Causal role of immune cells in major depressive disorder and bipolar disorder: Mendelian randomization (MR) study

Major depressive disorder (MDD) and bipolar disorder (BD) are prevalent psychiatric conditions linked to in�ammatory processes. This study employs a two-sample Mendelian randomization (MR) approach to investigate the relationship between immune cells and the risk of MDD and BD. Rigorous sensitivity analyses are conducted to assess the reliability, heterogeneity, and horizontal pleiotropy of the �ndings. The study reveals a signi�cant association between immune cell pro�les and the susceptibility to MDD and BD. Speci�cally, 29 immunophenotypes exhibit a substantial connection with MDD risk, including CD27 on IgD + CD38 − unswitched memory B cell (inverse variance weighting (IVW): odds ratio(OR) [95%]: 1.017[1.007 to 1.027], p = 0.001), CD27 on IgD + CD24 + (IVW: OR [95%]: 1.021 [1.011 to 1.031], p = 4.821×10 − 5), CD4 on Central Memory CD4 + T cell(IVW: OR [95%]:0.979 [0.963 to 0.995], p = 0.011) and the other 25 immunophenotypes. Additionally, 35 immunophenotypes demonstrate a signi�cant association with BD risk, such as CD33br HLA DR + AC AC (IVW: OR [95%]: 1.022[1.007 to 1.036], p = 0.007), CD28 + CD45RA-CD8br %T cell RC (IVW: OR [95%]: 1.024 [1.008 to 1.041], p = 0.004) CD62L on CD62L + myeloid DC MFI (IVW: OR [95%]:0.926 [0.871 to 0.985], p = 0.014)), and the other 32 immunophenotypes. This MR study provides robust evidence supporting a causal relationship between immune cells and the susceptibility to MDD and BD, offering valuable insights for future clinical investigations.


INTRODUCTION
Major depressive disorder (MDD) and bipolar disorder (BD), are highly prevalent and debilitating conditions associated with high suicide rates and a heavy social burden 1 .MDD is widely recognized as one of the most pressing mental health problems.The global number of incident cases has increased almost 50% over the past 30 years, and more than 264 million people of all ages are currently a icted 2 .Several clinical symptoms may be present, including depressed mood, anhedonia, fatigue or loss of energy nearly every day, feelings of worthlessness, and diminished ability to think or concentrate 3 .BD is a severe psychiatric condition characterized by recurrent episodes of elevated (mania or hypomania) or depressed mood and interepisodic periods with no or few symptoms 4,5 , which affects about 45 million people in the world 6 and shares overlapping depressive symptoms and genes with major depressive disorder 7,8 , indicative of shared aetiological mechanisms.Low-grade systemic in ammation/immune dysregulation could be one such mechanism 9 .
At the epidemiologic level, early life infection is linked with autoimmune disease and subsequent mental disorders in adults [10][11][12][13][14] .Complex immune-brain interactions (that affect neural development, survival, and function) might have causal and therapeutic implications for disorders including psychiatric illness including MDD and BD [15][16][17][18][19] , while cytokines play a signi cant role in these psychiatric disorders 20,21 .T helper type 1 (Th1) cells are known to mediate cellular immune reactions and including production of the cytokines (IL-1, IL-6, IL-12), IFN-γ, and TNF-α; T helper type 2 (Th2) cells enhance antibody-mediated immune reactions and induce the production of cytokines IL-4 IL-5 and IL-10 while T helper type 3 (Th3) cells heighten the production of TGF-beta 18 .The higher levels of TNFα and IFN-γ in in vitro-stimulated CD8 + T cells isolated from patients with depression and IFN-γ levels correlate with the severity of the condition 22,23 .Imbalances between IL-6 and TGF-β can lead to depression 24 .For BD patients, there was a signi cant increase at the Treg cell levels 25 , while soluble IL-2 receptor regulates the immune response by regulating the proliferation of Treg cells through IL-2 26 .Two recent meta-analyses listed increased serum levels of interleukin (IL)-4, IL-10, TNF-α, sIL-2R, sIL-6R and sTNF-R1 as the most consistent ndings in patients with BD 27,28 .In a study of 27 female BD patients, measures of Th1, Th2 and T helper type 17 (Th17) cytokines from stimulated lymphocytes showed a decrease in regulatory T (Treg) cells, coupled to increased Th1 (IL-2, IL-6) and Th-17 (IL-17) cytokines 29 .After confounding factors such as age, sex, and BMI were adjusted, the level of MCP-1, a secretion of monocytes, in the CSF of BD patients was signi cantly higher than that of healthy controls 30 .Although these studies have controlled for various confounding factors, residual confounding due to limited sample size, awed study design, and beyond the scope of existing studies remains an important alternative explanation.Therefore, we can use Mendelian randomization (MR) to control for these residual confounding factors.
MR serves as an observational study design, utilizing genetic variants as instrumental variables to estimate the causal impact of risk factors on health outcomes 31 .Given its reduced susceptibility to confounding variables, resistance to measurement errors, and the ability to mitigate bias resulting from reverse causality, MR has emerged as a dependable method for obtaining robust estimates of the causal effects of diverse risk factors on health outcomes 32 .In this study, an exhaustive two-sample MR analysis was conducted to discern the causal association between immune cell signatures and depression as well as bipolar disorder.

