Advertisement

RBFOX1 and working memory: from genome to transcriptome revealed post-transcriptional mechanism separate from ADHD

  • Author Footnotes
    # These authors contribute equally.
    Yuanxin Zhong
    Footnotes
    # These authors contribute equally.
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Author Footnotes
    # These authors contribute equally.
    Na Zhang
    Footnotes
    # These authors contribute equally.
    Affiliations
    Department of Biology, School of Life Science, Southern University of Science and Technology, Shenzhen, 518055, China

    Department of Biological Science, National University of Singapore, Singapore
    Search for articles by this author
  • Author Footnotes
    # These authors contribute equally.
    Feng Zhao
    Footnotes
    # These authors contribute equally.
    Affiliations
    Department of Biology, School of Life Science, Southern University of Science and Technology, Shenzhen, 518055, China
    Search for articles by this author
  • Suhua Chang
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Wei Chen
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Qingjiu Cao
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Li Sun
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Yufeng Wang
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Zhiyuan Gong
    Affiliations
    Department of Biological Science, National University of Singapore, Singapore
    Search for articles by this author
  • Lin Lu
    Correspondence
    Correspondence: Lin Lu, Peking University Sixth Hospital, Huayuan Bei Road 51, Haidian District, Beijing, 100191, China.
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China

    Peking-Tsinghua Center for Life Sciences, Peking University, Beijing 100191, China; International Data Group/Mcgovern Institute for Brain Research at Peking University, Peking University, Beijing 100191, China
    Search for articles by this author
  • Dong Liu
    Correspondence
    Correspondence: Dong Liu, Department of Biology, School of Life Sciences, Southern University of Science and Technology, Shenzhen, 518055, China. Tel: +86-755-88018487.
    Affiliations
    Department of Biology, School of Life Science, Southern University of Science and Technology, Shenzhen, 518055, China
    Search for articles by this author
  • Li Yang
    Correspondence
    Correspondence: Li Yang, Peking University Sixth Hospital, Huayuan Bei Road 51, Haidian District, Beijing, 100191, China. Tel: +86-13651220707.
    Affiliations
    Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China
    Search for articles by this author
  • Author Footnotes
    # These authors contribute equally.
Open AccessPublished:September 01, 2022DOI:https://doi.org/10.1016/j.bpsgos.2022.08.006

      Abstract

      BACKGROUND

      Many psychiatric disorders share working memory (WM) impairment phenotype, yet the genetic causes remain unclear. Here, we generate the genetic profiles of WM deficits using attention-deficit/hyperactivity disorder (ADHD) samples and validate the results in the zebrafish models.

      METHODS

      We have employed two relatively large ADHD cohorts, 799 and 776 samples, respectively. WM impairment was characterized by the Rey Complex Figure Test. Firstly, association analyses were conducted at single variant, gene-based, and gene-set levels. Deeper insights into the biological mechanism were gained from further functional exploration by bioinformatic analyses and zebrafish models.

      RESULTS

      Genomic analyses have identified and replicated a locus with rs75885813 as the index SNP to be significantly associated with WM defects but not with ADHD. Functional feature exploration found these SNPs may regulate the expression level of RBFOX1 through the chromatin interaction. Further pathway enrichment analysis of potential associated SNPs revealed the involvement of post-transcription regulation that affects mRNA stability and/or alternative splicing. Zebrafish with functionally knocked-down or genome-edited rbfox1 exhibited WM impairment but no hyperactivity. Transcriptome profiling of rbfox1 defective zebrafish indicated that alternative exon usages of snap25a and exon-skipping of grin1 genes separately yet partially lead to reduced WM learning of larval zebrafish.

      CONCLUSION

      The locus with rs75885813 of RBFXO1 was identified to be associated with working memory. Rbfox1 regulates synaptic and long-term potentiation related genes to adjust WM at the post-transcriptional level.

      Keywords

      Introduction

      Working memory (WM) is the ability to maintain and manipulate information in the brain/CNS to guide goal-directed behaviors, requiring gene expression control. WM impairments are a common trait of many psychiatric disorders, especially attention deficit/hyperactivity disorder (ADHD) (
      • Ramos A.A.
      • Hamdan A.C.
      • Machado L.
      A meta-analysis on verbal working memory in children and adolescents with ADHD.
      ). With an estimated prevalence of 5.9% in school-age children and core symptoms of inattention, hyperactivity and impulsivity, ADHD children usually exhibit impairments in working memory (
      • Pievsky M.A.
      • McGrath R.E.
      The Neurocognitive Profile of Attention-Deficit/Hyperactivity Disorder: A Review of Meta-Analyses.
      ). WM defects are the leading cause of academic failure in ADHD patients(
      • Fried R.
      • Chan J.
      • Feinberg L.
      • Pope A.
      • Woodworth K.Y.
      • Faraone S.V.
      • et al.
      Clinical correlates of working memory deficits in youth with and without ADHD: A controlled study.
      ), yet the underlying genetic and neurobiological bases are still not fully understood.
      To date, studies of the etiology of ADHD have identified only a few variants of candidate genes, largely due to the heterogeneous nature of clinical phenotypes (
      • Arias-Vásquez A.
      • Altink M.E.
      • Rommelse N.N.
      • Slaats-Willemse D.I.
      • Buschgens C.J.
      • Fliers E.A.
      • et al.
      CDH13 is associated with working memory performance in attention deficit/hyperactivity disorder.
      ). Endophenotypes, on the other hand, may play important roles in unveiling the psychopathological processes of most psychiatric disorders. In this sense, WM defect is a reliable and promising endophenotype that could combine various pathways, including neurotransmission systems, ion channels, transcription regulators, and neurodevelopmental genes (
      • Pfeiffer P.
      • Egorov A.V.
      • Lorenz F.
      • Schleimer J.H.
      • Draguhn A.
      • Schreiber S.
      Clusters of cooperative ion channels enable a membrane-potential-based mechanism for short-term memory.
      ,
      • Shimshek D.R.
      • Bus T.
      • Schupp B.
      • Jensen V.
      • Marx V.
      • Layer L.E.
      • et al.
      Different Forms of AMPA Receptor Mediated LTP and Their Correlation to the Spatial Working Memory Formation.
      ,
      • Rashid A.J.
      • Cole C.J.
      • Josselyn S.A.
      Emerging roles for MEF2 transcription factors in memory.
      ,
      • Zhang C.S.
      • Bertaso F.
      • Eulenburg V.
      • Lerner-Natoli M.
      • Herin G.A.
      • Bauer L.
      • et al.
      Knock-in mice lacking the PDZ-ligand motif of mGluR7a show impaired PKC-dependent autoinhibition of glutamate release, spatial working memory deficits, and increased susceptibility to pentylenetetrazol.
      ,
      • Eagle A.L.
      • Gajewski P.A.
      • Robison A.J.
      Role of hippocampal activity-induced transcription in memory consolidation.
      ). In neurotransmission system, presynaptic components, such as calcineurin, Dystrobrevin Binding Protein 1 (DTNBP1), dysbindin, Glutamate Decarboxylase 1 (GAD1) and Synaptosome Associated Protein 25 (SNAP25) (
      • Straub R.E.
      • Lipska B.K.
      • Egan M.F.
      • Goldberg T.E.
      • Callicott J.H.
      • Mayhew M.B.
      • et al.
      Allelic variation in GAD1 (GAD67) is associated with schizophrenia and influences cortical function and gene expression.
      ,
      • Kircher T.
      • Markov V.
      • Krug A.
      • Eggermann T.
      • Zerres K.
      • Nöthen M.M.
      • et al.
      Association of the DTNBP1 genotype with cognition and personality traits in healthy subjects.
      ,
      • Spangaro M.
      • Bosia M.
      • Zanoletti A.
      • Bechi M.
      • Cocchi F.
      • Pirovano A.
      • et al.
      Cognitive dysfunction and glutamate reuptake: effect of EAAT2 polymorphism in schizophrenia.
      ,
      • Jentsch J.D.
      • Trantham-Davidson H.
      • Jairl C.
      • Tinsley M.
      • Cannon T.D.
      • Lavin A.
      Dysbindin modulates prefrontal cortical glutamatergic circuits and working memory function in mice.
      ,
      • Gao Q.
      • Liu L.
      • Chen Y.
      • Li H.
      • Yang L.
      • Wang Y.
      • et al.
      Synaptosome-related (SNARE) genes and their interactions contribute to the susceptibility and working memory of attention-deficit/hyperactivity disorder in males.
      ,
      • Cottrell J.R.
      • Levenson J.M.
      • Kim S.H.
      • Gibson H.E.
      • Richardson K.A.
      • Sivula M.
      • et al.
      Working memory impairment in calcineurin knock-out mice is associated with alterations in synaptic vesicle cycling and disruption of high-frequency synaptic and network activity in prefrontal cortex.
      ); postsynaptic proteins, such as AKT Serine/Threonine Kinase 1 (AKT1), Glutamate Ionotropic Receptor NMDA Type Subunit 1 (GRIN1), Glutamate Ionotropic Receptor NMDA Type Subunit 2B (GRIN2B), Nitric Oxide Synthase 1 (NOS1) and Neuregulin 1 (NRG1) (
      • Pergola G.
      • Di Carlo P.
      • Andriola I.
      • Gelao B.
      • Torretta S.
      • Attrotto M.T.
      • et al.
      Combined effect of genetic variants in the GluN2B coding gene (GRIN2B) on prefrontal function during working memory performance.
      ,
      • Thomas E.H.X.
      • Bozaoglu K.
      • Rossell S.L.
      • Gurvich C.
      The influence of the glutamatergic system on cognition in schizophrenia: A systematic review.
      ,
      • Reif A.
      • Herterich S.
      • Strobel A.
      • Ehlis A.C.
      • Saur D.
      • Jacob C.P.
      • et al.
      A neuronal nitric oxide synthase (NOS-I) haplotype associated with schizophrenia modifies prefrontal cortex function.
      ,
      • Schmitt W.B.
      • Sprengel R.
      • Mack V.
      • Draft R.W.
      • Seeburg P.H.
      • Deacon R.M.
      • et al.
      Restoration of spatial working memory by genetic rescue of GluR-A-deficient mice.
      ,
      • Reisel D.
      • Bannerman D.M.
      • Schmitt W.B.
      • Deacon R.M.
      • Flint J.
      • Borchardt T.
      • et al.
      Spatial memory dissociations in mice lacking GluR1.
      ), their interacting factors, such as D-Amino Acid Oxidase (DAO), and Olfactomedin 3 (OLFM3) (
      • Roussos P.
      • Giakoumaki S.G.
      • Adamaki E.
      • Georgakopoulos A.
      • Robakis N.K.
      • Bitsios P.
      The association of schizophrenia risk D-amino acid oxidase polymorphisms with sensorimotor gating, working memory and personality in healthy males.
      ,
      • Heck A.
      • Fastenrath M.
      • Ackermann S.
      • Auschra B.
      • Bickel H.
      • Coynel D.
      • et al.
      Converging genetic and functional brain imaging evidence links neuronal excitability to working memory, psychiatric disease, and brain activity.
      ,
      • Goldberg T.E.
      • Straub R.E.
      • Callicott J.H.
      • Hariri A.
      • Mattay V.S.
      • Bigelow L.
      • et al.
      The G72/G30 gene complex and cognitive abnormalities in schizophrenia.
      ), and some transcriptional regulators, such as Pro-Apoptotic WT1 Regulator (PAWR), T-Box Transcription Factor 1 (TBX1), Zinc Finger Protein 804A (ZNF804A), DGCR8 Microprocessor Complex Subunit (DGCR8), and Cytoplasmic FMR1 Interacting Protein 1 (CYFIP1) (
      • Boku S.
      • Izumi T.
      • Abe S.
      • Takahashi T.
      • Nishi A.
      • Nomaru H.
      • et al.
      Copy number elevation of 22q11.2 genes arrests the developmental maturation of working memory capacity and adult hippocampal neurogenesis.
      ,
      • Linden D.E.
      • Lancaster T.M.
      • Wolf C.
      • Baird A.
      • Jackson M.C.
      • Johnston S.J.
      • et al.
      ZNF804A genotype modulates neural activity during working memory for faces.
      ) are all associated with WM. Such a research strategy to pick up candidate genes related to WM defects has shed light on understanding that defects in the neurotransmission system contribute significantly to WM deficiency.
      Visual working memory is crucial in processing visual information. Its deficiencies are linked to general dysfunction in cognition (
      • Fukuda K.
      • Vogel E.
      • Mayr U.
      • Awh E.
      Quantity, not quality: the relationship between fluid intelligence and working memory capacity.
      ), and presented as one of the important symptoms in several psychiatric disorders (
      • Anticevic A.
      • Repovs G.
      • Corlett P.R.
      • Barch D.M.
      Negative and Nonemotional Interference with Visual Working Memory in Schizophrenia.
      ,

