Genetic variation in the Major Histocompatibility Complex and association with depression

Background The prevalence of depression is higher in individuals suffering from autoimmune diseases, but the mechanisms underlying the observed comorbidities are unknown. Epidemiological findings point to a bi-directional relationship - that depression increases the risk of developing an autoimmune disease, and vice-versa. Shared genetic etiology is a plausible explanation for the overlap between depression and autoimmune diseases. In this study we tested whether genetic variation in the Major Histocompatibility Complex (MHC), which is associated with risk for autoimmune diseases, is also associated with risk for depression. Method We fine-mapped the classical MHC (chr6: 29.6-33.1 Mb), imputing 216 Human Leukocyte Antigen (HLA) alleles and four Complement Component 4 (C4) haplotypes in studies from the Psychiatric Genomics Consortium (PGC) Major Depressive Disorder (MDD) working group and the UK Biobank (UKB). In the 26 PGC-MDD studies, cases met a lifetime diagnosis of MDD, determined by a structured diagnostic interview. In the UKB, cases and controls were identified from an online mental health questionnaire. The total sample size was 45,149 depression cases and 86,698 controls. We tested for association between depression status and imputed MHC variants in each study and performed an inverse-variance weighted meta-analysis across the PGC-MDD and UKB samples, applying both a conservative region-wide significance threshold (3.9-e6) and a candidate threshold (1.6e-4). Results No HLA alleles or C4 haplotypes were associated with depression at the conservative threshold in the PGC, UKB or meta-analysis. HLA-B*08:01 was associated with modest protection for depression at the candidate threshold in the meta-analysis. Under the conservative threshold, 70 SNPs were detected in the UKB and 143 SNPs were detected in the meta-analysis, mirroring previous findings from highly powered GWAS of depression. Discussion We found no evidence that HLA alleles, which play a major role in the genetic susceptibility to autoimmune diseases, or C4 haplotypes, which are strongly associated with schizophrenia, confer risk for depression. These results indicate that autoimmune diseases and depression do not share common risk loci of moderate or large effect in the MHC.


Introduction
Depression is a debilitating psychiatric disorder with an estimated lifetime prevalence of 15% 1 , making it the leading cause of global disability 2 . The disorder is characterised by heterogeneous symptoms profiles 3 and variable treatment outcomes 4 . Developing effective pharmaceutical treatments relies on uncovering the etiology of a disorder 5 , and the field of psychiatric genetics has made great progress toward this objective in the last decade by devoting substantial resources to the study of depression 6,7 . Despite this progress, the underlying biology of depression is still not fully understood. Comorbid psychiatric and physical traits may indicate shared biological pathways and provide a path to uncovering the etiology of idiopathic psychiatric disorders 8 . Here, we focus on comorbid autoimmune diseases and depression, consider the mechanisms that could drive the overlap, and test for evidence of shared genetic influences in the Major Histocompatibility Complex (MHC).
Epidemiological studies indicate that individuals with a history of autoimmune disease are at greater risk of developing mood disorders compared to individuals without a history of autoimmune disease [9][10][11][12] . For example, a Danish Registry study 9 identified individuals who had hospital contact for autoimmune diseases or mood disorders over a thirty year period and showed that the risk of developing a mood disorder increased following onset of any autoimmune disease (Incident Rate Ratio (IRR) = 1.45; 95% CI = 1.39-1.52).
One interpretation is that the distress arising from autoimmune afflictions is causal to the onset of a mood disorder. However, other evidence indicates that the relationship is bi-directional 13,14 . For example, another Danish Registry study 13 showed that individuals with a history of depression were at increased risk for developing any autoimmune disease, compared to those without a history of depression (IRR = 1.25, 95% CI = 1. 19-1.31), and that this increase remained relatively stable across the first decade after diagnosis of depression.
There are a number of plausible explanations for the observed overlap between depression and autoimmunity. Shared environmental influences may increase risk for both disorders, e.g. stress is a risk factor for autoimmune disease 15 , and there is a phenotypic and genetic correlation between anxiety and depression 16 . Another view is that shared genetic influences act on autoimmune disease and depression through common immune pathways. Inflammation is a hallmark characteristic of autoimmune disease 17 and elevated levels of pro-inflammatory cytokines have been observed in some individuals with depression 18 . Despite increasing interest in the role of inflammatory pathways in the pathogenesis of depression 19 , the role of genetic variation in the MHC, which plays a crucial role in human immunity 20 , has not been thoroughly interrogated in the context of depression.
