Skip to main content
Journal of Big Data
Download PDF
Download ePub

Inhibitory neuron links the causal relationship from air pollution to psychiatric disorders: a large multi-omics analysis

Journal of Big Data volume 11, Article number: 127 (2024) Cite this article

Abstract

Psychiatric disorders are severe health challenges that exert a heavy public burden. Air pollution has been widely reported as related to psychiatric disorder risk, but their casual association and pathological mechanism remained unclear. Herein, we systematically investigated the large genome-wide association studies (6 cohorts with 1,357,645 samples), single-cell RNA (26 samples with 157,488 cells), and bulk-RNAseq (1595 samples) datasets to reveal the genetic causality and biological link between four air pollutants and nine psychiatric disorders. As a result, we identified ten positive genetic correlations between air pollution and psychiatric disorders. Besides, PM2.5 and NO2 presented significant causal effects on schizophrenia risk which was robust with adjustment of potential confounders. Besides, transcriptome-wide association studies identified the shared genes between PM2.5/NO2 and schizophrenia. We then discovered a schizophrenia-derived inhibitory neuron subtype with highly expressed shared genes and abnormal synaptic and metabolic pathways by scRNA analyses and confirmed their abnormal level and correlations with the shared genes in schizophrenia patients in a large RNA-seq cohort. Comprehensively, we discovered robust genetic causality between PM2.5, NO2, and schizophrenia and identified an abnormal inhibitory neuron subtype that links schizophrenia pathology and PM2.5/NO2 exposure. These discoveries highlight the schizophrenia risk under air pollutants exposure and provide novel mechanical insights into schizophrenia pathology, contributing to pollutant-related schizophrenia risk control and therapeutic strategies development.

Graphical Abstract

Introduction

Psychiatric disorders were the most mysterious diseases in medicine for their unknown genetic mechanism and casual risky factors, raising heavy public burdens. These could be attributed to the neurophysiological complexity and the lack of effective research approaches [1]. However, their pathogenic factors or risky conditions are being consistently investigated. Air pollution has been a severe public health concern during the past decades and has been linked to the increased risk of various chronic diseases such as cardiovascular disease and cancers [2,3,4,5]. Urban dwellers are currently exposed to many detected outdoor air pollution gradients, including fine particulate matter (PM2.5), particulate matter of ≤ 10 mm in diameter (PM10), nitrogen oxides (NOx), and indoor agents like nitrogen dioxide (NO2) as reported [6], their long-term exposure was associated with destroyed lung function or higher asthma incidence during adulthood, as well as psychiatric disorders [7]. Air pollution exposures have raised much concerns and research interests in psychiatric disorder risk [8]. For instance, depression risk was observed to increase by pregnancy PM2.5 exposure [9], and the association between air pollution and schizophrenia risk has been proposed [10]. Whereas the non-causal observational studies and unpractical randomized controlled trials blocked the potential causality investigation and prevented public health decision-making [11].

Mendelian randomization (MR) is a Genome-wide association studies (GWAS)-based epidemiological approach. It utilizes randomized alleles (genetic variants) allocation to simulate randomized grouping in prospective randomized controlled trials. It maximally avoids confounders and reveals the causal relationship between the exposure and the outcome [12]. In principle, the MR analyses rely on three basic assumptions: First, the genetic variants should present a robust association with the exposure. Second, the genetic association between the exposure and outcome should be independent of confounders. Third, the genetic variants affect the outcome exclusively via the exposures [13, 14]. On these bases, MR avoids measurement errors due to the well-defined and stable genetic instrument variants (IVs) and could eliminate reverse causation because the disease cannot affect genotype. Besides, the environmental exposure proxying genetic variants is unlikely to be confounded by other factors [15]. These advantages make MR an appropriate and advanced approach to investigating causal associations between air pollution exposure and psychiatric disorder risk.

Despite inferring the pollutants to psychiatric risk causality, understanding their biological mechanism was also urgently required. Previous reports have shown their correlations with specific genetic patterns [16, 17]. YWHAB polymorphic locus rs6031849 could strengthen cumulative PM2.5’s associations with schizophrenia relapse [17]. Also, the extent of PM2.5 exposure’s influence on depression-related neural function networks could differ by polygenic risk in gene-by-environment interactions [16]. Gene expression transfers gene-level information to biological effects [18,19,20,21]. Experimentally, PM2.5 exposure increased Adra2b levels in the mice’s brains, and Adra2b overexpression, in turn, could enhance the anxiety-like behavior under PM2.5 exposure [22]. Meanwhile, PM2.5 increased the microglia-related neuroinflammatory transcription [23] to potentially promote mental disorders progression [24]. However, the current understanding of how air pollutants affect psychiatric disorder pathology is still insufficient. The emerging high-throughput approaches offered effective investigation into gene-trait correlations. Transcriptome-wide association study (TWAS) links population gene expression with phenotypes [25], while bulk RNA-seq and single-cell RNA (scRNA) have provided more detailed expression-traits associations and cell-level clues [26]. Large bulk RNA-seq analyses discovered enriched excitatory and inhibitory neuron pathways associated with schizophrenia risk [27], and scRNA further identified schizophrenia populations with specific excitatory and inhibitory neuronal cell states [28]. Therefore, integrating RNA-based approaches with MR is beneficial in inferring air pollutants’ causal and biological effects on psychiatric disorders.

In this study, we applied a linkage disequilibrium score regression (LDSC) and a two-sample MR (TSMR) to explore the genetic and causal association between the exposures of common air pollution gradients and nine psychiatric disorder risks and used multivariable Mendelian randomization (MVMR) to exclude the bias of common confounding factors (Fig. 1). Furthermore, the biological mechanism investigation was performed via the TWAS, scRNA, and RNA-seq cohort analyses. We aimed to determine whether air pollutants function as the casual risk of psychiatric disorders and reveal potential biological mechanisms, thereby benefiting public health decision-making and therapeutic management.

Fig. 1
figure 1

The MR analyses design and data included in this study

Full size image

(a) LDSC and TSMR identifies genetic and causal associations between air pollution exposure and nine psychiatric disorders, and their confounders-adjusted association was examined by MVMR. (b) TWAS converged from GWAS to investigate the potential genes and biological processes involved in the association between air pollution exposure and nine psychiatric disorders. (c) ScRNA and RNA-seq cohort analyses validated the abnormally expressed genes from TWAS and their involved pathways. Created by Biorender.com.

Methods

MR Study design and data sources

MR analysis has been revealed as an important tool to link environmental exposure and psychiatric outcome risk with casual associations [29, 30]. Therefore, we used MR analysis to investigate the causality between air pollution exposure and psychiatric disorder risk. The flow chart of our study design is shown in Fig. 1. The summary-level GWAS data we used were collected from publicly available databases (summarized in Additional Table S1, sTable 1). No restriction of gender, age, income, or education level was set for these GWAS.

The GWAS data of participants under diverse air pollution exposure levels were derived from UK Biobank [31,32,33] and collected from the MRC IEU database (https://gwas.mrcieu.ac.uk/) [34, 35]. The residential air pollution range was evaluated in different locations in Great London with a land use regression for the annual average of 2010. The mean PM10 level was 16·24 ± 1.90micro-g/m3, from 11.78 to 31.39 micro-g/m3, and the mean PM2.5 level was 9·99 ± 1.06 micro-g/m3, from 8·17–21.31 micro-g/m3. The summary-level GWAS of PM10 and PM2.5 contained 423,796 individuals and 9,851,867 single-nucleotide polymorphisms (SNPs). The mean NO2 level was 26·71 ± 7·58 micro-g/m3, from 12·93–108·49 micro-g/m3 The mean NOx level was 44·11 ± 15·53 micro-g/m3, ranging 19·74–265·94 micro-g/m3. The summary-level GWAS of NO2 and NOx included 456,380 individuals and 9,851,867 SNPs.

We also retrieved the GWAS data for analyzing the potential confounders, and these include body mass index (BMI) [31], alcohol intake frequency [31], number of cigarettes previously smoked daily [31], education level (years of schooling) [34], and income level (average total household income before tax) [31]. These potential confounders’ GWAS data included 336,109, 462,346, 108,946, 766,345, and 397,751 participants respectively.

To avoid the bias generated by sample overlapping, the GWAS data of the psychiatric outcomes were obtained from databases outside the UK biobank. The psychiatric disorders were all diagnosed by ICD-10. The GWAS data for major depression (170,756 cases, 329,443 controls, 8,481,298 SNPs) [36], schizophrenia (52,017 cases, 75,889 controls, 7,659,768 SNPs) [27], anorexia nervosa (3495 cases, 10,982 controls, 10,641,224 SNPs), [37] and obsessive-compulsive disorder (OCD) (26,888 cases, 7037 controls, 8,409,517 SNPs) [38] were obtained from the PGC consortium, bipolar disorder (4501 cases, 192,220 controls, 16,380,409 SNPs), post-traumatic stress disorder (PTSD) (1103 cases, 198,110 controls, 16,380,382 SNPs), anxiety (20,992 cases, 166,584 controls, 16,380,449 SNPs), generalized anxiety disorder (GAD) (2163 cases, 198,110 controls, 16,380,388 SNPs) and phobic anxiety disorder (PAD) (2200 cases, 198,110 controls, 16,380,394 SNPs) were obtained from FinnGen (round 5) [39, 40]. This research utilized publicly available data, which waived the ethical approval requirement. Each study contributing to the GWAS contains details for ethical approval and participant consent in their original publications. This research requires no specific ethical approval.

