Abstract
Background
Comorbidities and genetic correlations between gastrointestinal tract diseases and psychiatric disorders have been widely reported, but the underlying intrinsic link between Alzheimer’s disease (AD) and inflammatory bowel disease (IBD) is not adequately understood.
Methods
To identify pathogenic cell types of AD and IBD and explore their shared genetic architecture, we developed Pathogenic Cell types and shared Genetic Loci (PCGL) framework, which studied AD and IBD and its two subtypes of ulcerative colitis (UC) and Crohn’s disease (CD).
Results
We found that monocytes and CD8 T cells were the enriched pathogenic cell types of AD and IBDs, respectively. By PCGL framework, there was a significant global genetic correlation between AD and each of IBD, UC, and CD. Especially, local genetic correlations between AD and IBD showed strong signals in chr6. Bidirectional two-sample MR Analyses also validated these. Cross-trait meta-analysis identified two key genetic loci rs660895 (on chr6) and rs917117 (on chr7), which have not been previously reported. Two loci are located on the genes HLA-DRB1 and JAZF1, respectively. MAGMA genome-wide gene-based analysis identified six overlapping genes including HLA-DRB1. Subsequently, for one thing, SMR analyses further validated six shared genes in specific tissues and monocytes. For another, pathway enrichment analysis revealed shared genes were enriched in several natural killer cell mediated cytotoxicity and chemokine signaling pathways.
Conclusions
PCGL not only revealed the significant genetic correlations underlying AD and IBDs but also identified enriched pathogenic cell types and new shared loci and genes. We highlighted the mediation of HLA-DRB1 effects in the comorbidity mechanisms.
Keywords: Alzheimer’s disease, inflammatory bowel disease, genome-wide association studies, pathogenic cell types, genetic loci, Mendelian randomization
Introduction
With the increase in the aging population worldwide, Alzheimer’s disease (AD) has become a rapidly increasing public health concern [1]. It is estimated that, by 2050, there will be 152 million people with AD and other dementias [2]. An increasing body of evidence suggests that neuroinflammation plays an important role in AD pathogenesis [3].
The comorbidities and associations between gastrointestinal tract diseases and psychiatric disorders have been widely reported in recent years. Inflammatory bowel disease is a chronic, non-specific gastrointestinal inflammatory disease that mainly includes ulcerative colitis and Crohn’s disease [4]. CD affects any part of the gastrointestinal tract, and its lesions are characterized by a focal and trans mural nature. UC is characterized by inflammation of the colonic mucosa, beginning in the rectum and with a continuous and circular extension [5]. Wang et al. [6] used IBD and AD as examples to demonstrate the possible link between chronic intestinal inflammation and central nervous system (CNS) inflammation, and a new intervention approach for AD was proposed. Xing et al. [7] performed a meta-analysis evaluating the associations of ulcerative colitis, Crohn’s disease and related medications with risk of AD and found that medications for the treatment of inflammatory bowel disease (IBD) might be associated with a lower risk of AD.
It is not difficult to find that there must be an association between AD and IBD, but studies on comorbidity mechanisms, genetic association and enriched pathogenic cell types between AD and IBD are still relatively rare.
In order to identify the causal cell types that uniquely and independently contribute to AD, IBD, UC, and CD pathogenesis via mediation of genetic effects and to examine the genetic correlations and potential causality between AD and each of IBD, UC and CD, we developed Pathogenic Cell types and shared Genetic Loci (PCGL) framework, which leveraged large-scale statistics data. Our PCGL framework is a novel integration and a scientifically generalizable pipeline. Specifically, we applied stratified linkage disequilibrium score regression (LDSC) [8] to estimate the enrichment of AD, IBD, UC, and CD GWAS in open chromatin regions (OCRs, i.e. ATAC-seq peaks) from each of the 16 hematopoietic cell types to identify the causal cell types that contribute to AD, IBD, UC and UC pathogenesis, respectively. We then estimate genetic correlations and local genetic correlations between AD and each of IBD, UC and CD by determining that ingle nucleotide polymorphism (SNP) heritability for AD, IBD, UC, and CD was enriched in cell types. We used bidirectional two-sample Mendelian randomization (MR) [9] to provide further evidence for the causal relationship between AD and each of IBD, UC, and CD. By demonstrating the correlations of AD-IBD, AD-CD, and AD-UC, and then we performed cross-trait meta-analyses to identify risk SNPs underlying the joint phenotypes AD-IBD, AD-UC, and AD-CD. Finally, MAGMA genome-wide gene-based analysis was conducted to show the overlapping genes among four diseases which we performed SMR analysis to further validate the causal relationship with, and then we wanted to demonstrate differential expression levels of specific overlapping genes for diseased and control groups in AD. At the same time, we applied an enrichment analysis on the overlapping genes to better understand the shared etiology among AD, IBD, UC, and CD by MAGMA. A general overview of the study is visualized in Fig. 1.
Figure 1.
The schematic illustration of PCGL framework.
Materials and methods
Study samples
Dataset for AD
We collected GWAS summary statistics data for AD from the published study in 2022 [10]. The data were downloaded from the European Bioinformatics Institute GWAS Catalog with accession number GCST90027158. Regarding the validation dataset for AD, we included individuals of European ancestry [11] with accession number GCST007320. A detailed description of these GWAS data sources was shown in Supplementary Table S1. AD snRNA-seq data were collected from the studies of Anderson AG et al. (accession GSE214979) [12].
GWAS datasets for IBD, UC and CD
The case sample for IBD comprises case samples in UC and CD GWAS [13]. All individuals were restricted to European ancestry to decrease the bias from population stratification. A detailed description of these included data sources was shown in Supplementary Table S1.
Immune cell ATAC-seq data
ATAC-seq data were collected from the studies of Corces et al.(accession GSE74912) [14] and Buenrostro et al. (accession GSE118189) [15] for 16 different human hematopoietic progenitor and terminal populations. Alignment of ATAC-seq data and peak-calling could be performed with reference to the study of Ulirsch et al. [16]. To identify cell-type-specific peaks for each cell type, we used ATAC-seq peaks for that cell type and removed any peaks that overlapped with a peak present in any one of the other 15 cell types.
Sources of expression quantitative trait loci data
We obtained cis-expression quantitative trait loci (eQTL) data for brain and intestinal tissues of European ancestry from website (https://yanglab.westlake.edu.cn/software/smr/#DataResource). AD tissue-specific genes were obtained from eQTL data for brain tissues, while IBD (UC, CD) tissue-specific genes were obtained from eQTL data for intestinal tissues from sigmoid colon, transverse colon, and terminal ileum from the GTEx Consortium (V8) [17]. The single-cell eQTL (sc-eQTL) data of monocyte cell were obtained from the 1 M-scBloodNL study by Oelen et al. [18] and downloaded from the eQTLGen database (https://eqtlgen.org/sc/).
Statistical analyses
Linkage disequilibrium score regression
To calculate enrichments of the AD, IBD, UC and CD GWAS data within annotations (e.g. ATAC-seq peaks), we applied stratified LDSC (https://github.com/bulik/ldsc) [8, 19], which was run on the summary statistics data for the AD, IBD, UC and CD GWAS.