The MHC is divided into three functionally distinct regions: class I and II regions contain highly polymorphic human leukocyte antigen (HLA) genes that are strongly associated with risk for autoimmune disease 17,21,22 , and the class III region contains complement component 4 genes (C4), which are strongly associated with risk for schizophrenia 23 . Three recent genome-wide association studies (GWAS) indicate that genetic variation within the MHC is involved in risk for depression. The largest GWAS of depression from the Psychiatric Genomics Consortium 24 identified 44 independent loci associated with depression including a SNP in the classical class I region. A GWAS of a broadly defined depression phenotype in the UK Biobank 25 , revealed a peak of association in the MHC, with the strongest evidence for association also coming from the classical class I region. A recent meta-analysis 26 combined summary statistics from the latter two studies 24,25 with estimates from a GWAS of depression including 23andMe data 27 , and identified a SNP within the extended class I region with strong evidence for association with depression.
Highly polymorphic loci and long-range linkage-disequilibrium in the MHC complicate the interpretation of SNP associations 24 . However, with the advent of methods for imputing HLA alleles 28 and C4 haplotypes 23 , there is an opportunity to dissect SNP signal in the region with fine-mapping techniques. We used this approach to test whether genetic variation associated with risk for autoimmune disease and schizophrenia is also associated with risk for depression. To our knowledge, this is the first study to leverage imputation to interrogate the involvement of HLA alleles and C4 haplotypes in depression. Our efforts should lead to a better understanding of the role of these loci in depression, and may provide insights into the mechanisms driving comorbid autoimmunity and depression.

Participants
Participant data came from a subset of the Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium (PGC-MDD) 24 and from the UK Biobank (UKB) 29 30 , the International Classification of Diseases, 9 th Edition (ICD-9) 31 , the International Classification of Diseases, 10 th Edition (ICD-10) 32 or the Composite International Diagnostic Interview Short Form (CIDI-SF) 33 . In most of the PGC-MDD studies, bipolar disorder, non-affective psychosis and substance use disorder were exclusion criteria in the cases, and controls were screened for absence of MDD and other psychiatric disorders. Ethical approvals were obtained by Principal Investigators of each study, with all participants giving full informed consent.
The UKB is a prospective cohort study that has collected genotype and phenotype data for over 500,000 individuals across the UK, aged between 40 and 69 at the time of recruitment 29  Using genotype data that had undergone preliminary QC by the UKB 29 , we created an inclusion list of individuals of European ancestry using 4-means clustering on the first two principal components provided by the UKB. We created an exclusion list of individuals using relatedness kinship (KING) estimates provided by the UKB, and removed one individual from each pair up to 3rd-degree relationships (KING r 2 > 0.044) 36 . In the remaining data we applied QC such that variants and individuals below the following thresholds were retained: SNP missingness (before individual HLA allele and C4 haplotype imputation HLA alleles were imputed using genotype data from the PGC studies using the SNP2HLA software 28  HLA alleles were imputed in the UKB by the core analytical team using the HLA*IMP:02 software 29 with multi-population reference panels 43  C4 haplotypes were imputed using genotype data in the PGC and UKB studies using the SNP2HLA software 28 Table 2). Individual scores for C4A brain expression were calculated by multiplying the dosage for each C4 haplotype by the corresponding value for C4A brain expression. We tested genetically predicted C4A brain expression for association with depression case/control status in the PGC and UKB samples, controlling for six principal components. Genetically predicted C4A brain expression was also used as a covariate in conditional analyses on HLA alleles.

Discussion
To further understand the mechanisms driving comorbid autoimmunity and depression, we investigated evidence for shared genetic influences in the MHC, a region harboring genetic risk for autoimmune diseases and psychiatric disorders. Our primary aim was to test HLA alleles and C4 haplotypes for association with depression. Under a conservative regionwide significance threshold, we found no evidence that HLA alleles, which play a major role in susceptibility to autoimmune diseases, or C4 haplotypes, which associate strongly with risk for schizophrenia, also confer risk for depression. However, under a candidate threshold, HLA-B*0801 had significant evidence for association with depression status.