Linkage disequilibrium score regression

We utilized the summarized GWAS data to perform the genetic correlations between the four air pollutant exposures and the nine psychiatric disorders via LDSC [40, 41]. LDSC evaluated the genetic correlation by leveraging the fact that the GWAS effect size estimation for a given SNP encompasses the effects of all SNPs in linkage disequilibrium (LD) with that SNP. First, all SNPs were harmonized with munge_sumstats.py. Then, the genetic correlations were estimated by the ldsc.py with pre-computed LD scores of 1000 Genome European data [42].

Selection for instrumental variables

A threshold (5e-6) was used to ensure sufficient IVs for robust analyses, which has been commonly used in MR research including psychiatric causality inference [43,44,45,46,47,48,49]. Then, we calculated F-statistics for each IV and excluded IVs with F-statistics < 10 to retain the strong instruments only [50, 51]. F statistics for each instrument in the exposures were calculated by \(\:\frac{\frac{R2}{K}}{\left[\left(1-R2\right)\left(N-K-1\right)\right]}\), where K is the number of SNP, N is the sample size, R2 is the variance explained by SNPs calculated by \(\:2*EAF*\left(1-EAF\right)*\left(\frac{Beta}{SE}\right)\)2 [51]. These approaches were sequentially conducted to ensure the first assumption (exposure correlation assumption) of MR analyses was obeyed. Then, linkage disequilibrium analyses (r 2 < 0.001, distance < 10 MB) were conducted to select independent IVs further, eliminating the linkage disequilibrium effects. Finally, IVs significantly correlated with the outcome were excluded. These filtrations were performed to ensure the third assumption was obeyed.

Two-sample mendelian randomization

Seven different methods [random-effect inverse-variance weighted (IVW) [52], weighted median, MR Egger [52,53,54], MR-RAPS [55] MR-PRESSO [56], Simple Mode, and Weighted mode] were conducted for two-sample MR. IVW was used as the main results, in which the weighted regression of the SNP-outcome effects and SNP-exposure effects were calculated with the intercept constrained to zero. IVW [52] had the optimal statistical power but under the assumption that all instruments were valid and without pleiotropy. Weighted median and MR egger were used for supplementary results due to their more robust estimates in broader conditions, although less efficient [52,53,54], and MR-RAPS and MR-PRESSO are advanced in tackling pleiotropy [55, 56]. Moreover, the TSMR examined the association between SNPs and outcome, and the significant SNP was removed after ensuring that its removal exerted no effect on TSMR results according to the leave-one-out approach. Heterogeneity was analyzed by Cochran’s Q test [57]. Steiger tests were conducted to examine the causal direction of SNPs [58]. Horizontal pleiotropy was analyzed by the MR Egger intercept test [59, 60], and leave-one-out analysis was used to evaluate whether a single SNP could affect the results, and this could detect the potential violation of the second and third assumptions that the genetic variants are independent of confounders and merely affect the outcome via the exposure [14].

Multivariate mendelian randomization

MVMR allowed for estimating the effects of multiple exposures on an outcome. The included exposures could be confounders, mediators, or colliders [61]. MVMR was also suitable for accounting for pleiotropic variants [62]. We used MVMR to estimate more direct effects of air pollution on psychiatric disorders, adjusting for BMI, alcohol intake frequency, the number of cigarettes previously smoked daily, education level, and income. This is also critical in identifying potential violence of the second assumption that the genetic associations are not correlated to potential confounders [63].

Transcriptome-wide association study (TWAS) and enrichment of biological pathways

To conduct transcriptomic imputation, we converted GWAS into TWAS by the FUSION method [64]. Expression quantitative trait loci (eQTL)-based linear model was used in FUSION to predict gene expression based on the reference panels of RNA-seq. European cortex samples of RNA-seq of Genotype-Tissue Expression version 8 (GTEx v8) [65], CommonMind Consortium’s (CMC), and splicing reference [66] were used as reference panels in this study. An Omnibus test was performed to evaluate the combined association of a single gene in multiple reference panels. Genes significantly associated with air pollution and psychiatric disorders were identified in TWAS results. Combined P values of TWAS for air pollutants and psychiatric disorders were calculated by Fisher’s Combined P-value (FCP) method. Additionally, we conducted biological pathway enrichment analyses for these genes based on the Gene Ontology database to further understand the potential biological mechanisms underlying air pollution and psychiatric disorders.

Single-cell RNA data processing and analyses

The 26 schizophrenia and control scRNA samples were obtained from gene expression omnibus (GEO, GSE228315) [67]. These samples were derived from either donor or engineered brain organoids. The control group contains eight iPSC samples derived from control donors and five control-engineered hPSC samples, and the SCZ group contains eight iPSC samples derived from SCZ patients and five SCZ-engineered hPSC samples. The quality control, doublets removal, data integration, reduction, and cell annotation were conducted according to the previously published study [67]. The level of the shared genes in each cell was estimated using “AddModuleScore” function [68], and the functional enrichment was performed using “AUCell”, “UCell”, “singscore”, and “ssgsea ” algorithms [69, 70] from the “irGSEA” R package (https://github.com/chuiqin/irGSEA) [71]. We evaluated the metabolic communications among different cell clusters by “MEBOCOST” [72]. The BayesPrism [73] deconvolution was used to estimate the cell proportions in each bulk RNA-seq sample.

RNA-seq cohort processing and analyses

We retrieved the TPM expression matrix ‘DER-02-PEC-Gene_expression_matrix’ and clinical information ‘PEC_capstone_data_map_clinical’ of schizophrenia and control cases from http://adult.psychencode.org/#Derived, and the merged PsychENCODE and GTEx TPM matrix was then transferred into log2(TPM + 1) matrix for the downstream analyses. The differentially expressed genes were identified using the “limma” R package, and the ssGSEA score of the shared genes was calculated by the “GSVA” [74] R package. To investigate the pathways associated with the shared genes in schizophrenia, we performed WGCNA [75] to filter genes with positive correlations with the shared genes and ssGSEA score. The gene module with the highest correlation was selected and the genes inside the module with gene significance > 0.5 and module membership > 0.9 [76] were harvested for functional enrichment. The functional enrichment was conducted using the “clusterProfiler” R package [77]. The correlations between shared genes, ssGSEA score, and the pathways were visualized using the “Cytoscape” software [78].

Statistical analyses

Analyses were conducted using the combination of ‘TwoSampleMR’ packages [79], ‘ggplot2’, ‘clusterProfiler’ [77], ‘enrichplot’, and ‘DOSE’ [80] in Rstudio version 4.2.2 and FUSION software. Normally-and non-normally-distributed parameters were tested by student’s t-test and Wilcoxon test, respectively. Comparsion among multiple groups of normally- and non-normally-distributed data were examined by Anova and Kruskal Wallis-test, respectively. The correlations between normally distributed parameters were calculated by Pearson’s correlation coefficient and were adjusted by the default ‘holm’ method. A false discovery rate (FDR < 0.05) was performed to correct for multiple independent tests in TSMR [23, 52,53,54,55,56], TWAS [64] analyses, gene expression differences comparison, and differentially enriched pathways identification. Results with FDR < 0.05 were regarded as statistically significant. Meanwhile, those P < 0.05 with FDR ≥ 0.05 were regarded as suggestive [81, 82]. We used raw P values rather than FDR for MVMR analyses, considering these analyses exploratory and robustness tests for the TSMRs, consistent with previous studies [63, 83,84,85].

Results

TSMR identified the air pollutants-psychiatric disorders pairs with causal association

After filtration, 29, 58, 84, and 75 instrument SNPs were selected for PM10, PM2.5, NO2, and NOx, respectively (Additional Table S1, sTables 2, 3, 4 and 5). We then performed LDSC to evaluate the genetic correlation between the four air pollutant exposures and nine psychiatric disorder outcomes (Additional Table S1, sTable 6). PM2.5, NO2, and NOx exhibit significant positive genetic correlations with schizophrenia, major depression, and anxiety. Also, a significant positive result between PM10 and major depression was observed (Fig. 2a). These preliminarily indicate their potential correlations. Further, we conducted TSMRs to calculate the effect size of four air pollutants on nine psychiatric disorders, thereby determining their causal correlations. As Fig. 2b shows, two significant positive causal associations were found, with PM2.5 & schizophrenia exhibiting the highest causal effect (OR: 1.870, 95%CI: 1.254-2·790, FDR < 0.05, P < 0.01), followed by NO2 & schizophrenia (OR: 1.708, 95% CI: 1.261-2·315, FDR < 0.05, P < 0.01). Besides, suggestive positive associations between NOx & schizophrenia (OR: 1.477, 95% CI: 1.077-2·026, P < 0.05), PM10 & GAD (OR: 3·804, 95% CI: 1.128-12·831, P < 0.05), PM10 & major depression (OR: 1.224, 95% CI: 1.008–1.486, P < 0.05), PM2.5 & major depression (OR: 1.268, 95% CI: 1.037–1.551, P < 0.05), and PM2.5 & bipolar disorder (OR: 1.711, 95% CI: 1.035-2·829, P < 0.05) (Table 1). The results of other associations were recorded in Additional Table S1, sTable 7, and 8.

