EPIGENETICS 2023, VOL. 18, NO. 1, 2207253 https://doi.org/10.1080/15592294.2023.2207253 RESEARCH ARTICLE DNA methylation at birth and fine motor ability in childhood: an epigenome-wide association study with replication Fadila Serdarevica,b,c*, Mannan Luoa,d,e*, Irma Karabegovićf, Anne-Claire Binterg,h,i, Silvia Alemanya,j,k, Ryan Mutzela, Monica Guxensa,g,h,i, Mariona Bustamanteg,h,i, Aida Hajdarpasicl, Tonya Whitea,m, Janine F Felixd,n, Charlotte A.M. Cecila,f,o, and Henning Tiemeiera,b a Department of Child and Adolescent Psychiatry, Erasmus MC, University Medical Centre, Rotterdam, the Netherlands; bDepartment of Social and Behavioral Science, Harvard T.H. Chan School of Public Health, Boston, MA, USA; cSarajevo Medical School, Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina; dThe Generation R Study Group, Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands; eDepartment of Psychology, Education and Child Studies, Erasmus University Rotterdam, Rotterdam, the Netherlands; fDepartment of Epidemiology, Erasmus University Medical Center, Rotterdam, the Netherlands; gISGlobal, Barcelona, Spain; h Universitat Pompeu Fabra, Barcelona, Spain; iSpanish Consortium for Research on Epidemiology and Public Health (CIBERESP), Instituto de Salud Carlos III, Madrid, Spain; jPsychiatric Genetics Unit, Group of Psychiatry Mental Health and Addiction, Vall d’Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain; kBiomedical Network Research Centre on Mental Health (CIBERSAM), Instituto de Salud Carlos III, Madrid, Spain; lDepartment of Medical Biology, and Genetics, Sarajevo Medical School, Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina; mDepartment of Radiology and Nuclear Medicine, Erasmus University Medical Centre, Rotterdam, the Netherlands; nDepartment of Pediatrics, Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands; o Molecular Epidemiology, Department of Biomedical Data Sciences, Leiden University Medical Center, Leiden, the Netherlands ABSTRACT ARTICLE HISTORY Lower fine motor performance in childhood has been associated with poorer cognitive develop­ Received 28 October 2022 ment and neurodevelopmental conditions such as autism spectrum disorder, yet, biological Revised 10 April 2023 underpinnings remain unclear. DNA methylation (DNAm), an essential process for healthy neuro­ Accepted 19 April 2023 development, is a key molecular system of interest. In this study, we conducted the first epigen­ KEYWORDS ome-wide association study of neonatal DNAm with childhood fine motor ability and further DNA methylation; cord examined the replicability of epigenetic markers in an independent cohort. The discovery study blood; fine motor was embedded in Generation R, a large population-based prospective cohort, including development; EWAS; a subsample of 924 ~ 1026 European-ancestry singletons with available data on DNAm in cord cognitive function blood and fine motor ability at a mean (SD) age of 9.8 (0.4) years. Fine motor ability was measured using a finger-tapping test (3 subtests including left-, right-hand and bimanual), one of the most frequently used neuropsychological instruments of fine motor function. The replication study comprised 326 children with a mean (SD) age of 6.8 (0.4) years from an independent cohort, the INfancia Medio Ambiente (INMA) study. Four CpG sites at birth were prospectively associated with childhood fine motor ability after genome-wide correction. Of these, one CpG (cg07783800 in GNG4) was replicated in INMA, showing that lower levels of methylation at this site were associated with lower fine motor performance in both cohorts. GNG4 is highly expressed in the brain and has been implicated in cognitive decline. Our findings support a prospective, repro­ ducible association between DNAm at birth and fine motor ability in childhood, pointing to GNG4 methylation at birth as a potential biomarker of fine motor ability. Introduction manipulate objects or perform specific tasks Fine motor skills involve the fine control of fingers appropriately. These motor movements also entail and hands and are important for activities such as the age-appropriate development of related physi­ playing an instrument, writing, cutting, and open­ cal skills, such as core trunk control and shoulder ing boxes. Efficient fine motor control requires strength, as these provide a stable base from which a cascade of neuronal activities in order to the arm and hand can then move with control, as CONTACT Fadila Serdarevic fserdarevic@gmail.com Erasmus Medical Center Rotterdam; Department of Child and Adolescent Psychiatry, Rotterdam 3000 CB, Netherlands *The first authorship shared. Supplemental data for this article can be accessed online at https://doi.org/10.1080/15592294.2023.2207253 © 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent. 2 F. SERDAREVIC ET AL. well as the smaller muscles of the hands. In turn, unclear. Epigenetic processes, such as DNA the efficiency of fine motor skills influences the methylation (DNAm), have emerged as quality of manual task outcomes as well as the a potential molecular system of interest, as they speed of task performance [1]. Children with regulate gene activity in response to both genetic poor fine motor skills have difficulties developing and environmental influences [31]. DNAm at birth appropriate independence in ‘life’ skills, for is particularly of interest, as it (i) may be a good instance, getting dressed and feeding themselves, proxy of genetic, biological, or prenatal environ­ which has social implications within the family mental factors relevant to the fine motor develop­ and peer relationships. ment [31,32], while the epigenetic signal becomes The early development of fine motor ability under­ ‘noisier’ in later life; and (ii) growing evidence lies and precedes cognitive functioning later in life [2– shows that DNAm at birth associates more 8]. Repetitive rapid finger tapping is a standard test of strongly with certain neurodevelopmental out­ fine motor control of the upper extremities. Hubel comes that also feature fine motor deficits, such et al. refer to the finger-tapping test (FTT) as an as ADHD and social communication deficits efficient and precise measure of tapping speed and [33,34], compared to DNAm patterns measured kinetics [9], and therefore one of the most frequently concurrently in childhood (i.e., prospective > used neuropsychological instruments of utility in cross-sectional associations). As such, DNAm pat­ research and clinical studies of motor performance terns at birth – but not in childhood – may also [10]. Lower performance in fine motor ability is mediate risk for developmental outcomes in later a neuropsychological deficit that transcends diagnos­ life (e.g., DNAm