vitro. Journal of 815 Neurophysiology 55, 739–750 (1986). 816 111. Malik, A. N. et al. Genome-wide idenZficaZon and characterizaZon of funcZonal neuronal 817 acZvity-dependent enhancers. Nat Neurosci 17, 1330–1339 (2014). 818 112. Ellenbroek, B. & Youn, J. Rodent models in neuroscience research: is it a rat race? Disease 819 Models & Mechanisms 9, 1079–1087 (2016). 820 113. Stranahan, A. M. SimilariZes and differences in spaZal learning and object recogniZon 821 between young male C57Bl/6J mice and Sprague-Dawley rats. Behav Neurosci 125, 791–795 822 (2011). 823 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 28 824 Figure 1. Overview of dataset collection and meta -analysis workflow. A. Workflow of 825 transcriptomic studies acquisition and filtering. B. Pipeline of RNA-seq dataset processing 826 and meta-analysis. A star indicates the features that can be explored in the online app. 827 Figure 1 A BStudy collection and filtering RNA-seq dataset processing and meta-analysis .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 29 828 Figure 2. Overview of mouse RNA-seq studies and meta-analysis. A. Heatmap showing 829 the number of detected genes shared between pairs of mouse RNA -seq studies. B. 830 Heatmap showing the number of DEGs shared between pairs of mouse RNA -seq studies. 831 C. PCA of mouse RNA -seq datasets. EE, Enriched Environment (red); SE, Standard 832 Environment (blue); VE, Voluntary Exercise (dark blue). N = 36 (EE), 37 (SE), 4 (VE). D. 833 Number of genes for which meta -analysis statistics were computed plotted against the 834 Figure 2 A B C D E F

Mouse Rat Intersection Fisher up 1044 25 5 Fisher down 525 167 11 Stouffer up 319 1 0 Stouffer down 48 7 0 REM up 26 1 0 REM down 7 1 0 Shared genes between mouse studies Shared DEGs between mouse studies n. shared genes n. shared DEGs PCA Number of significant genes Metavolcano (REM) -1 Number of genes Minimum number of datasets in which a gene is detected log2 FC - log10 adj p-value Number of genes detected in at least X datasets Sign consistency 0 1 0 1 2 3 15 10 5 0 -5 -10 -15 Perez, 2024 (2nd cohort) Espeso-Gil, 2021 (RIB) Barker, 2021 Privitera, 2020 More, 2023 Espeso-Gil, 2021 (POL) Schmidt, 2022 (2nd GEO set) Schmidt, 2022 (1st GEO set) Perez, 2024 (1st cohort) Herrera-Rivero, 2022 (C57) Herrera-Rivero, 2022 (b6d) Wassouf, 2018 Gregoire, 2018 Feng, 2023 Zhang, 2018 Caradonna, 2022 Methi, 2024 Alaiz-Noya, 2025 Gregoire, 2018 Barker, 2021 Alaiz-Noya, 2025 Espeso-Gil, 2021 (RIB) Herrera-Rivero, 2022 (b6d) Caradonna, 2022 Methi, 2024 Perez, 2024 (2nd cohort) Schmidt, 2022 (1st GEO set) Espeso-Gil, 2021 (POL) Schmidt, 2022 (2nd GEO set) Wassouf, 2018 Feng, 2023 Privitera, 2020 Zhang, 2018 More, 2023 Herrera-Rivero, 2022 (C57) Perez, 2024 (1st cohort) Perez, 2024 (2nd cohort)Espeso -Gil, 2021 (RIB)Barker, 2021Privitera, 2020 More, 2023 Espeso -Gil, 2021 (POL) Schmidt, 2022 (2 nd GEO set) Schmidt, 2022 (1 st GEO set) Perez, 