Fig. 2
figure 2

LDSC genetic correlation and TSMR effect for each association between air pollutants and psychiatric disorders. (a) The LDSC dot map indicates the genetic correlations by dot size and statistical significance by color. (b) The TSMR dot map indicates the effect size by dot size and statistical significance by colors.

Full size image
Table 1 Significant TSMR results for air pollution and psychiatry disorders
Full size table

In the sensitivity analyses, the causal directions of the associations in TSMR were examined using Steiger tests. The results showed that the causal direction for all significant and suggestive positive associations identified by TSMR was correct (Additional Table S1, sTable 9). Additionally, heterogeneity was observed in some associations, while the main method, random-effect IVW we used, fitted the presence of heterogeneity (Additional Table S1, sTable 10) [86]. The leave-one-out analyses showed no peculiar SNPs and the robustness of the results in all TSMR results (Additional Figure S1-S4). Moreover, pleiotropy was also not observed in all significant TSMR results (FDR < 0.05), as the P-values for pleiotropy tests were higher than 0.05 (Additional Table S1, sTable 11), showing that the basic assumptions of MR were obeyed.

MVMR identified the causal association without common confounder disruptions

The aforementioned TSMR methods have filtered out the likely causal associations between four air pollutants and nine psychiatric disorders. However, the effects of common confounders might disrupt these causal effects. Hence, we applied the MVMR approach to further identify the robust causal correlations after including the number of daily cigarettes, BMI, alcohol intake, education, and income level as confounders. We observed that PM2.5 and NO2 exhibit consistent causal effects on schizophrenia after all confounding factors were included in MVMR analyses (Fig. 3; Table 2). In contrast, some suggestive effects of previously identified causal exposure in TSMR were interfered with by the inclusion of confounders. The aforementioned suggestive causal effect of PM10 on GAD attenuated towards null with the inclusion of BMI, alcohol intake, income, and number of cigarettes smoked daily, and that of PM10 and PM2.5 on major depression also failed to pass the MVMR as affected by either cigarette, BMI, income, education or alcohol inclusion (Additional Table S1, sTable 12). Similarly, PM2.5’s causal effects on bipolar disorder were also adjusted as not robust after integrating cigarette, BMI, and education. Shortly, these results demonstrate that schizophrenia harbored the most robust independent cause by air pollutants without violating the MR assumptions mediated by confounders, at least for the five common confounding factors BMI, alcohol, cigarette intakes, income, and education.

Fig. 3
figure 3

MVMR of each causal association adjusted by confounders. The forest map describes the adjusted causal effect size of four air pollutants on nine psychiatric disorders by cigarette, BMI, alcohol, education, and income. The triangle and circle dot indicate that the effect size was significant and not significant after the confounder adjustment, respectively

Full size image
Table 2 Significant MVMR results for air pollution and psychiatry disorders
Full size table

TWAS discovered shared genes significantly associated with air pollution and psychiatric disorders

We converted the GWAS results to TWAS in order to discover the genes with significant correlations with both air pollution and psychiatric disorders. The detailed gene information and statistical summary of significant associations identified by TSMR and MVMR are listed in Additional Table S1 (sTables 13–17). We depicted the chromosomal location identified of shared genes between PM2.5 & schizophrenia, NO2 & schizophrenia, respectively, and those between the two associations (Fig. 4A, 4B). As a result, 13 shared genes were found between PM2.5 and schizophrenia, and 15 shared genes were found between NO2 and schizophrenia in the same direction. Seven shared genes (SULT1A1, ZNF680, RPAP2, NT5C2, KIAA1109, FANCL, and ALG1L11P) were found in both PM2.5 & schizophrenia and NO2 & schizophrenia with the same direction. Notably, these seven shared genes between the two significant associations are located on different chromosomes, suggesting their independent effects in exposure-outcome interactions (Fig. 4a). In order to further investigate the biological processes in which these shared genes are potentially involved, we input these genes to conduct functional enrichment analyses based on the database of GO terms. The results showed that synapse-related and histone arginine methylation processes were enriched by 13 shared genes between PM2.5 & schizophrenia (Fig. 4c), while macrophage and endoplasmic reticulum-related processes were also observed by 15 shared genes between NO2 & schizophrenia (Fig. 4d). The detailed enrichment results were recorded in Additional Table S1, sTables 18–19. Importantly, nucleoside and alcohol metabolic processes enrichment were found in both two associations (Fig. 4c, 4d), implying that their metabolic disturbance might be the mechanism underlying pollutants-mediated schizophrenia risk.

Fig. 4
figure 4

The shared genes involved in the significant causal association between air pollutants and psychiatric disorders. (a) The chromosome graph shows the location of shared genes recognized by TWAS within each significant association. (b) The Venn diagram depicts the shared genes and the common shared genes between both PM2.5 & schizophrenia and NO2 schizophrenia. (c) Functional enrichment of shared genes within PM2.5 & schizophrenia (c) and NO2 & schizophrenia association (d), respectively

Full size image

To further examine whether the IVs used in TSMR were located in the shared genes, we annotated the location of each IV (Additional Table S1, sTable 2–5). None of the IV was located at the shared genes, except rs10094026, which is located at the intron of MFHAS1, significantly correlated with NO2 (P = 7·70E-7) but not significantly correlated with schizophrenia (P = 4·58E-1). After excluding this IV, NO2 still showed significant correlation and minor alteration in TSMR by IVW (OR: 1.739, 95% CI:1.280-2·362, P < 0.01). Besides, leave-one-out tests also confirmed the robustness of the result after leaving rs10094026 (Additional Fig. 3).

ScRNA analyses identified abnormally expressed shared genes and their involved pathways in a schizophrenia-derived neuron subtype

The abnormal shared genes between air pollution and schizophrenia risk indicate an influenced cell population during the pathological development. To investigate the potential cell type that turns abnormal and might mediate the causality, we analyzed the transcriptional expression of shared genes in schizophrenia/control scRNA samples. The scRNA cells were annotated to 27 defined clusters with corresponding cell-specific markers, consistent with a previous publication [67] (Additional Figure S5a), the completed cell annotation, as well as the group allocation, were shown in Additional Figure S5b and S5c, respectively. We first compared the expression differences of the shared genes on each cell type between the healthy donor and schizophrenia-derived samples. We found that five shared genes that show a positive genetic correlation with air pollution & schizophrenia (RPAP2, KIAA1109, CDC42BPA, EVI5, and PXK) were upregulated in the IN2 cluster (inhibitory neuron subtype2) from the schizophrenia group (Fig. 5a), suggesting that the IN2 cells were affected in pollutants-induced schizophrenia development. To dive into the IN2 cells, we extracted and clustered the IN2 cells according to their high variable features first, and divided them into two levels according to the median scores calculated by the shared genes’ expression. We noticed that after the IN2 cells were clustered into C0 and C1, the C1 cluster exhibited a remarkably higher proportion of high-level cells, as well as a higher expression of the five genes (Fig. 5b). This implies that the five genes contribute greatly to the cell fate influence of IN2. Further, we compared the biological heterogeneity between the low-level and high-level IN2 cells, the high-level cells presented elevated biological processes including neuron-neuron synaptic transmission, plus end-directed organelle transport along microtubule etc. and protein export, valine leucine and isoleucine biosynthesis pathways etc. (Fig. 5c), indicating their abnormal synaptic activities and metabolic process. Hence, we therefore compared their metabolic crosstalk differences with other cell clusters. We found that the high-level IN2 exhibits a higher number of metabolite-sensor communication with several cell clusters like oRG and RG-like cells etc. (Fig. 5d,e). Specifically, we identified that the receptor genes SLC3A2 and SLC38A1 are highly and upregulated in the high-level IN2 cells (Fig. 5f). When final metabolite-sensor communications were estimated, we noticed that high-level IN2 cells showed significantly higher levels of L-Glutamine-SLC3A2-mediated communications from cycling ventral NEC, non-cycling NEC, cycling NEC, intermediate cells, IN6, and RG-like cells, compared to the low-level IN2 cells. Additionally, they also received significantly stronger L-Glutamine-SLC38A1-mediated communications from oRG, IN7, and CN3 cells (Fig. 5g), suggesting that synaptic and metabolic abnormality might be caused in IN2 cells expressing higher expression of the five shared genes.

