Multi-omics Neuroimaging Genetics Study

Part 1: Prefrontal Volume Mediates Effect of COMT Polymorphism on Interference Resolution Capacity in Healthy Male AdultsObjective:There exist gender differences in the modulation of catechol-Omethyltransferase(COMT)Val158Met polymorphism on cognitive performance;however,the underlying gene-anatomy-cognition pathways remain unknown.Here we hypothesize that prefrontal volume may mediate the modulation of COMT Val158 Met polymorphism on interference resolution capacity in a gender-dependent manner.Subjects and Methods:1.Interference Resolution CapacityThis study collected a total of 261 healthy young subjects(mean age: 22.76 ± 2.5 years;119 males).There is no significant difference(P > 0.05)in age between the male and female volunteers.The Stroop color-word block-design task consisted of 4 alternating blocks: word reading blocks and color reading blocks.During the task,the participants were instructed to select the colorword semantic meaning or color by pressing the corresponding key.Finally,the accuracy and reaction time of each block would be recorded.2.GenotypingWe genotyped the COMT rs4680(G?A)in each subject using PCR and LDR method.3.NeuroimagingThe preprocessing procedures for high-resolution structural images included segmentation of grey and white matter,spatial normalization and smoothing that were performed using the VBM8 implemented in SPM8,and those for f MRI data included slice timing,realignment,spatial normalization,resampling,smooth and filter that were performed using DPARSFA.The high-resolution structural images were used to calculate grey matter volume(GMV)maps and f MRI data were were used to calculate functional measure maps.4.Statistical analysisThe multiple regression was applied to test the GMV of brain regions correlated with interference resolution capacity.A full factor ANOVA analysis(two-way)was used to test interaction effect between COMT and gender on brain structure and function.Finally,mediation analysis was applied to assess the gene-anatomycognition pathways in males and females,respectively.Results:1.Interference resolution capacity differences between COMT genotypesThe 2-way ANOVA showed a significant gene × gender interaction on reaction delay of word interference(RD-W)(P = 0.023).Post hoc analysis revealed that the male Val/Val subjects had a greater RD-W(poorer interference resolution capacity)than the male Met carrier subjects.2.GMV differences between COMT genotypesIn the whole brain analysis,we found a significant COMT × gender interaction on the GMV in the left lateral frontal pole,although there was no significant main effect of COMT.Within the executive control network mask which involved in interference resolution capacity,we found a much more significant COMT × gender interaction.Post hoc comparisons showed that Val/Val males exhibited decreased GMV than Met carrier males.3.Functional measure differences between COMT genotypesThe left lateral FP cluster with significant COMT × gender interaction on GMV in the whole brain analysis was extracted as the ROI for analyses of the following functional measures.For the ALFF and Re Ho of the left lateral FP,we did not find any significant main effect of COMT genotype4.Association between brain GMV and interference resolution capacityThe whole brain voxel-wise partial correlation analysis showed that males and females showed significant correlations between interference resolution capacity and GMV in different brain regions.The interference resolution capacity was mainly correlated with the dorsal anterior cingulate cortex in females,but with the prefrontal cortex in males.5.Mediation analysisThe mediation analysis was performed in males and females,respectively.In male subjects,we found a significant indirect effect suggesting that the GMV of the left frontal pole may mediate the association between COMT and interference resolution capacity.Conclusion:Using a mediation analysis,we identified a gene-anatomy-cognition pathway to explain how COMT Val158 Met polymorphism affects interference resolution capacity via modulating the GMV of the left lateral FP in healthy male subjects.This pathway may not only provide us a better understanding the genetic modulation of cognitive control in healthy subjects,but also provide a reference frame for investigating the mechanisms of executive impairments in brain disorders.Part 2: Neurobiological substrates underlying the effect of genomic risk for depression on the conversion of amnestic mild cognitive impairmentBackground:The amnestic mild cognitive impairment(a MCI)is a state of cognitive deficit that is not severe enough to fulfill the criteria of dementia and shows a much higher probability of developing into Alzheimer's disease(AD).Identifying biological