Study design
The overall design used for this work is illustrated in Fig. 1.We employed a MR analysis to investigate the causal associations of 731 immune cell signatures in depression and bipolar disorder.Three assumptions are required for a valid genetic instrument: (i) it has a causal relationship with the exposure; (ii) it is independent of confounders; (iii) it is only associated with the outcome through the exposure.
Immunity-wide GWAS data sources GWAS summary statistics for each immune trait are publicly available from the GWAS Catalog (accession numbers from GCST90001391 to GCST90002121) 33 .A total of 731 immunophenotypes were included in the analysis, which comprised of absolute cell (AC) counts (n = 118), median uorescence intensities (MFI) representing surface antigen levels (n = 389), morphological parameters (MP) (n = 32), and relative cell (RC) counts (n = 192).The AC, MFI, and RC features encompassed B cells, CDCs, mature stages of T cells, monocytes, myeloid cells, TBNK (T cells, B cells, natural killer cells), and Treg panels.The MP feature consisted of CDC and TBNK panels 34 .The original GWAS on immune traits utilized data from 3,757 European individuals, and there were no overlapping cohorts.Approximately 22 million SNPs were genotyped with high-density arrays and imputed using the Sardinian sequencebased reference panel 35 .Associations were tested while adjusting for covariates such as sex, and age.

Genome-wide association study (GWAS) data sources for MDD
The GWAS summary statistics for MDD were obtained from the Psychiatric Genomics Consortium (PGC) and UKBiobank ( https://gwas.mrcieu.ac.uk/datasets/ieu-b-102/).The study performed a meta-analysed data on 807,553 individuals (246,363 cases and 561,190 controls) from the three largest GWAS of depression 36 .This research identi ed 102 independent variants, 269 genes, and 15 gene-sets associated with depression, including both genes and gene-pathways associated with synaptic structure and neurotransmission.

Genome-wide association study (GWAS) data sources for BD
The GWAS summary statistics for BD were obtained from the PGC ( https://gwas.mrcieu.ac.uk/datasets/ieu-b-5110/). The summary statistics for bipolar disorder were derived from a GWAS meta-analysis of 57 cohorts collected in Europe, North America and Australia, totaling 41,917 BD cases and 371,549 controls of European descent 37 .In this research, researchers identi ed 64 associated genomic loci, 33 of which are novel discoveries.

Selection of instrumental variables
Based on recent research, we set the signi cance level for each immune trait IV at 1 × 10-5.To obtain independent IVs, we performed clumping (r 2 < 0.001 within a 1,0000-kb distance) based on the linkage disequilibrium (LD) reference panel of the 1000 Genomes Project.For MDD and BD, we adjusted the signi cance level to 5 × 10 − 8. Additionally, to avoid bias brought by weak instruments, we considered IVs with F statistics > 10 as strong instruments and reserved them for the following analysis.
We then extracted IVs from outcome MDD and BD summary statistics and removed those harboring potential pleiotropic effects on MDD and BD (p < 5× 10-8) (as previously reported 38 ).We harmonized the exposure and outcome SNPs to ensure that effect estimates were aligned for the same effect allele.