Fig. 5
figure 5

Single-cell analyses identify schizophrenia-specific cells with abnormal levels of shared genes, pathways, and metabolic patterns. (a) The stacked violin plot compares the expression differences of shared genes between schizophrenia and control cases in all cell clusters. (b) The scatter plot depicts the different levels of five shared genes and their integrated score between IN2 subtypes. (c) Comparison of functional heterogeneity between the high-level and low-level IN2 cells. (d) Comparison of the number of metabolite-sensor communication among all cell clusters. (e) The circle plot describes the frequencies of communication events estimated from other cell clusters to the high-level and low-level IN2 cells. The arrows indicate the communication direction. (f) Comparison of metabolite sensor expressions between the high-level and low-level IN2 cells. (b) The communication network of metabolite-sensor communication from other cells to the high-level and low-level IN2 cells

Full size image

RNA-seq cohort analyses validated the abnormal expression of five shared genes and their involved pathways in schizophrenia cases

Finally, we employed an RNA-seq cohort with a large number of schizophrenia and control individual cases to validate the abnormal level of the shared genes, pathways, and the schizophrenia-related IN2 subtypes in schizophrenia patients. We confirmed that RPAP2, KIAA1109 (BLTP1), CDC42BPA, EVI5, and PXK are consistently upregulated in schizophrenia patients compared to control cases (Fig. 6a,b). Meanwhile, we evaluated the cell proportions in each sample, and we noticed that the high-level IN2 cell proportion was higher in schizophrenia patients, while the low-level IN2 cell proportion was lower (Fig. 6c). After dividing schizophrenia patients into IN2_high and IN2_low groups according to the high-level IN2 cell level, we confirmed that patients with higher high-level IN2 cell proportion harbored elevated expression of RPAP2, KIAA1109, CDC42BPA, EVI5, and PXK (Fig. 6d), indicating their close association with IN2 cells. To further verify the pathways that are associated with the five genes and their ssGSEA score, we subsequently conducted weighted gene co-expression network analysis (WGCNA) to identify genes that are highly associated with RPAP2, KIAA1109, CDC42BPA, EVI5, PXK, and ssGSEA level. To obtain the most relevant genes, we collected the genes from the turquoise module with the highest correlation that harbored gene significance > 0.5 and module membership > 0.9 (Fig. 6e). Subsequently, functional enrichment analysis was performed according to the harvested genes. We noticed that pathways related to the endoplasmic reticulum and neuron projection were enriched (Fig. 6f) and their correlations with the five shared genes and the ssGSEA score were shown (Fig. 6g). We then calculated the correlation between the five genes, pathways levels, and the high-level IN2 cell proportion and their strong positive correlations were confirmed (Fig. 6h), implying that IN2 cells with abnormal transcriptional alterations were associated with schizophrenia development.

Fig. 6
figure 6

RNA-seq validates the abnormality of shared genes, cell types, and functional enrichment in schizophrenia patients. (a) The volcano plot shows the differentially expressed genes between normal and schizophrenia cases in the RNA-seq cohort. (b) Comparison of the five shared genes between normal control and schizophrenia cases. (c) Deconvolution of scRNA cell proportion in bulk RNA-seq sample from schizophrenia and control patients. (d) Comparison of five shared genes between the IN2-high and IN2-low groups pf schizophrenia patients. (e) WGCNA modules and their correlations with the five shared genes and their ssGSEA score. (f) The functional annotation of the harvested genes from WGCNA. (g) The correlation network exhibits the correlations between the five shared genes, ssGSEA score, and the enriched pathways. The line width and color depth indicate the correlation value, the solid line indicates a positive correlation and the dashed line indicates a negative correlation. (h) The correlations of IN2_high cell proportion with five shared genes and pathways

Full size image

Discussion

Air pollution has been raised as a severe environmental problem found to affect multisystem dysfunctions, including cardiovascular [87, 88], respiratory diseases [89], neurologic diseases [90], cancers [91, 92], and autoimmune diseases [93], and its correlations with psychiatric disorders have also been widely observed. Though much observational research has consistently supported the association between air pollutant exposure and mental disorders, their clear causal associations are still unrevealed due to intrinsic approache limitations. The results from observational research were usually biased by multiple confounders like BMI, cigarette smoking, and alcohol drinking [94], and spurious associations could thus be concluded. To overcome this limitation, we conducted MR to explore their causal associations. We found five suggestive positive as well as two significant positive causal associations via TSMR and excluded common confounders in the causal effects on schizophrenia by PM2.5 and NO2 using MVMR. Robust causal correlations between other psychiatric disorders (such as anxiety, PAD, and PTSD) and air pollution were not found. Our results accorded with but also contradicted some previous preclinical and systematic review reports [10, 94], indicating the potential bias or interspecies biological/pathological heterogeneity might have disrupted the true associations.

We compared our MR results with previously published research to clarify which exposure to outcome associations are robustly developed and which are not. A recent systemic review including 13 Asia articles has proposed the concerns of short-term air pollution exposure’s risk for schizophrenia. They found that PM2.5, PM10, and NO2 exposure correlated to increased schizophrenia risk with consideration of age, country, pollutant concentration, and temperature. [10] However, we noticed that PM10 does not present causal effects on schizophrenia during TSMR analyses, suggesting that the MR analyses might provide the possibility to overcome potential biases. While for PM2.5 and NO2, two recently published MR studies proposed causality clue between PM2.5 and schizophrenia [95, 96], which is consistent with our results. Moreover, we discovered NO2 as another independent risk exposure to schizophrenia and further validated PM2.5’s and NO2’s causal effects on schizophrenia without bias of BMI, smoking, education, incomes, and alcohol intake. However, limitations should be presented in that we did not include Asian cases, and the contradictory discoveries were not adjusted by case area. This might have differentiated pollutants’ effects on schizophrenia risk. Isobel et al. [94] have also conducted a meta-analysis to investigate the correlation between air pollution exposure and depression, anxiety, and bipolar disorder. Their study showed that long-term PM2.5 has a significant positive association with depression, while we found it not significant after adjusting the TSMR via multiple testing corrections. However, we recommended that attention should still be paid to these suggestive associations because of the possibility of overcorrection from traditional MR and adjustment of P-values [97].

A subsequent TWAS analysis was performed to present the transcriptional explanation for genetic casual associations. For schizophrenia, many genes are involved in its associations with either PM2.5 or NO2. Notably, four genes (SULT1A1, INO80E, TAOK2, and DOC2A) were located on 16p11.2, whose copy number variant was associated with schizophrenia risk [98]. SULT1A1 encodes a sulfotransferase that inactivates dopamine via sulfation. It has been detected as a candidate psychosis suppressor in the methamphetamine-treated mice brain, which resembles a positive symptom of schizophrenia [99]. This was consistent with our results. INO80E, a chromatin remodeling INO80 complex subunit coding gene, has been identified as a schizophrenia-associated gene [98]. The alteration of DNA open chromatin region affected by chromatin-remodeling-complex was associated with schizophrenia risk SNPs, and the CRISPR editing of these SNPs affected neurodevelopment [100], indicating that chromatin remodeling is important during schizophrenia development. However, the role of INO80E in schizophrenia pathology is still unclear due to the limited research. Dendritic spine maturation is critical for synapse integrity. Synapse dysfunction could disrupt synaptic transmission and contribute to schizophrenia development [101]. TAOK2 is a serine/threonine kinase coding gene and its alteration blocked the phosphorylation the cytoskeletal GTPase Septin7, which prevents dendritic spines maturation [102] and thereby probably promotes schizophrenia. DOC2A also codes proteins involved in calcium-dependent neurotransmitter release, and its depletion or duplication was found in schizophrenia cases [103], suggesting its abnormality promotes the schizophrenia process. Interestingly, PM2.5 has been discovered to induce synapse damage and dysfunction [104, 105], and this shows its probability of causing schizophrenia progression by disturbing synapse function. But its association with DOC2A or TAOK2 remains unclear. On other chromosome regions, PRMT7 encodes a protein arginine methyltransferase with histone methylation capability. Histone methylation is a well-recognized epigenetic abnormality in schizophrenia because it disrupts oligodendrocytes and myelination function [106]. Long-term PM2.5 exposure was found to increase the H3K4 and H3K9 methylation in macrophages, decreasing their IL-6 and IFN-β secretion [107], indicating PRMT7’s role in PM2.5-mediated schizophrenia risk. Also, PRMT7 was abnormally expressed in schizophrenia tissues and was highly associated with schizophrenia [108]. NT5C2 encodes a cytosolic 5’-nucleotidase to regulate purine/pyrimidine balance and it was a target of either schizophrenia risk miRNA variants or risk SNP cis-regulation [109, 110]. Additionally, the level of CDC42 Binding Protein Kinase alpha and UDP-glucose/glycoprotein glucosyltransferase 2, encoded by CDC42BPA and UGGT2, respectively, were found elevated in the schizophrenia samples [111, 112]. Shortly, we discovered genes potentially involved in schizophrenia development, as previous evidence suggests, while their roles in PM2.5- or NO2-mediated schizophrenia risk have not been noticed. Hence, we provided novel clues for the genetic engagements in air pollution and schizophrenia interaction.