Conditioning on genetically predicted C4A brain expression did not diminish signal from HLA-B*0801, indicating independence from C4 haplotypes.
We further explored common HLA alleles associated with autoimmune diseases that have evidence of a bi-directional relationship with depression. The strongest evidence for association with depression was HLA-B*08:01, followed by HLA-DQB1*02:01 and HLA-DRB1*03:01. Previous studies have shown that all three HLA alleles are risk increasing for SLE 54,55 , HLA-DRB1*03:01 is also risk increasing for multiple sclerosis 48,49 and primary adrenocortical insufficiency 50,51 , and HLA-B*08:01 is also risk increasing for psoriasis 52,53 . In contrast, our findings indicate that HLA-B*08:01, HLA-DQB1*02:01 and HLA-DRB1*03:01 have modest protective effects in depression, indicating these alleles do not harbor shared risk for autoimmune disease and depression.
Imputation of C4 haplotypes identified four common haplotypes, none of which were associated with risk for depression in the PGC studies, in the UKB or the meta-analysis.
These results are in stark contrast to schizophrenia where association with C4 haplotypes accounts for most of the observed SNP association in the HLA region. Our results suggest the C4 association does not contribute to the common genetic susceptibility between depression and schizophrenia, observed in a genetic correlation of rG=0. 34 (p = 7.7e-40) between these disorders 24 .
At the level of region-wide significance, we detected 70 SNPs associated with depression in the UKB sample, and 143 in the meta-analysis. In each case, the top SNP was in moderate to strong LD with other significant variants, indicating a single peak of independent association. We found consistency in SNP signal between our study and the PGC-MDD GWAS of depression 24 , with the top SNPs in each study showing moderate to strong LD.
This was not unexpected given that our study is a subset of the studies included in the PGC-MDD meta-analysis 24 .
The true identity of causal variants within the MHC remains unresolved, and fine-mapping within the MHC is challenging due to the high density of genetic variation and strong LD.
Our results strongly suggest that the association signal observed in the MHC in depression 24,25 does not arise from HLA alleles or C4 haplotypes. These results suggest that any associations with these variants are either rare or have very modest effect sizes.
We note that Howard, et al. 25 increased power by leveraging a broader phenotyping approach. It is interesting to speculate that the broader depression phenotype captures individuals distressed by physical disease. This would go some way to explaining signal in the MHC, which has evidence for association with more diseases than another other region of the genome 20 . However, a more parsimonious explanation could be that MHC signal in depression maps to SNPs or to other genetic loci not imputed in this study. This is highly plausible in light of the fact that the MHC contains more genes than any other region in the human genome 20 . Under this scenario, large sample sizes and sequencing may be required to dissect SNP signal within the MHC.
Although our findings do not support a role for HLA alleles within the MHC in risk for depression, it is possible that shared genetic risk for depression and autoimmune diseases is situated outside the MHC. Efforts to identify genome-wide pleiotropy were undertaken in the recent PGC-MDD GWAS using LD Score to estimate genetic correlations between depression and several autoimmune diseases 24 . There was no evidence for significant cross-trait correlations; the strongest observed was between depression and inflammatory bowel disease (rG = 0.07, p = 0.06). In other efforts to detect genome-wide pleiotropy, Euesden et al. 14 found no evidence that polygenic risk scores (PRS) for rheumatoid arthritis predicted depression status in an independent sample, nor did PRS for depression predict autoimmune disease status.
One possibility is that there is a sub-group of individuals enriched for depression and autoimmune risk alleles. Under this scenario, there may be insufficient power to detect the relationship. Identifying, for example, a sub-group of individuals with depression, who are also enriched for autoimmune risk alleles, would go some way to explaining the observed comorbidity between these traits. Furthermore, identifying a sub-type of depression, e.g. an 'immune related' depression group, would help to dissect heterogeneity in the depression phenotype.
In summary, this study is the first to interrogate the involvement of HLA alleles and C4 haplotypes in depression risk, and we find no evidence that either type of genetic variant plays a major role in susceptibility for depression. In contrast, the three HLA alleles that showed nominal significance in our study conferred modest protective effects for depression. Furthermore, the strong association for C4 alleles seen in schizophrenia is absent in depression. Large sample sizes and regional sequence data may be required to dissect SNP signal within the MHC.