alterations at birth may impact tic boundaries. Abnormal fine motor skills have been early neurodevelopment and downstream beha­ associated with a range of neurodevelopmental dis­ vioural phenotypes that persist despite changes in orders, such as autism spectrum disorder (ASD) and DNAm patterns per se). attention deficit hyperactivity disorder (ADHD), as Previous epigenetic studies have linked DNAm well as cognitive decline in ageing adults [11–16]. of candidate genes at birth, such as the glucocorti­ Mental disorders, including ASD and ADHD, can coid receptor gene NR3C1, to poorer motor per­ be considered as the extremes of quantitative traits formance during early development [35–37]. rather than dichotomizing the continuous variation of However, evidence from epigenome-wide associa­ symptoms [17–19]. Identifying the underlying tion studies (EWASs) is limited. We are aware of mechanisms that give rise to individual differences only one EWAS, which identified associations in fine motor ability, in particular during develop­ between cord blood DNAm at a cytosine- mentally sensitive periods of childhood, provides phosphate-guanine (CpG) site in SPTBN4 and opportunities to understand typical neurodevelop­ infant motor function, measured using ment and cognitive function. It can also lead to new a composite score of general (fine and gross) insights into the molecular processes underlying risks motor skills [38]. Given that research to date for related neurodevelopmental disorders in the gen­ implicates more fine then gross motor skills as eral population [20,21]. indicators of cognitive development [39,40], it is Previous research has shown that fine motor important to focus specifically on fine motor abil­ ability is partially under genetic influence, with ity. Yet, to our knowledge, no study has examined twin and familial heritability ranging from 41% epigenome-wide DNAm patterns associated with to 86% on different fine motor tests [22,23]. fine motor ability in childhood. Furthermore, it is Environmental factors that shape motor develop­ unclear to what extent current findings replicate in ment may occur very early in life, for example, independent samples, which is important for iden­ exposure to adverse intra-uterine environments, tifying reliable biomarkers. such as higher placental vascular resistance, mal­ In light of these gaps, we performed an EWAS nutrition, and stress, are associated with poor examining long-term prospective associations motor ability, including FTT performance [24– between DNAm at birth (cord blood) and fine 30]. Yet, how these influences affect early fine motor ability in childhood, measured by FTT at motor development at a molecular level remains age 10 years, leveraging data from a large EPIGENETICS 3 population-based prospective cohort, the 1 (full methylation). CpG sites with values outside the Generation R Study (GenR). We further tested 25th percentile − 3 × interquartile range (IQR) and whether our findings can be replicated in an inde­ the 75th percentile+3 × IQR of the distribution were pendent birth cohort. To verify the relevance of identified as outliers and winsorized. FTT to higher-order cognition in the general pae­ diatric population, we also examined the prospec­ Finger Tapping Test tive association between FTT performance and In GenR, the fine motor ability was measured using a later cognitive function in adolescence (age 14 computerized finger-tapping task during an assess­ years). Finally, we investigated whether slower ment visit at the age-10 years. The task was pro­ FTT performance mediated the association grammed in Python using modules from the between lower DNAm at birth and lower cognitive PsychoPy toolbox (version 1.90.3), as previously functioning in the general paediatric population. described [45,46]. Briefly, children were instructed to perform five 10-second finger tapping trials in which they tapped their index fingers on a button Methods as quickly as possible using their (1) right, (2) left, (3) Study population both (alternating), (4) right and (5) then left hand. For the right- and left-hand conditions, we calcu­ Detailed information for each cohort is described lated the average number of taps across two trials on in the Supplementary Methods. Primary analyses the same hand. For the alternating condition, the were conducted using data from GenR, total number of taps was calculated as bimanual a population-based prospective cohort from early outcome. foetal life onwards, based in Rotterdam, the In INMA, fine motor ability was measured dur­ Netherlands. The design and sample characteris­ ing the age-7 assessment visit using a computerized tics of GenR have been described in detail else­ finger-tapping task administered with E-Prime 2.0 where [41]. The EWAS analyses included [47], first completing two trials with the preferred a subsample of 1,396 European-ancestry singletons hand for practice, then two more trials with the with available DNAm data at birth. Of these, 924 non-preferred hand. The children used ~ 1069 children had data on fine motor ability at a standardized method by pressing a key in a mean (SD) age of 9.8 (0.3) years and relevant a joystick board; it was attached to the desk in covariates. In addition, we included 2813 partici­ order to avoid slips while doing the task. We calcu­ pants with cognitive assessment at a mean (SD) lated the mean number of taps of same-hand trials. age of 13.6 (0.3) years. A flowchart of sample selection is described in Figure S1. Cognitive Functions The replication analyses were performed in an Cognitive functions were measured by the independent population-based prospective cohort, Wechsler Intelligence Scale for Children-Fifth the INfancia Medio Ambiente (INMA) study in Edition (WISC-V) at 14 years of age in GenR Spain [42]. We included 326 European singletons [48]. Four core subtests (i.e., Vocabulary, Matrix with complete data on cord blood DNAm at birth Reasoning, Digit Span, and Coding) were selected and fine motor ability at a mean (SD) age of 6.8 (0.4) to produce four domain-specific index scores: ver­ years. bal comprehension, fluid reasoning, working memory, processing speed. Trained research assis­ DNA methylation tants administered four subtests. T-scores based Briefly, DNAm from cord blood was measured using on Dutch norm for all subtests were summed the Illumina Infinium HumanMethylation450 and converted to an estimated Full-Scale BeadChip in both cohorts. Pre-processing and nor­ Intelligence Quotient (FSIQ) score. malization were performed in R with the CPACOR [43] workflow in GenR and the Bioconductor package Covariates minfi [44] in INMA. DNAm levels are characterized Covariates for epigenetic associations in both by beta (β) values