Given that transcriptional changes can enhance the association between genetic abnormalities and genotypic transformation, we analyzed the shared genes in bulk tissues and single-cell clusters. Five shared genes (BLTP1/KIAA1109, CDC42BPA, EVI5, PXK, and RPAP2) are upregulated in the schizophrenia bulk cases and IN2 cell cluster, a subcluster of GABAergic neurons. BLTP1/KIAA1109 gene variation was associated with neuron migration and embryonic abnormality [113], while whether it leads to BLTP1/KIAA1109-deficiency phenotype remains unknown. Overexpressed CDC42BPA protein level [111] was observed in schizophrenia samples, and its association with ATF4-mediated endoplasmic reticulum (ER) stress was recently reported [114] in the Alsheimer disease model, this is similar to what we obtained that CDC42BPA was positively correlated with the endoplasmic reticulum activities in schizophrenia cases, indicating CDC42BPA could participate in ER-mediated schizophrenia risk. Pxk encodes a secretory protein MONaKA that binds to and limits the function of Na⁺, K⁺-ATPase, it was found to cause Na⁺ extrusion efficiency thus disrupting the hippocampal neural energy balance [115], suggesting that transcriptional change of Pxk might exert widespread damage on neuron function like synaptic activities, which is consistent with our elevated enrichment result in the high-level IN2 neurons. GABAergic neuron dysfunction in schizophrenia pathology has gained increasing attention [116], recent large data reports have discovered clues that suggest the involvement of inhibitory neuron activity in schizophrenia risk [27, 28]. Interestingly, we also discovered that the high-level IN2 neurons highly expressed SLC3A2 and SLC38A1, and received a high level of communication mediated via L-Glutamine. As GABAergic neurotransmission relies on the system-A-mediated glutamine transmission [117], the regulated SLC38A1 might indicate that the activity of IN2 was disordered. Importantly, a higher level of this GABAergic neuron subtype was observed in schizophrenia patients, and its association with abnormal pathways and shared genes was verified, demonstrating that it links the genetic causality from air pollution to schizophrenia. So far, no direct evidence has been published regarding the associations between transcriptional changes of shared genes and schizophrenia pathology, especially in inhibitory neurons, our discoveries provide novel clues for the biological mechanism of schizophrenia risk.

We should emphasize that the results of MR analyses were strictly based on three assumptions: (1) The instrument variants present strong associations with exposures. (2) The genetic association should be free of confounding factors. (3) The instrument variants affect the outcomes exclusively on exposures. Therefore, our study has applied a series of approaches to ensure our results obeyed these assumptions. We excluded IVs with associations with outcomes and with F < 10 to confirm that the retained IVs are strongly associated with air pollution exposure. Besides, we also include linkage disequilibrium analyses and horizontal pleiotropy analyses to ensure the second and their assumptions were not violated. Considering the existence of potential confounders, BMI, cigarette, and alcohol drinking, in the pollutants-disorder association, we applied MVMR to exclude their confounding effects while evaluating the causal correlations, as a previous study shows [63]. We finally validated the shared genes and pathways in scRNA and bulk-RNA data at the transcriptional level to strengthen the association between genetic variation and schizophrenia phenotype. However, there are still some limitations in this study. The major one is that some unavoidable confounding factors are retained. For instance, the population from different cohorts might introduce biases from population heterogeneity. Although we include as many confounding factors in MVMR analyses as we can, including BMI, cigarette smoking, alcohol drinking, education, and income levels, many other potential confounders might also act as potential confounders and need to be further revealed, such as diet factors. Additionally, the scarcity of transcription data pertaining to air pollution and the unrealizable prospective or experimental verification have constrained the exploration of potential links between schizophrenia-related genes and air pollution.

Comprehensively, we used LDSC, TSMR, and MVMR to identify robust causal associations between four common air pollutant exposures and the common psychiatric disorder outcomes, as well as their effect size. We found that the independent causal effects of schizophrenia by PM2.5 and NO2 could be established. Further, we applied TWAS analyses and discovered the shared genes and pathways between PM2.5 & schizophrenia and NO2 & schizophrenia risk, respectively. Finally, the scRNA and bulk RNA-seq data then identified an inhibitory neuron subtype with abnormal level of the shared genes and pathways. These findings confirmed the risk of air pollution exposure for schizophrenia and identified a critical neuron cell type participating in the pathological process. Therefore, more attention should be paid to schizophrenia risk control under air pollution exposure.

Data availability

The datasets analyzed during the current study are available in the [UK Biobank] repository [https://biobank.ndph.ox.ac.uk], [PGC consortium] repository [https://pgc.unc.edu/], [FinnGen] repository [https://github.com/FINNGEN/pheweb/], CommonMind Consortium’s repository [https://www.nimhgenetics.org/resources/commonmind], GTEx repository [https://www.gtexportal.org/home/], GEO repository [https://www.ncbi.nlm.nih.gov/gds/?term= ], and PsychEncode repository [http://adult.psychencode.org/#Derived]. The GWASs used in this study were available in the [MRC IEU] repository [https://gwas.mrcieu.ac.uk/] etc. Detailed information on UK, PGC, and FinnGene cohorts can be found in sTable 1 of Additional Table S1. Code availability statementAll code during the current study are available from the corresponding author on reasonable request.

Abbreviations

PM:

Particulate Matter

NOx:

Nitrogen Oxides

NO2:

Nitrogen Dioxide

IVs:

Instrument Variants

LDSC:

Linkage Disequilibrium Score Regression

TSMR:

TWO-SAMPLE MR

MVMR:

MULTIVARIABLE MENDELIAN RANDOMIZATION

TWAS:

Transcriptome-Wide Association Study

ScRNA:

SINGLE-CELL RNA SEQUENCE

SNPs:

Single-Nucleotide Polymorphisms

GWAS1:

Genome-Wide Association Studies

BMI:

Body Mass Index

OCD:

Obsessive-Compulsive Disorder

PTSD:

Post-Traumatic Stress Disorder

GAD:

Generalized Anxiety Disorder

PAD:

Phobic Anxiety Disorder

LD:

Linkage Disequilibrium

IVW:

Random-Effect Inverse-Variance Weighted

eQTL:

Expression Quantitative Trait Loci

GTEx:

Genotype-Tissue Expression

CMC:

COMMONMIND CONSORTIUM

FCP:

FISHER’S COMBINED P-VALUE

GEO:

Gene Expression Omnibus

IN2:

Inhibitory Neuron Subtype 2

References

  1. Sullivan PF, Geschwind DH. Defining the genetic, genomic, Cellular, and Diagnostic architectures of Psychiatric disorders. Cell. 2019;177(1):162–83.

    Article  Google Scholar 

  2. Copat C, Cristaldi A, Fiore M, et al. The role of air pollution (PM and NO(2)) in COVID-19 spread and lethality: a systematic review. Environ Res. 2020;191:110129.

    Article  Google Scholar 

  3. de Bont J, Jaganathan S, Dahlquist M, Persson Å, Stafoggia M, Ljungman P. Ambient air pollution and cardiovascular diseases: an umbrella review of systematic reviews and meta-analyses. J Intern Med. 2022;291(6):779–800.

    Article  Google Scholar 

  4. Yang BY, Qian Z, Howard SW, et al. Global association between ambient air pollution and blood pressure: a systematic review and meta-analysis. Environ Pollution (Barking Essex: 1987). 2018;235:576–88.

    Article  Google Scholar 

  5. Zare Sakhvidi MJ, Lequy E, Goldberg M, Jacquemin B. Air pollution exposure and bladder, kidney and urinary tract cancer risk: A systematic review. Environmental pollution (Barking, Essex: 1987) 2020; 267: 115328.

  6. Eguiluz-Gracia I, Mathioudakis AG, Bartel S, et al. The need for clean air: the way air pollution and climate change affect allergic rhinitis and asthma. Allergy. 2020;75(9):2170–84.

    Article  Google Scholar 

  7. Buoli M, Grassi S, Caldiroli A, et al. Is there a link between air pollution and mental disorders? Environ Int. 2018;118:154–68.

    Article  Google Scholar 

  8. Ioannidis JPA. Air pollution as cause of mental disease: Appraisal of the evidence. PLoS Biol. 2019;17(8):e3000370.

    Article  Google Scholar 

  9. Newbury JB, Heron J, Kirkbride JB, et al. Air and noise Pollution exposure in early life and Mental Health from Adolescence to Young Adulthood. JAMA Netw open. 2024;7(5):e2412169.

    Article  Google Scholar 

  10. Song R, Liu L, Wei N et al. Short-term exposure to air pollution is an emerging but neglected risk factor for schizophrenia: a systematic review and meta-analysis. Sci Total Environ 2022: 158823.

  11. Aldinger F, Schulze TG. Environmental factors, life events, and trauma in the course of bipolar disorder. J Neuropsychiatry Clin Neurosci. 2017;71(1):6–17.

    Google Scholar 

  12. Zhu G, Zhou S, Xu Y, et al. Mendelian randomization study on the causal effects of COVID-19 on childhood intelligence. J Med Virol. 2022;94(7):3233–9.