2024 (1st cohort) Herrera -Rivero, 2022 (C57) Herrera -Rivero, 2022 (b6d) Wassouf, 2018Gregoire, 2018 Feng, 2023Zhang, 2018 Caradonna, 2022 Methi, 2024 Alaiz -Noya, 2025 Gregoire, 2018 Barker, 2021 Alaiz -Noya, 2025 Espeso -Gil, 2021 (RIB) Herrera -Rivero, 2022 (b6d)Caradonna, 2022 Methi, 2024 Perez, 2024 (2nd cohort) Schmidt, 2022 (1 st GEO set) Espeso -Gil, 2021 (PO L) Schmidt, 2022 (2 nd GEO set) Wassouf, 2018 Feng, 2023 Privitera, 2020Zhang, 2018More, 2023 Herrera -Rivero, 2022 (C57) Perez, 2024 (1st cohort) .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 30 minimum number of studies in which each gene was detected. E. Number of significant 835 genes up - or down -regulated in EE vs SE identified by meta -analysis of mouse and rat 836 datasets, and their intersection. A minimum of 6 (mouse) and 2 (rat) datasets with detected 837 gene expression values was required for a gene to be included in the corresponding meta -838 analysis. REM, Fisher and weighted Stouffer, adj p-value < 0.01. F. Volcano plot from REM 839 meta-analysis of mouse RNA -seq studies. A minimum of 6 datasets with detected gene 840 expression values was required for a gene to be included. Error bars represent 95% 841 confidence intervals. Genes significantly upregulated and downregulated under EE are 842 coloured in red and blue, respectively (adj p-value < 0.01). 843 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 31 844 Figure 3. EE modulates key genes and complexes operating at the synapse. A, B. 845 Bubble plots illustrating Gene Ontology (GO) enrichment analysis for upregulated and 846 downregulated genes obtained with Fisher’s method. A minimum of six datasets with 847 detected gene expression values was required for a gene to be included. The 15 most 848 Figure 3 Fisher UP A C Fisher DOWNFisher UP Fisher DOWN D -log10 q-value-log10 q-value nd ns 2 3 4 5 6 7 8 nd ns 2 3 4 5 6 7 8 9 > 10 non-synaptic gene synaptic gene Gabra5 Perp Ywhag Serpinh1 Vwf Slc16a8 Atp6v0c Aqp1 Ap2m1 Cfl1 Abcc4 Rab3a Nebl Kremen1 Hba-a2 Gabbr2 Dsg2 Ddit4 Col8a1 Wnt2 Slc16a4 Atp5d mt-Nd1 Cd63 Smarcc2 Mmp2 Prrt2 Nr4a1 Klk8 Hbb-bt Scara5 Doc2b Hip1r Col8a2 Car14 Col4a2 Slc9a3r1 Mapk8ip1 Plat Arid1b Lrp6 Sf3b3 Spag16 Klhl40 Hbb-bs Ftl1 Dnajc3 Cx3cl1 Col18a1 Car12 Eng Atp2b2 Sort1 Anxa11 Nckap1 Hspa5 Srrm2 Skil Klhl21 Hba-a1 Pdzd2 Dnaja4 Hk1 F5 Drd1 Fst Atp2b3 Cst3 Aox1 Dnm1 Mmp19 Usp25 Prpf6 Thra Wnt3 Rbm45 Frmpd3 Nlgn2 Ctsd Ctsc Caly Col9a3 Snta1 Klc2 Phlpp1 Ezr Lrp5 Uchl1 Prkcg Ncoa2 Tnpo1 H6pd Mitf Kalrn Msx1 Folr1 Erc2 Vegfa Bsg Hsd17b10 Mapk3 Polr2a A2m Ube2g2 Robo2 Mt2 Kifc2 Kmt2d Fkbp5 Nrxn2 Csrnp1 Sec24c Mylk3 Col1a1 Atp2b1 Cryab Calml4 Ss18 Tuba1c Tubb4b Mt1 Kif17 Nfic Kl Dlgap2 Creld2 Ptk2b Cnn2 Pik3r1 Wfs1 Trpc6 Calm3 Ppp1r9b Tubb2b Tubb2a Mpp7 Kcnc3 H4c11 Pik3r2 Iqsec2 Crb3 Lman2 Nrgn Dock3 Atp1a3 Numbl Akt2 Mt3 Tacc1 Tuba1b Timp2 Kcnh3 H2bc4 Fgf1 Sipa1l1 Crb2 Ctsz Itpka Maged1 Fxyd1 Nfasc Slc4a2 Gsn Stra6 Ppp1r9a Mmp15 