Translational Psychiatry www.nature.com/tp ARTICLE OPEN Comprehensive analysis of omics data identifies relevant gene networks for Attention-Deficit/Hyperactivity Disorder (ADHD) ✉ Judit Cabana-Domínguez 1,2,3,4 , María Soler Artigas 1,2,3,4, Lorena Arribas1,2, Silvia Alemany1,2, Laura Vilar-Ribó 1,2,3 , Natalia Llonga , Christian Fadeuilhe1,2,3,5, Montse Corrales1,2,3,5, Vanesa Richarte1,2,3,5, Josep Antoni Ramos-Quiroga 1,2 1,2,3,5 and ✉ Marta Ribasés 1,2,3,4 © The Author(s) 2022 Attention-deficit/hyperactivity disorder (ADHD) is a highly prevalent neurodevelopmental disorder that results from the interaction of both genetic and environmental risk factors. Genome-wide association studies have started to identify multiple genetic risk loci associated with ADHD, however, the exact causal genes and biological mechanisms remain largely unknown. We performed a multi-step analysis to identify and characterize modules of co-expressed genes associated with ADHD using data from peripheral blood mononuclear cells of 270 ADHD cases and 279 controls. We identified seven ADHD-associated modules of co-expressed genes, some of them enriched in both genetic and epigenetic signatures for ADHD and in biological pathways relevant for psychiatric disorders, such as the regulation of gene expression, epigenetics and immune system. In addition, for some of the 1234567890();,: modules, we found evidence of potential regulatory mechanisms, including microRNAs and common genetic variants. In conclusion, our results point to promising genes and pathways for ADHD, supporting the use of peripheral blood to assess gene expression signatures in psychiatric disorders. Furthermore, they highlight that the combination of multi-omics signals provides deeper and broader insights into the biological mechanisms underlying ADHD. Translational Psychiatry (2022)12:409 ; https://doi.org/10.1038/s41398-022-02182-8 INTRODUCTION alterations in the neuronal circuits found in psychiatric disorders Attention-deficit/hyperactivity disorder (ADHD) is a highly pre- like ADHD [6]. Recently, the first epigenome-wide association study valent neurodevelopmental disorder that affects around 5–6% of (EWAS) in peripheral blood mononuclear cells (PBMCs) from adults children and adolescents worldwide, and in 40–65% of cases with ADHD was published, identifying four regions differentially persist into adulthood [1]. It is mainly characterized by inattention methylated and located in genes previously related to auto- and/or hyperactivity and high levels of impulsivity. immune disorders, cancer, or neuroticism [7]. Additional EWAS in ADHD is a complex disorder that results from the interaction of saliva and whole blood have been performed both in adults and both genetic and environmental risk factors, with an estimated children with ADHD diagnosis or ADHD symptoms, however, heritability of 70–80% throughout the lifespan [2]. Several studies results among studies are not consistent and further studies with support the role of both common and rare genetic variants in the larger sample sizes are needed [8–12]. development of ADHD, although its etiology and pathogenesis still Although genetic and epigenetic factors that contribute to the remain largely unknown [2]. The first genome-wide association etiology of ADHD have started to be identified through GWAS and study (GWAS) meta-analysis identifying genetic risk variants for EWAS, their biological relevance is difficult to characterize, in part, ADHD (20,183 cases and 35,191 controls) was published in 2019 because genetic risk loci were usually associated with the nearest [3]. They identified 12 independent ADHD risk loci and estimated gene, which may not be necessarily the true causal one. In that common variants account for 22% of the total ADHD contrast, the analysis of gene expression profiles provides a closer heritability. In addition, very recently, a larger GWAS meta- physiological picture of the disorder that is easier to interpret, and analysis on ADHD reported 21 new loci and a reduced estimated reduces the burden of multiple testing. Transcriptome studies in SNP heritability (h2SNP = 14%) [4]. These data highlight that part of ADHD, nevertheless, are limited by the inaccessibility of brain the genetic variance still needs to be explained, which may be samples and have been focused on whole blood or PBMCs. To accounted, in part, for gene by environment interactions [5]. date, eight transcriptomic studies on ADHD have been performed, Epigenetic processes (i.e. histone modifications, DNA methylation highlighting alterations in genes involved in several neuronal and microRNAs) are potential mechanisms by which environmen- functions and the immune system [13–20]. However, the studies tal risk factors lead to changes on gene expression and long-lasting performed so far were based on differential gene expression 1 Psychiatric Genetics Unit, Group of Psychiatry, Mental Health and Addiction, Vall d’Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain. 2 Department of Mental Health, Hospital Universitari Vall d’Hebron, Barcelona, Spain. 3Biomedical Network Research Centre on Mental Health (CIBERSAM), Madrid, Spain. 4 Department of Genetics, Microbiology, and Statistics, Faculty of Biology, Universitat de Barcelona, Barcelona, Spain. 5Department of Psychiatry and Forensic Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain. ✉email: judit.cabana@vhir.org; marta.ribases@vhir.org Received: 5 May 2022 Revised: 8 September 2022 Accepted: 13 September 2022 J. Cabana-Domínguez et al. 2 analyses between ADHD cases and controls, which assume that every gene acts as an independent unit in the expression landscape and select genes based on statistical significance. In contrast, gene co-expression network analyses use an unsuper- vised framework to identify groups of genes with similar expression patterns (co-expressed genes) independently of any phenotype and then correlate these gene modules with a phenotype of interest. This approach has been widely used to characterize patterns of co-expression in normal brain [21, 22] and both in brain and blood samples from several psychiatric disorders [23–28]. In the present study, we aimed to perform a multi-step analysis to identify and characterize modules of co-expressed genes associated with ADHD using expression data from PBMCs of ADHD cases and controls. To further understand the biological relevance and provide a more accurate picture of the regulatory mechan- isms, we performed