    Article  Google Scholar 

  13. Sanderson E, Glymour MM, Holmes MV, et al. Mendelian randomization. Nat Reviews Methods Primers. 2022;2(1):6.

    Article  Google Scholar 

  14. Li GH, Cheung CL, Chung AK, et al. Evaluation of bi-directional causal association between depression and cardiovascular diseases: a mendelian randomization study. Psychol Med. 2022;52(9):1765–76.

    Article  Google Scholar 

  15. Smith GD. Mendelian randomization for strengthening causal inference in Observational studies: application to gene × environment interactions. Perspect Psychol Sci. 2010;5(5):527–45.

    Article  Google Scholar 

  16. Li Z, Yan H, Zhang X et al. Air pollution interacts with genetic risk to influence cortical networks implicated in depression. Proc Natl Acad Sci USA 2021; 118(46).

  17. Lei Q, Huang X, Li T, et al. Effects of PM(2.5) pollution and single nucleotide polymorphisms of neurotrophin signaling pathway genes acting together on schizophrenia relapse. Int Arch Occup Environ Health. 2023;96(4):629–37.

    Article  Google Scholar 

  18. Cui HK, Tang CJ, Gao Y, Li ZA, Zhang J, Li YD. An integrative analysis of single-cell and bulk transcriptome and bidirectional mendelian randomization analysis identified C1Q as a novel stimulated risk gene for atherosclerosis. Front Immunol. 2023;14:1289223.

    Article  Google Scholar 

  19. Huang W, Jiang R, Li S, et al. Investigating shared genetic architecture between inflammatory bowel diseases and primary biliary cholangitis. JHEP Reports: Innov Hepatol. 2024;6(6):101037.

    Article  Google Scholar 

  20. Li J, Zhang Y, You Y, et al. Unraveling the mechanisms of NK cell dysfunction in aging and Alzheimer’s disease: insights from GWAS and single-cell transcriptomics. Front Immunol. 2024;15:1360687.

    Article  Google Scholar 

  21. Wang Y, Wang X, Yuan Z, Liu F, Luo X, Yang J. Identifying potential drug targets for keloid: A Mendelian randomization study. The Journal of investigative dermatology 2024.

  22. Rao X, Asico LD, Zanos P, et al. Alpha2B-Adrenergic receptor overexpression in the Brain Potentiate Air Pollution-induced behavior and blood pressure changes. Toxicol Sci. 2019;169(1):95–107.

    Article  Google Scholar 

  23. Kim RE, Shin CY, Han SH, Kwon KJ, Astaxanthin Suppresses. PM2.5-Induced Neuroinflammation by regulating akt phosphorylation in BV-2 microglial cells. Int J Mol Sci 2020; 21(19).

  24. Cui W, Ning Y, Hong W, Wang J, Liu Z, Li MD. Crosstalk between inflammation and glutamate system in Depression: signaling pathway and molecular biomarkers for ketamine’s antidepressant effect. Mol Neurobiol. 2019;56(5):3484–500.

    Article  Google Scholar 

  25. Wainberg M, Sinnott-Armstrong N, Mancuso N, et al. Opportunities and challenges for transcriptome-wide association studies. Nat Genet. 2019;51(4):592–9.

    Article  Google Scholar 

  26. Liu H, Huang S, Yang L, et al. Conventional dendritic cell 2 links the genetic causal association from allergic asthma to COVID-19: a mendelian randomization and transcriptomic study. J Big Data. 2024;11(1):26.

    Article  Google Scholar 

  27. Trubetskoy V, Pardiñas AF, Qi T, et al. Mapping genomic loci implicates genes and synaptic biology in schizophrenia. Nature. 2022;604(7906):502–8.

    Article  Google Scholar 

  28. Ruzicka WB, Mohammadi S, Fullard JF, et al. Single-cell multi-cohort dissection of the schizophrenia transcriptome. Sci (New York NY). 2024;384(6698):eadg5136.

    Article  Google Scholar 

  29. McGrath JJ, Mortensen PB, Visscher PM, Wray NR. Where GWAS and epidemiology meet: opportunities for the simultaneous study of genetic and environmental risk factors in schizophrenia. Schizophr Bull. 2013;39(5):955–9.

    Article  Google Scholar 

  30. Wootton RE, Jones HJ, Sallis HM. Mendelian randomisation for psychiatry: how does it work, and what can it tell us? Mol Psychiatry. 2022;27(1):53–7.

    Article  Google Scholar 

  31. Bahcall OG. UK Biobank - a new era in genomic medicine. Nat Rev Genet. 2018;19(12):737.

    Article  Google Scholar 

  32. Eeftens M, Beelen R, de Hoogh K, et al. Development of Land Use Regression models for PM(2.5), PM(2.5) absorbance, PM(10) and PM(coarse) in 20 European study areas; results of the ESCAPE project. Environ Sci Technol. 2012;46(20):11195–205.

    Article  Google Scholar 

  33. Beelen R, Hoek G, Vienneau D, et al. Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe – The ESCAPE project. Atmos Environ. 2013;72:10–23.

    Article  Google Scholar 

  34. Lee JJ, Wedow R, Okbay A, et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat Genet. 2018;50(8):1112–21.

    Article  Google Scholar 

  35. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018;7. https://doi.org/10.7554/eLife.34408

  36. Howard DM, Adams MJ, Clarke T-K, et al. Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions. Nat Neurosci. 2019;22(3):343–52.

    Article  Google Scholar 

  37. Duncan L, Yilmaz Z, Gaspar H, et al. Significant locus and Metabolic Genetic Correlations Revealed in Genome-Wide Association Study of Anorexia Nervosa. Am J Psychiatry. 2017;174(9):850–8.

    Article  Google Scholar 

  38. Revealing the complex. Genetic architecture of obsessive-compulsive disorder using meta-analysis. Mol Psychiatry. 2018;23(5):1181–8.

    Article  Google Scholar 

  39. Kurki MI, Karjalainen J, Palta P, et al. Author correction: FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;615(7952):E19.

    Article  Google Scholar 

  40. Kurki MI, Karjalainen J, Palta P, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613(7944):508–18.

    Article  Google Scholar 

  41. Bulik-Sullivan BK, Loh PR, Finucane HK, et al. LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet. 2015;47(3):291–5.

    Article  Google Scholar 

  42. Auton A, Brooks LD, Durbin RM, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74.

    Article  Google Scholar 

  43. Zhou H, Ji Y, Sun L, et al. Exploring the causal relationships and mediating factors between depression, anxiety, panic, and atrial fibrillation: a multivariable mendelian randomization study. J Affect Disord. 2024;349:635–45.

    Article  Google Scholar 

  44. Xiang M, Wang Y, Gao Z, et al. Exploring causal cforrelations between inflammatory cytokines and systemic lupus erythematosus: a mendelian randomization. Front Immunol. 2022;13:985729.

    Article  Google Scholar 

  45. Li S, Gong M. Mendelian randomisation analysis reveals the possible causal relationship between infections, microbiota and clinical disease. Gut 2023.

  46. Kwok MK, Kawachi I, Rehkopf D, Schooling CM. The role of cortisol in ischemic heart disease, ischemic stroke, type 2 diabetes, and cardiovascular disease risk factors: a bi-directional mendelian randomization study. BMC Med. 2020;18(1):363.

    Article  Google Scholar 

  47. Kjeldgaard Jensen S, Tingskov Pedersen CE, Fischer-Rasmussen K et al. Genetic predisposition to high BMI increases risk of early life respiratory infections and episodes of severe wheeze and asthma. The European respiratory journal 2024.

  48. Ge P, Luo Y, Chen H. Occupational exposure to diesel exhausts and pancreatic cancer. Int J Surg (London England) 2024.

  49. Chen H, Liu ZJIC, Images M. Iron Deficiency Anemia Contributes To The Development Of Multiple Cardiovascular Diseases: A Two-Sample Mendelian Randomization Study. 2024.

  50. Zhao SS, Holmes MV, Zheng J, Sanderson E, Carter AR. The impact of education inequality on rheumatoid arthritis risk is mediated by smoking and body mass index: mendelian randomization study. Rheumatology (Oxford). 2022;61(5):2167–75.

    Article  Google Scholar 

  51. Burgess S, Thompson SG. Avoiding bias from weak instruments in mendelian randomization studies. Int J Epidemiol. 2011;40(3):755–64.

    Article  Google Scholar 

  52. Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37(7):658–65.

    Article  Google Scholar 

  53. Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent estimation in mendelian randomization with some Invalid instruments using a weighted median estimator. Genet Epidemiol. 2016;40(4):304–14.

    Article  Google Scholar 

  54. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512–25.

    Article  Google Scholar 

  55. Zhao Q, Wang J, Hemani G, Bowden J, Small DS. Statistical inference in two-sample summary-data mendelian randomization using robust adjusted profile score. 2020; 48%J The Annals of Statistics(3): 1742–69, 28.

  56. Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693–8.

    Article  Google Scholar 

  57. Greco MF, Minelli C, Sheehan NA, Thompson JR. Detecting pleiotropy in mendelian randomisation studies with summary data and a continuous outcome. Stat Med. 2015;34(21):2926–40.