ranging from 0 (no methylation) to cohorts included: child sex, maternal age at 4 F. SERDAREVIC ET AL. delivery (in years), smoking during pregnancy follows. (1) We examined the concordance (binary categorization of ‘no smoking/quit in between methylation in peripheral blood and early pregnancy’ vs. ‘smoked throughout preg­ brain using two independent online tools based nancy’), gestational age at delivery (in weeks), on post-mortem data: the blood–brain concor­ child age at the fine motor assessment (in dance tool (http://epigenetics.essex.ac.uk/blood years), and estimated cell-type proportions esti­ brain/) from 75 adults characterizing whole blood mated using a cord blood cell type reference versus 4 brain regions (superior temporal gyrus, panel [49]. In addition, we adjusted for technical prefrontal cortex, entorhinal cortex, and cerebel­ covariates (i.e., batch effect only in GenR, as lum); and BECon (https://redgar598.shinyapps.io/ ComBat was applied to remove batch effect in BECon/) from 16 adults with 3 brain regions INMA), and handedness as measured by the (inferior temporal gyrus, anterior prefrontal cor­ Edinburgh Handedness Inventory (EHI) in tex, and parietal cortex). (2) Gene expression pro­ GenR. In INMA, handedness was defined as files were probed across 53 human tissues using a laterality index calculated with the relative dif­ GTEx database from the FUMA portal (https:// ference between scores obtained with preferred fuma.ctglab.nl/). (3) Look-up of methylation quan­ and non-preferred hand, expressed as a percent titative trait loci (mQTL) was performed using the of the sum. largest mQTL database to date to explore potential genetic influences (https://mqtldb.godmc.org.uk/). Additionally, to verify the relevance of FTT to Further details on DNAm, finger tapping test, cognitive functions, we examined the associations cognitive functions, and handedness are of the FTT score with FSIQ and 4 subdomain provided in Supplementary Methods index scores of cognition using linear regression. Statistical analysis The analyses were adjusted for sex, age of cogni­ tive assessment, handedness, child ethnicity, Epigenome-wide association study (EWAS) analysis maternal age, and education. Last, a candidate Robust linear regression was applied to investigate gene follow-up analysis was performed on results the prospective associations between neonatal extracted from a previous EWAS of infants’ motor DNAm and fine motor ability in childhood. ability for CpG sites annotated to SPTBN4, in DNA methylation β values at each CpG site were order to examine associations with DNAm at this specified as the predictor and right-, left-hand or previously implicated gene (n = 36 CpGs). P-values bimanual FTT score as the outcome adjusted for were adjusted for gene-level significance using the all covariates. In order to DNAm values were Benjamini-Hochberg method for false discovery z-standardized to facilitate interpretation and rate (FDR), and only methylation sites with FDR- comparison across two cohorts. Both DNAm β corrected q < 0.05 were considered statistically value and FTT outcomes were z-standardized to significant. facilitate interpretation and comparison across two cohorts. Probes were annotated using the meffil Replication analyses in INMA R package [50], enhanced using the University of Last, we attempted to replicate our findings in an California Santa Cruz (UCSC) Genome Browser, independent sample (n = 326) with complete data based on genome build hg19. P-values were on cord blood DNAm at birth and FTT at age 7 adjusted for genome-wide significance using false from INMA. Robust linear models were run in discovery rate (FDR) adjustment using the INMA on the top CpGs identified in GenR, adjust­ Benjamini–Hochberg method [51] and only sites ing for covariates as described above. with FDR-corrected q < 0.05 were considered sta­ tistically significant. Mediation analysis In a post hoc analysis, we examined whether fine Follow-up analyses motor development mediates the association Follow-up analyses were first performed to further between DNAm at birth (specifically in CpG sites characterize identified top CpGs (FDR q < 0.05) as showing significant replication in INMA) and EPIGENETICS 5 cognition in adolescence. To this end, we ran who were excluded. Children from INMA showed, a mediation model using R (package mediation) on average, higher left- and right-hand FTT per­ with 99% bias-corrected bootstraps confidence formance compared to children in GenR, likely intervals (CIs) using 1000 bootstrapped samples. due to differences in the duration of FTT assess­ We partitioned the total effect of replicated CpGs ment between cohorts (15-second per trial in on cognition into a direct and an indirect effect INMA and 10-second per trial in GenR). through fine motor development. The indirect effect is the estimated effect of DNAm on cogni­ tion through fine motor development, while the Epigenome-wide Association Analysis direct effect represents the effect of DNAm on At birth, lower methylation at one CpG cognition scores that is independent of fine (cg07783800) site was prospectively associated with motor development. slower childhood right and left hand FTT after FDR correction (q < 0.05; Table 2 and Figure 1). The top Data availability site is located in the gene body of GNG4, which is The summary statistics of the EWAS will be made a member of the G-protein family and has been available on figshare https://figshare.com/. associated with cognitive decline [52]. For left-hand FTT, four CpGs surpassed the FDR-correction, including (a) the same site cg07783800 in GNG4; Results (b) cg05136476, annotated to the gene body of PRKCD, a gene previously linked to several psychia­ Participants characteristics tric disorders by genome-wide association studies Sample characteristics are described in Table 1. (GWAS), including ASD, ADHD, and schizophrenia Differences in characteristics between children [53,54]; (c) cg07638797, annotated to the promotor included in our EWAS sample and those who of DYRK1A, a gene localized in the Down syndrome were not included were evaluated using chi- critical region of chromosome 21; variations in this square tests for the categorical and independent gene have also been linked to neuroticism and t-tests for the continuous variables. Children in the Parkinson’s disease by GWAS at a genome-wide current EWAS sample had a higher gestational age significant level [55,56]; (d) cg16705073, annotated and lower performance on left-hand and bimanual to ST3GAL2, a sialyltransferase gene responsible for FTT than those not included. Mothers of children sialylation of gangliosides and glycoproteins; knock­ in the study were on average older, had higher out