a comprehensive characterization of genes in each module and combined genomic and transcriptomic data to identify loci that may regulate the ADHD-associated co- expressed genes. MATERIALS AND METHODS Study design Fig. 1 Flowchart of the study. Modules of co-expressed genes were A comprehensive and multi-step approach was applied to identify and identified from peripheral blood mononuclear cells (PBMCs) of characterize modules of co-expressed genes in PBMCs. In the first step we processed transcriptomic data from 270 ADHD cases and 279 ran a Weighted Gene Correlation Network Analysis (WGCNA) on the controls by using Weighted Gene Correlation Network Analysis processed transcriptomic data from 270 ADHD cases and 279 controls and (WGCNA). Then, we assessed the association of the resulting assessed the association of the resulting co-expression modules with modules with the ADHD status and investigated their biologically ADHD status. Subsequently, we disentangled the biological relevance of relevance by (i) performing enrichment analyses in brain expression the ADHD-associated co-expression modules by (i) performing enrichment (ABAenrichment R package), functional pathways, druggable genes analyses in brain expression, functional pathways, druggable genes and and miRNA target genes using WebGestAlt webtool; (ii) integrating miRNA target genes, (ii) combining results with ADHD genetic, transcrip- ADHD transcriptomic, genetic and epigenetic data from GWAS tomic, and epigenetic signatures, and (iii) running a co-expression module meta-analysis [3] and EWAS [7] on ADHD; and (iii) running a co- eQTL analysis to identify loci regulating the ADHD-associated modules of expression module eQTL analysis to identify loci regulating the co-expressed genes (Fig. 1). ADHD-associated modules of co-expressed genes. Enrichment analyses in the ADHD-associated co-expression Participants Analysis of co-expression modules was performed in an in-house sample of modules We assessed whether genes in each ADHD-associated co-expression 270 ADHD cases (59.3% male, mean age = 34.2 years, s.d = 11.7) and 279 module were expressed in specific brain regions at different develop- controls (56.9% male, mean age = 36.6 years, s.d = 9.9). All subjects were mental stages using data from the Allen Human Brain Atlas with of European ancestry. Clinical assessment was conducted by structured ABAEnrichment R-package [30] (additional details in Supplementary interviews and self-reported questionnaires as previously described [19]. Information). Then, enrichment analyses with the webtool WebGestAlt Detailed information is available in Supplementary Information. The study (WEB-based GEne SeT AnaLysis Toolkit, http://www.webgestalt.org/) [31] was approved by the Clinical Research Ethics Committee (CREC) of Hospital were performed on: (i) Gene Ontology non-redundant Biological Process Universitari Vall d’Hebron, methods were performed in accordance with (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome, (ii) the relevant guidelines and regulations and written informed consent was target genes of pharmacological drugs based on the information from obtained from all subjects before inclusion in the study. PharmGKB and (iii) miRNA target genes. False discovery rate P-value (PFDR) < 0.05 was set as the significance threshold. Transcriptome profiling and weighted gene correlation In addition, the correlation between the identified miRNA and its network analysis (WGCNA) corresponding module eigengene was tested using the non-parametric RNA from PBMCs was isolated, hybridized to GeneChip Human Gene 1.1 ST Spearman rank correlation test in a subset of 310 individuals included in 96-Array plate (Affymetrix) and data were analyzed as previously described the WGCNA (60% overlap; 150 ADHD cases and 160 controls) from whom [19] (Supplementary Information). Modules of co-expressed genes were miRNA expression profile data from PBMC were available as described in identified from processed transcriptomic data by the WGCNA R-package Sanchez-Mora et al. [16]. Expression was available and retrieved from a [29]. A soft-thresholding power of 4 was selected (Fig. S1) and one-step total of 27 mature miRNAs and Bonferroni correction was used to adjust for network construction and module detection was performed considering multiple testing (P < 0.05/27 tests <1.85E−03). an unsigned network type with default values (additional details in Supplementary Information). Gene expression for each module was Integrative analysis of ADHD-associated co-expression represented by a module eigengene, derived from its first principal component and treated as a quantitative trait in the downstream analyses. modules and ADHD omics data ADHD transcriptomic signatures. After quality control and sample proces- The association between the module eigengenes and ADHD status or sing, differential gene expression profiles between the 270 ADHD cases potential confounding factors (age, sex, RNA integrity number (RIN), and and 279 controls used in the WGCNA analysis were obtained with Limma batch) was tested using regression analyses. Bonferroni correction was R-package [32]. Only genes with PFDR < 0.05 and fold change (FC) > |1.15| applied to correct for multiple testing considering the overall number of were considered differentially expressed and were used to test for co-expression modules constructed (P < 0.05/27 modules <1E-03). enrichment in the ADHD-associated co-expression modules using a Within each module we examined the correlation between module F-Fisher test and Bonferroni correction across all modules (P < 0.05/7 membership (an indicator of the intramodular connectivity of a gene modules <7.1E−03). based on the association between its expression and the module eigengene) and gene significance (effect size of the association between each gene and ADHD) using Pearson correlation (Fig. S2). ADHD genetic signatures. The identified ADHD-associated co-expression modules were used as gene sets to test for enrichment in ADHD genetic Translational Psychiatry (2022)12:409 J. Cabana-Domínguez et al. 3 signatures, considering the European ancestry GWAS summary statistics on the ADHD-associated co-expression modules. For six out of the ADHD described by Demontis et al. [3]. Gene-based analyses were run in seven modules we identified enrichment in target genes of at MAGMA_v1.08 [33] using the SNP-wise mean model, and SNPs were least one drug, being Antiinfective for systemic use and Anti- assigned to