Lepr H2aj Ptpn11 Kcnab2 Stx3 Cnih2 Iqgap2 Nedd9 Slc9a1 Notch2 Ahcyl2 Actn1 Wnk4 U2af2 Mical2 Ints6 H1f4 Fasn Shank1 Stxbp5l Tjp3 Unc13a Bcar3 Atp1b1 Penk Grin2a Pfn1 Stk39 Rprd2 Pcx Itpk1 H1f2 Ptgfrn Srcin1 Cplx2 F11r Ubr4 Gps1 Atp1a1 Chchd10 Homer1 Wasf3 Steap2 Pml Mccc1 Inpp5j Gulp1 Tubb4a Dlgap3 mt-Nd4 Cldn2 Mbp Wdr6 Xbp1 Apoo Dlg4 Actb Steap1 Syne3 Mast3 Il17re Gstp2 Ppl Tinagl1 mt-Nd5 Efnb1 Rasgrf1 Basp1 Atf6 Cd59a Epb41l1 Cpt1a Sparc Pkm Unc5a Il17rc Gstp1 Ncor2 Kcnj4 mt-Nd2 Cldn1 Kcnq2 Bambi Junb Ttr Agap2 Acsl6 Smpd1 Rab8b Ptprs Igfbp7 Gstm1 Esrra Htr2c mt-Nd4l Ckb Inf2 Hspb8 Fosl2 Bdnf Supt6 Gm13304 Strip2 Pex5l Lrrc4b Igfbp2 Grp Myo7a Map1a Ndufa1 Hsph1 Ccnd3 Hspa1b Fos Gpc1 Aff3 Ackr4 Slmap Pdgfra Sostdc1 Igfbp6 Ncdn Lrg1 Slc17a7 mt-Cytb Cetn2 Trpv4Sdf2l1 Fosb Lcat Rtn4r Nectin1 Trpc4 Pdgfd Lox Igf2 Rnh1 Eme2 Sorcs3 Cox7b Snap25 Rgs4 Bag3 Atf3 Lsr Elmo2 Cd82 Slc38a5 Pcdh7 Syne1 Manf Gm28729 Rab11fip1 Rhob Cox8b Cenpf Lrrc7 Lgals3 H3c7 Apoe Adgrb1 Ackr1 Slc38a3 Pkd1 Ttyh3 Tubb3 Mcm5 Ehd1 Kcnj10 Cox6a1 Nr4a2 Nefl BC067074 Ash1l App Npas2 Cers4 Slc22a5 P4ha2 Mag Loxl2 Gins2 Egr2 Git1 Sst Rhoq Cacna1i Eif4g1 Per1 Aplp1 Adcy1 Acer2 Sipa1l3 P4ha1 Lingo1 Nr2f1 Gfod1 Egr3 Efnb3 Cort Cdc42bpb Cacnb3 Atp7a Arntl Dab2 Inhba Furin Sept5 Pdyn Lgr6 Nr1d1 Tyro3 Egr1 Grasp Slc8a2 Icam1Cacnb2 Rragd Npas4 Hip1 Bmp6 Ace Sept10 Oprd1 Lgals3bp Hey2 Gas6 Plxnb2 Kndc1 Col4a3 Sptbn2 Scn3a Atp6v0e Epas1 Igf2r Tgfbr2 Tnr Scube3 Odc1 Sinhcaf Kcne2 Mdh1 Efemp2 Dgki Col4a5 Cd55 Scn1b Atp6v0a2 Hsp90ab1Reps2 Acvr1c Bgn Rpl7a Nqo1 Slc22a17 Hcn2 Gapdh Gfap Prkca Col4a4 Cdkn1a Scn4b Atp5g2 Arnt2 Ubb Bmp7 Ccn2 Rpl18a Nptxr Lcn2 Rps2 Gadd45g Eef1a2 Ddr1 Thbs1 Cdkn1c Scn3b Kdm6b Arhgap39 Fzd4 Acvr1 Acan Stk10 Nptx2 Sntb2 Uba52 Gadd45b Dusp4 Wwc1 Col6a2 Pacsin2 Cacna1h Atp5g1 Arc Ubc Rhod Aldh1a1 Syp Trim2 Krt8 Rps15 Hap1 Dusp14 Ddn Col6a1 Ccnd1 Gnas Atp6v1g2 Slc2a1 Irs2 Plec Brinp1 Scg2 Rin1 Krt18 Rps3a1 Kcnj6 3 STRING network for Fisher UP (0.05) BGO - Cellular Compartment GO - Biological Process SynGO - Cellular Compartment SynGO - Biological Process Fisher UP (1044 genes) Fisher DOWN (525 genes) Gabra5 Perp Ywhag Serpinh1 Vwf Slc16a8 Atp6v0c Aqp1 Ap2m1 Cfl1 Abcc4 Rab3a Nebl Kremen1 Hba-a2 Gabbr2 Dsg2 Ddit4 Col8a1 Wnt2 Slc16a4 Atp5d mt-Nd1 Cd63 Smarcc2 Mmp2 Prrt2 Nr4a1 Klk8 Hbb-bt Scara5 Doc2b Hip1r Col8a2 Car14 Col4a2 Slc9a3r1 Mapk8ip1 Plat Arid1b Lrp6 Sf3b3 Spag16 Klhl40 Hbb-bs Ftl1 Dnajc3 Cx3cl1 Col18a1 Car12 Eng Atp2b2 Sort1 Anxa11 Nckap1 Hspa5 Srrm2 Skil Klhl21 Hba-a1 Pdzd2 Dnaja4 Hk1 F5 Drd1 Fst Atp2b3 Cst3 Aox1 Dnm1 Mmp19 Usp25 Prpf6 Thra Wnt3 Rbm45 Frmpd3 Nlgn2 Ctsd Ctsc Caly Col9a3 Snta1 Klc2 Phlpp1 Ezr Lrp5 Uchl1 Prkcg Ncoa2 Tnpo1 H6pd Mitf Kalrn Msx1 Folr1 Erc2 Vegfa Bsg Hsd17b10 Mapk3 Polr2a A2m Ube2g2 Robo2 Mt2 Kifc2 Kmt2d Fkbp5 Nrxn2 Csrnp1 Sec24c Mylk3 Col1a1 Atp2b1 Cryab Calml4 Ss18 Tuba1c Tubb4b Mt1 Kif17 Nfic Kl Dlgap2 Creld2 Ptk2b Cnn2 Pik3r1 Wfs1 Trpc6 Calm3 Ppp1r9b Tubb2b Tubb2a Mpp7 Kcnc3 H4c11 