Given genomic annotations 
. Firstly, compute the LD score of SNP 
to category 
,
 |
(1) |
where 
is the squared correlation between SNPs 
and 
in the population. Then, we model the causal effect of SNP 
on phenotype 
as drawn from a distribution with mean 0 and variance
 |
(2) |
we can call 
the per-SNP heritability of SNP 
. For the cell type and cell type group analyses described in this manuscript, estimating 
is the goal. To estimate the 
, we first estimate 
from a reference panel, and then perform weighted regression 
on 
. 
is the GWAS sample size.
If effect sizes for both phenotypes are drawn from a bivariate normal distribution, then the optimal regression weights for genetic covariance estimation are:
 |
(3) |
Where 
denotes the 
score for study 
and SNP 
, 
is the sample size for study 
, 
is the genetic covariance, 
is the LD Score, 
is the heritability explained by SNPs in
, 
denote a set of 
SNPs, 
is the number of individuals included in both studies and 
is the phenotypic correlation among the 
overlapping samples.
We estimate genetic covariance by regressing 
against 
.We applied partitioned heritability analyses using LDSC under two different models based on the European ancestry samples of the 1000 Genomes Project Phase 1 [20] which was restricted to HapMap3 SNPs [21]. Details are provided in the Supplementary Note.
We used bivariate LDSC [22] to estimate genetic correlations (
, i.e. the proportion of genetic variance shared by two traits divided by the square root of the product of their SNP heritability estimates) between AD and each of IBD, UC, and CD, as well as between UC and CD.
Estimation of local genetic correlations using ρ-HESS
We first utilized single-trait LDSC with the GWAS summary statistics data of AD and each of IBD, CD, and UC to estimate SNP-based heritability of each of these four traits, and used genome-wide genetic correlation to quantify the extent of the shared genetic basis of these two diseases [22]. We performed ρ-HESS [23] to estimate the local SNP heritability per trait and genetic covariance between traits based on the 1000 Genomes European reference of hg19 genome build. Local genetic correlation estimates were then calculated from the local single-trait SNP heritability and local cross-trait genetic covariance estimates to quantify shared genetic foundation due to genetic variation at a small region in the genome.
Mendelian randomization analyses of AD and each of IBD, CD, and UC
MR analysis uses genetic variants as instrumental variables (IVs) to estimate the causal association between exposure and outcome. We used a strict threshold (
) in the selection of significant SNPs for exposures. Using the 1000 Genomes Project Phase 3 (European) as the reference panel and significant SNPs, we performed LD clumping to identify IVs (r2 < 0.001 within 10 000 kb).
We conducted a bidirectional two-sample MR analyses for AD and each of IBD, CD and UC. Inverse variance weighted (IVW) was utilized under a multiplicative random-effects model as the primary MR analysis [24, 25]. The potential horizontal pleiotropy was accessed by MR-Egger regression based on its intercept term [26]. Forest and scatter plots were used to visualize combined results of single and multi-SNP analyses. In addition, MR pleiotropy residual sum and outlier tests were used to identify horizontal pleiotropic outliers [27]. Moreover, to examine whether an individual IVs has a large influence on the regression coefficients, we conducted a leave-one-out approach.
Multi-trait analysis of GWAS
To identify risk SNPs associated with joint phenotypes comprising AD and each of IBD, UC, and CD, we implemented cross-trait meta-analysis of GWAS summary statistics using Multi-Trait Analysis of GWAS (MTAG) [28]. Unlike conventional inverse-variance meta-analysis, MTAG jointly analyzes GWAS summary statistics across multiple traits, accounting for sample overlap and exploiting genetic correlation to improve power. We implemented MTAG with options assuming: equal SNP heritability (
) for each trait and perfect genetic covariance(
)between traits. The upper bound for the false discovery rate (‘maxFDR’) was calculated to examine the assumptions on the equal variance–covariance of shared SNP effect sizes underlying the traits. Because summary data for each trait obtained using MTAG can substitute the summary GWAS data for individual traits. The genome-wide significance level for MTAG results of AD-IBD was set at 
.
Cross-phenotype association
We further used R package Cross Phenotype Association (CPASSOC) [29] to analyze the GWAS cross-trait association for AD-IBD, AD-UC, and AD-CD as a sensitivity analysis complementary to MTAG. CPASSOC jointly analyzes GWAS summary statistics and has two estimation models, SHom (for homogenous data) and SHet (for heterogeneous data). CPASSOC assumes the presence of heterogeneous effects across traits and estimates the cross-trait statistic SHet and p-value through a sample size weighted meta-analysis of GWAS summary data. We used SHet to combine summary statistics of AD and each of IBD, UC, and CD, followed by clumping in PLINK v1.9 [30].
Verification of shared genes and the four diseases based on SMR
To further validate the causal relationship between shared disease genes and the four diseases, we performed Mendelian randomization analysis based on summary data using the SMR software [31]. This method integrates eQTL and GWAS summary data to identify causal genes that affect traits through expression levels.
Gene set-based enrichment analysis and pathway analysis
We performed gene analysis using multi-marker analysis of genomic annotation (MAGMA) [32] to prioritize risk genes at susceptibility loci for AD, IBD, CD, and UC. By using the LD information of the European population and summary statistics from AD, IBD, UC, CD as input, all SNPs located between the transcription start and end sites were aggregated to that gene to calculate the gene p value based on a multiple regression model. Taking into consideration the vital regulatory function of peripheral genes with smaller effects in AD, IBD, UC, and CD, 
was considered statistically significant and defined as a threshold. The MAGMA results were tested for enrichment in gene sets for pathways or other biological processes to explore the potential genetic pathways contributing to the comorbidity of AD, IBD, UC and CD including those defined by Kyoto Encyclopedia of Genes and Genomes (KEGG) (Molecular Signatures Database [MsigDB] c2), gene ontology (GO) biological processes (MsigDB c5). The visualization of the pathway enrichment results was visualized using the R package ‘ggplot2’ in the R software (version 3.6.2) [33].
Results
AD, IBD, UC, and CD GWAS associations are enriched in progenitor and terminal peripheral immune cells
LDSC has the distinct advantage in that it leverages the genome-wide polygenic signal in the GWAS summary statistics data rather than selecting variants based on p-value thresholds or fine-mapping posterior probabilities.
Applying LDSC, we observed a statistically significant enrichment in 16 hematopoietic cell populations (enrichment p-value threshold of 
) (Fig. 2). The strongest enrichment of AD GWAS was observed in the OCR from the monocyte cell, it indicates monocytes constitute the major players involved in AD etiology. Lessons learned from many recent studies highlight the known and emerging roles of monocytes in AD pathogenesis and their role as therapeutic targets for AD [34]. According to the enrichment results of IBD, UC, and CD, the strongest enrichment was observed in the OCR from CD8 T cells, this demonstrates that CD8 T cells are involved in the pathogenesis of IBD, a complex multifactorial chronic disease. The function and plasticity of different subsets of CD8 T cells in health and IBD remain to be further investigated in a challenging field due to the limited availability of mucosal samples and adequate controls [35]. More details are provided in the Supplementary Note.