      Schecklmann M, Dresler T, Beck S, Jay JT, Febres R, Haeusler J, et al. (2011): Reduced prefrontal oxygenation during object and spatial visual working memory in unpolar and bipolar depression. 194:378-384.

      ). Zebrafish, a kind of vertebrate, has high similarity in physiological structure, growth and development process with human beings, and has shown high conservatism in evolution, making it a hot model organism in the field of biomedical research in recent years (
      • Cassar S.
      • Adatto I.
      • Freeman J.L.
      • Gamse J.T.
      • Iturria I.
      • Lawrence C.
      • et al.
      Use of Zebrafish in Drug Discovery Toxicology.
      ,

      Adamson KI, Sheridan E, Grierson AJ (2018): Use of zebrafish models to investigate rare human disease. 55:641-649.

      ).Although the central nervous system of zebrafish is different from that of mammals, several key brain regions of zebrafish are homologous with that of mammals (
      • Randlett O.
      • Wee C.L.
      • Naumann E.A.
      • Nnaemeka O.
      • Schoppik D.
      • Fitzgerald J.E.
      • et al.
      Whole-brain activity mapping onto a zebrafish brain atlas.
      ). The behavioral patterns of zebrafish and mammals are also quite similar. Zebrafish can show high-level behavior and neural integration, including memory, conditioned reflex and social behavior (
      • Spence R.
      • Gerlach G.
      • Lawrence C.
      • Smith C.
      The behaviour and ecology of the zebrafish, Danio rerio.
      ,
      • Lieschke G.J.
      • Currie P.D.
      Animal models of human disease: zebrafish swim into view.
      ). In this study, we performed genomic analyses of visual spatial WM in a relatively large number of ADHD children and validated the genetic variants significantly associated with WM abnormality. We further tested the affected gene in zebrafish, taking advantage of this highly tractable model system (
      • Cassar S.
      • Adatto I.
      • Freeman J.L.
      • Gamse J.T.
      • Iturria I.
      • Lawrence C.
      • et al.
      Use of Zebrafish in Drug Discovery Toxicology.
      ,

      Adamson KI, Sheridan E, Grierson AJ (2018): Use of zebrafish models to investigate rare human disease. 55:641-649.

      ).

      Methods and Materials

      Subjects: discovery and replication cohorts

      Two consecutive ADHD samples were recruited from our child psychiatric clinics, which included 1040 cases in discovery cohort and 1192 cases in replication cohort. Both cohorts were medication-free and aged between 6 and 16 years old. All cases met DSM-IV ADHD diagnostic criteria based on a semi-structured interview by senior child and adolescent psychiatrists using the Clinical Diagnostic Interview Scale (CDIS). The other inclusion criteria were as follows: full-scale intelligence quotient (FSIQ) ≥ 70, and both biological parents were of Han descent. Those comorbidities with major neurological or psychiatric disorders, such as schizophrenia, bipolar disorder, major depression disorder, pervasive development disorder and epilepsy were excluded. Individuals without WM measures (n = 200 and 342 respectively), age (n = 0 and 2 respectively) or IQ (n = 27 and 37 respectively) were excluded from genetic analyses (see the pipeline in Fig. S11). Majority of the cases were medication-naïve. If the child had been medicated, a wash-out period for at least 1 month was necessary before the recruitment. This work was approved by our Institute Review Board. Written informed consent was obtained from parents.

      Visuospatial working memory task: Rey complex figure test Delayed component (REYD)

      The subject was instructed to observe a complex figure designed by Rey (
      • Shuai L.
      • Chan R.C.
      • Wang Y.
      Executive function profile of Chinese boys with attention-deficit hyperactivity disorder: different subtypes and comorbidity.
      ), then was asked to draw the figure from memory onto a blank sheet of paper after 30s. After a 20-min delay, the subject was asked to recall and draw the figure from memory again. The test was scored according to the structure (0–6) and detail accuracy (0–36). The delayed structure and detail scores were the primary variables of interest in this study for their higher cognitive load. We conducted principal component analyses (PCAs) to extract common features from delayed structure and detail accuracy to produce an index of REY complex figure test Delayed component (REYD), which assess delayed working memory.