of this gene results in cognitive deficits in mice education, and were less likely to smoke through­ [57]. We did not observe any significant association out pregnancy compared to mothers of children for bimanual FTT scores (Table S1). Table 1. Descriptive statistics in two cohorts. GenR INMA (N = 925) (N = 326) Child characteristics Sex, % girls 50.8 48.2 Gestational age in weeks, mean (SD) 40.2 (1.5) 39.8 (1.4) Prematurity, % 2.38 2.45 Age at measure of FTT, years, mean (SD) 9.8 (0.3) 6.8 (0.4) Handedness, laterality quotient, mean (SD) 0.7 (0.5) 0.8 (0.5) Finger-tapping test (FTT), number of taps, mean (SD) Left-hand FTT 38.2 (5.6) 56.6 (8.8) Right-hand FTT 41.1 (6.6) 66.0 (8.1) Bimanual FTT (alternating) 35.1 (11.9) NA Maternal characteristics Age at intake/delivery, years, mean (SD) 32.1 (4.1) 31.7 (4.1) Sustained smoking during pregnancy, % 11.2 12.6 Educational level, % Lower (below university degree) 32.5 68.0 Higher (university degree and above) 67.5 32.0 Note. Prematurity is defined as neonates at less than 37 weeks gestation. 6 F. SERDAREVIC ET AL. Table 2. Top 10 CpGs derived from the EWAS of the association between neonatal blood DNAm and right and left hand finger tapping test at 10 years in GenR; sorted by ascending p-value (N = 925). CpGs Chr Pos Nearest gene Effect SE p FDR Outcome: right-hand FTT cg07783800 chr1 235803662 GNG4 0.17 0.03 5.11E–08 0.023 cg14660125 chr4 89378460 HERC5 0.14 0.03 6.95E–07 0.159 cg20790998 chr1 68290436 GNG12 −0.18 0.04 1.31E–06 0.171 cg09469355 chr1 2161886 SKI 0.16 0.03 1.61E–06 0.171 cg26657045 chr5 60132712 ELOVL7 0.17 0.04 2.20E–06 0.171 cg13964781 chr19 14552350 PKN1 0.14 0.03 2.29E–06 0.171 cg26095658 chr2 71017887 FIGLA −0.13 0.03 2.65E–06 0.171 cg11849213 chr14 65210900 PLEKHG3 0.16 0.03 3.28E–06 0.171 cg04477675 chr6 33168449 SLC39A7;RXRB −0.14 0.03 3.35E–06 0.171 cg23468128 chr2 61108270 REL −0.15 0.03 4.35E–06 0.200 Outcome: left-hand FTT cg05136476 chr3 53190709 PRKCD* 0.15 0.03 1.20E–07 0.026 cg07783800 chr1 235803662 GNG4 0.16 0.03 1.70E–07 0.026 cg07638797 chr21 38792116 DYRK1A 0.19 0.04 2.12E–07 0.026 cg16705073 chr16 70472678 ST3GAL2 0.17 0.03 2.29E–07 0.026 cg10194627 chr13 74568601 KLF12 0.21 0.04 7.25E–07 0.067 cg09004351 chr7 5515555 FBXL18 0.15 0.03 2.20E–06 0.168 cg02019988 chr3 50311211 SEMA3B −0.13 0.03 9.38E–06 0.404 cg19321991 chr1 213124472 VASH2 −0.20 0.04 1.02E–05 0.404 cg02960777 chr7 77562848 PHTF2 −0.14 0.03 1.05E–05 0.404 cg19870626 chr1 175889703 RFWD2* 0.13 0.03 1.07E–05 0.404 Note. FTT: finger-tapping test; Chr: chromosome; Pos: position; Effect, standardized regression coefficients, SE, standardized error. The full model is adjusted for sex, gestational age, age at FTT assessment, handedness, maternal age in take/delivery, maternal smoking during pregnancy, batch effects, estimated cell-type proportions.*Annotation based on University of California Santa Cruz Known Gene fills in the nearest gene within 10 MB. Figure 1. Manhattan plots (Fig.1a. Left-hand; Fig.1b. Right-hand). showing EWAS between cord blood DNA methylation and finger tapping test at 10 years (N = 925). The horizontal line indicates a FDR threshold of q < 0.05. Top CpG sites are highlighted in green. EPIGENETICS 7 Follow-up analyses with the lower total score of cognitive function in adolescence (right-hand, β = .20, p < .001); left- Blood–brain concordance hand, β = .20, p < .001). Slower performance in Using two independent tools, we found modest right-hand FTT was associated with less-optimal and inconsistent DNAm patterns across blood functioning in all cognitive subdomains except for versus different brain regions for the four identi­ verbal comprehension, while lower performance in fied CpG sites (Table S2). For example, left-hand FFT was associated with poorer proces­ cg07783800 showed a positive correlation between sing speed only (Table S4). blood and entorhinal cortex (r = 0.30) from the blood–brain DNAm comparison tool, but nega­ Mediation analysis tive correlations across blood versus parietal cor­ We examined the potential mediation (indirect tex and inferior temporal gyrus (r = -0.39) from effect) and found that the association between lower BECon. Despite partial overlap between frontal methylation at the cg07783800 site and lower cogni­ cortex and anterior prefrontal cortex across tive functioning was mediated by less optimal fine tools, cg05136476 showed a positive correlation motor development; the indirect effect was signifi­ between blood and frontal cortex (r = 0.11), but cant for the right (β = 0.027, 95% CI: 0.01, 0.05, p = a negative correlation between blood and anterior 0.009) and for the left hand (β = 0.019, 95% CI: 0.01, prefrontal cortex (r = −0.23). These findings sug­ 0.03, p = 0.007). We did not observe a significant gest that DNAm patterns on FTT-associated direct effect of cg07783800 on cognition. CpGs vary across tissues (blood vs. brain tissue) and across specific brain regions. Candidate gene follow-up analysis In candidate gene follow-up analysis, none of the Gene expression profile across tissues CpGs annotated to SPTBN4 reached gene-level Next, we assessed gene expression levels across 53 significance after FDR correction (Table S5). tissues including blood and several brain regions from public GTEx data (Figure S2). The GNG4 gene showed higher expression in the ovary, pitui­ Replication analyses in INMA tary gland, and brain, but with low specificity in We sought to replicate the four top CpG sites brain regions. PRKCD and DYRK1A were highly identified in GenR in an independent birth cohort. expressed across different tissues, with lower but Lower methylation at the Cg07783800 site (see consistent expression in brain-related tissues. Figure 2) was associated with a lower FTT score in both the left- and right-hand in GenR (βleft = Methylation quantitative trait loci (mQTL) 0.16, SEleft = 0.03, pleft = 1.70E–07; βright = 0.17, Of the four significant CpGs identified at birth, SEright = 0.03, pright = 5.11E–08) and was signifi­ three (cg05136476 in PRKCD, cg16705073 in cantly associated with right-hand FTT in the ST3GAL2, cg07638797 in the promoter of same direction in INMA (β = 0.12, SE = 0.05, p = DYRK1A) were all associated with several known 0.023). This means that a 1 SD decrease in methy­ mQTLs (Table S3). Only cg07783800 (GNG4) was lation at cg07783800 was associated with a 0.17 SD unrelated to known mQTLs. This was further sup­ lower right-hand FTT score in the discovery ported by a heritability tool based on data from cohort and a 0.12 SD lower FTT score in the monozygotic and dizygotic twins [58], showing replication cohort. No associations were replicated low additive genetic and low shared environmental with left-hand FTT in INMA (Table S6). influences on DNAm