Figure 2.
LDSC enrichment results of GWAS from different diseases in OCRs of 16 hematopoietic cell types.
Genetic correlations between AD and IBDs
We performed a preliminary genome-wide association analysis for AD as shown in Fig. 3(A) and Supplementary Fig. S2A. We first applied stratified LDSC (S-LDSC) [8] with the baseline-LD model [36] to estimate the liability-scale SNP heritability for AD and each of IBD, UC and CD (Table 1). Single-trait LDSC shows SNP heritability estimates of 0.0217 (SE = 0.002) for GWASAD, 0.3259 (SE = 0.0302) for GWASIBD, 0.4836 (SE = 0.0533) for GWASCD and 0.2745 (SE = 0.0315) for GWASUC. Mean χ2 statistics for GWASAD, GWASIBD, GWASCD and GWASUC are 1.2769, 1.2837, 1.2322 and 1.1905, respectively.
Figure 3.
Genetic correlations between AD and IBD, UC, and CD.
Table 1.
Single-trait LDSC of AD GWAS, IBD GWAS, CD GWAS and UC GWAS.
| Phenotype | Observed-scale h2 (SE) | Mean chi2 | Lambda GC | Intercept (SE) |
|---|---|---|---|---|
| AD | 0.0217 (0.002) | 1.2769 | 1.2038 | 1.0684 (0.01) |
| IBD | 0.3259 (0.0302) | 1.2837 | 1.1747 | 1.0615 (0.0077) |
| CD | 0.4836 (0.0533) | 1.2322 | 1.1428 | 1.0312 (0.0089) |
| UC | 0.2745 (0.0315) | 1.1905 | 1.127 | 1.0453 (0.0082) |
Pairwise LDSC shows negative genetic correlation (without constrained intercept) between AD and each of IBD, CD, and UC, −0.0423 (SE = 0.0632), −0.0467 (SE = 0.0564), and − 0.0478 (SE = 0.0738), respectively. For comparison, the genetic correlation between UC and CD was also shown in Fig. 3(B), 0.7057 (SE = 0.0567).
When studying the validation dataset for AD, we also found a negative genetic correlation between AD and IBD (including UC and CD. For more details, please see the Supplementary Fig. S5.
Local genetic correlations between AD and IBDs
We used the heritability estimation from summary statistics (ρ-HESS) method [23] to evaluate local genetic correlations across the genome between AD and each of IBD, UC and CD. The local genomic regions around individual AD loci from GWAS showed signals of genetic overlap with UC (Fig. 3(D), Supplementary Fig. S1), and we also validated in the validation dataset for AD (Supplementary Fig. S6). Figure 3(C) shows a contrast plot of degrees of polygenicity between AD and UC. Contrast polygenicity plots for contrasting degrees of polygenicity among other traits are shown in Supplementary Fig. S2 and Supplementary Fig. S7. We identified a region (chr6:31571218-32,682,664, 
) that show genome wide significant positive local genetic correlation between AD and UC (Fig. 3(D)). Although we observed no local correlation for specific genomic regions of AD and IBD (Supplementary Fig. S1), we found strong signals in chromosome 6 (31571218-32682664; p = 2.68 × 10−4; Table S1).
Bidirectional two-sample Mendelian randomization analysis for causality of each of IBD, UC, and CD on AD
We used bidirectional two-sample MR Analysis to provide evidence for the causal relationship between AD and IBD, AD and UC, and AD and CD. In this study, a negative genetic causal relationship was found between AD and IBD (including UC and CD). This is consistent with the direction of the genetic correlations calculated above. In the reverse MR studies, there is a negative genetic causal relationship between IBD and AD, but there is no evidence of a genetic causal association between UC and CD and AD.
First, we identified a significant SNP from AD, IBD, UC, and CD GWAS as the IV in bidirectional two-sample MR analysis. For two-sample MR, we default to MR Egger, weighted median, IVW, simple mode and weighted mode. We also conducted heterogeneity test, gene pleiotropy test and leave-one-out sensitivity analysis, and the results showed that there was heterogeneity among these IVs, there was no horizontal pleiotropy, and gene pleiotropy did not affect the reliability of results (Supplementary Table S2), and our results are also reliable in the leave-one-out sensitivity analysis.
In the primary weighted median analysis, genetic liability to AD is significantly associated with an decreased IBD risk (odds ratio [OR] = 0.5251, 95% confidence interval [CI]: 0.3411- 0.8082, P = 3.4 × 10−3). In the primary Weighted median analysis, genetic liability to AD is significantly associated with an decreased UC risk (OR = 0.3921, 95% CI: 0.2447- 0.6284, P = 1 × 10−4). In the primary Weighted median analysis, genetic liability to AD is significantly associated with an decreased CD risk (OR = 0.3960, 95% CI: 0.2563–0.6118, P = 3.00 × 10−05). Bidirectional two-sample shows genetic liability to IBD is significantly associated with an decreased AD risk (OR = 0.8648, 95% CI: 0.8069–0.9268, P = 3.99 × 10−5). All of them are validated by Inverse variance weighted, Simple mode and Weighted mode approaches (Table 2, Supplementary Fig. S3). However, bidirectional MR shows genetic liability to UC that was found not to impart an influence on AD risk (OR = 1.0144, 95% CI: 0.9437-1.0904, P = .6980).
Table 2.
Mendelian randomization analysis for the causality of AD on inflammatory bowel diseases.