genes based on a positional-approach and eMAGMA [34, 35]. neoplastic and immunomodulating agents the most common Competitive gene-set analysis was performed using P-values obtained Anatomical Therapeutic Chemical classification categories across from each gene-based analysis. Bonferroni correction was applied to correct for multiple testing (P < 7.1E-03; Supplementary Information) all modules (Table 1 and S7-8). Interestingly, module M5, enriched in genes involved in the immune system and inflammatory ADHD methylation signatures. To test for enrichment in ADHD epigenetic response, showed enrichment in drugs from all Anatomical marks in modules of co-expressed genes we used the summary statistics of Therapeutic Chemical categories, especially those related to the an EWAS on PBMCs from 103 ADHD patients and 100 controls [7] (90% immune response, as expected. sample overlap with the in-house sample used in WGCNA), setting the Enrichment in miRNA target genes was identified in modules unadjusted P-value < 0.01 to select differentially methylated proves M1 and M7. Genes in M1 were targeted by 24 families of miRNAs, (n = 3967 CpG sites). We considered enrichment in epigenetic signatures resulting on 40 mature miRNAs, and genes in M7 were targeted by in a co-expression module when the three approaches used (methylglm, five mature miRNAs (Table 1 and S9). Consistently, a significant methylRRA and gsameth [36]) were significant after applying the Bonferroni correlation between the eigengene profile of module M1 and the correction for multiple comparison corrections (P < 7.1E−03; Supplemen- tary Information). expression of four out of 27 of these miRNAs (hsa-miR-142–5p, hsa-miR-181a-5p, hsa-miR-192–5p, and hsa-miR-215–5p) was found in a subset of 310 individuals (150 ADHD cases and 160 Gene-module eQTL analysis and functional annotation controls) from which miRNA and gene expression from PBMCs Genetic information was available from a subset of 231 ADHD subjects and was available (Fig. S4 and Table S10). 264 controls included in the WGCNA. A GWAS with each module eigengene as the dependent variable were performed to identify genetic Then, we further explored the ADHD-associated co-expression variants associated with each co-expression module. After ascertain modules by integrating transcriptomics with genetic and epige- normality of module eigengene (Table S1), seven gene-module eQTL netic data on ADHD. We explored whether genes differentially analyses were run under an additive linear regression model using expressed between ADHD cases and controls were grouped in any PLINK_1.09, adding as covariates the first 10 principal components, sex, of the identified ADHD-associated modules of co-expressed genes, age, and the genotyping wave. Lead SNPs were identified in each eQTL and found a significant enrichment in module M5 (P < 2.2E−16), analysis considering a P-value < 1E−06 and functionally annotated using which also remained significant when considering only highly the FUMA protocol (Functional Mapping and Annotation of Genome-Wide connected genes (module membership > 0.8; Table 1). In addition, Association Studies, https://fuma.ctglab.nl/) [37] (Supplementary we found module M4 significantly enriched in genetics (PMAGMA = Information). Raw data from this article are not publicly available because of 1.8E−03; PeMAGMA = 4.2E−03) and epigenetics (Pgsameth = 1.6E limitations in ethical approvals and the summary data will be available −04; PmethylRRA = 1.1E−03; Pmethylglm = 8.1E−05) signatures for upon request. ADHD, using data from GWAS meta-analysis [3] and EWAS [7] on ADHD (Table 1 and S11). We performed a co-expression module eQTL analysis to identify RESULTS loci regulating ADHD-associated modules in a subset of 495 The WGCNA identified a total of 27 modules of co-expressed individuals included in the WGCNA (91.3%) from whom genomic genes with size ranging from 33 to 2191 genes (Fig. S1). 42.7% of and gene expression profiles were available. After strict quality genes (N = 8114) were not assigned to any module and remained control criteria, we ran a GWAS on module eigengenes of each of in the module M0. Seven co-expression modules were associated the seven ADHD-associated co-expression modules independently with ADHD after multiple testing correction (modules M1–M7, (M1–M7; Fig. S5). QQ plots indicate minimal effects of genomic Table S2). No association between module eigengenes and inflation, and consequently population substructure, on the potential confounders, including age, sex, RIN, or batch, was analyses (Fig. S6). No SNP overcame the genome-wide significance detected for any module (Table S2). All modules were consistent threshold, but 12 independent genomic loci showed suggestive across samples and have characteristic band structures suggestive evidence of association (P < 1E−06) with different module of well-defined modules (Fig. S2). Interestingly, modules M1, M3, eigengenes (Table 2 and Fig. S7). Functional annotations revealed and M6 showed high module membership—gene significance that these loci lay on regions of open chromatin and that most of correlation (r2 > 0.4), suggesting that the higher the connectivity the signals were intergenic or intronic (Fig. 2). Several SNPs in of a gene within the module, the stronger the association with these genomic risk loci were likely to affect the binding of ADHD (Fig. S3). transcription factors (RBD score = 2b; rs73866266, rs59928606, Different patterns of gene expression in the brain at different rs10830974 and rs36098630), had CADD scores > 12.37, suggest- developmental stages were found across ADHD-associated co- ing high deleteriousness (rs73170573, rs13408514, rs1508617, expression modules. M2 genes are broadly expressed in the whole rs9565360, and rs10830974; Fig. 2, Table 2 and S12) or were brain during the lifespan, while genes in M7 are expressed in a located in regulatory regions of the brain (rs73170578, specific brain area, the M1C_primary motor cortex, only during the rs62096513, and rs12583109), according to the information of prenatal stage. Besides, genes in modules M1 and M4 show enhancer and promotor histone marks from the HaploReg broader expression in different areas from the telencephalon webtool [39] (Table 2). In addition, four SNPs, rs62096513, during the prenatal period and are mainly expressed in the rs6707596, rs66506812 and rs2462337, lay in nearby genes cerebellar cortex after birth (Table 1 and S3). encoding transcription factors (ZSCAN30, SP3, CSRNP3 and CUX1, To explore the biological relevance of ADHD-associated co- respectively; Table 2). Of them, rs6707596 nearby SP3 and expression modules further, we performed a functional enrich- rs62096513 located in intron 1 of ZSCAN30 were cis-eQTL of ment analysis in genes in each module and found that several of these genes in PBMCs in our sample (Fig. 3). Interestingly, the co- them were enriched in genes involved in pathways previously expression module M1, which showed suggestive evidence of related to psychiatric disorders [38], including the posttranscrip- association with rs62096513 that lies in blood and brain regulatory tional regulation of gene expression and epigenetics (M1 and M7), regions of ZSCAN30 and is cis-eQTL in PBMCs, is enriched in target covalent chromatin remodeling (M4) or immune system and genes for this specific transcription factor (P = 1.27E-07), which inflammatory response (M5), among others (Table 1 and S4-6). suggest that ZSCAN30 may be upstream regulator of the M1 We also performed an enrichment analysis in druggable genes in module of co-expressed genes. Translational Psychiatry (2022)12:409 4 Table 1. Summary of the main results from the analyses on the ADHD-associated modules of co-expressed genes. Module No. genes Association Brain Main biological function ATC1 drug miRNA ADHD ADHD genetic ADHD with ADHDa expression categories target genes transcriptomic signatureb epigenetic signature signaturec M1 1546 Effect= 9.53 Prenatal— Posttranscriptional A, B, C, L 40 miRNAs (24 – – – P = 9.3E−06 telencephalon regulation of gene miR families) Postnatal— expression cerebellum Methylation M2 1239 Effect = −6.49 Whole brain mRNA processing A, B, C, D, J, L – – – mglm = 5.2E P = 1.7E−03 Metabolism −03 M3 63 Effect= 7.87 – – – – – – gsam = 0.0269 P = 1.8E−03 mglm = 7.9E−03 M4 885 Effect = −7.82 Prenatal - Regulation of gene B, L - - MA = 1.8E−03 gsam = 1.6E P = 1.8E−03 telencephalon expression eM = 4.2E−03 −04 Postnatal - Covalent chromatin mRRA = 1.1E cerebellum modification −03 mglm = 8.1E −05 M5 133 Effect = −11.17 − Immune response A, B, C, D, H, J, − P = 2.2E−16 − − P = 2.6E-06 L, M, N, R, S M6 258 Effect= 11.58 – mRNA processing J − − − − P = 2.1E−07 M7 185 Effect= 8.66 Prenatal - Posttranscriptional B 5 miRNAs - - gsam = 0.0157 J. Cabana-Domínguez et al. P = 4.4E-05 M1C regulation of gene expression Epigenetics In bold results from gene-set analyses that overcome multiple testing corrections for each method. ATC1 (Anatomical Therapeutic Chemical) categories: A = Alimentary system and metabolism; B = Blood and hematopoietic organs; C = Heart therapy; D = Dermatics; H = Hormone preparations excl.sex; J = Antiinfective for systemic use; L = Antineoplastic and immunomodulating agents; M = musculo-skeletal system; N = Nervous system; R = Respiratory tract; S = Sensory organs. MA = MAGMA; eM = eMAGMA; mglm = methlyglm; gsam = gsameth; mRRA = methylRRA; DE = differentially expressed; M1C = human primary motor cortex. a Logistic regression Effect = log(OR). b Demontis et al. Nat Genet. 2019;51(1):63–75. c Rovira et al. Transl Psychiatry. 2020;10(1):199. Translational Psychiatry (2022)12:409 Table 2. Top hits from GWAS on the module eigengenes from the seven ADHD-associated co-expression modules. Module SNP GWAS MAF intra/ Nearby genes TFa Enrichment on Regulatory region inc eQTL in eQTL whole P-value intergenic target genesb PBMCsd bloode blood brain M1 rs73170578 2.17E−07 0.061 intra CNTNAP2 – – – SNP in LD – – (rs73170573) M1 rs62096513 5.40E−07 0.066 intra ZSCAN30 YES YES YES YES YES – M3 rs75370437 2.56E−07 0.055 intra ACSM1, ACSM3 – – – – – ACSM1 M4 rs6707596 5.53E−07 0.344 intra SP3 YES NO SNP in LD – YES AC106900.6 (rs13010038) Translational Psychiatry (2022)12:409 M4 rs72873859 5.65E−07 0.296 intra HEATR5B – – – – – STRN, PRKD3, RP11–288C18.1 M5 rs66506812 3.35E−07 0.055 intra CSRNP3 YES NA – – – – M5 rs10830974 7.21E−07 0.360 inter SLC36A4, MTNR1B – – – – – – M6 rs73866245 1.35E−07 0.023 inter PCOLCE2, PAQR9, – – – – – – U2SURP M6 rs114943986 1.37E−07 0.037 inter PRR16 – – – – – – M6 rs2462337 3.52E−07 0.271 inter CUX1 YES NO YES – – – M6 rs72806699 6.39E−07 0.010 intra MEAK7 – – – – – – M7 rs12583109 6.00E−07 0.153 intra NBEA, MAB21L1 – – – SNP in LD – – (rs9565360) a TF: the nearby gene is a transcription factor. b Enrichment analysis of transcription factor’s target genes in their corresponding module using F-Fisher test. NA: list of TF target genes not available. c Haploreg v4.1 tool (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php), when one SNP in high LD (r2 > 0.8) is located in a regulatory region is indicated. d Results from the eQTL analysis in our samples. e GTEx v8 (https://www.gtexportal.org/home/). J. Cabana-Domínguez et al. 5 J. Cabana-Domínguez et al. 6 Fig. 2 Functional categories, Regulome DB scores, and minimum chromatin states for independent risk loci associated to any module eigengene. Regulome DB score predicts likelihood of regulatory functionality, lower scores indicate higher likelihood. Further information can be found in Boyle et al. [68]. Minimum Chromatin State across 127 tissue and cell types, lower scores indicate higher accessibility, with states 1–7 referring to open chromatin states. DISCUSSION and synaptic function [48], and genetic variants in genes encoding In the present study, we used a network-based approach to these proteins have been related to rare neurodevelopmental identify novel ADHD-associated modules of co-expressed genes in disorders [49], as well as common psychiatric disorders like PBMCs. To further investigate the biological significance of the schizophrenia [50]. ADHD-associated networks identified, we performed a compre- Module M5 was enriched in genes involved in immune system hensive characterization of each module by performing enrich- and inflammatory response, pathways known to play an important ment analysis in biological pathways and drug or miRNA target role in the development of neuropsychiatric disorders [38, 51, 52], genes. We also performed an integrative analysis by combining particularly in ADHD [53]. Moreover, genes in module M5, and to a transcriptomic, genetic and epigenetic data on ADHD and run an less extent in module M2, are targeted by a great variety of known eQTL analysis to identify genetic variants that could regulate the therapeutic drugs, especially by those that target the immune ADHD-associated modules of co-expressed genes. Our results system (including the Anatomical