levels at this site (r = 0.04 and 0.10, respectively), but high nonshared envir­ Discussion onmental influences (r = 0.86). This is the first study to characterize epigenome- Associations between FTT and cognitive functions wide associations between DNA methylation at We found that slower FTT performance for both birth and fine motor ability in childhood. We high­ right- and left-hand at age 10 years were associated light three key findings. First, we identified 4 CpG 8 F. SERDAREVIC ET AL. Figure 2. Regional Association Plot for the Top DNA Methylation (CpG) Site cg07783800. Note. On the top graph, the x-axis depicts the position in base pair (bp) (hg19) for the region of GNG4. The y-axis indicates the strength of association from EWAS with FTT. Each circle represents a CpG site. Red dashed line within the graph indicates the genome-wide significance threshold. The regulatory information and correlation matrix of other CpG sites in the region with the top hit are shown below the x-axis. Color intensity marks the strength of the correlation and color indicates the direction of the correlation. The figure was made using the R package coMET. sites in cord blood that were prospectively asso­ cohort approaches to replicate epigenetic associa­ ciated with a child’s fine motor ability, as measured tions and reduce the risk of false positive discov­ by the Finger Tapping Test (FTT), after genome- eries [59]. Third, we found an association between wide correction in a large population-based birth slower FTT and worse cognitive function measured cohort. These FTT-associated CpGs map to genes at 14 years of age, supporting the functional rele­ that have been implicated in cognitive functions, vance of this neuropsychological measure of fine neurodevelopmental and neurodegenerative dis­ motor control to higher-order cognitive function eases. Second, one CpG annotated to the GNG4 in the general paediatric population. Furthermore, gene, cg07783800, was replicated in an independent slower FTT performance in childhood mediated the birth cohort, suggesting it might be a valid early association between lower DNAm at birth of our epigenetic marker of fine motor ability. Our find­ replicated CpG site and delayed cognition in ings highlight the importance of employing multi- adolescence. EPIGENETICS 9 Our EWAS at birth identified four CpGs after were mapped are expressed in the brain and have genome-wide correction. For the top site, been linked to neurodevelopmental and neurode­ cg07783800, we observed that the lower DNAm generative diseases, including autism and Down in this site was associated with less optimal fine syndrome. Impaired neuromotor development is motor performance on both hands, which was implicated in all of these neurodevelopmental supported by the replication for right hand results disorders. in an independent cohort. This CpG is annotated We observed a relationship between fine motor to the body of the GNG4 gene, a modulator and skills in childhood and cognitive performance in transducer of several transmembrane signalling adolescence. This converges with previous epide­ systems that play a role in haemostasis and gluca­ miological studies as well as other lines of evi­ gon response. It has been suggested that its expres­ dence, for example neuroimaging studies, which sion variability in utero influences cognitive have shown that maturation of the dorsolateral trajectories in ageing [52] GNG4 is mostly prefrontal cortex and cerebellum may underlie expressed in the cerebellum, hypothalamus, hypo­ both motor experience and active exploration of physis (pituitary gland), amygdala, basal ganglia, the world shaping these cognitive functions [62]. and cortex. In addition, GNG4 is expressed in the In samples of typically developing children, neu­ ovaries and may play a role in the hypothalamus- romimaging studies have also shown that regions pituitary-ovarian axis. GNG4 expression changes of the brain that underpin motor functioning are over time, with the highest expression during foe­ among the first to mature [63]. While difficult to tal development, decreasing with ageing [52]. tease apart the directionality of associations, sev­ Interestingly, cg07783800 was unrelated to known eral researchers have posited that fine motor abil­ mQTLs, and as such unlikely to be under strong ities may precede the development of cognitive genetic control – a finding further supported by abilities, starting with Piaget’s theory of develop­ twin data: variation in this site was mainly mental stages and recently supported by explained by non-shared environmental influ­ a longitudinal cross-lagged study [7,8]. It is possi­ ences. Future studies are needed to elucidate ble that epigenetic mechanisms partly explain the whether DNAm at this site reflects and potentially association between less optimal fine development propagates the influence of prenatal environmen­ and cognitive functioning. In line with this tal exposures on fine motor development. hypothesis, we found that DNAm levels at birth The other three top CpGs identified in GenR of our replicated CpG (cg07783800) prospectively did not replicate in INMA. We thus only briefly associated with cognition in adolescence through mention them here. The CpG cg05136476 is fine motor development. Specifically, slower FTT located in the PRKCD gene, which regulates the mediated the association between lower DNAm at processing of the amyloid precursor protein birth and poorer cognitive functioning during (APP), a central element in Alzheimer’s disease adolescence. pathophysiology; however, PRKCD has also been Overall, these findings point to GNG4 methy­ implicated in ASD [60]. Cg07638797 is annotated lation as a promising candidate for future studies to the promotor of DYRK1A, a gene localized in investigating the development of fine motor abil­ the Down syndrome critical region of chromo­ ities. It will be important in future to test some 21. This gene plays a critical function in whether GNG4 methylation may be causally the central nervous system during development related to motor and cognitive function, through and ageing. It controls the differentiation of pre­ for example the use of advanced causal inference natally formed neurons and has effects on synaptic approaches in humans (e.g., Mendelian randomi­ plasticity. Cg16705073 is annotated to ST3GAL2 zation) as well as experimental models (e.g., epi­ gene products sialyltransferase, which are largely genetic editing). Furthermore, it will be useful to responsible for ganglioside terminal α2–3 sialyla­ examine whether information on GNG4 methyla­ tion in the brain, synthesizing the major brain tion may add predictive value on top of known gangliosides GD1a and GT1b [61]. Taken