    Article  MathSciNet  Google Scholar 

  58. Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13(11):e1007081.

    Article  Google Scholar 

  59. Burgess S, Thompson SG. Interpreting findings from mendelian randomization using the MR-Egger method. Eur J Epidemiol. 2017;32(5):377–89.

    Article  Google Scholar 

  60. Burgess S, Thompson SG. Erratum to: interpreting findings from mendelian randomization using the MR-Egger method. Eur J Epidemiol. 2017;32(5):391–2.

    Article  Google Scholar 

  61. Sanderson E, Davey Smith G, Windmeijer F, Bowden J. An examination of multivariable mendelian randomization in the single-sample and two-sample summary data settings. Int J Epidemiol. 2019;48(3):713–27.

    Article  Google Scholar 

  62. Carter AR, Sanderson E, Hammerton G, et al. Mendelian randomisation for mediation analysis: current methods and challenges for implementation. Eur J Epidemiol. 2021;36(5):465–78.

    Article  Google Scholar 

  63. Gagnon E, Mitchell PL, Manikpurage HD, et al. Impact of the gut microbiota and associated metabolites on cardiometabolic traits, chronic diseases and human longevity: a mendelian randomization study. J Transl Med. 2023;21(1):60.

    Article  Google Scholar 

  64. Gusev A, Ko A, Shi H, et al. Integrative approaches for large-scale transcriptome-wide association studies. Nat Genet. 2016;48(3):245–52.

    Article  Google Scholar 

  65. The GTEx. Consortium atlas of genetic regulatory effects across human tissues. Science. 2020;369(6509):1318–30.

    Article  Google Scholar 

  66. Fromer M, Roussos P, Sieberts SK, et al. Gene expression elucidates functional impact of polygenic risk for schizophrenia. Nat Neurosci. 2016;19(11):1442–53.

    Article  Google Scholar 

  67. Alboni S, van Dijk RM, Poggini S, et al. Fluoxetine effects on molecular, cellular and behavioral endophenotypes of depression are driven by the living environment. Mol Psychiatry. 2017;22(4):552–61.

    Article  Google Scholar 

  68. Butler A, Hoffman P, Smibert P, Papalexi E, Satija R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat Biotechnol. 2018;36(5):411–20.

    Article  Google Scholar 

  69. Andreatta M, Carmona SJ, UCell. Robust and scalable single-cell gene signature scoring. Comput Struct Biotechnol J. 2021;19:3796–8.

    Article  Google Scholar 

  70. Bhuva DD, Cursons J, Davis MJ. Stable gene expression for normalisation and single-sample scoring. Nucleic Acids Res. 2020;48(19):e113.

    Article  Google Scholar 

  71. Fan C, Chen F, Chen Y, Huang L, Wang M, Liu Y, et al. irGSEA: the integration of single-cell rank-based gene set enrichment analysis. Brief Bioinform. 2024;25(4). https://doi.org/10.1093/bib/bbae243

  72. Zheng R, Zhang Y, Tsuji T, Zhang L, Tseng Y-H, Chen K. MEBOCOST: metabolic cell-cell communication modeling by single cell transcriptome. 2022: 2022.05.30.494067.

  73. Chu T, Wang Z, Pe’er D, Danko CG. Cell type and gene expression deconvolution with BayesPrism enables bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology. Nat cancer. 2022;3(4):505–17.

    Article  Google Scholar 

  74. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.

    Article  Google Scholar 

  75. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.

    Article  Google Scholar 

  76. Lin S, Xu H, Zhang A, et al. Prognosis analysis and validation of m(6)a signature and Tumor Immune Microenvironment in Glioma. Front Oncol. 2020;10:541401.

    Article  Google Scholar 

  77. Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7.

    Article  Google Scholar 

  78. Doncheva NT, Morris JH, Gorodkin J, Jensen LJ. Cytoscape StringApp: Network Analysis and Visualization of Proteomics Data. J Proteome Res. 2019;18(2):623–32.

    Article  Google Scholar 

  79. Hemani G, Zheng J, Elsworth B et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife 2018; 7.

  80. Yu G, Wang LG, Yan GR, He QY. DOSE: an R/Bioconductor package for disease ontology semantic and enrichment analysis. Bioinf (Oxford England). 2015;31(4):608–9.

    Google Scholar 

  81. Lord J, Jermy B, Green R et al. Mendelian randomization identifies blood metabolites previously linked to midlife cognition as causal candidates in Alzheimer’s disease. Proc Natl Acad Sci U S A 2021; 118(16).

  82. Chen X, Kong J, Pan J, et al. Kidney damage causally affects the brain cortical structure: a mendelian randomization study. EBioMedicine. 2021;72:103592.

    Article  Google Scholar 

  83. Kelly DM, Georgakis MK, Franceschini N, Blacker D, Viswanathan A, Anderson CD. Interplay between chronic kidney Disease, Hypertension, and stroke: insights from a multivariable mendelian randomization analysis. Neurology. 2023;101(20):e1960–9.

    Article  Google Scholar 

  84. Zhang Z, Wang M, Gill D, Zhu W, Liu X. Genetically predicted sleep traits and functional Outcome after ischemic stroke: a mendelian randomization study. Neurology. 2023;100(11):e1159–65.

    Article  Google Scholar 

  85. Seyedsalehi A, Warrier V, Bethlehem RAI, Perry BI, Burgess S, Murray GK. Educational attainment, structural brain reserve and Alzheimer’s disease: a Mendelian randomization analysis. Brain 2022.

  86. Papadimitriou N, Dimou N, Tsilidis KK, et al. Physical activity and risks of breast and colorectal cancer: a mendelian randomisation analysis. Nat Commun. 2020;11(1):597.

    Article  Google Scholar 

  87. Liu CX, Liu YB, Peng Y, Peng J, Ma QL. Causal effect of air pollution on the risk of cardiovascular and metabolic diseases and potential mediation by gut microbiota. Sci Total Environ. 2024;912:169418. https://doi.org/10.1016/j.scitotenv.2023.169418

  88. Xia Z, Liu Y, Liu C, Dai Z, Liang X, Zhang N, et al. The causal effect of air pollution on the risk of essential hypertension: a Mendelian randomization study. Front Public Health 2024;12. https://doi.org/10.3389/fpubh.2024.1247149

  89. Liu C, Peng J, Liu Y, Peng Y, Kuang Y, Zhang Y, et al. Causal relationship between particulate matter 2.5 (PM2.5), PM2.5 absorbance, and COVID-19 risk: a two-sample Mendelian randomisation study. J Glob Health. 2023;13. https://doi.org/10.7189/jogh.13.06027

  90. Li Z, Wen J, Wu W, Dai Z, Liang X, Zhang N, et al. Causal relationship and shared genes between air pollutants and amyotrophic lateral sclerosis: a large-scale genetic analysis. CNS Neurosci Ther: 2024;30(7):e14812. https://doi.org/10.1111/cns.14812

  91. Jiang F, Zhao J, Sun J, Chen W, Zhao Y, Zhou S, et al. Impact of ambient air pollution on colorectal cancer risk and survival: insights from a prospective cohort and epigenetic Mendelian randomization study. EBioMedicine. 2024;103:105126. https://doi.org/10.1016/j.ebiom.2024.105126

  92. Li W, Wang W. Causal effects of exposure to ambient air pollution on cancer risk: insights from genetic evidence. Sci Total Environ. 2024;912:168843.

    Article  Google Scholar 

  93. Wen J, Zhang J, Zhang H, Zhang N, Lei R, Deng Y, et al. Large-scale genome-wide association studies reveal the genetic causal etiology between air pollutants and autoimmune diseases. J Transl Med. 2024;22(1). https://doi.org/10.1186/s12967-024-04928-y

  94. Braithwaite I, Zhang S, Kirkbride JB, Osborn DPJ, Hayes JF. Air Pollution (Particulate Matter) exposure and associations with Depression, anxiety, bipolar, psychosis and suicide risk: a systematic review and Meta-analysis. Environ Health Perspect. 2019;127(12):126002.

    Article  Google Scholar 

  95. Li C, Chen H, Gu Y, Chen W, Liu M, Lei Q, et al. Causal effects of PM2.5 exposure on neuropsychiatric disorders and the mediation via gut microbiota: a Mendelian randomization study. Ecotoxicol Environ Saf. 2024;275:116257. https://doi.org/10.1016/j.ecoenv.2024.116257

  96. Zhu Y, Hu F, Zhou X, Xue Q. Estimating the causal effect of air pollution on mental disorders: a two-sample Mendelian randomization study. J Neurorestoratology. 2024;12(2):100114. https://doi.org/10.1016/j.jnrt.2024.100114

  97. Kain J, Owen KA, Marion MC, Langefeld CD, Grammer AC, Lipsky PE. Mendelian randomization and pathway analysis demonstrate shared genetic associations between lupus and coronary artery disease. Cell Rep Med. 2022;3(11):100805.

    Article  Google Scholar 

  98. Vysotskiy M, Zhong X, Miller-Fleming TW, Zhou D, Cox NJ, Weiss LA. Integration of genetic, transcriptomic, and clinical data provides insight into 16p11.2 and 22q11.2 CNV genes. Genome Med. 2021;13(1):172.