Pik3r2 Iqsec2 Crb3 Lman2 Nrgn Dock3 Atp1a3 Numbl Akt2 Mt3 Tacc1 Tuba1b Timp2 Kcnh3 H2bc4 Fgf1 Sipa1l1 Crb2 Ctsz Itpka Maged1 Fxyd1 Nfasc Slc4a2 Gsn Stra6 Ppp1r9a Mmp15 Lepr H2aj Ptpn11 Kcnab2 Stx3 Cnih2 Iqgap2 Nedd9 Slc9a1 Notch2 Ahcyl2 Actn1 Wnk4 U2af2 Mical2 Ints6 H1f4 Fasn Shank1 Stxbp5l Tjp3 Unc13a Bcar3 Atp1b1 Penk Grin2a Pfn1 Stk39 Rprd2 Pcx Itpk1 H1f2 Ptgfrn Srcin1 Cplx2 F11r Ubr4 Gps1 Atp1a1 Chchd10 Homer1 Wasf3 Steap2 Pml Mccc1 Inpp5j Gulp1 Tubb4a Dlgap3 mt-Nd4 Cldn2 Mbp Wdr6 Xbp1 Apoo Dlg4 Actb Steap1 Syne3 Mast3 Il17re Gstp2 Ppl Tinagl1 mt-Nd5 Efnb1 Rasgrf1 Basp1 Atf6 Cd59a Epb41l1 Cpt1a Sparc Pkm Unc5a Il17rc Gstp1 Ncor2 Kcnj4 mt-Nd2 Cldn1 Kcnq2 Bambi Junb Ttr Agap2 Acsl6 Smpd1 Rab8b Ptprs Igfbp7 Gstm1 Esrra Htr2c mt-Nd4l Ckb Inf2 Hspb8 Fosl2 Bdnf Supt6 Gm13304 Strip2 Pex5l Lrrc4b Igfbp2 Grp Myo7a Map1a Ndufa1 Hsph1 Ccnd3 Hspa1b Fos Gpc1 Aff3 Ackr4 Slmap Pdgfra Sostdc1 Igfbp6 Ncdn Lrg1 Slc17a7 mt-Cytb Cetn2 Trpv4Sdf2l1 Fosb Lcat Rtn4r Nectin1 Trpc4 Pdgfd Lox Igf2 Rnh1 Eme2 Sorcs3 Cox7b Snap25 Rgs4 Bag3 Atf3 Lsr Elmo2 Cd82 Slc38a5 Pcdh7 Syne1 Manf Gm28729 Rab11fip1 Rhob Cox8b Cenpf Lrrc7 Lgals3 H3c7 Apoe Adgrb1 Ackr1 Slc38a3 Pkd1 Ttyh3 Tubb3 Mcm5 Ehd1 Kcnj10 Cox6a1 Nr4a2 Nefl BC067074 Ash1l App Npas2 Cers4 Slc22a5 P4ha2 Mag Loxl2 Gins2 Egr2 Git1 Sst Rhoq Cacna1i Eif4g1 Per1 Aplp1 Adcy1 Acer2 Sipa1l3 P4ha1 Lingo1 Nr2f1 Gfod1 Egr3 Efnb3 Cort Cdc42bpb Cacnb3 Atp7a Arntl Dab2 Inhba Furin Sept5 Pdyn Lgr6 Nr1d1 Tyro3 Egr1 Grasp Slc8a2 Icam1Cacnb2 Rragd Npas4 Hip1 Bmp6 Ace Sept10 Oprd1 Lgals3bp Hey2 Gas6 Plxnb2 Kndc1 Col4a3 Sptbn2 Scn3a Atp6v0e Epas1 Igf2r Tgfbr2 Tnr Scube3 Odc1 Sinhcaf Kcne2 Mdh1 Efemp2 Dgki Col4a5 Cd55 Scn1b Atp6v0a2 Hsp90ab1Reps2 Acvr1c Bgn Rpl7a Nqo1 Slc22a17 Hcn2 Gapdh Gfap Prkca Col4a4 Cdkn1a Scn4b Atp5g2 Arnt2 Ubb Bmp7 Ccn2 Rpl18a Nptxr Lcn2 Rps2 Gadd45g Eef1a2 Ddr1 Thbs1 Cdkn1c Scn3b Kdm6b Arhgap39 Fzd4 Acvr1 Acan Stk10 Nptx2 Sntb2 Uba52 Gadd45b Dusp4 Wwc1 Col6a2 Pacsin2 Cacna1h Atp5g1 Arc Ubc Rhod Aldh1a1 Syp Trim2 Krt8 Rps15 Hap1 Dusp14 Ddn Col6a1 Ccnd1 Gnas Atp6v1g2 Slc2a1 Irs2 Plec Brinp1 Scg2 Rin1 Krt18 Rps3a1 Kcnj6 1 Atp1b1 Bcar3 Unc13a Tjp3 Stxbp5l Shank1 Fasn H1f2 Itpk1Pcx Rprd2 Stk39 Pfn1 Grin2a Penk Slc9a1 Nedd9 Iqgap2 Cnih2 Stx3 Kcnab2 Ptpn11 H1f4 Ints6 Mical2 U2af2 Wnk4 Actn1 Ahcyl2 Notch2 Fxyd1 Maged1 Itpka Ctsz Crb2 Sipa1l1 Fgf1 H2aj Lepr Mmp15 Ppp1r9a Stra6 Gsn Slc4a2 Nfasc Atp1a3 Dock3 Nrgn Lman2 Crb3 Iqsec2 Pik3r2 H2bc4 Kcnh3 Timp2 Tuba1b Tacc1 Mt3 Akt2 Numbl Wfs1 Pik3r1 Cnn2 Ptk2b Creld2 Dlgap2 Kl H4c11 Kcnc3 Mpp7 Tubb2a Tubb2b Ppp1r9b Calm3 Trpc6 Atp2b1 Col1a1 Mylk3 Sec24c Csrnp1 Nrxn2 Fkbp5 Nfic Kif17 Mt1 Tubb4b Tuba1c Ss18 Calml4 Cryab Bsg Vegfa Erc2 Folr1 Msx1 Kalrn Mitf Kmt2d Kifc2 Mt2 Robo2 Ube2g2 A2m Polr2a Mapk3 Hsd17b10 Snta1 Col9a3 Caly Ctsc Ctsd Nlgn2 Frmpd3H6pd Tnpo1 Ncoa2 Prkcg Uchl1 Lrp5 Ezr Phlpp1 Klc2 Atp2b3 Fst Drd1 F5 Hk1 Dnaja4 Pdzd2Rbm45 Wnt3 Thra Prpf6 Usp25 Mmp19 Dnm1 Aox1 Cst3 Atp2b2 Eng Car12 Col18a1 Cx3cl1 Dnajc3 Ftl1 