measures with the potential to predict the conversion from a MCI to AD is clinically important for early interventions of AD.Among these measures,a lifetime history of major depressive disorder(MDD),the presence of depressive symptom,or the coexistence of a diagnosis of MDD has been found to increase the conversion risk from a MCI to AD,despite of non-significant findings.However,few studies have investigated why depression could increase the conversion risk from a MCI to AD.Objective:Here,we aimed to investigate the predictive effect of the polygenic risk scores of MDD-specific genetic variants(PRSs MDD)on the conversion from non-depressed a MCI to AD,and its underlying neurobiological mechanisms.Subjects and Methods:1.PRS constructionThe PRS calculation requires a discovery sample and a target sample.The discovery sample was used to identify effect size of a set of genetic variants that were nominally associated with the disease status at a predefined P value.And then the PRS was calculated for each subject in the target sample to estimate cumulative genetic risk of this subject for the disease.Summary statistics data of the Psychiatric Genomics Consortium(PGC)and International Genomics of Alzheimer's Project(IGAP)were used as the discovery samples to calculate PRSMDD,PRSAD,PRSs MDD,PRSs AD,and PRSs MDD+AD.The target sample included 322 non-depressed(with a geriatric depression scale < 6)a MCI patients provided by the first stage of Alzheimer's disease Neuroimaging Initiative(ADNI-1).According to the follow-up results in July 2015 released by ADNI,a MCI subjects were divided into the conversion(a MCI-C,N= 187)and stable(a MCI-S,N=135)groups.Here,the final diagnosis for patients with follow-up loss was based on the last clinical evaluation.The sample was used to test whether PRS could predict the conversion from a MCI to AD(up to 108 months follow up).2.PRS predict a MCI convert to ADHere the logistic regression,Cox proportional hazard model,LD score regression and colocalization analysis by estimating the Bayesian posterior probability were used to test the predictive effect of each PRS measurement on the conversion of a MCI;The voxel-wise multiple regression analysis was performed to identify brain regions whose GMVs were significantly correlated with PRS using SPM8,while controlling for the effect of neuroimaging sites.Multiple comparisons were corrected using a voxel-level family wise error(FWE)method(Pc < 0.05,cluster size > 200 voxels).The GMVs of brain regions with significant correlations with the PRS were extracted for further analysis.Mediation analysis and Medenlian Randomization were used to test the causal effect from PRS to brain to a MCI conversion.3.Fine-mapping MDD-specific genetic variants into genesThe identified PRSs MDD genetic variants were fine-mapped into genes using different strategies,and then enrichment and protein-protein interaction network analyses were performed to identify potential functions of these genes.Temporaland cell type-specific expression analyses were finally used to explore in which developmental periods and cell types these genes are significantly expressed in hippocampal tissue.Results:1.PRSs MDD predict the conversion from a MCI to ADWe found the PRSMDD coud predict the conversion from a MCI to AD.To exclude the possibility that the predictive effect of the PRSMDD on the conversion is driven by genetic variants common to MDD and AD,we only used genetic variants specific to MDD to calculate PRSs MDD by excluding common genetic variants of the two disorders.We found that the PRSs MDD could independently predict the conversion from a MCI to AD,and a MCI patients with high risk scores(65.38%)showed 16.25% higher conversion rate than those with low risk(49.13%).2.Neuroanatomy substrates underlying the prediction effect of PRSs MDDThe whole brain voxel-wise correlation analysis showed that the PRSs MDD was correlated with the left hippocampal volume.Mediation analysis and Mendelian randomization analysis demonstrated the causal interference from the PRSs MDD to left hippocampal volume and the conversion of a MCI.3.Functional annotation of PRSs MDD related genesThe MDD-specific genetic variants were mapped into genes using different strategies,and then enrichment analyses and protein-protein interaction network analysis revealed that these genes were involved in developmental process and amyloid-beta binding.4.Hippocampal temporal and cell type-specific expression analysisThey showed temporal-specific expression in the hippocampus in middle and late fetal developmental periods.Cell type-specific expression analysis of these genes demonstrated significant over-representation in the pyramidal neurons and interneurons in the hippocampus.Conclusion:In this study,we found that the PRSs