    Article  Google Scholar 

  99. Niculescu AB 3rd, Segal DS, Kuczenski R, Barrett T, Hauger RL, Kelsoe JR. Identifying a series of candidate genes for mania and psychosis: a convergent functional genomics approach. Physiol Genom. 2000;4(1):83–91.

    Article  Google Scholar 

  100. Forrest MP, Zhang H, Moy W et al. Open chromatin profiling in hiPSC-Derived neurons prioritizes functional Noncoding Psychiatric risk variants and highlights neurodevelopmental loci. Cell Stem Cell 2017; 21(3): 305 – 18.e8.

  101. Buck SA, Quincy Erickson-Oberg M, Logan RW, Freyberg Z. Relevance of interactions between dopamine and glutamate neurotransmission in schizophrenia. Mol Psychiatry. 2022;27(9):3583–91.

    Article  Google Scholar 

  102. Yadav S, Oses-Prieto JA, Peters CJ, et al. TAOK2 kinase mediates PSD95 Stability and dendritic spine maturation through Septin7 phosphorylation. Neuron. 2017;93(2):379–93.

    Article  Google Scholar 

  103. Glessner JT, Reilly MP, Kim CE, et al. Strong synaptic transmission impact by copy number variations in schizophrenia. Proc Natl Acad Sci USA. 2010;107(23):10584–9.

    Article  Google Scholar 

  104. Ku T, Li B, Gao R, et al. NF-κB-regulated microRNA-574-5p underlies synaptic and cognitive impairment in response to atmospheric PM(2.5) aspiration. Part Fibre Toxicol. 2017;14(1):34.

    Article  Google Scholar 

  105. Liu J, Liu B, Yuan P, et al. Role of PKA/CREB/BDNF signaling in PM2.5-induced neurodevelopmental damage to the hippocampal neurons of rats. Ecotoxicol Environ Saf. 2021;214:112005.

    Article  Google Scholar 

  106. Li M, Xiao L, Chen X. Histone acetylation and methylation underlie oligodendroglial and myelin susceptibility in Schizophrenia. Front Cell Neurosci. 2022;16:823708.

    Article  Google Scholar 

  107. Ma JH, Song SH, Guo M, et al. Long-term exposure to PM2.5 lowers influenza virus resistance via down-regulating pulmonary macrophage Kdm6a and mediates histones modification in IL-6 and IFN-β promoter regions. Biochem Biophys Res Commun. 2017;493(2):1122–8.

    Article  Google Scholar 

  108. Fiorica PN, Wheeler HE. Transcriptome association studies of neuropsychiatric traits in African americans implicate PRMT7 in schizophrenia. PeerJ. 2019;7:e7778.

    Article  Google Scholar 

  109. Hauberg ME, Holm-Nielsen MH, Mattheisen M, et al. Schizophrenia risk variants affecting microRNA function and site-specific regulation of NT5C2 by miR-206. Eur Neuropsychopharmacology: J Eur Coll Neuropsychopharmacol. 2016;26(9):1522–6.

    Article  Google Scholar 

  110. Duarte RRR, Troakes C, Nolan M, Srivastava DP, Murray RM, Bray NJ. Genome-wide significant schizophrenia risk variation on chromosome 10q24 is associated with altered cis-regulation of BORCS7, AS3MT, and NT5C2 in the human brain. Am J Med Genet Part B Neuropsychiatric Genetics: Official Publication Int Soc Psychiatric Genet. 2016;171(6):806–14.

    Article  Google Scholar 

  111. Konopaske GT, Balu DT, Presti KT, Chan G, Benes FM, Coyle JT. Dysbindin-1 contributes to prefrontal cortical dendritic arbor pathology in schizophrenia. Schizophr Res. 2018;201:270–7.

    Article  Google Scholar 

  112. Kim P, Scott MR, Meador-Woodruff JH. Abnormal expression of ER quality control and ER associated degradation proteins in the dorsolateral prefrontal cortex in schizophrenia. Schizophr Res. 2018;197:484–91.

    Article  Google Scholar 

  113. Kane MS, Diamonstein CJ, Hauser N, Deeken JF, Niederhuber JE, Vilboux T. Endosomal trafficking defects in patient cells with KIAA1109 biallelic variants. Genes Dis. 2019;6(1):56–67.

    Article  Google Scholar 

  114. Kumar A, Karuppagounder SS, Chen Y, et al. 2-Deoxyglucose drives plasticity via an adaptive ER stress-ATF4 pathway and elicits stroke recovery and Alzheimer’s resilience. Neuron. 2023;111(18):2831–e4610.

    Article  Google Scholar 

  115. Liebmann T, Fritz N, Kruusmägi M et al. Regulation of neuronal Na,K-ATPase by Extracellular Scaffolding proteins. Int J Mol Sci 2018; 19(8).

  116. Yan Z, Rein B. Mechanisms of synaptic transmission dysregulation in the prefrontal cortex: pathophysiological implications. Mol Psychiatry. 2022;27(1):445–65.

    Article  Google Scholar 

  117. Brown MN, Mathews GC. Activity- and age-dependent modulation of GABAergic neurotransmission by system A-mediated glutamine uptake. J Neurochem. 2010;114(3):909–20.

    Article  Google Scholar 

Download references

Acknowledgements

The authors appreciate the public databases, Biorender online tool (Biorender.com), websites, and software used in the study, as well as their contributors, and thank the High-Performance Computing Center of Central South University for the support of this work. The CommonMind Consortium’s bio-samples and/or data for this publication were obtained from the NIMH Repository & Genomics Resource, a centralized national biorepository for genetic studies of psychiatric disorders. This research received no grant from any funding agency in the public, commercial, or not-for-profit sectors.

Funding

Not applicable.

Author information

Author notes

  1. Xisong Liang and Jie Wen contributed equally to this work.

Authors and Affiliations

  1. Department of Neurosurgery, Xiangya Hospital, Central South University, Changsha, 410008, Hunan, P. R. China

    Xisong Liang, Jie Wen, Chunrun Qu, Ziyu Dai, Zhixiong Liu, Fan Fan & Quan Cheng

  2. Hypothalamic Pituitary Research Centre, Xiangya Hospital, Central South University, Changsha, 410008, Hunan, P. R. China

    Xisong Liang, Jie Wen, Chunrun Qu, Nan Zhang, Ziyu Dai, Hao Zhang, Peng Luo, Ming Meng, Zhixiong Liu, Fan Fan & Quan Cheng

  3. National Clinical Research Center for Geriatric Disorders, Xiangya Hospital, Central South University, Changsha, 410008, Hunan, P. R. China

    Xisong Liang, Jie Wen, Chunrun Qu, Ziyu Dai, Zhixiong Liu, Fan Fan & Quan Cheng

  4. College of Life Science and Technology, Huazhong University of Science and Technology, Wuhan, 430074, Hubei, P. R. China

    Nan Zhang

  5. Department of Neurosurgery, The Second Affiliated Hospital of Chongqing Medical University, Chongqing, 400010, P. R. China

    Hao Zhang

  6. Department of Oncology, Zhujiang Hospital, Southern Medical University, Guangzhou, 510000, Guangdong, P. R. China

    Peng Luo

  7. Department of Neurosurgery, The First Affiliated Hospital of Kunming Medical University, Kunming, Yunnan, 650032, P. R. China

    Ming Meng

Authors
  1. Xisong Liang
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jie Wen
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Chunrun Qu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Nan Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Ziyu Dai
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Hao Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Peng Luo
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Ming Meng
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Zhixiong Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Fan Fan
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Quan Cheng
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

X.L: Resources, Methodology, Software, Formal Analysis, Writing - Original Draft, and Writing - Review & Editing. J.W: Resources, Methodology, Software, Formal Analysis, Writing - Review & Editing, and Writing - Review & Editing. C.Q: Methodology, Visualization, and Writing - Review & Editing. N.Z: Visualization, Methodology, and Writing - Review & Editing. Z.D: Methodology, Software, and Writing - Review & Editing. H.Z: Validation, Software, and Writing - Review & Editing. P.L: Methodology and Writing - Review & Editing. M.M: Methodology, Validation, and Writing - Review & Editing. Z.L: Validation and Writing - Review & Editing. F.F: Conceptualization, Supervision, Project administration, and Writing - Review & Editing. Q.C: Resources, Conceptualization, Supervision, and Writing - Review & Editing. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Fan Fan or Quan Cheng.

Ethics declarations

Ethical approval and consent to participate

This research utilized publicly available data, which waived the ethical approval requirement. Each study contributing to the GWAS contains details for ethical approval and participant consent in their original publications. This research requires no specific ethical approval.

Consent for publication

All authors consented to the submission and publication of this study.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Liang, X., Wen, J., Qu, C. et al. Inhibitory neuron links the causal relationship from air pollution to psychiatric disorders: a large multi-omics analysis. J Big Data 11, 127 (2024). https://doi.org/10.1186/s40537-024-00960-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40537-024-00960-3

Keywords