Hba-a1 Klhl21 Skil Srrm2 Hspa5 Nckap1 Anxa11 Sort1 Slc9a3r1 Col4a2 Car14 Col8a2 Hip1r Doc2b Scara5 Hbb-bs Klhl40 Spag16 Sf3b3 Lrp6 Arid1b Plat Mapk8ip1 Atp5d Slc16a4 Wnt2 Col8a1 Ddit4 Dsg2 Gabbr2 Hbb-bt Klk8 Nr4a1 Prrt2 Mmp2 Smarcc2 Cd63 mt-Nd1 Atp6v0c Slc16a8 Vwf Serpinh1 Ywhag Perp Gabra5 Hba-a2 Kremen1 Nebl Rab3a Abcc4 Cfl1 Ap2m1 Aqp1 Atp6v1g2 Gnas Ccnd1 Col6a1 Ddn Dusp14 Kcnj6 Rps3a1 Krt18 Rin1 Scg2 Brinp1 Plec Irs2 Slc2a1 Atp5g1 Cacna1h Pacsin2 Col6a2 Wwc1 Dusp4 Hap1 Rps15 Krt8 Trim2 Syp Aldh1a1 Rhod Ubc Arc Kdm6b Scn3b Cdkn1c Thbs1 Ddr1 Eef1a2 Gadd45b Uba52 Sntb2 Nptx2 Stk10 Acan Acvr1 Fzd4 Arhgap39 Atp5g2 Scn4b Cdkn1a Col4a4 Prkca Gfap Gadd45g Rps2 Lcn2 Nptxr Rpl18a Ccn2 Bmp7 Ubb Arnt2 Atp6v0a2 Scn1b Cd55 Col4a5 Dgki Efemp2 Gapdh Hcn2 Slc22a17 Nqo1 Rpl7a Bgn Acvr1c Reps2 Hsp90ab1 Atp6v0e Scn3a Sptbn2 Col4a3 Kndc1 Plxnb2 Mdh1 Kcne2 Sinhcaf Odc1 Scube3 Tnr Tgfbr2 Igf2r Epas1 Rragd Cacnb2 Icam1 Slc8a2 Grasp Egr1 Gas6 Hey2 Lgals3bp Oprd1 Sept10 Ace Bmp6 Hip1Npas4 Atp7a Cacnb3 Cdc42bpb Cort Efnb3 Egr3 Tyro3 Nr1d1 Lgr6 Pdyn Sept5 Furin Inhba Dab2 Arntl Eif4g1 Cacna1i Rhoq Sst Git1 Egr2 Gfod1 Nr2f1 Lingo1 P4ha1 Sipa1l3 Acer2 Adcy1 Aplp1 Per1 BC067074 Nefl Nr4a2 Cox6a1 Kcnj10 Ehd1 Gins2 Loxl2 Mag P4ha2 Slc22a5 Cers4 Npas2 App Ash1l Lgals3 Lrrc7 Cenpf Cox8b Rhob Rab11fip1 Mcm5 Tubb3 Ttyh3 Pkd1 Slc38a3 Ackr1 Adgrb1 Apoe H3c7 Bag3 Rgs4 Snap25 Cox7b Sorcs3 Eme2 Gm28729 Manf Syne1 Pcdh7 Slc38a5 Cd82 Elmo2 Lsr Atf3 Sdf2l1 Trpv4 Cetn2 mt-Cytb Slc17a7 Lrg1 Rnh1 Igf2 Lox Pdgfd Trpc4 Nectin1 Rtn4r Lcat Fosb Hspa1b Ccnd3 Hsph1 Ndufa1 Map1a Myo7a Ncdn Igfbp6 Sostdc1 Pdgfra Slmap Ackr4 Aff3 Gpc1 Fos Hspb8 Inf2 Ckb mt-Nd4l Htr2c Esrra Grp Igfbp2 Lrrc4bPex5l Strip2 Gm13304 Supt6 Bdnf Fosl2 Bambi Kcnq2 Cldn1 mt-Nd2 Kcnj4 Ncor2 Gstm1 Igfbp7 PtprsRab8b Smpd1 Acsl6 Agap2 Ttr Junb Basp1 Rasgrf1 Efnb1 mt-Nd5 Tinagl1 Ppl Gstp1 Il17rc Unc5a Pkm Sparc Cpt1a Epb41l1 Cd59a Atf6 Wdr6 Mbp Cldn2 mt-Nd4 Dlgap3 Tubb4a Gstp2 Il17re Mast3 Syne3 Steap1 Actb Dlg4 Apoo Xbp1 Gps1 Ubr4 F11r Cplx2 Srcin1 Ptgfrn Gulp1 Inpp5jMccc1 Pml Steap2 Wasf3 Homer1 Chchd10 Atp1a1 Atp1b1 Bcar3 Unc13a Tjp3 Stxbp5l Shank1 Fasn H1f2 Itpk1Pcx Rprd2 Stk39 Pfn1 Grin2a Penk Slc9a1 Nedd9 Iqgap2 Cnih2 Stx3 Kcnab2 Ptpn11 H1f4 Ints6 Mical2 U2af2 Wnk4 Actn1 Ahcyl2 Notch2 Fxyd1 Maged1 Itpka Ctsz Crb2 Sipa1l1 Fgf1 H2aj Lepr Mmp15 Ppp1r9a Stra6 Gsn Slc4a2 Nfasc Atp1a3 Dock3 Nrgn Lman2 Crb3 Iqsec2 Pik3r2 H2bc4 Kcnh3 Timp2 Tuba1b Tacc1 Mt3 Akt2 Numbl Wfs1 Pik3r1 Cnn2 Ptk2b Creld2 Dlgap2 Kl H4c11 Kcnc3 Mpp7 Tubb2a Tubb2b Ppp1r9b Calm3 Trpc6 Atp2b1 Col1a1 Mylk3 Sec24c Csrnp1 Nrxn2 Fkbp5 Nfic Kif17 Mt1 Tubb4b Tuba1c Ss18 Calml4 Cryab Bsg Vegfa Erc2 Folr1 Msx1 Kalrn Mitf Kmt2d Kifc2 Mt2 Robo2 Ube2g2 A2m Polr2a Mapk3 Hsd17b10 Snta1 Col9a3 Caly Ctsc Ctsd Nlgn2 Frmpd3H6pd Tnpo1 Ncoa2 Prkcg Uchl1 Lrp5 Ezr Phlpp1 Klc2 Atp2b3 Fst Drd1 F5 Hk1 Dnaja4 Pdzd2Rbm45 Wnt3 Thra Prpf6 Usp25 Mmp19 Dnm1 Aox1 Cst3 Atp2b2 Eng Car12 Col18a1 Cx3cl1 Dnajc3 Ftl1 Hba-a1 Klhl21 Skil Srrm2 Hspa5 