Therapeutic Chemical categories identified seven ADHD-associated modules of co-expressed genes Antiinfective for systemic use and Antineoplastic and immunomo- and support that the study of gene correlation networks may dulating agents), pointing to genes in these co-expression improve our understanding of the complex molecular systems networks as potential therapeutic targets. Importantly, a recent underling ADHD. study that explored the druggable genome in ADHD also pointed Two of the ADHD-associated co-expression modules identified to drugs to treat autoimmune disorders and malignancies as a (M1 and M7), were enriched in genes involved in posttranscrip- potential novel path for the treatment of ADHD [54]. Besides, in tional regulation of gene expression and epigenetic modifications, module M5 we also found an enrichment in genes differentially two relevant pathways in the pathogenesis of ADHD [6]. In the expressed in ADHD patients compared with controls, suggesting same line, we found enrichment in target genes for several that differentially expressed genes in ADHD cases are co- miRNAs in these two modules. In particular, the expression of four expressed and participate in the same biologic pathways. of them (hsa-miR-142–5p, hsa-miR-181a-5p, hsa-miR-192–5p and Furthermore, this enrichment was also significant when consider- hsa-miR-215–5p) also correlates with the eigengene profile of ing only highly connected genes, highlighting that the genes module M1, pointing them as potential upstream regulatory differentially expressed are central nodes highly connected in this mechanisms underlying the M1 co-expression network. Some of network, reinforcing their relevance in the pathophysiology these miRNAs have been previously related to ADHD, like miR- of ADHD. 192–5p upregulated in PBMCs of ADHD patients [16], and The integrative analysis of transcriptomics, genomics, and comorbid psychiatric disorders, such as miR-192–5p and miR- epigenomics data on ADHD revealed that genes in module M4, 215–5p that were differentially expressed in the dorsolateral also involved in the regulation of gene expression and epigenetic prefrontal cortex of major depression patients [40] or miR-181a-5p mechanisms, were enriched in both genetic and epigenetic extensively related to drug addiction both in mice and human signatures previously described for ADHD [3, 7]. We used two studies [41–45]. Interestingly, these miRNAs share many target complementary approaches to assign ADHD-associated SNPs to genes, suggesting a complex and redundant regulatory system, genes, based on position or eQTL results, and found consistent particularly in the case of miR-291–5p and miR-215–5p which results. PNPLA2 and IQSEC1 were the central genes in the module recognize the same seed sequence. Several of these miRNAs may more significantly associated with ADHD using both methods. regulate a number of central genes (those with high intramodular PNPLA2 encodes an enzyme involved in the hydrolysis of connectivity) from module M1, such as CPSF6 encoding a subunit triglycerides in adipose tissue, and has been related to obesity of a cleavage factor required for the RNA cleavage and [55], a highly comorbid disorder in ADHD [56]. In addition, a recent polyadenylation processing, which was previously related to study pointed PNPLA2 as one of the most high-confidence causal externalizing behaviors including ADHD [46], and RICTOR, which genes for ADHD, after combining GWAS, eQTL and gene plays an essential role during the neurodevelopment and has expression data [57]. IQSEC1 encodes a guanine nucleotide been associated with hyperactivity and reduced anxiety-like exchange factor, essential for the maintenance of glutamatergic behavior in conditional knock-out mice in the dorsal neural synapses [58], one of the key neurotransmitter systems involved in progenitor cells [47]. the pathophysiology of ADHD in combination with dopamine Module M1, as well as M2 and M6, were also enriched in genes [59, 60]. that encode proteins involved in the processing of messenger RNA The eQTL analysis did not reveal any genetic variant that (mRNA), which includes any process related to the conversion of a overcame the genome-wide significance threshold, but we found primary mRNA transcript into one or more mature mRNAs. mRNA 12 independent genomic loci that showed suggestive evidence of processing and alternative splicing are key processes for both the association (P < 1e−06) with the different module eigengenes. We diversification of protein isoforms and the spatio-temporal control identified a genetic variant associated with the co-expression of transcripts, essential for the neuronal development, maturation, module M1, rs62096513, which is located in a blood and brain Translational Psychiatry (2022)12:409 J. Cabana-Domínguez et al. 7 Fig. 3 Regional plots and cis-eQTL analyses in PBMCs in our in-house sample of 270 ADHD cases and 279 controls. In the top figure, regional plots for module eigengenes for the (A) rs62096513 and (B) rs6707596 loci and in the bottom figure, boxplots showing the effect of rs62096513 in ZSCAN30 and rs6707596 in SP3 gene expression. Results from the linear regression p-value are shown: *P < 0.5, **P < 0.01, ***P < 0.001. regulatory region of a transcription factor, ZSCAN30, and regulates cannot discard that these conditions may have influenced the its expression in PBMCs. Interestingly, module M1 was enriched in results of the present study. So, further studies in the same cell target genes for ZSCAN30 that is also included in the same type are required to confirm our results. Additionally, the module, suggesting that this transcription factor is an upstream identified modules were based on expression data from PBMCs, regulator of the co-expressed genes in the module. Besides, we a non-invasive peripheral tissue whose expression profile has identified another genetic variant associated with the M1 module been proposed as a surrogate for expression profiling in the eigengene, rs73170578, located in CNTNAP2, which encodes a central nervous system [67], and further evidence in the brain is neuronal transmembrane protein member of the neurexin super- required to confirm their role in the pathophysiology of the family that function as cell adhesion molecules and receptors. disorder. Both rare and common genetic variants in CNTNAP2 have been In summary, we conducted a multi-step analysis to identify and associated with neurodevelopmental disorders [61, 62], with a