      Genomic and bioinformatics analyses of working memory in ADHD patients

      DNA extraction and genotyping

      Genomic DNA was extracted directly from peripheral blood sample of each subject. For the first cohort, all participants were genotyped in our ADHD GWAS project using Affymetrix 6.0 array. For the second one, genotyping was performed by Illumina Infinium PsychArray. Genotypes were called by GENOME STUDIO calling algorithm with the human reference genome (hg19). The same quality control steps were performed. Individuals with per-individual autosomal heterozygosity S.D larger than the mean, gender inconsistent with site reports, a per-individual call rate < 95%, and the lower call rate in a pair of individuals with proportion identity by descent (IBD) PI_HAT >0.185 were excluded (
      • Anderson C.A.
      • Pettersson F.H.
      • Clarke G.M.
      • Cardon L.R.
      • Morris A.P.
      • Zondervan K.T.
      Data quality control in genetic case-control association studies.
      ). Furthermore, the relatedness between two cohorts was checked (PI_HAT < 0.05). Then, the variants were filtered based on per-SNP call rate < 95%, deviation from Hardy-Weinberg equilibrium with P < 0.001 or a minor allele frequency (MAF) < 1% (
      • Anderson C.A.
      • Pettersson F.H.
      • Clarke G.M.
      • Cardon L.R.
      • Morris A.P.
      • Zondervan K.T.
      Data quality control in genetic case-control association studies.
      ). After quality control, 1026 samples with 644166 SNPs and 1147 samples with 284176 SNPs remained for the two cohorts respectively (the pipeline shown in Fig. S12). We imputed non-genotyped SNPs of two cohorts using IMPUTE2, with 2186 phased haplotypes from the full 1000 Genomes Project Integrated Phase 3 Release (
      • Sudmant P.H.
      • Rausch T.
      • Gardner E.J.
      • Handsaker R.E.
      • Abyzov A.
      • Huddleston J.
      • et al.
      An integrated map of structural variation in 2,504 human genomes.
      ) as the reference panel. We removed imputed SNPs with a squared correlation with the true genotypes r2 < 0.9 or with minor allele frequency < 0.01. Finally, 6552994 and 5468003 SNPs were included for two cohorts respectively after imputation.

      Genetic Single-marker analyses

      We performed association analyses on single SNP for REYD using PLINK version 1.9 (
      • Chang C.C.
      • Chow C.C.
      • Tellier L.C.
      • Vattikuti S.
      • Purcell S.M.
      • Lee J.J.
      Second-generation PLINK: rising to the challenge of larger and richer datasets.
      ). By multi-dimensional scaling (MDS), no substantial population stratification was found. As WM is correlated with age, sex, and IQ (

      Eve M Valera, Ph.D. , Ariel Brown, Ph.D. , Joseph Biederman, M.D. , Stephen V. Faraone, Ph.D. , Nikos Makris, M.D., Ph.D. , Michael C. Monuteaux, Sc.D. , et al. (2010): Sex Differences in the Functional Neuroanatomy of Working Memory in Adults With ADHD. 167:86-94.

      ,

      Soares PSM, de Oliveira PD, Wehrmeister FC, Menezes AMB, Rohde LA, Gonçalves H (2022): Does IQ Influence Association Between Working Memory and ADHD Symptoms in Young Adults? 26:1097-1105.

      ), we included them as parts of the covariates. Linear regression models were used with age, sex, IQ and the top 10 eigenvectors from the genetic principal component analysis as covariates for pruned SNPs (r2 = 0.2) by EIGENSOFT 4.2 (
      • Price A.L.
      • Patterson N.J.
      • Plenge R.M.
      • Weinblatt M.E.
      • Shadick N.A.
      • Reich D.
      Principal components analysis corrects for stratification in genome-wide association studies.
      ). When we performed phenotypic association analyses for ADHD (i.e. ADHD diagnosis and inattention, impulsivity, and hyperactivity measures based on CDIS), all covariates mentioned above were included except IQ. Bonferroni correction was used and P < 5e-8 was regarded as whole-genome-wide significance. Significant loci were tested in the replication sample (Pcorrected < 0.05). Meta-analyses were also implemented by PLINK (
      • Chang C.C.
      • Chow C.C.
      • Tellier L.C.
      • Vattikuti S.
      • Purcell S.M.
      • Lee J.J.
      Second-generation PLINK: rising to the challenge of larger and richer datasets.
      ). All reported P-values were two-sided.

      Gene-based and Pathway enrichment analyses

      We further conducted gene-based analyses for both REYD and ADHD by MAGMA (
      • de Leeuw C.A.
      • Mooij J.M.
      • Heskes T.
      • Posthuma D.
      MAGMA: generalized gene-set analysis of GWAS data.
      ). Then we performed pathway enrichment analyses using SNPs with P < e-4 from the meta-analyses as implemented in MAGMA (
      • de Leeuw C.A.
      • Mooij J.M.
      • Heskes T.
      • Posthuma D.
      MAGMA: generalized gene-set analysis of GWAS data.
      ) with a 35 Kb upstream and 10Kb downstream window around genes as the default setting. A total of 10185 Gene Ontology (GO) and 186 KEGG gene sets (obtained from MSigDB(46)) were included. Competitive p values were computed and interpreted. MAGMA’s built-in Bonferroni correction was used for multiple testing corrections.

      Polygenic Risk Score

      Then we performed polygenic analyses to detect shared genetic basis between working memory and ADHD symptoms. Based on the association analysis on the REYD of the discovery sample, we used PRSice (
      • Choi S.W.
      • O'Reilly P.F.
      PRSice-2: Polygenic Risk Score software for biobank-scale data.
      ) to perform the polygenic risk score analyses to select the most precise threshold for the p value that predicted ADHD symptoms in both discovery and validation cohorts respectively, with step increased p-value thresholds (0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.02, 0.03, 0.04, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, and 1). An empirical multiple testing correction implemented in PRSice was applied, which is based on a permutation procedure. The significance of the regression results was corrected by a permutation test with 10,000 replicates and alpha = 0.001 as suggested was used (
      • Choi S.W.
      • O'Reilly P.F.
      PRSice-2: Polygenic Risk Score software for biobank-scale data.
      ). We also explored whether genetic components of glutamate receptor (gluR) or long-term potentiation (LTP)-related pathways shared between working memory and ADHD symptoms. We included 236 and 75 genes that are belonged to gluR and LTP-related pathways according to MSigDB (
      • Liberzon A.
      • Birger C.
      • Thorvaldsdóttir H.
      • Ghandi M.
      • Mesirov J.P.
      • Tamayo P.
      The Molecular Signatures Database (MSigDB) hallmark gene set collection.
      ) and performed the same PRS analyses as mentioned above.

      Regulatory feature analyses and network construction

      We obtained the regulatory features of the significant SNPs from rSNPBase3.0 (
      • Guo L.
      • Wang J.
      rSNPBase 3.0: an updated database of SNP-related regulatory elements, element-gene pairs and SNP-based gene regulatory networks.
      ) and HaploReg v4.1 (
      • Ward L.D.
      • Kellis M.
      HaploReg v4: systematic mining of putative causal variants, cell types, regulators and target genes for human complex traits and disease.
      ), two user-friendly graphic interface web tools which integrate comprehensive information related to genomic regulation. We also searched the expression quantitative loci in the Genotype-Tissue Expression (GTEx) (
      GTEx Consortium
      The Genotype-Tissue Expression (GTEx) project.
      ). The expression plot was generated using THE HUMAN PROTEIN ATLAS. Using STRING (
      • Szklarczyk D.
      • Morris J.H.
      • Cook H.
      • Kuhn M.
      • Wyder S.
      • Simonovic M.
      • et al.
      The STRING database in 2017: quality-controlled protein-protein association networks, made broadly accessible.
      ), the interacting genes network were constructed for RBFOX1 and those enriched genes from pathway analyses.

      Functional analyses of rbfox1 in zebrafish

      Zebrafish husbandry and care

      Zebrafish were raised at 28 °C with a density of 8-10 fish/L and experimented with the established standards (

      Westerfield M (2000): The zebrafish book. A guide for the laboratory use of zebrafish (Danio rerio). 4 ed. Eugene: University of Oregon Press.

      ). The wild-type (WT) zebrafish used in this study was Oregon AB strain/line. All zebrafish experiments were conducted according to the guidelines approved by the Institutional Animal Care and Use Committee of Southern University of Science and Technology (SUSTech).

      Zebrafish loss-of-function studies

      Loss-of-function (knockdown or knockout) experiments conducted in present study were achieved by Morpholino antisense oligo technique (Gene Tools Inc., Corvallis, OR), four guide RNAs/Cas9 technique, or Cas9 ribonucleoproteins (RNPs) technique. To obtain mutants of zebrafish rbfox1 to evaluate if the gene affects VSM, we employed synthetic crRNA:tracrRNA duplex guide RNAs and Cas9 (Hoshijima et al., 2019). Two target sequences in RRM of Rbfox1 were chosen and each created indels (found in each copy of the target locus of rbfox1) in over 80% injected embryos (Supplementary Fig. S4). Since most injected F0 embryos could be treated as true null mutants (Hoshijima et al., 2019), we used 5dpf injected larval zebrafish, and Morpholino anti-sense oligo-injected morphant to conduct behavior tests. In case of snap25a, four gRNAs (Supplementary Table S5, http://crispor.tefor.net/) were co-injected with Cas9 nuclease, and the resulted F0 fish larvae were called knock-out fish or snap25a 4gRNA fish.