Nckap1 Anxa11 Sort1 Slc9a3r1 Col4a2 Car14 Col8a2 Hip1r Doc2b Scara5 Hbb-bs Klhl40 Spag16 Sf3b3 Lrp6 Arid1b Plat Mapk8ip1 Atp5d Slc16a4 Wnt2 Col8a1 Ddit4 Dsg2 Gabbr2 Hbb-bt Klk8 Nr4a1 Prrt2 Mmp2 Smarcc2 Cd63 mt-Nd1 Atp6v0c Slc16a8 Vwf Serpinh1 Ywhag Perp Gabra5 Hba-a2 Kremen1 Nebl Rab3a Abcc4 Cfl1 Ap2m1 Aqp1 Atp6v1g2 Gnas Ccnd1 Col6a1 Ddn Dusp14 Kcnj6 Rps3a1 Krt18 Rin1 Scg2 Brinp1 Plec Irs2 Slc2a1 Atp5g1 Cacna1h Pacsin2 Col6a2 Wwc1 Dusp4 Hap1 Rps15 Krt8 Trim2 Syp Aldh1a1 Rhod Ubc Arc Kdm6b Scn3b Cdkn1c Thbs1 Ddr1 Eef1a2 Gadd45b Uba52 Sntb2 Nptx2 Stk10 Acan Acvr1 Fzd4 Arhgap39 Atp5g2 Scn4b Cdkn1a Col4a4 Prkca Gfap Gadd45g Rps2 Lcn2 Nptxr Rpl18a Ccn2 Bmp7 Ubb Arnt2 Atp6v0a2 Scn1b Cd55 Col4a5 Dgki Efemp2 Gapdh Hcn2 Slc22a17 Nqo1 Rpl7a Bgn Acvr1c Reps2 Hsp90ab1 Atp6v0e Scn3a Sptbn2 Col4a3 Kndc1 Plxnb2 Mdh1 Kcne2 Sinhcaf Odc1 Scube3 Tnr Tgfbr2 Igf2r Epas1 Rragd Cacnb2 Icam1 Slc8a2 Grasp Egr1 Gas6 Hey2 Lgals3bp Oprd1 Sept10 Ace Bmp6 Hip1Npas4 Atp7a Cacnb3 Cdc42bpb Cort Efnb3 Egr3 Tyro3 Nr1d1 Lgr6 Pdyn Sept5 Furin Inhba Dab2 Arntl Eif4g1 Cacna1i Rhoq Sst Git1 Egr2 Gfod1 Nr2f1 Lingo1 P4ha1 Sipa1l3 Acer2 Adcy1 Aplp1 Per1 BC067074 Nefl Nr4a2 Cox6a1 Kcnj10 Ehd1 Gins2 Loxl2 Mag P4ha2 Slc22a5 Cers4 Npas2 App Ash1l Lgals3 Lrrc7 Cenpf Cox8b Rhob Rab11fip1 Mcm5 Tubb3 Ttyh3 Pkd1 Slc38a3 Ackr1 Adgrb1 Apoe H3c7 Bag3 Rgs4 Snap25 Cox7b Sorcs3 Eme2 Gm28729 Manf Syne1 Pcdh7 Slc38a5 Cd82 Elmo2 Lsr Atf3 Sdf2l1 Trpv4 Cetn2 mt-Cytb Slc17a7 Lrg1 Rnh1 Igf2 Lox Pdgfd Trpc4 Nectin1 Rtn4r Lcat Fosb Hspa1b Ccnd3 Hsph1 Ndufa1 Map1a Myo7a Ncdn Igfbp6 Sostdc1 Pdgfra Slmap Ackr4 Aff3 Gpc1 Fos Hspb8 Inf2 Ckb mt-Nd4l Htr2c Esrra Grp Igfbp2 Lrrc4bPex5l Strip2 Gm13304 Supt6 Bdnf Fosl2 Bambi Kcnq2 Cldn1 mt-Nd2 Kcnj4 Ncor2 Gstm1 Igfbp7 PtprsRab8b Smpd1 Acsl6 Agap2 Ttr Junb Basp1 Rasgrf1 Efnb1 mt-Nd5 Tinagl1 Ppl Gstp1 Il17rc Unc5a Pkm Sparc Cpt1a Epb41l1 Cd59a Atf6 Wdr6 Mbp Cldn2 mt-Nd4 Dlgap3 Tubb4a Gstp2 Il17re Mast3 Syne3 Steap1 Actb Dlg4 Apoo Xbp1 Gps1 Ubr4 F11r Cplx2 Srcin1 Ptgfrn Gulp1 Inpp5jMccc1 Pml Steap2 Wasf3 Homer1 Chchd10 Atp1a1 2 1 3 E Atp1b1 Bcar3 Unc13a Tjp3 Stxbp5l Shank1 Fasn H1f2 Itpk1Pcx Rprd2 Stk39 Pfn1 Grin2a Penk Slc9a1 Nedd9 Iqgap2 Cnih2 Stx3 Kcnab2 Ptpn11 H1f4 Ints6 Mical2 U2af2 Wnk4 Actn1 Ahcyl2 Notch2 Fxyd1 Maged1 Itpka Ctsz Crb2 Sipa1l1 Fgf1 H2aj Lepr Mmp15 Ppp1r9a Stra6 Gsn Slc4a2 Nfasc Atp1a3 Dock3 Nrgn Lman2 Crb3 Iqsec2 Pik3r2 H2bc4 Kcnh3 Timp2 Tuba1b Tacc1 Mt3 Akt2 Numbl Wfs1 Pik3r1 Cnn2 Ptk2b Creld2 Dlgap2 Kl H4c11 Kcnc3 Mpp7 Tubb2a Tubb2b Ppp1r9b Calm3 Trpc6 Atp2b1 Col1a1 Mylk3 Sec24c Csrnp1 Nrxn2 Fkbp5 Nfic Kif17 Mt1 Tubb4b Tuba1c Ss18 Calml4 Cryab Bsg Vegfa Erc2 Folr1 Msx1 Kalrn Mitf Kmt2d Kifc2 Mt2 Robo2 Ube2g2 A2m Polr2a Mapk3 Hsd17b10 Snta1 Col9a3 Caly Ctsc Ctsd Nlgn2 Frmpd3H6pd Tnpo1 Ncoa2 Prkcg Uchl1 Lrp5 Ezr Phlpp1 Klc2 Atp2b3 Fst Drd1 F5 Hk1 Dnaja4 Pdzd2Rbm45 Wnt3 Thra Prpf6 Usp25 Mmp19 Dnm1 Aox1 Cst3 Atp2b2 Eng Car12 Col18a1 Cx3cl1 Dnajc3 Ftl1 Hba-a1 Klhl21 Skil Srrm2 Hspa5 Nckap1 Anxa11 Sort1 