| Exposure | Outcome | Methods | SNPs | OR(95%CI) | OR(95%ICdo) | OR95(%ICup) | B | P |
|---|---|---|---|---|---|---|---|---|
| AD | IBD | MR Egger | 39 | 1.307855907 | 0.204696019 | 8.356230281 | 0.268389084 | 0.778267975 |
| Weighted median | 39 | 0.525058626 | 0.341131072 | 0.808154353 | −0.644245354 | 0.003410458 | ||
| IVW | 39 | 0.713900315 | 0.410930397 | 1.240243273 | −0.337011941 | 0.231718254 | ||
| Simple mode | 39 | 0.248467712 | 0.037203184 | 1.659433359 | −1.392442373 | 0.158835745 | ||
| Weighted mode | 39 | 0.284999493 | 0.140268094 | 0.579067618 | −1.255267878 | 0.001309882 | ||
| AD | UC | MR Egger | 29 | 1.403103521 | 0.15594085 | 12.62465539 | 0.338686584 | 0.764848663 |
| Weighted median | 29 | 0.392121797 | 0.244672364 | 0.628430203 | −0.936182782 | 0.000100078 | ||
| IVW | 29 | 0.619738952 | 0.303206871 | 1.266713934 | −0.478456935 | 0.189591385 | ||
| Simple mode | 29 | 0.214934392 | 0.085095208 | 0.542883601 | −1.537422451 | 0.002982861 | ||
| Weighted mode | 29 | 0.271864983 | 0.148348359 | 0.498223031 | −1.302449721 | 0.000235935 | ||
| AD | CD | MR Egger | 42 | 0.963066889 | 0.165430806 | 5.606560555 | −0.037632411 | 0.96680947 |
| Weighted median | 42 | 0.395937543 | 0.256252997 | 0.6117647 | −0.9264988 | 3.00E-05 | ||
| IVW | 42 | 0.676536701 | 0.374528726 | 1.222074239 | −0.390768582 | 0.195233109 | ||
| Simple mode | 42 | 0.178908058 | 0.091532858 | 0.34968965 | −1.720883248 | 1.01E-05 | ||
| Weighted mode | 42 | 0.215081902 | 0.123846594 | 0.373528437 | −1.536736383 | 2.55E-06 | ||
| IBD | AD | MR Egger | 230 | 1.032439987 | 0.919592872 | 1.159135047 | 0.03192492 | 0.589317875 |
| Weighted median | 230 | 0.864777727 | 0.806867254 | 0.926844551 | −0.145282768 | 3.99E-05 | ||
| IVW | 230 | 1.027883876 | 0.965374642 | 1.094440664 | 0.0275022 | 0.390256094 | ||
| Simple mode | 230 | 1.962001715 | 1.687702077 | 2.280882853 | 0.673965235 | 4.09E-16 | ||
| Weighted mode | 230 | 0.844461009 | 0.784648874 | 0.908832497 | −0.169056715 | 1.03E-05 | ||
| UC | AD | MR Egger | 199 | 1.0404237 | 0.944174294 | 1.1464848 | 0.039628034 | 0.424597484 |
| Weighted median | 199 | 1.014409471 | 0.943696351 | 1.090421272 | 0.014306641 | 0.6979636 | ||
| IVW | 199 | 1.006032939 | 0.951713555 | 1.063452621 | 0.006014814 | 0.831801123 | ||
| Simple mode | 199 | 1.724769663 | 1.187746535 | 2.504600355 | 0.545093513 | 0.004633826 | ||
| Weighted mode | 199 | 0.985339512 | 0.943813461 | 1.028692632 | −0.014769015 | 0.502182934 | ||
| CD | AD | NA | NA | NA | NA | NA | NA | NA |
Identification of risk SNPs from cross-trait GWAS meta-analysis of AD and IBD
After excluding SNPs that are genome-wide significant in the respective single-trait GWAS, we identified one SNP (rs660895) which is mapped to HLA-DRB1 gene and HLA-DQA1 associated with the joint phenotype AD-IBD and AD-UC. A SNP (rs917117) uniquely associated in the cross-trait GWAS meta-analyses of AD-CD (Table 3). In previous study, it has been confirmed that the HLA-DRB1/DQA1 gene is a risk gene belonging to AD [37]. The human leukocyte antigen (HLA) allele group HLA-DQA1*05 predisposes to UC. Infliximab has become a key IBD medication globally with high spending, but most patients develop anti-infliximab antibodies that may lead to treatment inefficacy. HLA-DQA1*05 is associated with the development of antibodies against infliximab in patients with IBD [38]. Previous studies have performed high-density SNP typing of MHC in IBD patients, with a primary role for HLA-DRB1*01:03 in both Crohn’s disease and ulcerative colitis [39]. The loci rs917117 (on chr7) is mapped to JAZF1 gene, the overexpression of which has an effect on proinflammatory cytokine levels, and variants within the gene region are significantly associated with AD [40].
Table 3.
Novel genetic variants associated with cross-trait AD and IBD (or UC or CD) revealed by MTAG and CPASSOC.
| Trait | SNP | P.AD | P.UC | Mtag_pval | P.CPASSOC |
|---|---|---|---|---|---|
| AD-UC | rs660895 |
|
|
|
|
| AD-IBD | rs660895 |
|
|
|
|
| AD-CD | rs917117 |
|
|
|
|
We validated our results on another AD GWAS dataset GCST007320. For details, see Supplementary Note.
Common pathways between AD, IBD, UC, and CD and differential expression analysis
MAGMA genome-wide gene-based analysis identified genes associated with AD, IBD, UC, and CD after Bonferroni correction is shown in Fig. 5(A), respectively. There are 6 overlapping genes between AD, IBD, UC and CD (Table 4). It can be seen from figure, 219 genes (PMAGMA < 3.23 × 10−7) are identified as significantly correlated with AD, 605 genes are related to IBD, 391 genes are associated with UC, 593 genes are related to CD. There are 12 overlapping genes between AD and IBD; 10 overlapping genes between AD and UC; 15 overlapping genes AD and CD; 6 overlapping genes between AD, IBD, UC and CD. We found that four of these six overlapping genes belonged to HLA class II genes, and that HLA-DRB1 and HLA-DQA1 were also genes mapped by the risk SNP identified by the above studies. Some evidence suggests that HLA mainly regulates the immune inflammatory responses, at least partly contributes to AD pathogenesis [41]. Meanwhile, the genes in the HLA are important in determining the susceptibility and phenotype of Crohn’s disease and ulcerative colitis [42].
Figure 5.
The genes were identified by MAGMA method, and further enrichment analysis and pathway analysis were performed.
Table 4.
Overlapping genes between AD, IBD, UC, and CD.
| Gene | Position | PMAGMA | |||
|---|---|---|---|---|---|
| AD | IBD | UC | CD | ||
| HLA-DQA1 | Chr6: 32587396-32,689,774 | 5.3044e-11 | 5e-10 | 5e-10 | 2.4817e-16 |
| HLA-DQB1 | Chr6: 32624464-32,716,689 | 4.7945e-08 | 1.6986e-14 | 5.3633e-14 | 5e-10 |
| HLA-DRA | Chr6: 32357619-32,447,823 | 1.1633e-10 | 2.7468e-14 | 5e-10 | 7.5781e-10 |
| HLA-DRB1 | Chr6: 32543769-32,639,836 | 1.7328e-10 | 5e-10 | 5e-10 | 1.8487e-13 |
| BTNL2 | Chr6: 32327513-32,424,900 | 1.4038e-11 | 1.0819e-14 | 5e-10 | 4.3042e-10 |
| TSBP1-AS1 | Chr6: 32205717-32,442,822 | 1.0456e-12 | 5.5423e-14 | 5e-10 | 1.1424e-10 |
We performed SMR analysis to further validate the causal relationship between shared disease genes (Tables 5, 6) and diseases with PSMR < 0.05. The results of tissue-specific SMR analysis showed that HLA-DRB1 is a causal gene shared by the four diseases (Table 5). HLA-DQA1 led to an increased risk of AD, IBD, and UC. BTNL2 gene expression increased the risk of AD and reduced the risk of IBD and UC. The expression levels of HLA-DQB1 and HLA-DRA genes in specific tissues were associated with an increased risk of IBD (UC, CD) and AD, respectively. The results of cell-specific SMR analysis (Table 6) also showed the expression levels of HLA-DRB1 in the monocyte cell were associated with both AD and IBD (UC) risk. HLA-DQA1 and HLA-DQB1 were associated with IBD (UC, CD) and AD, respectively. We used another snRNA-seq dataset of AD to validate that the gene expression levels of HLA-DQB1 and HLA-DRB1 differ between the diseased and unaffected groups (Fig. 4) using the Wilcoxon test [43]. Our results (Table 7) further support the involvement of HLA-DRB1 and HLA-DQB1 in the pathogenesis of AD. For more details, see Supplementary Note.