      Genotyping

      Genomic DNA was extracted from 24 hpf zebrafish embryo(s) using TIANamp Genomic DNA Kit (TIANGEN). The aimed region(s) of the gene locus was amplified by gene-specific PCR primers, and the expected DNA fragment was purified by QIAquick Gel Extraction Kit (QIAGEN). The purified PCR product(s) was cloned for sequencing purpose (Sangon Biotech). The cloning primers sequence were listed in Supplementary Table S6.

      Quantitative real-time PCR

      Total RNA was extracted from 50 wild-type or mutant larvae at 5 dpf after fixing them in TRIzol reagent (Invitrogen). A total of 1 μg RNA was reverse-transcribed to the first strand of cDNA with the random primer using a cDNA synthesis kit (Promega). qRT-PCR primers for snap25a, grin1a, grin1b and β-actin gene were listed in Supplementary Table S7. qRT-PCR was performed in an ABI 7500 Real-Time PCR instrument with the SYBR green detection system, and results were normalized with β-actin expression using ΔΔCt method.

      Transcriptome sequencing analysis

      Total RNA was separately prepared from 50 larvae (5dpf) of WT or rbfox1 morphant and gone through the DNBSEQ platform, with the NCBI accession number, GCF_000002035.6_GRCz11 as the zebrafish genome reference. The sequencing data was filtered with SOAPnuke (v1.5.2), then the clean reads were mapped to genome using HISAT2 (v2.0.4). After aligning the clean reads to genes via Bowtie2 (v2.2.5), BGI created a database to include all annotated coding transcripts with actual number of reads/counts calculated by RSEM (v1.2.12). Based on Hypergeometric test and corrected by Q value with a rigorous threshold (Q value ≤ 0.05), respectively, the GO and KEGG enrichment analysis of differential gene expressions and significant levels of terms and pathways were obtained.

      Zebrafish behavioral tests: working memory

      Working memory can be represented by habituation in zebrafish. Short-term habituation could be analyzed through repeated acoustic or visual stimulations. Free swimming 5 dpf zebrafish larvae were dispersed in a 96-well plate to sit in the observation chamber (DanioVision, Noldus). Response velocity to acoustic/vibrational or dark flash stimuli was detected and calculated by using EthoVision XT13 video-tracking software (Noldus). For acoustic habituation assay,10 stimuli with the acoustic intensity of 90 dB were delivered with 1 sec inter-stimulus interval (ISI) as baseline, followed by 20 stimuli with 1 sec ISI. Acoustic habituation was indicated as: % acoustic habituation= [1-(velocity of stimuli 21-30)/ (response velocity of baseline)] *100. For visual habituation assay, 4-block training protocol was performed without any break between blocks. Each block consisted 120 dark flash with a 15 sec ISI, and each dark flash lasted for 1 sec. Habituation was indicated as: % Visual Habituation= [1-Block4/Block1] *100 (Supplementary Fig. S7).

      Statistical analyses of zebrafish experiments

      GraphPad Prism version 7.00 was used for statistics analysis of zebrafish experiments. For two-group comparison, one-tailed t test with 95% confidence interval was used. For three or more groups comparison, One-way ANOVA with 95% confidence interval was used.

      Results

      Genome-wide association of working memory and the relationship with ADHD symptoms at the single-variant and polygenic levels
      We first performed a genome-wide analyses of visuospatial WM in ADHD children. A total of 799 and 776 cases, were ultimately qualified for the discovery and replication stages, respectively (MDS plots are available in Supplementary Material). The demographic description, Intelligence Quotients, and cognitive phenotype data for the discovery and replication samples are presented in Supplementary Table S1.
      We used delayed component (REYD) which was the first PC axis score accounting for 94.1% and 91.1% variance in the discovery and validation cohorts respectively, in the following analyses. In the discovery stage we identified five significant SNPs in whole genome-wide association analysis, among which rs75885813 was ranked on top (Pcorrected = 3.83e-9). The other four SNPs were in high linkage disequilibrium with rs75885813 (r2 > 0.9, Table 1). Quantile-quantile (QQ) plot for SNP associations is presented in Supplementary Fig. S2. All five SNPs were located on 16p13.3 (Chr16:7120001-7160000, hg19) within the RBFOX1 gene (Fig. 1). They reached statistical significance in both the replication stage (Pcorrected < 0.05) and meta-analyses (Pcorrected < 5e-9). However, associations between the RBFOX1 SNPs and core symptoms of ADHD (i.e., inattention, hyperactivity, impulsivity, and overall symptoms) were not statistically significant (Pcorrected > 0.05, data available upon request), nor was the association for Polygenic Risk Score (PRSs) weighted by WM for ADHD patients (Pcorrected > 0.05; Supplementary Table S2).
      Table 1Identified significant SNPs from genome-wide analysis of REYD in discovery and replication samples and meta-analyses.


      CHR
      SNPPosA1LD with rs75885813BETASESTATP
      discoveryreplicationdiscoveryreplicationdiscoveryreplicationdiscoveryreplicationmeta-analysis
      16rs758858137141263T--0.669-0.2760.1120.130-5.96-2.123.83e-090.0343.67e-09
      16rs1148916717142203C0.98-0.669-0.3520.1130.135-5.90-2.605.56e-090.00946.04e-10
      16rs1157125987142236T0.98-0.669-0.3520.1130.135-5.90-2.605.56e-090.00946.04e-10
      16rs1161837077142287A0.98-0.669-0.3520.1130.135-5.90-2.605.56e-090.00946.04e-10
      16rs72059957140331G0.98-0.667-0.350.1130.135-5.88-2.596.05e-090.00976.69e-10
      Five whole-genome significant SNPs were identified which are in high linkage disequilibrium (r2 > 0.8). SE, standard error; STAT, coefficient t-statistic.
      Figure thumbnail gr1
      Figure 1Regional plot of significant locus for genome-wide association of REYD. The grey horizontal line represents the threshold for genome-wide significant association (P = 5 × 10-8). rs75885813, Pcorrected = 3.83e-9, was the top-ranked SNP. The other four SNPs significant were in high linkage disequilibrium with rs75885813 (r2 > 0.8). All significant SNPs were located within RBFOX1.

      Associated variants may affect transcriptional regulation

      Given the strong association between rs75885813 (i.e., an intron variant of RBFOX1) and REYD, we further examined the regulatory features of this variant, as well as its interaction networks and co-expression genes. The chromatin state analyses showed that rs75885813 is in functional regulatory regions in various tissues indicated by the enhancer-specific H3K4me1 and promotor-specific H3K4m3 markers. It also alters the regulatory motifs of some transcription factors in the brain (Supplementary Table S3). The Hi-C data suggested that a chromatin loop can form between the genomic region (16:7120001-7160000, hg19) where rs75885813 is located and the promoter of RBFOX1 (16:6040001-6080000, hg19), also suggesting a possible regulatory effect (Supplementary Fig. S3) on RBFOX1 transcriptional regulation. We further constructed a network to include its interacting partners and co-expressing genes. As shown in Supplementary Fig. S4, the functional protein interaction analyses indicated that most of them (8/11) are RNA-binding proteins and may play roles in RNA alternative splicing events.