MDD could independently predict the conversion from a MCI to AD,and the joint use of the PRSs MDD and PRSs AD could select a MCI patients with a much high risk for the conversion.The predictive effect of the PRSs MDD on the conversion may be mediated by the hippocampus via affecting its early developmental process and amyloid-beta binding.Part 3: Genetic-epigenetic-transcriptomic causal pathways for hippocampal volume in Alzheimer's diseaseBackground:The hippocampal atrophy is one of the prominent changes of Alzheimer's disease(AD),which is with great heritability.More and more evidence has demonstrated that regulatory elements play a key role in normal neurodevelopmental processes and heterogeneous pathogenesis of brain disorders,such as AD.Therefore,clarifying causal pathways(genetic-epigenetic-transcriptome)underlying hippocampal volume will improve our understanding about the neurobiological substrates of the hippocampal changes in the process of AD.Objective:Here,we aimed to investigate whose gene expression and Cp G site DNA methylation levels can affect the hippocampal volume based on blood and hippocampus tissue.Then the causal interference test was used to explore whether the effect of a regulatory SNP on hippocampal volume was mechanistically mediated through its impact on gene expression or DNA methylation.Finally,we explored associations of SNPs-DNA methylation-gene expression-hippocampal volume pathway in AD and healthy old men,respectively.Subjects and Methods:1.Association between molecular phenotype and hippocampal volumeWhole-genome sequenced data were generated from 818 European descent individuals of ADNI(128 AD,415 MCI,267 controls and 8 of uncertain diagnosis).Among these individuals,gene expression(N = 811),DNA methylation(N = 653)were derived from blood samples.After strict quality control,based on 585 subjects with both gene expression and hippocampal volume data and 607 subjects with both DNA methelation and hippocampal volume data,linear regression model was applied to test the association between gene expression/DNA methylation and mean volume of the bilateral hippocampus.Cis-e QTL(N = 735)and cis-m QTL(N = 604)mapping was performed using the nominal pass function.The mapping window was defined as 1 Mb up and downstream of the transcription start site/Cp G site position.The Bonferroni method was applied to correct for multiple comparisons in the cise QTL(P < 1.06e-6)and m QTL mapping(P < 7.0e-8).Integrating cis-e QTL,cism QTL and integrative SMR analysis was applied to validate the association between molecular phenotype and hippocampal volume.2.Causal interference testTo test the causal relationship between a regulatory SNP,genes expression and genes methylation,causal inference test(CIT)was applied in the 571 subjects with genotype,gene expression and gene methylation data to test the causal model(the effect of a regulatory SNP on genes expression is mediated by DNA methylation).Moreover,the significant causal pathway regulatory SNP-DNA methylation-genes expression model would be validated in the human brain hippocampus datasets using two sample Mendelian Randomsize.3.Causal pathway analysis in AD cases and controlsIt is well-known that the volumetric reduction in the hippocampus is the most prominent early pathological feature of AD.The cascading effect of regulatory SNP on hippocampal volume in molecular phenotype may have a role in pathogenesis of AD.Here two-way ANOVA and Pearson correlation were applied to explore whether the hippocampal volume-associated genes are differentially expressed and methylated in AD cases compared with controls.Results:1.Idenifying 18 genes whose expression and methylation were associated with hippocampal volumeA total of 122 overlapped genes were identified in linear regression model,indicating that their expression and methylation were associated with hippocampal volume.Of these genes,only 18 genes were validated using cis-e QTL,cis-m QTL and SMR boinformatic analyses.2.Causal relationship between regulatory SNP,gene expression and DNA methylation based on blood and hippocampus tissueThe availability of multi-omic data from the same individuals enabled us to go beyond overlapping gene analyses and to further investigate the causal relationship between gene expression and DNA methylation in the cascading effect of regulatory SNPs on hippocampal volume.This analysis was performed on 18 genes and their 43 regulatory SNPs,and thus resulted in 258 SNPs-expression-methylation association sets.Using Bonferroni correction with the casual inference test(P < 0.05/258),38 association sets were conformed to the methylation-mediated model,2 sets conformed to the transcription-mediated model,205 sets conformed to the independent model.Here three genes(ANKRD37,PCMT1 and