alto­ risk factors for poor motor development. Of gether, all genes to which the four CpG sites note, the identified effect sizes were small, and 10 F. SERDAREVIC ET AL. it is unlikely that information on single CpG sites relevant brain regions, and the reliance on mainly will sufficiently improve predictive performance, adult post-mortem samples, which may not gen­ in the same way that single genetic variants are eralize to neurodevelopmental processes such as insufficient to predict complex traits or disor­ fine motor ability and cognition in childhood. ders. However, the use of polyepigenetic scores We also note that peripheral DNAm patterns (also referred to as methylation risk scores – akin may still be useful and valid markers even in the to polygenic risk scores) may help in future to absence of significant cross-tissue concordance. To capturing broader DNAm patterns associated illustrate, similarly to our study, DNAm at the with motor ability and help improve predictive cg05575921 site in blood (annotated to the power. AHRR) – a strong, widely replicated biomarker of smoking exposure which is associated with poor cognitive function and structural brain integ­ Strengths and limitations rity in adults [64] – shows null-to-weak positive This study presents a number of strengths, includ­ associations with DNAm in the brain using the ing its prospective design, objective performance- blood–brain comparison tool, whereas it shows based measures of child fine motor ability, and the negative correlations with the brain regions tagged replication in an independent cohort. However, by the BECon tool. Finally, this study focused on several limitations should be noted. Since subjects DNA methylation, which is one of several types of included in the discovery phase were not repre­ epigenetic mechanisms (e.g., histone or RNA mod­ sentative of the whole cohort concerning maternal ifications), which may contribute to motor devel­ education and ethnicity, the generalizability of opment and the pathogenesis of motor findings from the current study may be limited. diseases [65]. However, we replicated the top CpG in an inde­ pendent cohort (INMA), where the education of mothers is quite different than in GenR. While not Conclusions impossible, reverse causality at this age is unlikely to explain our results, as hand fine motor ability To our knowledge, we report the first EWAS of only manifests at later stages of child development. fine motor ability. At four CpGs, lower DNAm at We could not perform causal inference analysis as birth was prospectively associated with poorer genome-wide association studies of motor devel­ fine motor ability in childhood, as indexed by opment are unavailable. In addition, epigenetic slower performance on the finger tapping test. mechanisms are tissue-specific, and the functional All genes corresponding to the identified CpGs relevance of blood-based DNAm for brain-based have been involved in brain pathways regulating phenotypes, such as motor ability, is unclear. As neurodevelopment and neurodegeneration. One we do not have individual gene expression data in site annotated to the GNG4 gene, cg07783800, Generation R, we utilized publicly available tools showed consistent results for both hands in the to explore the expression of the annotated genes in Generation R study and was replicated for the other relevant tissues (e.g., the brain). The sample right hand in the INMA study, showing potential sizes of these tissue databases are often small, as an epigenetic marker of fine motor develop­ limited to adults, and available post-mortem tis­ ment. Furthermore, lower DNAm at cg07783800 sue; therefore, caution is warranted when inter­ was associated with poor cognitive functioning preting these results. through non-optimal fine motor development. Regarding the blood–brain correlations, it is Our findings highlight the importance of employ­ important to note that although currently available ing multi-cohort approaches to replicate epige­ blood–brain tools are helpful in exploring cross- netic associations and reduce the risk of false tissue concordance, they have several caveats that positive discoveries. The results from this study limit their interpretation, including the use of contribute to a better understanding of epigenetic small samples with mixed clinical presentation factors associated with fine motor development in and incomplete phenotyping, limited data on childhood and later cognitive function. EPIGENETICS 11 Acknowledgments 00241, Fundació La marató de TV3 (090430), EU Commission (261357-MeDALL: Mechanisms of the Development of The Generation R Study is conducted by the Erasmus Medical ALLergy), and European Research Council (268479-BREATHE: Center in close collaboration with the School of Law and BRain dEvelopment and Air polluTion ultrafine particles in Faculty of Social Sciences at Erasmus University Rotterdam, scHool childrEn). the Municipal Health Service Rotterdam area, the Rotterdam Homecare Foundation, and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond (STAR-MDC), Rotterdam. We gratefully acknowledge the contribution of children and par­ References ents, general practitioners, hospitals, midwives, and pharma­ cies in Rotterdam. The generation and management of the [1] Seo SM. The effect of fine motor skills on handwriting Illumina 450K methylation array data (EWAS data) for the legibility in preschool age children. J Phys Ther Sci. Generation R Study was executed by the Human Genotyping 2018;30(2):324–327. Facility of the Genetic Laboratory of the Department of [2] Emck C, Bosscher R, Beek P, et al. Gross motor per­ Internal Medicine, Erasmus MC, the Netherlands. We thank formance and self-perceived motor competence in chil­ Mr. Michael Verbiest, Ms. Mila Jhamai, Ms. Sarah Higgins, dren with emotional, behavioural, and pervasive Mr. Marijn Verkerk and Dr. Lisette Stolk for their help in developmental disorders: a review. Dev Med Child creating the EWAS database. We thank Dr. A.Teumer for his Neurol. 2009;51(7):501–517. work on the quality control and normalization scripts. The [3] Martin R, Tigera C, Denckla MB, et al. Factor structure of general design of the Generation R Study is made possible by paediatric timed motor examination and its relationship financial support from the Erasmus Medical Center with IQ. Dev Med Child Neurol. 2010;52(8):e188–194. Rotterdam, Erasmus University Rotterdam, the Netherlands [4] Alamiri B, Nelson C, Fitzmaurice GM, et al. Organization for Health Research and Development and the Neurological soft signs and cognitive performance in Ministry of Health, Welfare and Sport. early childhood. Dev Psychol. 