Table 5.
The SMR results of causal relationships between tissue-specific genes and diseases.
| Gene | β | SE | P SMR | Diseases | Tissue |
|---|---|---|---|---|---|
| HLA-DQA1 | 0.0303 | 0.0087 | 0.0005 | AD | Brain |
| HLA-DQA1 | 0.4206 | 0.0700 | 1.8305E-09 | IBD | Transverse colon |
| HLA-DQA1 | 0.9803 | 0.1462 | 2.0137E-11 | UC | Transverse colon |
| HLA-DQA1 | −0.0610 | 0.0450 | 0.1748 | CD | Transverse colon |
| HLA-DQB1 | 0.0017 | 0.0094 | 0.8528 | AD | Brain |
| HLA-DQB1 | 0.3248 | 0.0450 | 5.2865E-13 | IBD | Sigmoid colon |
| HLA-DQB1 | 0.3657 | 0.0462 | 2.4369E-15 | IBD | Transverse colon |
| HLA-DQB1 | 0.2158 | 0.0419 | 2.5190E-07 | IBD | Terminal ileum |
| HLA-DQB1 | 0.5163 | 0.0691 | 7.7133E-14 | UC | Sigmoid colon |
| HLA-DQB1 | 0.5814 | 0.0704 | 1.4536E-16 | UC | Transverse colon |
| HLA-DQB1 | 0.4517 | 0.0785 | 8.7296E-09 | UC | Terminal ileum |
| HLA-DQB1 | 0.1762 | 0.0325 | 5.6109E-08 | CD | Sigmoid colon |
| HLA-DQB1 | 0.1984 | 0.0348 | 1.1264E-08 | CD | Transverse colon |
| HLA-DQB1 | −0.0394 | 0.0291 | 0.1760 | CD | Terminal ileum |
| HLA-DRA | 0.0862 | 0.0190 | 6.0635E-06 | AD | Brain |
| HLA-DRB1 | 0.0458 | 0.0096 | 1.7614E-06 | AD | Brain |
| HLA-DRB1 | 0.3223 | 0.0611 | 1.3052E-07 | IBD | Transverse colon |
| HLA-DRB1 | 0.2101 | 0.0458 | 4.4561E-06 | IBD | Terminal ileum |
| HLA-DRB1 | 0.7941 | 0.1210 | 5.2518E-11 | UC | Transverse colon |
| HLA-DRB1 | 0.7934 | 0.1320 | 1.8280E-09 | UC | Terminal ileum |
| HLA-DRB1 | −0.1764 | 0.0575 | 0.0021 | CD | Transverse colon |
| HLA-DRB1 | −0.2903 | 0.0595 | 1.0585E-06 | CD | Terminal ileum |
| BTNL2 | 0.0703 | 0.0157 | 7.6238E-06 | AD | Brain |
| BTNL2 | −0.1598 | 0.0315 | 4.0985E-07 | IBD | Terminal ileum |
| BTNL2 | −0.3675 | 0.0626 | 4.3282E-09 | UC | Terminal ileum |
| BTNL2 | 0.0259 | 0.0231 | 0.2622 | CD | Terminal ileum |
| TSBP1-AS1 | 0.0035 | 0.0136 | 0.7946 | AD | Brain |
Table 6.
The SMR results of causal relationships between cell-specific genes for monocytes and diseases.
| Gene | β | SE | P SMR | Diseases |
|---|---|---|---|---|
| HLA-DQA1 | −0.0221 | 0.0120 | 0.0660 | AD |
| HLA-DQA1 | −0.1302 | 0.0299 | 1.3502E-05 | IBD |
| HLA-DQA1 | −0.4915 | 0.0893 | 3.6577E-08 | UC |
| HLA-DQA1 | 0.1798 | 0.0391 | 4.2642E-06 | CD |
| HLA-DQB1 | 0.0362 | 0.0112 | 0.0012 | AD |
| HLA-DRB1 | 0.0537 | 0.0127 | 2.2354E-05 | AD |
| HLA-DRB1 | −0.1328 | 0.0209 | 1.9197E-10 | IBD |
| HLA-DRB1 | −0.3196 | 0.0415 | 1.3067E-14 | UC |
| HLA-DRB1 | 0.0291 | 0.0169 | 0.0856 | CD |
Figure 4.
The different expression levels of HLA-DQB1 and HLA-DRB1 genes in Alzheimer’s and Unaffected groups.
Table 7.
The results of differential expression analysis.
| Gene | P_value | Avg_log2FC | AD | Unaffected |
|---|---|---|---|---|
| HLA-DQB1 | 0.0045 | 1.6110 | 0.0980 | 0.0470 |
| HLA-DRB1 | 0.0049 | 0.6353 | 0.1960 | 0.1230 |
Then we performed an enrichment analysis on six shared risk genes that were enriched for some biological processes related to the regulation of immune responses as well as antigen processing and presentation (Fig. 5(D)). In order to better understand the shared etiology among AD, IBD, UC and CD, based on the MsigDB GO dataset and Kyoto Encyclopedia of Genes and Genomes (KEGG) dataset, Gene set enrichment analysis and pathway analysis for AD, IBD, UC, and CD were performed by MAGMA. The resulting GO term and KEGG pathways are shown in Fig. 5(B) and Supplementary Fig. S4. As can be seen from the figure, 16 shared GO term exist between AD, IBD, and UC and CD, including 12 items in the biological process group, 2 in the cellular component group, and 2 in the molecular function group (Fig. 5(B), (E)). The shared cellular components were enriched in MHC protein complex and N-methyl-D-aspartate (NMDA) selective glutamate receptor complex. The shared molecular functions were relevant to NMDA glutamate receptor activity and peptide antigen binding. A total of 5 shared related pathways were obtained (Fig. 5(C), (F)). For more details, see Supplementary Note.
Discussion
The study based on our PCGL framework not only identifies the pathogenic cell types for AD, IBD, UC, and CD, but also provides new insights into the shared genetic architecture among them.