characterize modules of co-expressed genes associated with special relevance in ADHD and autism [63, 64]. In addition, module ADHD using expression data from PBMCs in ADHD cases and M4 was associated with rs6707596, that is an eQTL of the SP3 gene controls. We identified seven ADHD-associated modules of co- in PBMCs, a transcription factor involved in synaptic plasticity [65]. expressed genes, some of them being enriched in both genetic Finally, we identified four genetic variants associated with M6 and epigenetic signatures for ADHD and on biological pathways module, among them, rs2462337 is located in a blood regulatory relevant for psychiatric disorders, such as the regulation of gene region upstream the CUX1 gene, a transcription factor involved in expression, epigenetic mechanisms and immune signaling. We the control of neuronal differentiation and the regulation of also found preliminary evidence for some potential regulatory dendritic branching, spine development, and synapse formation mechanisms, including microRNAs and genetic variants, for some in cortical neurons [66]. of the ADHD-associated modules of co-expressed genes identified. Gene networks analyses reduce the dimensionality of genome- These results pinpoint promising genes and pathways for ADHD, wide gene expression data without losing important biological support the use of peripheral blood to assess gene expression information and alleviate the multiple testing burden associated signatures for the disorder and highlight that the combination of with the traditional gene-based methods. Similar network-based multi-omics signals provides deeper and broader insights into the studies have been performed using gene expression data in both biological mechanisms underlying the disorder. brain and blood in several psychiatric disorders like autism, schizophrenia and bipolar disorder [23–28]. These studies were usually performed in small sample sizes (n < 100 individuals), REFERENCES limiting their statistical power. In contrast, we improved the 1. Sayal K, Prasad V, Daley D, Ford T, Coghill D. ADHD in children and young people: resolution and robustness of gene networks by considering more prevalence, care pathways, and service provision. Lancet Psychiatry. than 500 subjects, which allowed the identification of seven 2018;5:175–86. 2. Faraone SV, Larsson H. Genetics of attention deficit hyperactivity disorder. Mol ADHD-associated modules enriched in relevant and highly Psychiatry. 2019;24:562–75. significant biological pathways. However, although our transcrip- 3. Demontis D, Walters RK, Martin J, Mattheisen M, Als TD, Agerbo E, et al. Discovery tomic analyses were performed mainly in medication-naive ADHD of the first genome-wide significant risk loci for attention deficit/hyperactivity patients without comorbid disorders (93.7% of all ADHD cases), we disorder. Nat Genet. 2019;51:63–75. Translational Psychiatry (2022)12:409 J. Cabana-Domínguez et al. 8 4. Demontis D, Walters GB, Athanasiadis G, Walters R, Therrien K, Farajzadeh L, et al. 26. Radulescu E, Jaffe AE, Straub RE, Chen Q, Shin JH, Hyde TM, et al. Identifi- Genome-wide analyses of ADHD identify 27 risk loci, refine the genetic archi- cation and prioritization of gene sets associated with schizophrenia risk by tecture and implicate several cognitive domains. medRxiv. 2022; https://doi.org/ co-expression network analysis in human brain. Mol Psychiatry. 10.1101/2022.02.14.22270780. 2018;25:791–804. 5. Palladino VS, McNeill R, Reif A, Kittel-Schneider S. Genetic risk factors and gene- 27. Jalbrzikowski M, Lazaro MT, Gao F, Huang A, Chow C, Geschwind DH, et al. environment interactions in adult and childhood attention-deficit/hyperactivity Transcriptome profiling of peripheral blood in 22q11.2 deletion syndrome reveals disorder. Psychiatr Genet. 2019;29:63–78. functional pathways related to psychosis and autism spectrum disorder. PLoS 6. Mirkovic B, Chagraoui A, Gerardin P, Cohen D. Epigenetics and attention-deficit/ ONE 2015;10. https://doi.org/10.1371/journal.pone.0132542. hyperactivity disorder: new perspectives? Front Psychiatry. 2020;11. https:// 28. Hess JL, Tylee DS, Barve R, de Jong S, Ophoff RA, Kumarasinghe N, et al. Tran- doi.org/10.3389/fpsyt.2020.00579. scriptomic abnormalities in peripheral blood in bipolar disorder, and dis- 7. Rovira P, Sánchez-Mora C, Pagerols M, Richarte V, Corrales M, Fadeuilhe C, et al. crimination of the major psychoses. Schizophr Res. 2020;217:124. Epigenome-wide association study of attention-deficit/hyperactivity disorder in 29. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network adults. Transl Psychiatry. 2020;10:1–12. analysis. BMC Bioinform. 2008;9:1–13. 8. Wilmot B, Fry R, Smeester L, Musser ED, Mill J, Nigg JT. Methylomic analysis of 30. Grote S, Prüfer K, Kelso J, Dannemann M. ABAEnrichment: an R package to test salivary DNA in childhood ADHD identifies altered DNA methylation in VIPR2. J for gene set expression enrichment in the adult and developing human brain. Child Psychol Psychiatry Allied Discip. 2016;57:152–60. Bioinformatics. 2016;32:3201–3. 9. Walton E, Pingault JB, Cecil CAM, Gaunt TR, Relton CL, Mill J, et al. Epigenetic 31. Liao Y, Wang J, Jaehnig EJ, Shi Z, Zhang B. WebGestalt 2019: gene set analysis profiling of ADHD symptoms trajectories: a prospective, methylome-wide study. toolkit with revamped UIs and APIs. Nucleic Acids Res. 2019;47:W199–W205. Mol Psychiatry. 2017;22:250–6. 32. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differ- 10. Mooney MA, Ryabinin P, Wilmot B, Bhatt P, Mill J, Nigg JT. Large epigenome-wide ential expression analyses for RNA-sequencing and microarray studies. Nucleic association study of childhood ADHD identifies peripheral DNA methylation Acids Res. 2015;43:e47. associated with disease and polygenic risk burden. Transl Psychiatry. 33. de Leeuw CA, Mooij JM, Heskes T, Posthuma D. MAGMA: generalized gene-set 2020;10:1–12. analysis of GWAS data. PLoS Comput Biol. 2015;11:e1004219. 11. Meijer M, Klein M, Hannon E, van der Meer D, Hartman C, Oosterlaan J, et al. 34. Gerring ZF, Mina-Vargas A, Gamazon ER, Derks EM. E-MAGMA: an eQTL-informed Genome-Wide DNA methylation patterns in persistent attention-deficit/hyper- method to identify risk genes using genome-wide association study summary activity disorder and in association with impulsive and callous traits. Front Genet. statistics. Bioinformatics. 2021;37:2245–9. 2020;11:16. 