Slc9a3r1 Col4a2 Car14 Col8a2 Hip1r Doc2b Scara5 Hbb-bs Klhl40 Spag16 Sf3b3 Lrp6 Arid1b Plat Mapk8ip1 Atp5d Slc16a4 Wnt2 Col8a1 Ddit4 Dsg2 Gabbr2 Hbb-bt Klk8 Nr4a1 Prrt2 Mmp2 Smarcc2 Cd63 mt-Nd1 Atp6v0c Slc16a8 Vwf Serpinh1 Ywhag Perp Gabra5 Hba-a2 Kremen1 Nebl Rab3a Abcc4 Cfl1 Ap2m1 Aqp1 Atp6v1g2 Gnas Ccnd1 Col6a1 Ddn Dusp14 Kcnj6 Rps3a1 Krt18 Rin1 Scg2 Brinp1 Plec Irs2 Slc2a1 Atp5g1 Cacna1h Pacsin2 Col6a2 Wwc1 Dusp4 Hap1 Rps15 Krt8 Trim2 Syp Aldh1a1 Rhod Ubc Arc Kdm6b Scn3b Cdkn1c Thbs1 Ddr1 Eef1a2 Gadd45b Uba52 Sntb2 Nptx2 Stk10 Acan Acvr1 Fzd4 Arhgap39 Atp5g2 Scn4b Cdkn1a Col4a4 Prkca Gfap Gadd45g Rps2 Lcn2 Nptxr Rpl18a Ccn2 Bmp7 Ubb Arnt2 Atp6v0a2 Scn1b Cd55 Col4a5 Dgki Efemp2 Gapdh Hcn2 Slc22a17 Nqo1 Rpl7a Bgn Acvr1c Reps2 Hsp90ab1 Atp6v0e Scn3a Sptbn2 Col4a3 Kndc1 Plxnb2 Mdh1 Kcne2 Sinhcaf Odc1 Scube3 Tnr Tgfbr2 Igf2r Epas1 Rragd Cacnb2 Icam1 Slc8a2 Grasp Egr1 Gas6 Hey2 Lgals3bp Oprd1 Sept10 Ace Bmp6 Hip1Npas4 Atp7a Cacnb3 Cdc42bpb Cort Efnb3 Egr3 Tyro3 Nr1d1 Lgr6 Pdyn Sept5 Furin Inhba Dab2 Arntl Eif4g1 Cacna1i Rhoq Sst Git1 Egr2 Gfod1 Nr2f1 Lingo1 P4ha1 Sipa1l3 Acer2 Adcy1 Aplp1 Per1 BC067074 Nefl Nr4a2 Cox6a1 Kcnj10 Ehd1 Gins2 Loxl2 Mag P4ha2 Slc22a5 Cers4 Npas2 App Ash1l Lgals3 Lrrc7 Cenpf Cox8b Rhob Rab11fip1 Mcm5 Tubb3 Ttyh3 Pkd1 Slc38a3 Ackr1 Adgrb1 Apoe H3c7 Bag3 Rgs4 Snap25 Cox7b Sorcs3 Eme2 Gm28729 Manf Syne1 Pcdh7 Slc38a5 Cd82 Elmo2 Lsr Atf3 Sdf2l1 Trpv4 Cetn2 mt-Cytb Slc17a7 Lrg1 Rnh1 Igf2 Lox Pdgfd Trpc4 Nectin1 Rtn4r Lcat Fosb Hspa1b Ccnd3 Hsph1 Ndufa1 Map1a Myo7a Ncdn Igfbp6 Sostdc1 Pdgfra Slmap Ackr4 Aff3 Gpc1 Fos Hspb8 Inf2 Ckb mt-Nd4l Htr2c Esrra Grp Igfbp2 Lrrc4bPex5l Strip2 Gm13304 Supt6 Bdnf Fosl2 Bambi Kcnq2 Cldn1 mt-Nd2 Kcnj4 Ncor2 Gstm1 Igfbp7 PtprsRab8b Smpd1 Acsl6 Agap2 Ttr Junb Basp1 Rasgrf1 Efnb1 mt-Nd5 Tinagl1 Ppl Gstp1 Il17rc Unc5a Pkm Sparc Cpt1a Epb41l1 Cd59a Atf6 Wdr6 Mbp Cldn2 mt-Nd4 Dlgap3 Tubb4a Gstp2 Il17re Mast3 Syne3 Steap1 Actb Dlg4 Apoo Xbp1 Gps1 Ubr4 F11r Cplx2 Srcin1 Ptgfrn Gulp1 Inpp5jMccc1 Pml Steap2 Wasf3 Homer1 Chchd10 Atp1a1 2 Fisher UP (1044 genes) Fisher DOWN (525 genes) extracellular space basolateral plasma memb. postsynapse axon dendritic tree dendrite cell body neuronal cell body glutamatergic synapse secretory vesicle distal axon apical part of cell apical plasma membrane basal plasma membrane myelin sheath postsynapse presynapse glutamatergic synapse cell body dendrite axon neuron to neuron synapse secretory vesicle asymmetric synapse postsyn. specialization postsynaptic density secretory granule transport vesicle memb. synaptic vesicle memb. exocityc vesicle memb. monoatomic ion transport synaptic signaling monoatomic cation transport trans-synaptic signaling chemical synaptic transmission anterograde trans-synaptic signaling cell junction organization metal ion transport monoatomic ion transm. transport behaviour monoatomic cation transm. transp. regulation of trans-synaptic signaling modulation of chemical synaptic transm. regulation