Compared to previous studies, our PCGL framework offers several methodological and interpretative advantages. Firstly, from the perspective of global and local correlation dimension, unlike studies such as Jansen et al. [11], which focused solely on genome-wide genetic correlations, we combined stratified LDSC and ρ-HESS to evaluate both global and local genetic correlations, allowing us to pinpoint specific genomic regions contributing to comorbidity between AD and IBDs. Secondly, from the perspective of cross-trait risk loci detection dimension, while most prior studies lacked formal cross-trait meta-analytic modeling, we incorporated MTAG and CPASSOC to jointly analyze GWAS summary statistics of AD and IBD subtypes, improving the sensitivity to detect pleiotropic SNPs and shared risk loci. Thirdly, from the perspective of casual inference dimension, Wang et al. [44] relied primarily on two-sample MR for causal inference between IBS and psychiatric disorders. Beyond traditional Mendelian randomization, our framework extended casual inference by applying SMR that integrates GWAS with tissue- and cell-type-specific eQTL data from the brain, gut, and monocytes, enabling casual gene identification with cell/tissue specific expression support. In terms of data diversity, previous research such as Zeng et al. [45], which focused on eQTL annotation to microglia or monocytes; in contrast, we analyzed genetic enrichment across 16 hematopoietic immune cell types and incorporated single-cell eQTL data from monocytes and bulk tissue eQTLs from brain and gut, providing a richer and disease-relevant regulatory context. Moreover, from the perspective of pathway-level interpretation dimension, prior studies often stopped at SNP or gene-level interpretation, with limited downstream functional analysis, while we performed MAGMA-based gene-level and pathway enrichment analysis, revealing shared immune-related pathways such as natural killer cell-mediated cytotoxicity and chemokine signaling, which may underlie AD–IBD comorbidity. Finally, from the perspective of disease scope and analytical integration dimension, our study not only expands the disease scope to include both UC and CD alongside AD, but also offers a multi-layered, integrative analytical pipeline—from SNPs to pathways—that strengthens both the resolution and interpretability of findings in comparison to earlier work.
However, several limitations are worth mentioning. First, our research findings were limited to the European population. Because of the heterogeneity of ethnic genomes, the conclusions may not be applicable to other races. Second, since every trait in a study is determined by multiple genetic variants, it is difficult to completely exclude pleiotropy or other direct causal pathways in any MR study. Finally, although our study identified novel loci and pathways associated with AD and IBD, more experimental studies are needed to understand the underlying mechanisms.
Future research should also consider integrating CNS-specific epigenomic and transcriptomic datasets, such as those from microglia or neurons, to better capture the tissue-specific regulatory mechanisms of AD. Extend to other ancestral groups to fully reveal the biological mechanisms underlying these two classes of diseases. Additionally, expanding the PCGL framework to single-cell multi-omics will allow us to characterize dynamic regulatory mechanisms and disease progression with higher resolution.
In summary, our study provides a robust analytical framework and novel biological insights into the shared genetics and pathogenic cell types of AD and IBD. With further validation and extension, PCGL may serve as a valuable tool for uncovering shared mechanisms and therapeutic targets in other disease domains.
Conclusions
The results obtained through PCGL demonstrated specific genetic correlations between AD and IBD (including UC and CD) and identified their enriched pathogenic cell types, as well as new shared loci and shared genes between them.
LDSC analysis identified monocytes as a major player in AD etiology in 16 progenitor cells and terminal peripheral immune cells, with IBD (including UC and CD) having the strongest enrichment on CD8 T cells.
Genome-wide genetic correlation analysis revealed negative genetic correlations between AD and IBD, UC and CD. MR result is consistent with the direction of the genetic correlations calculated above. However, our study does not support a causal relationship between UC and CD and AD.
Cross-trait GWAS meta-analysis identified two SNP rs660895 and rs917117 shared between AD and IBD, UC, and CD, which have not been previously reported. These two loci are located on the genes HLA-DRB1 and JAZF1, respectively. Previous reports have mentioned that HLA-DRB1 is closely associated with AD and IBD, and that the overexpression of JAZF1 also affects the levels of pro-inflammatory cytokines. Thus, these two SNP may be involved in regulating the shared pathway between AD and IBD.
Our MAGMA genome-wide gene-based analysis showed that six overlapping genes including HLA-DQA1, HLA-DQB1, HLA-DRA, HLA-DRB1, BTNL2, and TSBP1-AS1 between AD, IBD, UC and CD, while four of these six overlapping genes which also included HLA-DRB1 belonged to HLA class II genes, and (HLA) class II gene was a potential therapeutic target for AD and IBDs.
On one hand, to further validate the causal relationship between six shared disease genes and the four diseases (AD, IBD, UC, CD), we performed SMR analyses to validate the shared genes in both tissue-specific cells and monocytes separately. SMR analysis results indicated that HLA-DRB1 is a causal gene shared by the four diseases and the expression levels of HLA-DRB1 which is also linked with (IBD, UC) risk and HLA-DQB1 in the driver cell type, monocytes, of AD as we mentioned at the beginning of our study were associated with AD risk. In monocytes of AD dataset, we showed HLA-DRB1 and HLA-DQB1 genes performed different expression levels between diseased and controlled groups. In a word, we support that HLA-DRB1 and HLA-DQB1 are involved in the pathogenesis of AD and HLA-DRB1 is shared by four diseases.
On the other hand, we used six overlapping genes to perform an enrichment analysis that the six shared risk genes were mainly involved in natural killer cell mediated cytotoxicity and chemokine signaling pathways. Inflammatory bowel disease is an autoimmune disease. In a study focused on CD, pathway enrichment analysis highlighted several immune- and disease-related pathways and these were largely driven by a core set of HLA genes common to all of these gene sets, primarily from MHC class II [46]. Through natural cytotoxicity, production of cytokines and chemokines, and migratory capacity, natural killer cells play a vital immunoregulatory role in the initiation and chronicity of inflammatory and autoimmune responses [47]. Increased cytokine-mediated cytotoxic natural killer cell activity is also demonstrated in Alzheimer’s disease patients [48]. Therefore, natural killer cell mediated cytotoxicity and chemokine signaling pathways may be the main comorbidity mechanisms of AD and IBD (including UC and CD).
These findings deepen our understanding of the common genetic mechanisms of AD and IBDs and may provide a genetic and cellular basis for the observed clinical comorbidity between neurodegenerative and intestinal inflammatory diseases, supporting the emerging concept of the gut-brain-immune axis. The PCGL framework can also be applied to the study of common genetic mechanisms in other diseases. While in this study we focus on AD and IBDs, each analytical component-such as LDSC, MR, cross-trait meta-analysis, and MAGMA-can be applied to other complex diseases or traits. The modular structure of PCGL allows adaption to other diseases or multi-trait studies, provided that high-quality GWAS summary statistics and relevant functional annotations are available.
Key points
To identify disease-causing cell types mediated by genetic effects and to examine genetic correlations and potential causality between different diseases, we developed Pathogenic Cell types and shared Genetic Loci framework, which is applicable to all diseases.
The enriched pathogenic cell types of Alzheimer’s disease (AD) and inflammatory bowel diseases (IBDs) and the genes they regulate were identified, translating genetic associations into biological mechanisms.
The shared genetic architecture between AD and IBDs was comprehensively evaluated and two new risk single nucleotide polymorphisms and six shared risk genes were identified, providing genetic and functional insights into the comorbid etiology of AD and IBDs.