SQRDL)were identified from the methylation-mediated model.After linkage disequilibrium(LD)analysis(R2>0.8),thus 4 unique pathways were found in the methylation-mediated model(causal model).Only rs10000869-ANKRD37 cg26741686 DNA methylationANKRD37 gene expression causal model pathway could be validateda based on human hippocampus tissue.3.Dysregulation of ANKRD37 in AD casesTo further explore the association between causal SNPs-methylationexpression-hippocampal volume pathway and disease status,we examined the pathway of ANKRD37 in healthy controls(N=193)and AD cases(N=83),respectively.We found that with the increasing allele of ANKRD37 rs10000869,ANKRD37 cg26741686 DNA methylation increased and ANKRD37 gene expression increased,which resulted in the decrease of the hippocampal volume.Conclusion:Based on both blood and hippocampus tissue,rs10000869-ANKRD37 cg26741686 DNA methylation-ANKRD37 gene expression-hippocampal volume causal pathway was validated.The causal SNP rs10000869 will increase ANKRD37 cg26741686 DNA methylation and ANKRD37 gene expression,which results in the decrease of the hippocampal volume.However,it showed more significant effect in AD cases compared to healthy controls.Part 4: Identifying vulnerable periods for the effects of global urbanization on adolescent brain development and mental health using remote sensing satellite imageryBackground:China has experienced rapid urbanization since the reform and opening up.The urbanization rate has increased from 17.9% in 1978 to 58.5% in 2017.As expected in 2030,Chinese urbanization rate will reach 71%.People living in urban will exceed 1 billion,and one in every eight people in the world will live in Chinses city.However,the rapid urbanization and industrialization have brought a varity of health challenges with the rich and modern civilization achievements to the people of the world.For example,mental illness is rising,such as schizophrenia,depression,and bipolar disorder,because of environmental pollution and lifestyle changes.Although all these negative changes can be attributed to rapid changes of the urban social environments,the neural mechanism underlying these changes are still unknown.In addition,most of current researches on global urbanization are limited to subjective questionnaire surveys,which can no longer meet the increasingly complex global urban development.Here our study characterizes global urbanization using multitemporal remote sensing satellite imagery and explore the neural mechanisms for the effect of golable urbaniazation on mental health based on Chinese and European adolescents.Objective:Here,we aimed to identify vulnerable periods for the effects of global urbanization on the adolensent brain development and mental health using remote sensing satellite imagery based on large multi-center Chinese and European datasets.Subjects and Methods:1.Urbanization index constructed by remote sensing satellite imageryUrbanization is generally analyzed with two principal dimensions: spatial manifestation of urban expansion and dynamics of demography.GHSL is a framework to produce global spatial information on population and on the physical size of settlements on the planet,of which population density(GHS-POP)is referred as degree of urbanization based on a classical theory of measuring urbanization.Product GHSL,Population Grid(P2016)for the epochs 1990-2000-2015 from Google earth engine(GEE)is applied in the participants from CHIMGEN and IMAGEN at 250m×250m resolution.Most of the following satellite images are extracted from GEE to measure different aspect of urbanization based on the acquired individual geography data from two projects.Here nighttime lights,normalized difference vegetation index(NDVI),normalized difference water index(NDWI),normalized difference building index(NDBI)were extracted from GEE platform.Global land cover mapping has been the most important variable on environmental changes and neighborhood surrounding resources.Here Climate Change Initiative Land Cover dataset(CCI-LC)from European Space Agency platform is used to extract land cover classes from 1992 to 2015.Here short trees,cropland and urban and built-up are selected.Similarly,based on the 1,202 longitudinal adolescent data of the European Imaging Genetics Consortium(IMAGEN)(incorporated 14-year-old data and follow up 2 19-year-old data),the same method was used to extract the above-mentioned population density indicators and 7 types of multi-temporal remote sensing satellite indicators.Confirmatory factor analysis(CFA)was applied for the above 7 remote sensing satellite variables to generate Chinese and European urbanization index,respectively.To consolidate the foundation of using remote sensing satellite to measure urbanicity,the urbanization index was correlated