2018;54(11):2043–2052. [5] Piek JP, Dawson L, Smith LM, et al. The role of early fine and gross motor development on later motor and Disclosure statement cognitive ability. Hum Mov Sci. 2008;27(5):668–681. [6] Bloch MH, Sukhodolsky DG, Dombrowski PA, et al. No potential conflict of interest was reported by the authors. Poor fine-motor and visuospatial skills predict persis­ tence of pediatric-onset obsessive-compulsive disorder into adulthood. J Child Psychol Psychiatr. 2011;52 Funding (9):974–983. DOI:10.1111/j.1469-7610.2010.02366.x The EWAS data was funded by a grant from the Netherlands [7] Piaget J. The development of time concepts in the Genomics Initiative (NGI)/Netherlands Organisation for child. Proc Annu Meet Am Psychopathol Assoc. Scientific Research (NWO) Netherlands Consortium for 1954-1955;34–44. discussion, 45-55. Healthy Aging (NCHA; project nr. 050-060-810), funds from [8] Mohring W, Frick A. Touching up mental rotation: the Genetic Laboratory of the Department of Internal Medicine, effects of manual experience on 6-month-old infants’ Erasmus MC, and a grant from the National Institute of Child mental object rotation. Child Dev. 2013;84(5):1554–1565. and Human Development (R01HD068437). HT was supported [9] Hubel KA, Yund EW, Herron TJ, et al. Computerized by a grant of the Dutch Ministry of Education, Culture, and measures of finger tapping: reliability, malingering and Science and the Netherlands Organization for Scientific traumatic brain injury. J Clin Exp Neuropsychol. Research (NWO grant No. 024.001.003, Consortium on 2013;35(7):745–758. Individual Development). FS was supported by a Royal [10] Rabin LA, Barr WB, Burton LA. Assessment practices Netherlands Academy of Science and Art (KNAW) Van of clinical neuropsychologists in the United States and Leersum fellowship. ML is supported by the scholarship from Canada: a survey of INS, NAN, and APA Division 40 the China Scholarship Council (201706990036). CC is supported members. Arch Clin Neuropsychol. 2005;20(1):33–65. by the European Research Council (ERC) under the European [11] Damme KSF, Schiffman J, Ellman LM, et al. Union’s Horizon 2020 Research and Innovation Programme Sensorimotor and activity psychosis-risk (SMAP-R) under grant agreements No 101039672 (TEMPO) and No scale: an exploration of scale structure with replication 848158 (EarlyCause). This project received funding from the and validation. Schizophr Bull. 2021;47(2):332–343. European Union’s Horizon 2020 research and innovation pro­ [12] Behere A, Shahani L, Noggle CA, et al. Motor functioning gramme (733206, LifeCycle).The epigenetic studies in INMA in autistic spectrum disorders: a preliminary analysis. were mainly funded by grants from Instituto de Salud Carlos J Neuropsychiatry Clin Neurosci. 2012;24(1):87–94. III (Red INMA G03/176, CB06/02/0041, CP18/00018), Spanish [13] Mostofsky SH, Rimrodt SL, Schafer JGB, et al. Atypical Ministry of Health (FIS-PI04/1436, FIS-PI08/1151 including motor and sensory cortex activation in FEDER funds, FIS-PI11/00610, FIS-FEDER-PI06/0867, FIS- attention-deficit/hyperactivity disorder: a functional FEDER-PI03-1615) Generalitat de Catalunya-CIRIT 1999SGR magnetic resonance imaging study of simple sequential 12 F. SERDAREVIC ET AL. finger tapping. Biological Psychiatry. 2006;59(1):48–56. [28] Trasti N, Vik T, Jacobsen G, et al. Smoking in preg­ DOI:10.1016/j.biopsych.2005.06.011 nancy and children’s mental and motor development at [14] Valera EM, Spencer RMC, Zeffiro TA, et al. Neural sub­ age 1 and 5 years. Early Hum Dev. 1999;55(2):137–147. strates of impaired sensorimotor timing in adult [29] Golding J, Emmett P, Iles-Caven Y, et al. A review of attention-deficit/hyperactivity disorder. Biological environmental contributions to childhood motor skills. Psychiatry. 2010;68(4):359–367. DOI:10.1016/j.biopsych. J Child Neurol. 2014;29(11):1531–1547. 2010.05.012 [30] Sammallahti S, Tiemeier H, Louwen S, et al. Fetal–placen­ [15] Travers BG, Bigler ED, Duffield TC, et al. Longitudinal tal blood flow and neurodevelopment in childhood: popu­ development of manual motor ability in autism spec­ lation-based neuroimaging study. Ultrasound Obstet trum disorder from childhood to mid-adulthood Gynecol. 2021;58(2):245–253. DOI:10.1002/uog.22185 relates to adaptive daily living skills. Dev Sci. 2017;20 [31] Barker ED, Walton E, Cecil CAM. AnnuaL research (4):e12401. DOI:10.1111/desc.12401 review: DNA methylation as a mediator in the association [16] Liou WC, Chan L, Hong CT, et al. Hand fine motor between risk exposure and child and adolescent skill disability correlates with dementia severity. Arch psychopathology. J Child Psychol Psychiatry. 2018;59 Gerontol Geriatr. 2020;90:104168. (4):303–322. [17] Plomin R, Haworth CMA, Davis OSP. Common disorders [32] Martin EM, Fry RC. Environmental Influences on the are quantitative traits. Nat Rev Genet. 2009;10(12):872–878. Epigenome: Exposure- Associated DNA methylation in [18] Lundström S, Chang Z, Råstam M, et al. Autism spec­ human populations. Annu Rev Public Health. 2018;39 trum disorders and autisticlike traits: similar etiology (1):309–333. in the extreme end and the normal variation. Arch Gen [33] Neumann A, Walton E, Alemany S, et al. Association Psychiat. 2012;69(1):46–52. DOI:10.1001/archgenpsy between DNA methylation and ADHD symptoms chiatry.2011.144 from birth to school age: a prospective meta-analysis. [19] Larsson H, Anckarsater H, Råstam M, et al. Childhood Transl Psychiatry. 2020;10(1):398. DOI:10.1038/ attention-deficit hyperactivity disorder as an extreme of s41398-020-01058-z a continuous trait: a quantitative genetic study of 8,500 [34] Rijlaarsdam J, Cecil CAM, Relton CL, et al. Epigenetic twin pairs. J Child Psychol Psyc. 2012;53(1):73–80. profiling of social communication trajectories and [20] Bernard JA, Mittal VA. Updating the research domain co-occurring mental health problems: a prospective, criteria: the utility of a motor dimension. Psychol Med. methylome-wide association study. Dev Psychopathol. 2015;45(13):2685–2689. 2022;34(3):1–10. [21] Mittal VA, Wakschlag LS. Research domain criteria [35] Bromer C, Marsit CJ, Armstrong DA, et al. Genetic (RDoC) grows up: strengthening neurodevelopment and epigenetic variation of the glucocorticoid receptor investigation within the RDoC framework. J Affect (NR3C1) in placenta and infant neurobehavior. Dev Disord. 2017;216:30–35. Psychobiol. 2013;55(7):673–683. [22] Francks C, Fisher SE, Marlow AJ, et al. Familial and [36] Marsit CJ, Maccani MA, Padbury JF, et al. Placental genetic effects on motor coordination, laterality, and 11-beta hydroxysteroid dehydrogenase methylation is reading-related cognition. Am J Psychiatry. 