Statistical analysis
The MR analyses were carried out utilizing the 'TwoSampleMR' package (version 0.5.8)within the R software environment (version 4.3.2).
The clumped summary-level data obtained from each GWAS were included with SNP rs identi er, βcoe cient or log odds ratio, standard errors or 95% con dence intervals (CI), effect allele, other allele, pvalue, effect allele frequency, and sample size.Subsequently, the exposure and outcome datasets were harmonized, and SNP effects and corresponding standard errors were obtained, while also eliminating palindromic SNPs with intermediate allele frequencies.In our study, we conducted a range of MR analyses, encompassing MR Egger, weighted median, inverse-variance weighted (IVW), simple mode, weighted mode, Simple mode and Weighted mode approaches.Among these, the IVW method is frequently employed 40 , which provided a high-powered estimate and relied on the assumption that all SNPs were valid genetic instruments 41 .If the p-value of IVW is less than 0.05, we think the exposure was considered to be causally related to the outcome factor.Then to evaluate the presence of variance, we performed heterogeneity examinations utilizing both the MR Egger and IVW techniques.The Cochrane's Q value was employed to appraise the variability of genetic instruments.To scrutinize the existence of horizontal pleiotropy, we utilized the MR Egger regression equation, where a p-value surpassing 0.05 indicates an absence of indications of horizontal pleiotropy 42 .Finally, we made a leaveone-out analysis to nd out whether an SNP could drive the bias of causal estimate.In addition, scatter plots, funnel plots and forest plots were used.

Exploration of the causal effect of immunophenotypes on MDD
In this study, we detected potential causal effects of 29 immunophenotypes on MDD: CD27 on IgD + The genetically CD27 on IgD + CD24 + showed a positive correlation with the risk of MDD, as evidenced by IVW method (OR [95%]:1.021[1.011 to 1.031], p = 4.821×10 − 5 ) (Fig. 2, Supplementary Tables 1-2, Supplementary Fig. 1).Additionally, the intercept of MR-Egger ruled out the possibility of horizontal pleiotropy for the 28 associations (Supplementary Table 3).There are some sensitivity analyses that demonstrated the robustness of the MR Results (Supplementary Fig. 2).Scatter plots and funnel plots also indicated the stability of the results (Supplementary Figs.3-4).

DISCUSSION
Utilizing extensive publicly available genetic data, we undertook MR analyses to examine potential causality and the direction of association of various immune cells concerning MDD and BD.Firstly, our ndings indicate compelling evidence for signi cant causal effects on MDD among four types of immune traits (MFI, RC, AC, and MP), with 29 immunophenotypes demonstrating statistical signi cance (P < 0.05).Secondly, within the context of four types of immune traits, 35 immunophenotypes exhibited noteworthy causal effects on BD (P < 0.05).

The potential role of immune cells in MDD
Our study revealed a correlation between elevated levels of CD3 on CD28 + CD45RA-CD8brs, CD127 on CD28 + CD4 + , CD39 + CD8br %T cell and CD127 on CD28 + CD4 + (Treg panel) with a decreased risk of MDD, while Activated Treg AC (Treg panel) with an increased risk of MDD.T cells can be categorized into three subsets: T helper cells, Cytotoxic T cells, and Treg cells.Depletion of Treg cells is associated with higher rate of depressive-like behaviors in mice, which is also associated with increased levels of Th1, Th2, Th17 cytokines, and reduced serotonin leve 21 .Treg cells are crucial for immune system regulation, homeostasis, and prevention of autoimmunity 43 .Depressive symptoms under chronic stress (after hip fracture) are associated with altered immunosuppressive phenotype in Treg lymphocytes 44 .The imbalances between IL-6 and TGF-β and between Th17 and Treg can lead to depression 24 .Treg promoter IL-2 alleviated depression and anxiety-like behaviors by restoring the balance between the two 24 .
Our ndings indicated that 14 types of B cells, namely CD27 on IgD + CD24 + , CD27 on IgD − CD38 − and other nine B cells, were associated with a causal relationship of MDD.Recent work has highlighted MDD as in ammatory conditions, involving both systemic and central immune cells.The involvement of immune cells in the process of MDD has been con rmed 45 .Depressive symptoms under chronic stress (after hip fracture) are associated with altered immunosuppressive phenotype in B lymphocytes 44 .In a recent review, the leave-one-out analysis of B cells showed that B cell counts were signi cantly higher in patients than in healthy controls after exclusion of study by Pavon et al 20,24 .
In our study, we discovered a correlation between CX3CR1 on CD14-CD16-(monocyte panel) and an increased risk of MDD.A growing body of evidence points towards the involvement of an abnormal in ammatory response system in the pathogenesis of MDD, and one reason may be that the upregulation of cluster 3 genes in monocytes of all MDD patients suggests a premature aging of the cells, i.e. mitochondrial apoptotic dysfunction and TNF "in ammaging" 46 .A systematic review proved that overall monocyte count was signi cantly higher in patients with depression compared to healthy controls 20 .