with population density from GHSL in rural,town,city and overall,respectively.2.Identifying vulnerable periods for the effect of urbanization index on structural brain developmentFirstly,voxel-wise multiple reggression was used to test the effect of overall exposure urbanicity before 18 years on structural brain in CHIMGEN.And then agesliding window voxel-based correlation was used to further identify vulnerable periods of the effect of Chinese urbanization index on adolescent structural brain development.Finally,IMAGEN dataset was used to validate the effect of European urbanization index on brain.3.Identifying vulnerable periods for the effect of urbanization index on neuropsychological assessments and mental healthWe aimed to further test the effect of Chinese and European urbanization index on a series of neuropsychological assessments(such as the episodic verbal memory,working memory,information processing speed,social cognition and executive control)and mental health(such as depression,state anxiery and trait anxiety).Similarly,the age-sliding window correlation was used to identify vulnerable periods for the effects of Chinese and European urbanization index on neuropsychological assessments and mental health.4.Urbanization-brain-health outcome pathwayMediation analysis was applied to assess the urbanization-brain-health outcome pathways in Chinese and European population.Results:1.Globle urbanization indexIn 1990,2000 and 2015,the Chinese and European urbanization index were significantly positively correlated with population density from GHSL datasets(the most significant r value is 0.72).These results demonstrated that remote sensing satellite data could be used as an effective measurement to assess the globle urbanization.2.Urbanization index was negatively correlated with left MSFC and positively correlated with cerebellum vermisVoxel-based correlation results demonstrated that overall exposure Chinese urbanization index was negatively correlated with left medial superior frontal cortex(MSFC)volume and positively correlated with cerebellar vermis(FWE correction,P < 0.05,cluster size > 100 voxels).Based on the IMAGEN follow up 2 19 years dataset,the overall exposure European urbanization index was also negatively correlated with left MSFC volume and positively correlated with cerebellar vermis under the same threshold.These results demonstrated that there were no differences between Chinese and European population in the effect of urbarnicy on brain.3.Identifying vulnerable periods for the effect of urbanization index on structural brain developmentAge sliding window voxel-based correlation analysis showed that Chinese urbanization index affected the left MSFC volume much more significant in the 11th-14 th age window of adolescent period and the cerebellum vermis volume much more significant in the 4th-7th age window of childhood period.Based on the IMAGEN baseline 14 years datasets,European urbanization index was also negatively correlated with left MSFC volume and positively correlated with cerebellum vermis under the same threshold.4.Urbanization index was positively correlated with perspective taking ability,depression and state anxietyIn the neuropsychological assessment and mental health analysis,we found that Chinese urbanization index was positively correlated with perspective taking of social cognition(accuracy: r = 0.124,P =0.002;reaction time: r =-0.245,P = 6.01e-7),depressive score(r =0.209,P = 1.44e-5)and state anxiety(r = 0.132,P = 0.001)of mental health.The correlation between urbanicity and perspective taking performance was strongest during mid-adolescence(age 13 to 16 years in accuracy and age 14 to 18 years in reaction time).The correlations between urbanicity with BDI and state anxiety were strongest in mid-adolescence: in BDI at age 12 to 16 years and in state anxiety at age 14-17.Based on the IMAGEN baseline 14 years datasets,European urbanization index was also significantly correlated with perspective taking(r = 0.103,P = 0.009)and depressive score(r = 0.118,P = 0.004).5.Urbanization index-brain-perspective taking pathwayMediation analysis showed that the left MSFC volume could mediate the effect of both Chinese and European urbanization index on perspective taking.Conclusion:In this study,Chinese and European urbanization which constructed by remote sensing satellite imagery were found to negatively affect the left MSFC volume much more significant in the adolescent period and positively affect the cerebellum vermis much more significant in the childhood period.In addition,both Chinese and European urbanization index were significantly correlated with social cognition and depression and the left MSFC voloume could mediate the effect of global urbanization index on perspective taking.