      Gene-based and pathway enrichment analyses

      No gene achieved significance after multiple corrections for neither REYD nor ADHD. We investigated subthreshold variants from the REYD association analysis using a pathway enrichment test. After removing gene sets that contained less than two genes defined in the GO analyses, 773 gene sets were included. Sixteen GO pathways reached significance (Pcorrected < 0.05, Fig. 2 and Table 2). All of these pathways are involved in post-transcription regulation, among which the mRNA metabolic process was ranked on top. The relationships of these significantly enriched pathways are presented in Supplementary Fig. S5. The extended interaction gene network showed the potentially interacted genes, many of which are overlapped with the risk genes for psychiatric disorders (Supplementary Fig. S6).
      Figure thumbnail gr2
      Figure 2Bubble diagram of the pathway analysis of potential associated variants for REY complex figure test Delayed component (REYD). The p value increased from top to bottom. All the pathways displayed in the figure were significant (Pcorrected < 0.05). The color represents p value, while the size represents the number of associated genes enriched in the particular pathway. REYD: REY complex figure test Delayed component.
      Table 2Enriched biological process by potential associated variants with working memory. The pathway enrichment analyses were conducted with SNPs with P < e-4 from the association analyses as implemented in MAGMA. GOBP, gene ontology biological processes; GOCC, gene ontology cellular components; GOMF, gene ontology molecular functions; NGENES/total, the number of genes enriched in the pathway and the total number of genes of the pathway; GENE list, the genes enriched in the pathway; BETA STD, beta standard; SE, standard error.
      FULL NAMENGENES/TOTALGENE LISTBETABETA STDSEP
      GO:0016071 GOBP MRNA METABOLIC PROCESS5/879RBFOX1, SMG6, TNFSF13, FXR2, RAVER10.6970.1440.09956.963E-10
      GO:1903311 GOBP REGULATION OF MRNA METABOLIC PROCESS3/334RBFOX1, TNFSF13, FXR21.5110.2440.2357.520E-09
      GO:0098791 GOCC GOLGI APPARATUS SUBCOMPARTMENT2/887RBFOX1, CNGB11.4540.1930.2301.189E-08
      GO:0003729 GOMF MRNA BINDING3/538RBFOX1, EIF4A1, FXR21.4160.2290.2312.538E-06
      GO:0006397 GOBP MRNA PROCESSING3/543RBFOX1, FXR2, RAVER11.3500.2180.2222.917E-08
      GO:0000375 GOBP RNA SPLICING VIA TRANSESTERIFICATION REACTIONS3/384RBFOX1, FXR2, RAVER11.3500.2180.2222.917E-08
      GO:0051236 GOBP ESTABLISHMENT OF RNA LOCALIZATION3/200NUP205, RBFOX1, SMG60.5900.09520.09793.882E-08
      GO:0006403 GOBP RNA LOCALIZATION3/233NUP205, RBFOX1, SMG60.5900.09520.09793.882E-08
      GO:0000381 GOBP REGULATION OF ALTERNATIVE MRNA SPLICING VIA SPLICEOSOME2/57RBFOX1, FXR21.3700.1810.2284.009E-08
      GO:0050684 GOBP REGULATION OF MRNA PROCESSING2/139RBFOX1, FXR21.3700.1810.2284.009E-08
      GO:0048024 GOBP REGULATION OF MRNA SPLICING VIA SPLICEOSOME2/100RBFOX1, FXR21.3700.1810.2284.009E-08
      GO:0043484 GOBP REGULATION OF RNA SPLICING2/144RBFOX1, FXR21.3700.1810.2284.009E-08
      GO:0035770 GOCC RIBONUCLEOPROTEIN GRANULE2/244RBFOX1, FXR21.3700.1810.2284.009E-08
      GO:0015931 GOBP NUCLEOBASE CONTAINING COMPOUND TRANSPORT4/252NUP205, SLC28A1, RBFOX1, SMG60.5110.09480.09464.537E-07
      GO:0003723 GOMF RNA BINDING9/1938RPN1, C7orf50, RALYL, RPP25L, RBFOX1, SMG6, EIF4A1, FXR2, RAVER10.4910.1340.09408.876E-07
      GO:0008380 GOBP RNA SPLICING5/478C2orf49, RBFOX1, USB1, FXR2, RAVER11.0910.2250.2192.360E-06

      Functional analyses of rbfox1 in zebrafish

      To test our hypothesis that a down regulation of Rbfox1 might lead to WM defect, we first established a measurement to monitor habituation learning behaviors in zebrafish larvae. We found that during acoustic/tapping stimulation with a total of 30 taps, larval zebrafish gradually reduced the extent of their startle responses (Supplementary Fig. S7A-B). After exposure to a massed/continued dark flash-training period, visual habituation was also evident (Supplementary Fig. S6C-D). Mutant-like larval zebrafish (5dpf), co-injected with chemically modified crRNAs/tracrRNAs and Cas9 protein against two sites of rbfox1 at the one-cell stage in wildtype (WT) fertilized eggs (Supplementary Fig. S8), exhibited a decrease in habituation (Fig. 3A-D). The knockdown of Rbfox1 function with anti-sense morpholino oligos that blocked either rbfox1 pre-RNA splicing or mRNA translation (ATG MO) resulted in similar habituation phenotypes (Fig. 3E-F), suggesting a relationship between Rbfox1 and the WM defect in zebrafish.
      Figure thumbnail gr3
      Figure 3rbfox1 is required for visual and acoustic habituation in larval zebrafish. (A) Schematic diagram of visual habituation. (B) Mean percentage of habituation during the visual habituation phase in the rbfox1 mutant. (C) Schematic diagram of acoustic habituation. (D) Mean percentage of habituation during the acoustic habituation phase in the rbfox1 mutant. (E) Mean percentage of habituation during the visual habituation phase in the rbfox1 MO injected fish. (F) Mean percentage of habituation during the acoustic habituation phase in the rbfox1 MO injected fish. The data are expressed as mean ± SE. n = 48 larval fish per group. ***p < 0.001, **p < 0.01, *p < 0.05 (paired t-test).

      Posttranscriptional regulation of WM genes

      To select zebrafish WM-related genes in an unbiased way, we profiled the transcriptome of rbfox1 Splicing MO injected larvae at 5dpf. The GO and KEGG enrichment analyses indicated that the most significantly changed pathways included the long-term potentiation (LTP), which was tightly related to memory formation, and TRP channels, which had been reported to define the hippocampal synaptic transmission and working memory (Fig. 4A). The most affected cellular components included pre- and post-synapse connections (Fig. 4B). Gene expression of proteins in postsynaptic density, the presynaptic membrane and the ionotropic glutamate receptor complex were all significantly changed. We found that grin1a, and grin1b were most abundantly expressed N-methyl-D-aspartate receptor genes in 5 dpf larvae, so were snap25a and snap25b that encode synaptosome-associated proteins (Supplementary Fig. S9). Zebrafish grin1 mutants exhibited a reduction of habituation (
      • Zoodsma J.D.
      • Chan K.
      • Bhandiwad A.A.
      • Golann D.R.
      • Liu G.
      • Syed S.A.
      • et al.
      A Model to Study NMDA Receptors in Early Nervous System Development.
      ) and MK-801 (NMDA receptor antagonist) treatment severely impacted larval zebrafish’s acoustic habituation (Supplementary Fig. S10A) also support the idea that these LTP genes are mainly responsible for a normal WM. Interestingly, in an attempt to learn how grin1a and grin1b were regulated in rbfox1 mutants/morphants, we found that post-transcriptional exon 4-skipping events in processing pre-mRNAs of both genes (Supplementary Fig. S10B-S10D). Thus, abnormal Rbfox1-mediated alternative splicing of grin1a and grin1b might in part result in the WM defect of rbfox1 mutant or morphant.
      Figure thumbnail gr4
      Figure 4Zebrafish rbfox1 regulates both presynaptic and postsynaptic functions. (A) KEGG pathway enrichment of DEGs in rbfox1 morphant. (B) GO enrichment of DEGs in rbfox1 morphant. All zebrafish larvae were collected at 5 dpf, and then whole transcriptome deep sequencing was performed.
      Among well-studied neurotransmission-related genes, functional protein association analyses revealed that SNAP25 is within a few closely related RNA processing components networked with RBFOX1 (Supplementary Fig. S4). Zebrafish Snap25a is the most abundant SNAP 5dpf larval, responsible for the presynaptic neurotransmitter release (Supplementary Fig. S9A). Zebrafish snap25a is a homolog of mouse Snap25, which is known to have two alternative splicing variants: Snap25a and Snap25b that correspond to two zebrafish splicing variants, snap25a-202 and snap25a-201 transcripts (Supplementary Fig. S9D), respectively during 24-96hpf. Only 11 amino acids are different between Snap25a-201 and -202, and these amino acids are all due to the alternative usage of exon-201/exon-202. Apparently, snap25a-201 is predominantly expressed during embryogenesis and early larval stage (Supplementary Fig. S9C).
      In rbfox1 ‘mutant’ and morphant, snap25a-201 transcripts were significantly decreased (p<0.01) while snap25a-202 expression was significantly increased (p<0.001) (Supplementary Fig. S9B). An Rbfox1 binding motif, (U)GCAUG is found in the exon 5b of rbfox1 and the overall reduced snap25a expression level also indicate that an imbalanced expression of two splicing variants most likely changes zebrafish VSM or acoustic/vibration-mediated memory.
      To obtain zebrafish snap25a mutants, we employed another gene knockout method that produces null F0 zebrafish with high probability (Wu et al., 2018). Using a mix of four sgRNAs against snap25a and Cas9 to inject the yolk of fertilized eggs, we evaluated learned memory of snap25a ‘mutant’ at 5dpf and discovered the significantly reduced acoustic/vibration and visual habituation (Fig. 5A), similar to those found in rbfox1 ‘mutant’ and morphant that were responsible for the impaired short-term memory.
      Figure thumbnail gr5
      Figure 5Posttranscriptional regulation of snap25a by Rbfox1. (A) Acoustic and visual habituation defects in snap25a-4sg mutant, tests performed on 5 dpf, n ≥48. (B) Changes of snap25a isoform ratios (exon5 alternative inclusion) in rbfox1 mutant. The data are expressed as mean ± SE. n = 48 larval fish per group. ***p < 0.001, **p < 0.01, *p < 0.05 (unpaired t-test).