The role of immune cells in BD
Our study indicated that ten types of Tregs, namely CD28 + CD45RA − CD8br %T cell, CD3 on secreting Treg, CD3 on CD8br and other seven Tregs, were associated with a causal relationship of BD.There is ample evidence that Tregs are capable of dampening the effects of pro-in ammatory Th1/Th17 cells and of pro-in ammatory monocytes/macrophages 47,48 .Th17 cells produce the in ammatory cytokine interleukin (IL) 17A, and the discovery of Th17 cell involvement in depression evolved from the well established link that IL-6, which is required for Th17 cell differentiation, contributes to the onset, and possibly maintenance, of depression 49 .The signi cantly higher levels of sCD25 and higher percentages of circulating CD4 + CD25 + FoxP3 + Treg cells can be seen in the circulation of BD patients 25 .BD patients had reduced proportions of natural Treg (CD4 + CD25 + FoxP3 + ) in parallel to higher cytokine production than healthy controls 29 .Additionally, a recent systematic analysis showed that the memory T helper cell count as well as the percentage of memory T helper cells were signi cantly higher in patients compared to healthy controls 20 .
In our research, there were 11 B cells associated with a causal relationship of BD.A research showed that about one-third of bipolar patients and one control had a high density of T and/or B cells by using an automated work ow for image acquisition and analysis and subsequent cluster analysis of T and B cell densities to assess lymphocyte in ltrates 50 .
There are two types of monocytes, namely CD64 on CD14 − CD16 − and CD14 on CD14 + CD16 + monocyte, were associated with a causal relationship of BD in our study.The role of CNS immune activation in BD has shown that BD is associated with the activation of monocytes and microglia 30 .

Limitation
We acknowledge certain inherent limitations in our study that warrant consideration.Firstly, we employed a p ≤ 10 − 5 cutoff for instrument selection due to the relatively small size of exposures Genome-Wide Association Studies (GWAS).While this may increase the risk of weak-instrument bias for individual genetic variants, our F-statistics indicated that these instruments possessed adequate strength (all > 10) (Supplementary Tables 7-8).Lowering the p-value threshold could potentially enhance the overall statistical power for Mendelian Randomization (MR) analysis.In the future, as larger GWAS for immune cell markers become available, more stringent selection criteria may be contemplated.Secondly, all GWAS were conducted on European samples, mitigating the risk of population strati cation bias but potentially limiting the generalizability of our ndings to other ethnic groups.Many of our results did not surpass stringent thresholds for multiple testing, necessitating replication of these MR ndings using larger GWAS.Lastly, it is crucial to recognize that MR Analysis cannot replace clinical trials; rather, it serves as a method to analyze the causal relationship between exposure and outcome.Therefore, additional studies are imperative to corroborate the potential association of immune cells with depression and the risk of coke oven exposure.

CONCLUSION
In our study, we demonstrated that CD62L − HLA DR ++ monocyte %monocyte, CD39 + CD8br %T cell, CD19 on IgD − CD38-and the other remaining 11 immune cells decrease the risk of MDD, while IgD + CD38 dim %B cell, Activated Treg AC, HLA DR on CD14 − CD16 − and the other remaining 12 immune cells may lead Figures Research overview and design of Mendelian randomization analysis.
The causal relationship between CD27 on IgD + CD24 + and MDD.