List of abbreviations
AD, Alzheimer’s disease; IBD, Inflammatory bowel disease; IBDs, Inflammatory bowel disease and its two subtypes -UC and CD; PCGL, Pathogenic Cell types and shared Genetic Loci; UC, Ulcerative colitis; CD, Crohn’s disease; GWAS, Genome-wide association studies; MR, Mendelian randomization; CNS, Central nervous system; ATAC-seq, Assay for transposase-accessible chromatin with high-throughput sequencing; LDSC, linkage disequilibrium score regression; SNP, Single nucleotide polymorphism; CPASSOC, Cross phenotype association; IVW, Inverse variance weighted; KEGG, Kyoto Encyclopedia of Genes and Genomes; GO, Gene ontology; MAGMA, Multi-marker analysis of genoMic annotation; S-LDSC, Stratified LDSC; ρ-HESS, Heritability estimation from summary statistics; MTAG, Multi-Trait Analysis of GWAS; HLA, Human leukocyte antigen; MsigDB, Molecular Signatures Database; NMDA, N-methyl-D-aspartate; IV, Instrumental variable.
Supplementary Material
Acknowledgements
Not applicable.
Peiluan Li, Professor in Bioinformatics, has been working on developing and applying machine learning methods in solving problems in biology and medicine, and has published many peer-reviewed articles in high impact journals such as Gut Microbe, Research, and Briefings in Bioinformatics.
Contributor Information
Jingjing Zhang, School of Mathematics and Statistics, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471000, China; Longmen Laboratory, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471003, China.
Yuqing Yan, Hubei Key Laboratory of Agricultural Bioinformatics, College of Informatics, Huazhong Agricultural University, No. 1, Shizishan Street, Hongshan District, Wuhan, Hubei 430074, China.
Liqin Han, School of Mathematics and Statistics, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471000, China; Longmen Laboratory, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471003, China.
Rui Qiao1, School of Mathematics and Statistics, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471000, China; Longmen Laboratory, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471003, China.
Xiaohui Niu, Hubei Key Laboratory of Agricultural Bioinformatics, College of Informatics, Huazhong Agricultural University, No. 1, Shizishan Street, Hongshan District, Wuhan, Hubei 430074, China.
Peiluan Li, School of Mathematics and Statistics, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471000, China; Longmen Laboratory, Henan University of Science and Technology, No. 263 Kaiyuan Avenue, Luolong District, Luoyang, Henan 471003, China.
Author contributions
J.J.Z, Y.Q.Y, and L.Q.H designed the research; J.J.Z, Y.Q.Y, and L.Q.H performed the research; J.J.Z, Y.Q.Y, L.Q.H, and R.Q analyzed and interpreted the data; J.J.Z and L.Q.H wrote the manuscript; P.L.L and X.H.N supervised and reviewed the manuscript. P.L.L supported the funding. All authors read and approved the final manuscript.
Conflict of interest: The authors declare that they have no competing interests.
Funding
This work was supported by National Natural Science Foundation of China (Nos. 61673008); the Young Backbone Teacher Funding Scheme of Henan (No. 2019GGJS079); Key R & D and Promotion Special Program of Henan Province (No. 212102310988); Natural Science Foundation of Henan Province (No. 242300420242); and Key Science and Technology Research Project of Henan Province (Nos. 242102310103,252102520024,252102240125).
Data availability
Publicly available datasets were analysed in this study. All data are available in the main text. The datasets analysed during the current study are available in the repository [https://www.ebi.ac.uk/gwas/, https://yanglab.westlake.edu.cn/software/smr/#DataResource, https://eqtlgen.org/sc/, and https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE214979]. The analyses were produced with standard code for software programs utilized, which can be made available from the corresponding author on reasonable request. All software used is freely available online.
Ethics approval and consent to participate
Not applicable.
Clinical trial number: not applicable.
Consent for publication
Our portions of the figure1 utilized images from Servier Medical Art (https://smart.servier.com/), licensed under Creative Commons Attribution 4.0 Unported License.
References
- 1. Li X, Feng X, Sun X. et al. Global, regional, and national burden of Alzheimer's disease and other dementias, 1990–2019. Front Aging Neurosci 2022;14:937486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Nichols E, Steinmetz JD, Vollset SE. et al. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the global burden of disease study 2019. Lancet Public Health 2022;7:e105–25. 10.1016/S2468-2667(21)00249-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Si ZZ, Zou CJ, Mei X. et al. Targeting neuroinflammation in Alzheimer’s disease: from mechanisms to clinical applications. Neural Regen Res 2023;18:708–15. 10.4103/1673-5374.353484. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Zhang SY, Chen LY, Wei C. et al. Correlation analysis between psychological symptoms and quality of life in patients with inflammatory bowel disease. Journal of Bengbu Medical College 2019;44:863–7. [Google Scholar]
- 5. Yanghong L, Xiaoxia Z, Beisha T. Research progress on the association between Parkinson’s disease and inflammatory bowel disease in genetics and gut microbiota. Chinese Journal of Neurology 2021;54:734–42. [Google Scholar]
- 6. Wang D, Zhang X, Du H. Inflammatory bowel disease: a potential pathogenic factor of Alzheimer's disease. Progress in Neuro-Psychopharmacology and Biological Psychiatry 2022;119:110610. 10.1016/j.pnpbp.2022.110610. [DOI] [PubMed] [Google Scholar]
- 7. Xing Y, Li P, Jia Y. et al. Association of inflammatory bowel disease and related medication exposure with risk of Alzheimer's disease: an updated meta-analysis. Front Aging Neurosci 2023;14:1082575. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Finucane HK, Bulik-Sullivan B, Gusev A. et al. Partitioning heritability by functional annotation using genome-wide association summary statistics. Nat Genet 2015;47:1228–35. 10.1038/ng.3404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Hemani G, Zheng J, Elsworth B. et al. The MR-base platform supports systematic causal inference across the human phenome. elife 2018;7:e34408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Bellenguez C, Kucukali F, Jansen IE. et al. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nat Genet 2022;54:412–36. 10.1038/s41588-022-01024-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Jansen IE, Savage JE, Watanabe K. et al. Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer’s disease risk. Nat Genet 2019;51:404–13. 10.1038/s41588-018-0311-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Anderson AG, Rogers BB, Loupe JM. et al. Single nucleus multiomics identifies ZEB1 and MAFB as candidate regulators of Alzheimer’s disease-specific cis-regulatory elements. Cell Genomics 2023;3:3. 10.1016/j.xgen.2023.100263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Liu JZ, Van Sommeren S, Huang H. et al. Association analyses identify 38 susceptibility loci for inflammatory bowel disease and highlight shared genetic risk across populations. Nat Genet 2015;47:979–86. 10.1038/ng.3359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Corces MR, Buenrostro JD, Wu B. et al. Lineage-specific and single-cell chromatin accessibility charts human hematopoiesis and leukemia evolution. Nat Genet 2016;48:1193–203. 10.1038/ng.3646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Buenrostro JD, Corces MR, Lareau CA. et al. Integrated single-cell analysis maps the continuous regulatory landscape of human hematopoietic differentiation. Cell 2018;173:1535–48e16. 