35. Gerring ZF, Gamazon ER, Derks EM. A gene co-expression network-based analysis 12. van Dongen J, Zilhão NR, Sugden K, Heijmans BT, ’t Hoen PAC, van Meurs J, et al. of multiple brain tissues reveals novel genes and molecular pathways underlying Epigenome-wide Association Study of attention-deficit/hyperactivity disorder major depression. PLoS Genet. 2019;15:e1008245. symptoms in adults. Biol Psychiatry. 2019;86:599–607. 36. Maksimovic J, Oshlack A, Phipson B. Gene set enrichment analysis for genome- 13. Marín-Méndez JJ, Patiño-García A, Segura V, Ortuño F, Gálvez MD, Soutullo CA. wide DNA methylation data. Genome Biol. 2021;22:1–26. Differential expression of prostaglandin D2 synthase (PTGDS) in patients with 37. Watanabe K, Taskesen E, Van Bochoven A, Posthuma D. Functional mapping attention deficit-hyperactivity disorder and bipolar disorder. J Affect Disord. and annotation of genetic associations with FUMA. Nat Commun. 2012;138:479–84. 2017;8:1826. 14. De Jong S, Newhouse SJ, Patel H, Lee S, Dempster D, Curtis C, et al. Immune 38. O’dushlaine C, Rossin L, Lee PH, Duncan L, Parikshak NN, Newhouse S, et al. signatures and disorder-specific patterns in a cross-disorder gene expression Psychiatric genome-wide association study analyses implicate neuronal, immune analysis. Br J Psychiatry. 2016;209:202–8. and histone pathways. Nat Neurosci. 2015;18:199. 15. Garcia-Martínez I, Sánchez-Mora C, Soler Artigas M, Rovira P, Pagerols M, Corrales M, 39. Ward LD, Kellis M. HaploReg v4: systematic mining of putative causal variants, cell et al. Gene-wide association study reveals RNF122 ubiquitin ligase as a novel sus- types, regulators and target genes for human complex traits and disease. Nucleic ceptibility gene for attention deficit hyperactivity disorder. Sci Rep. 2017;7:1–13. Acids Res. 2016;44:D877–D881. 16. Sánchez-Mora C, Soler Artigas M, Garcia-Martínez I, Pagerols M, Rovira P, Richarte 40. Yoshino Y, Roy B, Dwivedi Y. Differential and unique patterns of synaptic miRNA V, et al. Epigenetic signature for attention-deficit/hyperactivity disorder: identi- expression in dorsolateral prefrontal cortex of depressed subjects. Neu- fication of miR-26b-5p, miR-185-5p, and miR-191-5p as potential biomarkers in ropsychopharmacology. 2021;46:900–10. peripheral blood mononuclear cells. Neuropsychopharmacology. 2019;44:890–7. 41. Sim MS, Soga T, Pandy V, Wu YS, Parhar IS, Mohamed Z. MicroRNA expression 17. Lorenzo G, Braun J, Muñoz G, Casarejos MJ, Bazán E, Jimenez-Escrig A. RNA-Seq signature of methamphetamine use and addiction in the rat nucleus accumbens. blood transcriptome profiling in familial attention deficit and hyperactivity dis- Metab Brain Dis. 2017;32:1767–83. order (ADHD). Psychiatry Res. 2018;270:544–6. 42. Zhao Y, Zhang K, Jiang H, Du J, Na Z, Hao W, et al. Decreased expression of 18. Nuzziello N, Craig F, Simone M, Consiglio A, Licciulli F, Margari L, et al. Integrated plasma MicroRNA in patients with methamphetamine (MA) use disorder. J analysis of microRNA and mRNA expression profiles: an attempt to disentangle Neuroimmune Pharm. 2016;11:542–8. the complex interaction network in attention deficit hyperactivity disorder. Brain 43. Xu W, Zhao M, Lin Z, Liu H, Ma H, Hong Q, et al. Increased expression of plasma Sci. 2019;9:288. hsa-miR-181a in male patients with heroin addiction use disorder. J Clin Lab Anal. 19. Mortimer N, Sánchez-Mora C, Rovira P, Vilar-Ribó L, Richarte V, Corrales M, et al. 2020;34. https://doi.org/10.1002/jcla.23486. Transcriptome profiling in adult attention-deficit hyperactivity disorder. Eur 44. Chandrasekar V, Dreyer JL. microRNAs miR-124, let-7d and miR-181a regulate Neuropsychopharmacol. 2020;41:160–6. Cocaine-induced Plasticity. Mol Cell Neurosci. 2009;42:350–62. 20. McCaffrey TA, St. Laurent G, Shtokalo D, Antonets D, Vyatkin Y, Jones D, et al. 45. Chandrasekar V, Dreyer JL. Regulation of MiR-124, Let-7d, and MiR-181a in the Biomarker discovery in attention deficit hyperactivity disorder: RNA sequencing accumbens affects the expression, extinction, and reinstatement of cocaine- of whole blood in discordant twin and case-controlled cohorts. BMC Med induced conditioned place preference. Neuropsychopharmacology. Genomics. 2020;13. https://doi.org/10.1186/s12920-020-00808-8. 2011;36:1149–64. 21. Hu Y, Pan J, Xin Y, Mi X, Wang J, Gao Q, et al. Gene expression analysis reveals 46. Karlsson Linnér R, Mallard TT, Barr PB, Sanchez-Roige S, Madole JW, Driver MN, novel gene signatures between young and old adults in human prefrontal cortex. et al. Multivariate analysis of 1.5 million people identifies genetic associations Front Aging Neurosci. 2018;10:259. with traits related to self-regulation and addiction. Nat Neurosci. 22. Oldham MC, Konopka G, Iwamoto K, Langfelder P, Kato T, Horvath S, et al. 2021;24:1367–76. Functional organization of the transcriptome in human brain. Nat Neurosci. 47. Carson RP, Fu C, Winzenburger P, Ess KC. Deletion of rictor in neural progenitor 2008;11:1271–82. cells reveals contributions of mTORC2 signaling to tuberous sclerosis complex. 23. Li X, Zhang Y, Wang L, Lin Y, Gao Z, Zhan X, et al. Integrated analysis of brain Hum Mol Genet. 2013;22:140–52. transcriptome reveals convergent molecular pathways in autism spectrum dis- 48. Furlanis E, Scheiffele P. Regulation of neuronal differentiation, function, and order. Front Psychiatry. 2019;10:706. plasticity by alternative splicing. Annu Rev Cell Dev Biol. 2018;34:451–69. 24. Maschietto M, Tahira AC, Puga R, Lima L, Mariani D, Da Silveira Paulsen B, et al. 49. Fliedner A, Kirchner P, Wiesener A, van de Beek I, Waisfisz Q, van Haelst M, et al. Co-expression network of neural-differentiation genes shows specific pattern in Variants in SCAF4 cause a neurodevelopmental disorder and are associated with schizophrenia Bioinformatic and algorithmical studies. BMC Med Genomics. impaired mRNA processing. Am J Hum Genet. 2020;107:544–54. 2015;8:1–15. 50. Chapman RM, Tinsley CL, Hill MJ, Forrest MP, Tansey KE, Pardiñas AF, et al. 25. Gerring ZF, Gamazon ER, Derks EM. Consortium for the MDDWG of the PG. A Convergent evidence that ZNF804A Is a regulator of pre-messenger RNA pro- gene co-expression network-based analysis of multiple brain tissues reveals cessing and gene expression. Schizophr Bull. 2019;45:1267–78. novel genes and molecular pathways underlying major depression. PLOS Genet. 