of system process regulation of synaptic plasticity synaptic signaling trans-synaptic signaling cellular component morphogenesis chemical synaptic transmission anterograde trans-synaptic signaling cell part morphogenesis cell projection morphogenesis plasma memb. bounded cell proj. morph. neuron projection morphogenesis cell morphogen. involved in neuron diff. regulation of transmembrane transport regulation of trans-synaptic signaling modulation of chemical synaptic transm. axon development amide transport .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 32 significant GO terms for Cellular Compartment (A) and Biological Process (B) are shown. 849 The bubble size and colour indicate the number of genes associated to the term and the 850 adjusted p-value, respectively. GO terms related to synapse are underlined in red. C, D. 851 SynGO analysis of Cellular Compartment (C) and Biological Process (D) for upregulated 852 and downregulated genes obtained with Fisher’s method. E. Protein-protein interaction 853 network of the upregulated genes identified with Fisher’s method ( adj p -value < 0.01). 854 Network analysis was performed using STRING (physical subnetwork; cut-off for combined 855 interaction score: 0.4) and visualized with Cytoscape. Edge thickness indicates the strength 856 of data support. Only proteins displaying at least one interaction are shown. Side Panels (1-857 3) are zoom-in of selected clusters within the network. 858 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint 33 859 Figure 4. EE upregulates the activity -induced transcriptional program. A. Venn 860 diagrams displaying the overlap between IEGs as identified in 37 and significant upregulated 861 and downregulated genes from meta -analysis of mouse RNA -seq datasets. A minimum of 862 six datasets with detected gene expression values was required for a gene to be included . 863 REM, Fisher and weighted Stouffer, adj p -value < 0.01. The hypergeometric probability for 864 the intersections is shown. B. Effect of EE on IEG expression levels compared to all genes 865 and to a size-matched random gene subset. C. GSEA of IEG gene set in REM meta-analysis 866 of mouse datasets. A minimum of six datasets with detected gene expression values was 867 required for a gene to be included. FDR q-value < 0.001. D. Forest plots of Dusp5, Fosb, 868 and Nptx2 from REM meta-analysis of mouse datasets. 869 Figure 4 A B D GSEA for IEGs p = 8.6 x 10-20 Fisher DOWN (525) Fisher UP (1044) IEGs (200) 518 ns p = 6.8 x 10-11 Stouffer DOWN (48) Stouffer UP (319) IEGs (200) 211 ns ns REM DOWN (7) REM UP (26) IEGs (200) 10 ns all genes random genes (144) IEGs (144) -1.0 -0.5 0.0 0.5 1.0 1.5 log2 FC EE/SELog2 FC EE/SE **** ***ns C Random genes (144) All genes IEGs (144) Dusp5 Fosb Nptx2 log2 FC log2 FC log2 FC .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 11, 2026. ; https://doi.org/10.64898/2026.05.10.724097doi: bioRxiv preprint