      Discussion

      The past few decades have witnessed studies of many candidate genes that are related to the genetic vulnerability of WM impairments. However, most picked candidate genes were limited to the hypotheses. Our genome-wide single-variant and polygenic analyses of WM defects in ADHD children identified five RBFOX1 intronic variants/SNPs that may affect RBFOX1 level in the brain. Zebrafish loss-of-function of rbfox1 experiments showed both visual and auditory habituation defects in larval fish. When rbfox1 was knocked down or out, snap25a, grin1a and grin1b were found down-regulated and mis-spliced. To our knowledge, this is the first study that uncovers RBFOX1/rbfox1 regulates WM post-transcriptionally.
      Our pathway enrichment analysis has also directed to transcription regulation that linked to WM in accordance with previous candidate gene, gene-set, and polygenic studies (
      • Boku S.
      • Izumi T.
      • Abe S.
      • Takahashi T.
      • Nishi A.
      • Nomaru H.
      • et al.
      Copy number elevation of 22q11.2 genes arrests the developmental maturation of working memory capacity and adult hippocampal neurogenesis.
      ,
      • Linden D.E.
      • Lancaster T.M.
      • Wolf C.
      • Baird A.
      • Jackson M.C.
      • Johnston S.J.
      • et al.
      ZNF804A genotype modulates neural activity during working memory for faces.
      ). Some of the regulators are important for the synthesis of proteins required during neurobiological process of WM and mRNA metabolic process. If the expression of more RBFOX1/rbfox1-regulated LTP/synaptic genes is found to be controlled at the pre-RNA splicing and mRNA stability stages, the fine and quick regulation of neural activities such as habituation, startle response and dependent transcription could be economically and efficiently achieved.
      RBFOX1/rbfox1 is a pleiotropic gene that has been associated with seven specific psychiatric disorders (
      Cross-Disorder Group of the Psychiatric Genomics Consortium
      Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight Psychiatric Disorders.
      ). The most significant SNP, rs75885813 SNP was previously associated with Alzheimer’s disease (AD) related phenotypes in AD patients whose characteristic traits include WM deficits (
      • Stopford C.L.
      • Thompson J.C.
      • Neary D.
      • Richardson A.M.
      • Snowden J.S.
      Working memory, attention, and executive function in Alzheimer's disease and frontotemporal dementia.
      ,
      • Herold C.
      • Hooli B.V.
      • Mullin K.
      • Liu T.
      • Roehr J.T.
      • Mattheisen M.
      • et al.
      Family-based association analyses of imputed genotypes reveal genome-wide significant association of Alzheimer's disease with OSBPL6, PTPRG, and PDCL3.
      ). The association between RBFOX1 rs7193263 and major depression is also evident in a GWAS (p = 9.73e-9) (
      • Nagel M.
      • Jansen P.R.
      • Stringer S.
      • Watanabe K.
      • de Leeuw C.A.
      • Bryois J.
      • et al.
      Meta-analysis of genome-wide association studies for neuroticism in 449,484 individuals identifies novel genetic loci and pathways.
      ). RBFOX1 and WM were never directly associated, yet in a genome-wide gene expression study, Rbfox1 was linked to RNA processing after memory retrieval (
      • Peixoto L.L.
      • Wimmer M.E.
      • Poplawski S.G.
      • Tudor J.C.
      • Kenworthy C.A.
      • Liu S.
      • et al.
      Memory acquisition and retrieval impact different epigenetic processes that regulate gene expression.
      ). However, in the present study, no association is found between hyperactivity, an ADHD core symptom, and RBFOX1/rbfox1 or snap25a, either in the population genetic analyses or animal experiments. We noticed that the association of RBFOX1 was mainly driven by major depression (MD) sample (GWAS p-value: 9.73e-9) among 7 psychiatric disorders (
      • Nagel M.
      • Jansen P.R.
      • Stringer S.
      • Watanabe K.
      • de Leeuw C.A.
      • Bryois J.
      • et al.
      Meta-analysis of genome-wide association studies for neuroticism in 449,484 individuals identifies novel genetic loci and pathways.
      ). The p-value for the association of ADHD was only 0.0065 (
      • Demontis D.
      • Walters R.K.
      • Martin J.
      • Mattheisen M.
      • Als T.D.
      • Agerbo E.
      • et al.
      Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
      ). Thus, the effect of RBFOX1 on WM is potentially not only confined to ADHD. The present study further implies that RBFOX1 may be the common susceptibility gene of psychiatric disorders via regulating WM. More association studies of RBFOX1-regulated neurotransmission system genes and WM may be needed to strengthen our view.
      In the functional protein association networks of RBFOX1, SNAP25 attracted our attention. A previous study indicated that SNAP25 is related to WM deficits in ADHD patients (
      • Gao Q.
      • Liu L.
      • Chen Y.
      • Li H.
      • Yang L.
      • Wang Y.
      • et al.
      Synaptosome-related (SNARE) genes and their interactions contribute to the susceptibility and working memory of attention-deficit/hyperactivity disorder in males.
      ). In our study, we found a reduction of snap25a and snap25b expression in zebrafish rbfox1 morphant, which showed WM defect but no hyperactivity. The decrease of snap25a expression was more pronounced than that of snap25b. Zebrafish snap25a is homologous to mouse Snap25, which is known to have two alternative splicing variants: Snap25a (exon 5a, expressed late, corresponding to zebrafish snap25a-202) and Snap25b (exon 5b, expressed early, corresponding to zebrafish snap25a-201). Mouse Snap25a and Snap25b are expressed during embryonic and early postnatal development respectively. Snap25 differentially affects interactions with other soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs) and SNARE-interacting proteins (
      • Valladolid-Acebes I.
      • Daraio T.
      • Brismar K.
      • Harkany T.
      • Ögren S.O.
      • Hökfelt T.G.
      • et al.
      Replacing SNAP-25b with SNAP-25a expression results in metabolic disease.
      ,
      • Daraio T.
      • Valladolid-Acebes I.
      • Brismar K.
      • Bark C.
      SNAP-25a and SNAP-25b differently mediate interactions with Munc18-1 and Gβγ subunits.
      ). The Rbfox1-/- mouse brain exhibited normal mRNA level, but decreased Snap25b transcripts and increased Snap25a transcripts (
      • Gehman L.T.
      • Stoilov P.
      • Maguire J.
      • Damianov A.
      • Lin C.H.
      • Shiue L.
      • et al.
      The splicing regulator Rbfox1 (A2BP1) controls neuronal excitation in the mammalian brain.
      ), consistent with our findings of a decreased snap25a-201/25a-202 ratio (Fig. 5B). It is worth exploring whether reversing the snap25a-201/25a-202 ratio or simply increasing snap25a-201 mRNA can rescue the WM defect in rbfox1 mutants in the future study.
      However, there were some limitations in this study. First, due to the limited sample sizes, the current study is underpowered to detect other WM-related variants and unveil its genetic structure more comprehensively. Enlarging the sample size and cooperation with other groups could strengthen the statistical power.
      In summary, our present findings have revealed possible causal pathways of post-transcription regulators that trigger WM deficits and alternative splicing events, mediated by Rbfox1, control SNARE and LTP genes to affect WM.

      Uncited reference

      Sakai C, Ijaz S, Hoffman EJ (2018): Zebrafish Models of Neurodevelopmental Disorders: Past, Present, and Future. 11.

      ,
      • Guo S.
      Using zebrafish to assess the impact of drugs on neural development and function.
      ,
      • Ijaz S.
      • Hoffman E.J.
      Zebrafish: A Translational Model System for Studying Neuropsychiatric Disorders.
      ,

      FETCHO JR, LIU KS (1998): Zebrafish as a Model System for Studying Neuronal Circuits and Behaviora. 860:333-345.

      .

      Acknowledgements

      This work was supported by grants from the Major State Basic Research Development Program of China (973 Program, 2014CB846100), National Key R&D Program of China (2016YFC1306103 and 2018YFA0801006) and National Natural Science Foundation of China (grant numbers:81671358, 31771618, 81761128035, 81873803).
      Disclosures
      The authors report no biomedical financial interests or potential conflicts of interest.
      Contributions
      We thank all the patients for participating in this study. YXZ and LY conceived the study. YXZ conducted the population genetic analysis and wrote the first draft of the manuscript. NZ and FZ generated the zebrafish models and performed the statistical analyses of the animal experiments. WC and SHC helped with the data analyses. WC and LS helped with data collection. WFY helped with data collection and provided intellectual input. DL conducted the animal experiments. DL, LY, and LL provided intellectual input, edited the manuscript, and approved the final version.