51. Jiang NM, Cowan M, Moonah SN, Petri WA. The impact of systemic inflammation 2019;15:e1008245. on neurodevelopment. Trends Mol Med. 2018;24:794–804. Translational Psychiatry (2022)12:409 J. Cabana-Domínguez et al. 9 52. Han VX, Patel S, Jones HF, Dale RC. Maternal immune activation and neuroin- RNA isolation and preparation of samples. JCD undertook the statistical analyses. JCD, flammation in human neurodevelopmental disorders. Nat Rev Neurol. LVR, NL, SA, MR, and MSA participated in the study design and the discussion of 2021;17:564–79. results. JCD, MR, and MSA participated in the manuscript preparation. All authors 53. Dunn GA, Nigg JT, Sullivan EL. Neuroinflammation as a risk factor for attention contributed to the interpretation of the findings and revised and approved the final deficit hyperactivity disorder. Pharmacol Biochem Behav. 2019;182:22–34. version of the manuscript. 54. Hegvik TA, Waløen K, Pandey SK, Faraone SV, Haavik J, Zayats T. Druggable genome in attention deficit/hyperactivity disorder and its co-morbid conditions. New avenues for treatment. Mol Psychiatry. 2021;26:4004–15. FUNDING 55. La Cruz-Color LD, Hernández-Nazará ZH, Maldonado-González M, Navarro-Muñíz This work was supported by the Agència de Gestió d’Ajuts Universitaris i de Recerca E, Domínguez-Rosales JA, Torres-Baranda JR, et al. Association of the PNPLA2, (AGAUR, 2017SGR-1461), the Instituto de Salud Carlos III (PI18/01788, P19/01224, SCD1 and leptin expression with fat distribution in liver and adipose tissue from PI20/00041 and FI18/00285 to L.V.R), the Ministry of Science, Innovation and obese subjects. Exp Clin Endocrinol Diabetes. 2020;128:715–22. Universities (IJC2018-035346-I to M.S.A), the Network Center for Biomedical Research 56. Cortese S. The association between ADHD and obesity: Intriguing, progressively (CIBER) to J.C.D.; the European Regional Development Fund (ERDF), the European more investigated, but still puzzling. Brain Sci. 2019;9. https://doi.org/10.3390/ Union H2020 Programme (H2020/2014-2020) under grant agreements no. 728018 brainsci9100256. (Eat2beNICE), no. 848228 (DISCOvERIE) and no. 2020604 (TIMESPAN) and the ECNP 57. Fahira A, Li Z, Liu N, Shi Y. Prediction of causal genes and gene expression Network ‘ADHD across the Lifespan’. analysis of attention-deficit hyperactivity disorder in the different brain region, a comprehensive integrative analysis of ADHD. Behav Brain Res. 2019;364:183–92. 58. Scholz R, Berberich S, Rathgeber L, Kolleker A, Köhr G, Kornau HC. AMPA receptor signaling through BRAG2 and Arf6 critical for long-term synaptic depression. CONFLICT OF INTEREST Neuron. 2010;66:768–80. J.A.R.Q. was on the speakers bureau and/or acted as consultant for Janssen-Cilag, 59. Huang X, Wang M, Zhang Q, Chen X, Wu J. The role of glutamate receptors in Novartis, Shire, Takeda, Bial, Shionogi, Sincrolab, Novartis, BMS, Medice, Technofarma, attention-deficit/hyperactivity disorder: From physiology to disease. Am J Med Rubió and Raffo in the last 3 years. He also received travel awards (air tickets + hotel) Genet Part B Neuropsychiatr Genet. 2019;180:272–86. for taking part in psychiatric meetings from Janssen-Cilag, Rubió, Shire, Takeda, 60. Cai Y, Xing L, Yang T, Chai R, Wang J, Bao J, et al. The neurodevelopmental role of Shionogi, Bial, and Medice. The Department of Mental Health chaired by him received dopaminergic signaling in neurological disorders. Neurosci Lett. 2021;741. unrestricted educational and research support from the following companies in the https://doi.org/10.1016/j.neulet.2020.135540. last 3 years: Janssen-Cilag, Shire, Oryzon, Roche, Psious, and Rubió. C.F. and V.R. have 61. Poot M. Connecting the CNTNAP2 networks with neurodevelopmental disorders. received fees to give talks for Shire/Takeda and Rubió. All other authors declare no Mol Syndromol. 2015;6:7–22. biomedical financial interests or conflicts of interest. 62. Rodenas-Cuadrado P, Ho J, Vernes SC. Shining a light on CNTNAP2: complex functions to complex disorders. Eur J Hum Genet. 2014;22:171. 63. Elia J, Gai X, Xie HM, Perin JC, Geiger E, Glessner JT, et al. Rare structural variants ADDITIONAL INFORMATION found in attention-deficit hyperactivity disorder are preferentially associated with Supplementary information The online version contains supplementary material neurodevelopmental genes. Mol Psychiatry. 2010;15:637. available at https://doi.org/10.1038/s41398-022-02182-8. 64. Vecchia ED, Mortimer N, Palladino VS, Kittel-Schneider S, Lesch KP, Reif A, et al. Cross-species models of attention-deficit/hyperactivity disorder and autism Correspondence and requests for materials should be addressed to Judit Cabana- spectrum disorder: Lessons from CNTNAP2, ADGRL3, and PARK2. Psychiatr Genet. Domínguez or Marta Ribasés. 2019;29:1–17. 65. Yamakawa H, Cheng J, Penney J, Gao F, Rueda R, Wang J, et al. The transcription Reprints and permission information is available at http://www.nature.com/ factor Sp3 cooperates with HDAC2 to regulate synaptic function and plasticity in reprints neurons. Cell Rep. 2017;20:1319–34. 66. Cubelos B, Sebastián-Serrano A, Beccari L, Calcagnotto ME, Cisneros E, Kim S, Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims et al. Cux1 and Cux2 regulate dendritic branching, spine morphology, and in published maps and institutional affiliations. synapses of the upper layer neurons of the cortex. Neuron. 2010;66:523–35. 67. Sullivan PF, Fan C, Perou CM. Evaluating the comparability of gene expression in blood and brain. Am J Med Genet - Neuropsychiatr Genet. 2006;141 B:261–8. 68. Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, Kasowski M, et al. Annotation of functional variation in personal genomes using RegulomeDB. Open Access This article is licensed under a Creative Commons Genome Res. 2012;22:1790–7. Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party ACKNOWLEDGEMENTS material in this article are included in the article’s Creative Commons license, unless The authors are grateful to patients and controls who kindly participated in this indicated otherwise in a credit line to the material. If material is not included in the research. The microarrays service was carried out at High Technology Unit (UAT) at article’s Creative Commons license and your intended use is not permitted by statutory Vall d’Hebron Research Institute (VHIR), Barcelona (Spain). regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http:// creativecommons.org/licenses/by/4.0/. AUTHOR CONTRIBUTIONS JCD, MR, and MSA conceived the project. CF, MC, VR, JARQ participated in the clinical assessment and in the recruitment of patients. JCD, LA, and LVR participated in the © The Author(s) 2022 Translational Psychiatry (2022)12:409