2003;160 associated with newborn growth and a measure of (11):1970–1977. DOI:10.1176/appi.ajp.160.11.1970 neurobehavioral outcome. PloS one. 2012;7(3):e33794. [23] Williams LRT, Gross JB. Heritability of motor skill. doi:10.1371/journal.pone.0033794. Acta geneticae medicae et gemellologiae: twin research. [37] Rygiel CA, Dolinoy DC, Bakulski KM, et al. DNA 1980;29(2):127–136. methylation at birth potentially mediates the associa­ [24] Mountford HS, Hill A, Barnett AL, et al. Genome-wide tion between prenatal lead (Pb) exposure and infant association study of motor coordination. Front Hum neurodevelopmental outcomes. Environ Epigenet. Neurosci. 2021;15:669902. 2021;7(1):dvab005. DOI:10.1093/eep/dvab005 [25] Doney R, Lucas BR, Jones T, et al. Fine motor skills in [38] Hüls A, Wedderburn CJ, Groenewold NA, et al. children with prenatal alcohol exposure or fetal alcohol Newborn differential DNA methylation and subcortical spectrum disorder. J Dev Behav Pediatr. 2014;35 brain volumes as early signs of severe neurodevelop­ (9):598–609. mental delay in a South African birth cohort study. [26] Darling AL, Rayman MP, Steer CD, et al. Association World J Biol Psychiatry. 2022;2022(8):1–12. DOI:10. between maternal vitamin D status in pregnancy and 1080/15622975.2021.2016955 neurodevelopmental outcomes in childhood: results [39] Roebers CM, Röthlisberger M, Neuenschwander R, from the avon longitudinal study of parents and chil­ et al. The relation between cognitive and motor per­ dren (ALSPAC). Br J Nutr. 2017;117(12):1682–1692. formance and their relevance for children’s transition [27] Crous-Bou M, Gascon M, Gispert JD, et al. Impact of to school: a latent variable approach. Hum Mov Sci. urban environmental exposures on cognitive performance 2014;33:284–297. and brain structure of healthy individuals at risk for [40] Suggate S, Stoeger H. Fine motor skills enhance lexical Alzheimer’s dementia. Environ Int. 2020;138:105546. processing of embodied vocabulary: a test of the EPIGENETICS 13 nimble-hands, nimble-minds hypothesis. Q J Exp disorder highlights a novel locus at 10q24.32 and Psychol. 2017;70(10):2169–2187. a significant overlap with schizophrenia. Mol Autism. [41] Kooijman MN, Kruithof CJ, van Duijn CM, et al. The 2017;8(1):21. Generation R Study: design and cohort update 2017. [54] Peyrot WJ, Price AL. Identifying loci with different allele Eur J Epidemiol. 2016;31(12):1243–1264. DOI:10.1007/ frequencies among cases of eight psychiatric disorders s10654-016-0224-9 using CC-GWAS. Nature Genet. 2021;53(4):445–454. [42] Guxens M, Ballester F, Espada M, et al. Cohort Profile: [55] Nalls MA, Blauwendraat C, Vallerga CL, et al. the INMA–INfancia y Medio Ambiente–(Environment Identification of novel risk loci, causal insights, and heri­ and Childhood) Project. Int J Epidemiol. 2012;41 table risk for Parkinson’s disease: a meta-analysis of gen­ (4):930–940. DOI:10.1093/ije/dyr054 ome-wide association studies. Lancet Neurol. 2019;18 [43] Lehne B, Drong AW, Loh M, et al. A coherent approach (12):1091–1102. DOI:10.1016/S1474-4422(19)30320-5 for analysis of the Illumina HumanMethylation450 [56] Wendt FR, Pathak GA, Lencz T, et al. Multivariate BeadChip improves data quality and performance in genome-wide analysis of education, socioeconomic sta­ epigenome-wide association studies. Genome Bio. tus and brain phenome. Nat Human Behav. 2021;5 2015;16(1):1–12. DOI:10.1186/s13059-015-0600-x (4):482–496. [44] Aryee MJ, Jaffe AE, Corrada-Bravo H, et al. Minfi: [57] Yoo SW, Motari MG, Susuki K, et al. Sialylation reg­ a flexible and comprehensive bioconductor package ulates brain structure and function. Faseb J. 2015;29 for the analysis of infinium DNA methylation (7):3040–3053. DOI:10.1096/fj.15-270983 microarrays. Bioinformatics. 2014;30(10):1363–1369. [58] Hannon E, Knox O, Sugden K, et al. Characterizing DOI:10.1093/bioinformatics/btu049 genetic and environmental influences on variable DNA [45] White T, Muetzel RL, El Marroun H, et al. Paediatric methylation using monozygotic and dizygotic twins. population neuroimaging and the Generation R Study: PLoS Genet. 2018;14(8):e1007544. DOI:10.1371/jour the second wave. Eur J Epidemiol. 2018;33(1):99–125. nal.pgen.1007544 DOI:10.1007/s10654-017-0319-y [59] Rijlaarsdam J, Pappa I, Walton E, et al. An [46] Lopez-Vicente M, Lamballais S, Louwen S, et al. White epigenome-wide association meta-analysis of prenatal matter microstructure correlates of age, sex, handedness maternal stress in neonates: a model approach for and motor ability in a population-based sample of 3031 replication. Epigenetics. 2016;11(2):140–149. DOI:10. school-age children. Neuroimage. 2021;227:117643. 1080/15592294.2016.1145329 [47] Lezak MD. Neuropsychological assessment, 3rd ed. [60] Park YH, Hodges A, Risacher SL, et al. Dysregulated Fc New York, NY, US: Oxford University Press; 1995. gamma receptor-mediated phagocytosis pathway in [48] Kaufman AS, Raiford SE, Coalson DL. Intelligent testing Alzheimer’s disease: network-based gene expression with the WISC-V. New Jersey: John Wiley & Sons; 2015. analysis. Neurobiol Aging. 2020;88:24–32. [49] Gervin K, Salas LA, Bakulski KM, et al. Systematic [61] Sturgill ER, Aoki K, Lopez PH, et al. Biosynthesis of the evaluation and validation of reference and library selec­ major brain gangliosides GD1a and GT1b. Glycobiology. tion methods for deconvolution of cord blood DNA 2012;22(10):1289–1301. DOI:10.1093/glycob/cws103 methylation data. Clin Epigenetics. 2019;11(1). [62] Diamond A. Close interrelation of motor development [50] Min JL, Hemani G, Davey Smith G, et al. Meffil: and cognitive development and of the cerebellum and efficient normalization and analysis of very large prefrontal cortex. Child Dev. 2000;71(1):44–56. DNA methylation datasets. Bioinformatics. 2018;34 [63] Casey BJ, Tottenham N, Liston C, et al. Imaging the (23):3983–3989. developing brain: what have we learned about cognitive [51] Benjamini Y, Hochberg Y. Controlling the false discovery development? Trends Cogn Sci. 2005;9(3):104–110. rate: a practical and powerful approach to multiple testing. [64] Corley J, Cox SR, Harris SE, et al. Epigenetic signatures J Royal Stat Soc Ser B (Methodological). 1995;57 of smoking associate with cognitive function, brain (1):289–300. structure, and mental and physical health outcomes [52] Bonham LW, Evans DS, Liu Y, et al. Neurotransmitter in the Lothian Birth Cohort 1936. Transl Psychiatry. pathway genes in cognitive decline during aging: evi­ 2019;9(1):248. DOI:10.1038/s41398-019-0576-5 dence for GNG4 and KCNQ2 Genes. Am J Alzheimers [65] Yang Y, Yamada T, Hill KK, et al. Chromatin remodel­ Dis Other Demen. 2018;33(3):153–165. ing inactivates activity genes and regulates neural [53] Anney RJL, Ripke S, Anttila V, et al. Meta-analysis of coding. Science. 2016;353(6296):300–305. DOI:10. GWAS of over 16,000 individuals with autism spectrum 1126/science.aad4225