      References

        • Ramos A.A.
        • Hamdan A.C.
        • Machado L.
        A meta-analysis on verbal working memory in children and adolescents with ADHD.
        Clin Neuropsychol. 2020; 34: 873-898
        • Pievsky M.A.
        • McGrath R.E.
        The Neurocognitive Profile of Attention-Deficit/Hyperactivity Disorder: A Review of Meta-Analyses.
        Archives of clinical neuropsychology : the official journal of the National Academy of Neuropsychologists. 2018; 33: 143-157
        • Fried R.
        • Chan J.
        • Feinberg L.
        • Pope A.
        • Woodworth K.Y.
        • Faraone S.V.
        • et al.
        Clinical correlates of working memory deficits in youth with and without ADHD: A controlled study.
        Journal of clinical and experimental neuropsychology. 2016; 38: 487-496
        • Arias-Vásquez A.
        • Altink M.E.
        • Rommelse N.N.
        • Slaats-Willemse D.I.
        • Buschgens C.J.
        • Fliers E.A.
        • et al.
        CDH13 is associated with working memory performance in attention deficit/hyperactivity disorder.
        Genes, brain, and behavior. 2011; 10: 844-851
        • Pfeiffer P.
        • Egorov A.V.
        • Lorenz F.
        • Schleimer J.H.
        • Draguhn A.
        • Schreiber S.
        Clusters of cooperative ion channels enable a membrane-potential-based mechanism for short-term memory.
        eLife. 2020; 9
        • Shimshek D.R.
        • Bus T.
        • Schupp B.
        • Jensen V.
        • Marx V.
        • Layer L.E.
        • et al.
        Different Forms of AMPA Receptor Mediated LTP and Their Correlation to the Spatial Working Memory Formation.
        Front Mol Neurosci. 2017; 10: 214
        • Rashid A.J.
        • Cole C.J.
        • Josselyn S.A.
        Emerging roles for MEF2 transcription factors in memory.
        Genes, brain, and behavior. 2014; 13: 118-125
        • Zhang C.S.
        • Bertaso F.
        • Eulenburg V.
        • Lerner-Natoli M.
        • Herin G.A.
        • Bauer L.
        • et al.
        Knock-in mice lacking the PDZ-ligand motif of mGluR7a show impaired PKC-dependent autoinhibition of glutamate release, spatial working memory deficits, and increased susceptibility to pentylenetetrazol.
        The Journal of neuroscience : the official journal of the Society for Neuroscience. 2008; 28: 8604-8614
        • Eagle A.L.
        • Gajewski P.A.
        • Robison A.J.
        Role of hippocampal activity-induced transcription in memory consolidation.
        Rev Neurosci. 2016; 27: 559-573
        • Straub R.E.
        • Lipska B.K.
        • Egan M.F.
        • Goldberg T.E.
        • Callicott J.H.
        • Mayhew M.B.
        • et al.
        Allelic variation in GAD1 (GAD67) is associated with schizophrenia and influences cortical function and gene expression.
        Molecular psychiatry. 2007; 12: 854-869
        • Kircher T.
        • Markov V.
        • Krug A.
        • Eggermann T.
        • Zerres K.
        • Nöthen M.M.
        • et al.
        Association of the DTNBP1 genotype with cognition and personality traits in healthy subjects.
        Psychological medicine. 2009; 39: 1657-1665
        • Spangaro M.
        • Bosia M.
        • Zanoletti A.
        • Bechi M.
        • Cocchi F.
        • Pirovano A.
        • et al.
        Cognitive dysfunction and glutamate reuptake: effect of EAAT2 polymorphism in schizophrenia.
        Neuroscience letters. 2012; 522: 151-155
        • Jentsch J.D.
        • Trantham-Davidson H.
        • Jairl C.
        • Tinsley M.
        • Cannon T.D.
        • Lavin A.
        Dysbindin modulates prefrontal cortical glutamatergic circuits and working memory function in mice.
        Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology. 2009; 34: 2601-2608
        • Gao Q.
        • Liu L.
        • Chen Y.
        • Li H.
        • Yang L.
        • Wang Y.
        • et al.
        Synaptosome-related (SNARE) genes and their interactions contribute to the susceptibility and working memory of attention-deficit/hyperactivity disorder in males.
        Progress in neuro-psychopharmacology & biological psychiatry. 2015; 57: 132-139
        • Cottrell J.R.
        • Levenson J.M.
        • Kim S.H.
        • Gibson H.E.
        • Richardson K.A.
        • Sivula M.
        • et al.
        Working memory impairment in calcineurin knock-out mice is associated with alterations in synaptic vesicle cycling and disruption of high-frequency synaptic and network activity in prefrontal cortex.
        The Journal of neuroscience : the official journal of the Society for Neuroscience. 2013; 33: 10938-10949
        • Pergola G.
        • Di Carlo P.
        • Andriola I.
        • Gelao B.
        • Torretta S.
        • Attrotto M.T.
        • et al.
        Combined effect of genetic variants in the GluN2B coding gene (GRIN2B) on prefrontal function during working memory performance.
        Psychological medicine. 2016; 46: 1135-1150
        • Thomas E.H.X.
        • Bozaoglu K.
        • Rossell S.L.
        • Gurvich C.
        The influence of the glutamatergic system on cognition in schizophrenia: A systematic review.
        Neuroscience and biobehavioral reviews. 2017; 77: 369-387
        • Reif A.
        • Herterich S.
        • Strobel A.
        • Ehlis A.C.
        • Saur D.
        • Jacob C.P.
        • et al.
        A neuronal nitric oxide synthase (NOS-I) haplotype associated with schizophrenia modifies prefrontal cortex function.
        Molecular psychiatry. 2006; 11: 286-300
        • Schmitt W.B.
        • Sprengel R.
        • Mack V.
        • Draft R.W.
        • Seeburg P.H.
        • Deacon R.M.
        • et al.
        Restoration of spatial working memory by genetic rescue of GluR-A-deficient mice.
        Nature neuroscience. 2005; 8: 270-272
        • Reisel D.
        • Bannerman D.M.
        • Schmitt W.B.
        • Deacon R.M.
        • Flint J.
        • Borchardt T.
        • et al.
        Spatial memory dissociations in mice lacking GluR1.
        Nature neuroscience. 2002; 5: 868-873
        • Roussos P.
        • Giakoumaki S.G.
        • Adamaki E.
        • Georgakopoulos A.
        • Robakis N.K.
        • Bitsios P.
        The association of schizophrenia risk D-amino acid oxidase polymorphisms with sensorimotor gating, working memory and personality in healthy males.
        Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology. 2011; 36: 1677-1688
        • Heck A.
        • Fastenrath M.
        • Ackermann S.
        • Auschra B.
        • Bickel H.
        • Coynel D.
        • et al.
        Converging genetic and functional brain imaging evidence links neuronal excitability to working memory, psychiatric disease, and brain activity.
        Neuron. 2014; 81: 1203-1213
        • Goldberg T.E.
        • Straub R.E.
        • Callicott J.H.
        • Hariri A.
        • Mattay V.S.
        • Bigelow L.
        • et al.
        The G72/G30 gene complex and cognitive abnormalities in schizophrenia.
        Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology. 2006; 31: 2022-2032
        • Boku S.
        • Izumi T.
        • Abe S.
        • Takahashi T.
        • Nishi A.
        • Nomaru H.
        • et al.
        Copy number elevation of 22q11.2 genes arrests the developmental maturation of working memory capacity and adult hippocampal neurogenesis.
        Molecular psychiatry. 2018; 23: 985-992
        • Linden D.E.
        • Lancaster T.M.
        • Wolf C.
        • Baird A.
        • Jackson M.C.
        • Johnston S.J.
        • et al.
        ZNF804A genotype modulates neural activity during working memory for faces.
        Neuropsychobiology. 2013; 67: 84-92
        • Fukuda K.
        • Vogel E.
        • Mayr U.
        • Awh E.
        Quantity, not quality: the relationship between fluid intelligence and working memory capacity.
        Psychonomic bulletin & review. 2010; 17: 673-679
        • Anticevic A.
        • Repovs G.
        • Corlett P.R.
        • Barch D.M.
        Negative and Nonemotional Interference with Visual Working Memory in Schizophrenia.
        Biological psychiatry. 2011; 70: 1159-1168
      1. Schecklmann M, Dresler T, Beck S, Jay JT, Febres R, Haeusler J, et al. (2011): Reduced prefrontal oxygenation during object and spatial visual working memory in unpolar and bipolar depression. 194:378-384.

