4. Discussion
This paper is the first report showing that alternative splicing processes HLA-DPA1 mRNA into multiple isoforms. The functional significance of these different isoforms of HLA-DPA1 in brain is not clear, but undoubtedly would alter the function of this gene since the IgG domain and transmembrane domain are both substantially altered, and both of these domains are highly conserved across vertebrates. Further work is required to functionally elaborate alterations of mRNA of 30–100 amino acids in the translated products.
Since HLA-DPA1 is part of a heteromeric complex, the alternatively spliced isoforms may act to inhibit complex formation and the presentation of antigens on the cell surface. Major histocompatibility complex class II molecules are expressed by immune antigen-presenting cells like B cells, dendritic cells, and monocytes/macrophages, and designed to stably bind and present fragments from exogenous proteins to the immune system. There are disorders involving MHC class II deficiency (MIM209920). Typically, patients with these types of disorders fail to fight off infections, have increased infections following viral, bacterial, and fungal exposures, and lack expression of different MHC Class II genes. Variations in the density of MHC class II molecules on antigen-presenting cells influence the intensity and the nature of T-cell response. Taken together, a lack of MHC Class II response results in less T-cell activation.
Patients failing to produce proper MHC II–peptide complexes do not mount efficient antibody responses to infection, and this is probably the underlying factor in differential hepatitis B viral resistance shown in association studies of HLA-DPA1 [23]. It was also shown that the same SNP (rs3077) was a strong eQTL for HLA-DPA1 expression in human liver samples (p = 10−48) [35], and LCLs [36]. This finding suggests that the expression of HLA-DPA1 is related to resistance to hepatitis B virus (HBV). However, we did not test association of HBV in our brain or blood samples in this experiment, but are suggesting that the rs3077 cis eQTL allele that increased expression of HLA-DPA1 was also associated with decreased HBV infection rates, thus indicating that decreased expression of the HLA-DPA1 gene could result in increased infection rates. The SNP used in our study (rs9277341) is in the same LD block (r2 > 0.8) as rs3077, and rs9277341 SNP is an eQTL across multiple brain regions for HLA-DPA1 expression as well shown in the present results. Further, shown in Figure 10 are the HLA-DPA1 SNP values for association with schizophrenia, demonstrating an individual SNP in this region with highly significant association (association p = 2.46 × 10−10). OR = 1.1956). Taken together, these genetic results and a strong eQTL in brain support a general susceptibility for SZ association with decreased HLA-DPA1 expression in multiple brain regions. In a survey of developmental cortical trajectory of gene expression, during fetal and infant brain development there is an absence of HLA-DPA1 expression in fetal brain, and very low levels in infant brain, contrasted to child, teen, adult, and elderly brains [34].
Besides this very strong eQTL at rs9277341 that we confirmed in brain and LCLs, there is an environmental factor that could perhaps also account for the findings of decreased expression, at least in LCLs of the MHC Class II molecules. This possibility is shown in the intriguing report that morphine and opioid agonists can reduce MHC II levels in circulating B-cells by 88% [36]. The underlying mechanism is the immunosuppressive effect of high cortisol levels induced by morphine. Circulating levels of cortisol are abnormally high in neuropsychiatric disorders [37,38], and could possibly account for the decreased expression of MHC Class II in the present LCL study. Although the cell lines were passaged and presumably independent of cortisol stimulation in the present study, there might be epigenetic changes associated near the HLA-DPA1 locus or in transcription factors that enhance HLA-DPA1 expression. A next logical step would be to study the interaction of cortisol suppression of HLA-DPA1 molecules by genotype interaction with the eQTL SNPs, and to determine the epigenetic landscape involved in the cis eQTL effects observed in this study. Finally, we would need to determine more precisely the brain localization of HLA-DPA1 protein in brain. Previously, MHC Class I genes were shown to be localized to neurons and are essential for development and plasticity of neurons [24,39,40,41]. Little is known about expression of MHC Class II genes in brain, although we show that it is much lower in brain compared with PBMCs, transformed lymphocytes (EBV), and whole blood levels (Figure 9 A,B,C). Although it is not clear presently if neurons produce MHC Class II mRNA in human brain, MHC Class I and II are found on differential subpopulations of activated microglia in brain [42]. Expression of MHC Class II molecules are found in microglia using a general OX-6 antibody. To date, there is no specific study by immunohistochemistry with a specific monoclonal HLA-DPA1 antibody that has been conducted.
Since expression of MHC Class II appears to be reduced in brain and periphery, this might suggest an increased infection rate in patients with neuropsychiatric disorders. Although a higher rate of HIV in schizophrenia is reported, it is thought that the use of drugs and duration of untreated illness contributes to this increase [43]. Since there are literally thousands of immune-related genes, a follow-up paper is being planned to expand these current findings into the broader immune landscape using an independent array expression platform and analysis of immune-relevant upstream and downstream genes, as well as additional subjects, which also supports the present results for MHC Class II molecules [44].
In this study, we chose to use microarrays over next-generation sequencing (NGS). Both microarray and NGS technologies have their advantages and disadvantages; the selection depends upon the experimental goal. NGS has a wide range of applications and is particularly useful to investigate de novo mutations, new splice variants and non-coding RNA. NGS offers high sensitivity, high accuracy, and a broad dynamic range with no cross-hybridization. However, there are some limitations to NGS that make the use of microarrays a good alternative strategy. NGS has a higher cost with short reads that do not always allow the study of large splicing variants, chromosomal rearrangements or gene fusions like tiling arrays [45]. The analysis pipelines required for the NGS can be quite expensive and usually requires dedicated bioinformatics support to develop meaningful results.
On the other hand, microarrays have been used and accepted for the past three decades and have established analysis tools to speed the normalization and interpretation. Microarrays are powerful for analysis of known alternative splicing events [46]. Microarray technology is fast and a fairly reliable tool for applications involving gene expression, genotyping, and detection of known splicing variants. It requires, however, prior knowledge of the genome to design reliable probes with no cross hybridization in order to identify differences in transcript levels. In addition to gene and exon level expression, microarrays can also be used to investigate coding SNPs [47], but the detection ranges are limited and they require exceptionally high-quality mRNA. In the context of our experiments, microarrays produced reliable and replicable results concerning gene expression levels and led to the discovery of alternative splicing events of HLA-DPA1 transcripts.