10.1016/j.cell.2018.03.074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Ulirsch JC, Lareau CA, Bao EL. et al. Interrogation of human hematopoiesis at single-cell and single-variant resolution. Nat Genet 2019;51:683–93. 10.1038/s41588-019-0362-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. GTEx Consortium . The GTEx consortium atlas of genetic regulatory effects across human tissues. Science 2020;369:1318–30. 10.1126/science.aaz1776. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Oelen R, de Vries DH, Brugge H. et al. Single-cell RNA-sequencing of peripheral blood mononuclear cells reveals widespread, context-specific gene expression regulation upon pathogenic exposure. Nat Commun 2022;13:3267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bulik-Sullivan BK, Loh PR, Finucane HK. et al. LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet 2015;47:291–5. 10.1038/ng.3211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. 1000 Genomes Project Consortium, Abecasis GR, Auton A. et al. An integrated map of genetic variation from 1,092 human genomes. Nature 2012;491:56–65. 10.1038/nature11632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Altshuler DM, Gibbs RA, Peltonen L. et al. Integrating common and rare genetic variation in diverse human populations. Nature 2010;467:52–8. 10.1038/nature09298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Bulik-Sullivan B, Finucane HK, Anttila V. et al. An atlas of genetic correlations across human diseases and traits. Nat Genet 2015;47:1236–41. 10.1038/ng.3406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Shi H, Mancuso N, Spendlove S. et al. Local genetic correlation gives insights into the shared genetic architecture of complex traits. The American Journal of Human Genetics 2017;101:737–51. 10.1016/j.ajhg.2017.09.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 2013;37:658–65. 10.1002/gepi.21758. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Palmer TM, Sterne JAC, Harbord RM. et al. Instrumental variable estimation of causal risk ratios and causal odds ratios in Mendelian randomization analyses. Am J Epidemiol 2011;173:1392–403. 10.1093/aje/kwr026. [DOI] [PubMed] [Google Scholar]
- 26. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through egger regression. Int J Epidemiol 2015;44:512–25. 10.1093/ije/dyv080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Verbanck M, Chen CY, Neale B. et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018;50:693–8. 10.1038/s41588-018-0099-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Turley P, Walters RK, Maghzian O. et al. Multi-trait analysis of genome-wide association summary statistics using MTAG. Nat Genet 2018;50:229–37. 10.1038/s41588-017-0009-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Zhu X, Feng T, Tayo BO. et al. Meta-analysis of correlated traits via summary statistics from GWASs with an application in hypertension. The American Journal of Human Genetics 2015;96:21–36. 10.1016/j.ajhg.2014.11.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Purcell S, Neale B, Todd-Brown K. et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. The American journal of human genetics 2007;81:559–75. 10.1086/519795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Zhu Z, Zhang F, Hu H. et al. Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat Genet 2016;48:481–7. 10.1038/ng.3538. [DOI] [PubMed] [Google Scholar]
- 32. de Leeuw CA, Mooij JM, Heskes T. et al. MAGMA: generalized gene-set analysis of GWAS data. PLoS Comput Biol 2015;11:e1004219. 10.1371/journal.pcbi.1004219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Wickham H. Ggplot2. Wiley Interdisciplinary Reviews: Computational Statistics 2011;3:180–5. [Google Scholar]
- 34. Thériault P, ElAli A, Rivest S. The dynamics of monocytes and microglia in Alzheimer’s disease. Alzheimer's Res Ther 2015;7:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Casalegno Garduño R, Däbritz J. New insights on CD8+ T cells in inflammatory bowel disease and therapeutic approaches. Front Immunol 2021;12:738762. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Gazal S, Finucane HK, Furlotte NA. et al. Linkage disequilibrium–dependent architecture of human complex traits shows action of negative selection. Nat Genet 2017;49:1421–7. 10.1038/ng.3954. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Zhang X, Zou M, Wu Y. et al. Regulation of the late onset Alzheimer’s disease associated HLA-DQA1/DRB1 expression. Am J Alzheimers Dis Other Demen 2022;37:15333175221085066. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Nowak JK, Glapa-Nowak A, Banaszkiewicz A. et al. HLA-DQA1* 05 associates with extensive ulcerative colitis at diagnosis: an observational study in children. Genes 2021;12:1934. 10.3390/genes12121934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Goyette P, Boucher G, Mallon D. et al. High-density mapping of the MHC identifies a shared role for HLA-DRB1* 01: 03 in inflammatory bowel diseases and heterozygous advantage in ulcerative colitis. Nat Genet 2015;47:172–9. 10.1038/ng.3176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Wang Y, Sarnowski C, Lin H. et al. Key variants via Alzheimer’s disease sequencing project whole genome sequence data. medRxiv 2023:2023.08.28.23294631. 10.1101/2023.12.20.23300300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Wang ZX, Wan Q, Xing A. HLA in Alzheimer’s disease: genetic association and possible pathogenic roles. Neuromolecular Med 2020;22:464–73. 10.1007/s12017-020-08612-4. [DOI] [PubMed] [Google Scholar]
- 42. Ahmad T, Marshall SE, Jewell D. Genetics of inflammatory bowel disease: the role of the HLA complex. World J Gastroenterol: WJG 2006;12:3628. 10.3748/wjg.v12.i23.3628. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Pullin JM, McCarthy DJ. A comparison of marker gene selection methods for single-cell RNA sequencing data. Genome Biol 2024;25:56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Wang F, Liu YL, Jiang CH. et al. Association between psychiatric disorders and irritable bowel syndrome: a bidirectional Mendelian randomization study. J Affect Disord 2025;368:865–71. 10.1016/j.jad.2024.09.030. [DOI] [PubMed] [Google Scholar]
- 45. Zeng L, White CC, Bennett DA. et al. Genetic insights into the association between inflammatory bowel disease and Alzheimer’s disease. medRxiv 2024;2023:17.23286845. [DOI] [PubMed] [Google Scholar]
- 46. Kong L, Pokatayev V, Lefkovith A. et al. The landscape of immune dysregulation in Crohn’s disease revealed through single-cell transcriptomic profiling in the ileum and colon. Immunity 2023;56:444–58e5. 10.1016/j.immuni.2023.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Yang Y, Day J, Souza-Fonseca Guimaraes F. et al. Natural killer cells in inflammatory autoimmune diseases. Clinical & translational immunology 2021;10:e1250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Solerte SB, Fioravanti M, Pascale A. et al. Increased natural killer cell cytotoxicity in Alzheimer’s disease may involve protein kinase C dysregulation. Neurobiol Aging 1998;19:191–9. 10.1016/S0197-4580(98)00050-5. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Publicly available datasets were analysed in this study. All data are available in the main text. The datasets analysed during the current study are available in the repository [https://www.ebi.ac.uk/gwas/, https://yanglab.westlake.edu.cn/software/smr/#DataResource, https://eqtlgen.org/sc/, and https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE214979]. The analyses were produced with standard code for software programs utilized, which can be made available from the corresponding author on reasonable request. All software used is freely available online.