        • Cassar S.
        • Adatto I.
        • Freeman J.L.
        • Gamse J.T.
        • Iturria I.
        • Lawrence C.
        • et al.
        Use of Zebrafish in Drug Discovery Toxicology.
        Chemical research in toxicology. 2020; 33: 95-118
      2. Adamson KI, Sheridan E, Grierson AJ (2018): Use of zebrafish models to investigate rare human disease. 55:641-649.

        • Randlett O.
        • Wee C.L.
        • Naumann E.A.
        • Nnaemeka O.
        • Schoppik D.
        • Fitzgerald J.E.
        • et al.
        Whole-brain activity mapping onto a zebrafish brain atlas.
        Nature methods. 2015; 12: 1039-1046
        • Spence R.
        • Gerlach G.
        • Lawrence C.
        • Smith C.
        The behaviour and ecology of the zebrafish, Danio rerio.
        Biological reviews of the Cambridge Philosophical Society. 2008; 83: 13-34
        • Lieschke G.J.
        • Currie P.D.
        Animal models of human disease: zebrafish swim into view.
        Nature Reviews Genetics. 2007; 8: 353-367
      3. Sakai C, Ijaz S, Hoffman EJ (2018): Zebrafish Models of Neurodevelopmental Disorders: Past, Present, and Future. 11.

        • Guo S.
        Using zebrafish to assess the impact of drugs on neural development and function.
        Expert opinion on drug discovery. 2009; 4: 715-726
        • Ijaz S.
        • Hoffman E.J.
        Zebrafish: A Translational Model System for Studying Neuropsychiatric Disorders.
        Journal of the American Academy of Child and Adolescent Psychiatry. 2016; 55: 746-748
      4. FETCHO JR, LIU KS (1998): Zebrafish as a Model System for Studying Neuronal Circuits and Behaviora. 860:333-345.

        • Shuai L.
        • Chan R.C.
        • Wang Y.
        Executive function profile of Chinese boys with attention-deficit hyperactivity disorder: different subtypes and comorbidity.
        Archives of clinical neuropsychology : the official journal of the National Academy of Neuropsychologists. 2011; 26: 120-132
        • Anderson C.A.
        • Pettersson F.H.
        • Clarke G.M.
        • Cardon L.R.
        • Morris A.P.
        • Zondervan K.T.
        Data quality control in genetic case-control association studies.
        Nature protocols. 2010; 5: 1564-1573
        • Sudmant P.H.
        • Rausch T.
        • Gardner E.J.
        • Handsaker R.E.
        • Abyzov A.
        • Huddleston J.
        • et al.
        An integrated map of structural variation in 2,504 human genomes.
        Nature. 2015; 526: 75-81
        • Chang C.C.
        • Chow C.C.
        • Tellier L.C.
        • Vattikuti S.
        • Purcell S.M.
        • Lee J.J.
        Second-generation PLINK: rising to the challenge of larger and richer datasets.
        GigaScience. 2015; 4: 7
      5. Eve M Valera, Ph.D. , Ariel Brown, Ph.D. , Joseph Biederman, M.D. , Stephen V. Faraone, Ph.D. , Nikos Makris, M.D., Ph.D. , Michael C. Monuteaux, Sc.D. , et al. (2010): Sex Differences in the Functional Neuroanatomy of Working Memory in Adults With ADHD. 167:86-94.

      6. Soares PSM, de Oliveira PD, Wehrmeister FC, Menezes AMB, Rohde LA, Gonçalves H (2022): Does IQ Influence Association Between Working Memory and ADHD Symptoms in Young Adults? 26:1097-1105.

        • Price A.L.
        • Patterson N.J.
        • Plenge R.M.
        • Weinblatt M.E.
        • Shadick N.A.
        • Reich D.
        Principal components analysis corrects for stratification in genome-wide association studies.
        Nature genetics. 2006; 38: 904-909
        • de Leeuw C.A.
        • Mooij J.M.
        • Heskes T.
        • Posthuma D.
        MAGMA: generalized gene-set analysis of GWAS data.
        PLoS computational biology. 2015; 11e1004219
        • Liberzon A.
        • Birger C.
        • Thorvaldsdóttir H.
        • Ghandi M.
        • Mesirov J.P.
        • Tamayo P.
        The Molecular Signatures Database (MSigDB) hallmark gene set collection.
        Cell Syst. 2015; 1: 417-425
        • Choi S.W.
        • O'Reilly P.F.
        PRSice-2: Polygenic Risk Score software for biobank-scale data.
        GigaScience. 2019; 8
        • Guo L.
        • Wang J.
        rSNPBase 3.0: an updated database of SNP-related regulatory elements, element-gene pairs and SNP-based gene regulatory networks.
        Nucleic acids research. 2018; 46: D1111-d1116
        • Ward L.D.
        • Kellis M.
        HaploReg v4: systematic mining of putative causal variants, cell types, regulators and target genes for human complex traits and disease.
        Nucleic acids research. 2016; 44: D877-881
        • GTEx Consortium
        The Genotype-Tissue Expression (GTEx) project.
        Nature genetics. 2013; 45: 580-585
        • Szklarczyk D.
        • Morris J.H.
        • Cook H.
        • Kuhn M.
        • Wyder S.
        • Simonovic M.
        • et al.
        The STRING database in 2017: quality-controlled protein-protein association networks, made broadly accessible.
        Nucleic acids research. 2017; 45: D362-d368
      7. Westerfield M (2000): The zebrafish book. A guide for the laboratory use of zebrafish (Danio rerio). 4 ed. Eugene: University of Oregon Press.

        • Zoodsma J.D.
        • Chan K.
        • Bhandiwad A.A.
        • Golann D.R.
        • Liu G.
        • Syed S.A.
        • et al.
        A Model to Study NMDA Receptors in Early Nervous System Development.
        The Journal of neuroscience : the official journal of the Society for Neuroscience. 2020; 40: 3631-3645
        • Cross-Disorder Group of the Psychiatric Genomics Consortium
        Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight Psychiatric Disorders.
        Cell. 2019; 179 (e1411): 1469-1482
        • Stopford C.L.
        • Thompson J.C.
        • Neary D.
        • Richardson A.M.
        • Snowden J.S.
        Working memory, attention, and executive function in Alzheimer's disease and frontotemporal dementia.
        Cortex; a journal devoted to the study of the nervous system and behavior. 2012; 48: 429-446
        • Herold C.
        • Hooli B.V.
        • Mullin K.
        • Liu T.
        • Roehr J.T.
        • Mattheisen M.
        • et al.
        Family-based association analyses of imputed genotypes reveal genome-wide significant association of Alzheimer's disease with OSBPL6, PTPRG, and PDCL3.
        Molecular psychiatry. 2016; 21: 1608-1612
        • Nagel M.
        • Jansen P.R.
        • Stringer S.
        • Watanabe K.
        • de Leeuw C.A.
        • Bryois J.
        • et al.
        Meta-analysis of genome-wide association studies for neuroticism in 449,484 individuals identifies novel genetic loci and pathways.
        Nature genetics. 2018; 50: 920-927
        • Peixoto L.L.
        • Wimmer M.E.
        • Poplawski S.G.
        • Tudor J.C.
        • Kenworthy C.A.
        • Liu S.
        • et al.
        Memory acquisition and retrieval impact different epigenetic processes that regulate gene expression.
        BMC genomics. 2015; 16: S5
        • Demontis D.
        • Walters R.K.
        • Martin J.
        • Mattheisen M.
        • Als T.D.
        • Agerbo E.
        • et al.
        Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
        Nature genetics. 2019; 51: 63-75
        • Valladolid-Acebes I.
        • Daraio T.
        • Brismar K.
        • Harkany T.
        • Ögren S.O.
        • Hökfelt T.G.
        • et al.
        Replacing SNAP-25b with SNAP-25a expression results in metabolic disease.
        Proceedings of the National Academy of Sciences of the United States of America. 2015; 112: E4326-4335
        • Daraio T.
        • Valladolid-Acebes I.
        • Brismar K.
        • Bark C.
        SNAP-25a and SNAP-25b differently mediate interactions with Munc18-1 and Gβγ subunits.
        Neuroscience letters. 2018; 674: 75-80
        • Gehman L.T.
        • Stoilov P.
        • Maguire J.
        • Damianov A.
        • Lin C.H.
        • Shiue L.
        • et al.
        The splicing regulator Rbfox1 (A2BP1) controls neuronal excitation in the mammalian brain.
        Nature genetics. 2011; 43: 706-711