None of the significantly associated SNPs from the GWAS analyses cause obvious functional changes, so it is difficult to postulate a plausible model for SZ pathogenesis based on the current associations. Yet, other SZ research indicates three pathways to pathogenesis in which the associated SNPs could plausibly have functional roles: (i) inflammatory/infectious pathways, (ii) auto-immune abnormalities, (iii) non-immune related functions.
Role for SZ Associated SNPs in Inflammatory/Infectious Pathways
Reciprocal regulation of the immune and the central nervous systems is well documented [Aloisi et al., 2000a, 2000b; Tracey, 2009]. For example, neurotransmitters secreted from nerve terminals modulate immune cell activity [Levite, 2008]. Conversely, functional receptors for several immune mediators (e.g., chemokines) are located on neurons [Besedovsky et al., 1983; Bajetto et al., 2001; Adler and Rogers, 2005]. Cytokine and chemokine receptors are also expressed on neurons and glia [Coughlan et al., 2000; Gardoni et al., 2011], and can thus regulate neurodevelopment [Bajetto et al., 2001; Smith et al., 2007], apoptosis [Bajetto et al., 2002], signal transduction [Adler and Rogers, 2005], neuroplasticity [Ben Menachem-Zidon et al., 2008; Goshen et al., 2008; Koo and Duman, 2008], and neurotransmission [Kitagami et al., 2003; Morón et al., 2003a, 2003b; Volterra and Meldolesi, 2005]. Elevated immune markers (e.g., IL-6, CRP) are associated independently with cognitive deficits [Marsland et al., 2006; Dickerson et al., 2007], and hippocampal volume reduction [Marsland et al., 2008], and through interaction with exposure to certain infectious agents [Dickerson et al., 2012; Prasad et al., 2012]. Thus, increased pro-inflammatory tone could affect diverse neurobiological processes that have been implicated in the pathophysiology of schizophrenia and related psychotic disorders.
Mounting evidence suggests altered immune functions as well as “neuro-inflammation” in SZ [Heath and Krupp, 1967; Heath et al., 1967a, 1967b; Bayer et al., 1999; Radewicz et al., 2000; Rothermundt et al., 2001; Brown et al., 2004; Adler and Rogers, 2005; Saetre et al., 2007; Smith et al., 2007; Meyer and Feldon, 2009; Bechter et al., 2010; Ellman et al., 2010; Müller and Schwarz, 2010; Meyer et al., 2011]. Elevated levels of inflammatory cytokines in the peripheral blood have been noted in SZ patients compared with healthy controls [Potvin et al., 2008]. Some post-mortem brain studies indicate neuro-inflammation in the form of activated microglia/macrophages [Bayer et al., 1999; Radewicz et al., 2000; Wierzba-Bobrowicz et al., 2005], increased expression of inflammatory markers in the dorsolateral prefrontal cortex neurons [Fillman et al., 2013] as well as altered cerebral microvasculature [Kim et al., 2008a]. Whether the absence of obvious gliosis argues against the occurrence of neuroinflammation in SZ has been debated [Roberts et al., 1986; Stevens et al., 1988; Casanova et al., 1990; Arnold et al., 1996; Harrison, 1999], but non-invasive whole brain scanning techniques also document evidence of neuroinflammation [van Berckel et al., 2008; Doorduin et al., 2009]. Indirect evidence supporting the concept of neuroinflammation comes from a drug trial in which non-specific anti-inflammatory drugs reduced psychotic symptoms severity [Müller et al., 2002; Muller et al., 2010; Sommer et al., 2012]. Together, these studies suggest a role for neuroinflammation in SZ pathogenesis and a framework for additional tests of the hypothesis [Müller and Schwarz, 2010].
How might these findings bear on the HLA associations with SZ? In view of the critical role for HLA molecules in antigen presentation during the immune response process, it is not surprising that numerous infections are associated with HLA variation [Hill, 2006]. Analyses of HLA variation, particularly the extensive genetic diversity has been explained by pathogen-driven-balancing selection, which can predict HLA genetic differentiation worldwide [Sanchez-Mazas et al., 2012]. Functional variations in HLA molecules can alter immune functions that may modify the susceptibility to exposure to certain infectious agents; they may also regulate the balance between pro-inflammatory and anti-inflammatory factors. Such an altered balance could in turn affect other biological processes that are not classically immunological in nature. We are not aware of any studies that investigated links between these processes and variation due to the GWAS associated SNPs in the HLA region. Admittedly, it is difficult to establish such links in view of the complex, inter-linked nature of inflammatory processes. Indeed, the reported alterations in biological processes may be compensatory mechanism for a primary immunological change.
In summary, convergent lines of evidence suggest chronic low grade neuro-inflammation in SZ, as well as an array of immunological abnormalities. There are no obvious associations between the SZ associated HLA SNPs and any of these variables.
Role for SZ Associated SNPs in Auto-Immune Abnormalities
National hospital registry based studies from Denmark indicate that individuals with auto-immune diseases such as type 1 diabetes mellitus, psoriasis, Sjogren's syndrome, autoimmune hepatitis and dermatopolymyositis have an increased risk for SZ, with an estimated 29% increased risk across all autoimmune diseases [Eaton et al., 2010; Benros et al., 2011]. The elevated SZ risk is observed even if the diagnoses of autoimmune dysfunction predate the SZ diagnosis and is also observed among individuals with a family history of autoimmune diseases [Benros et al., 2011, 2012]. Conversely, a national case-registry based study of SZ patients from Taiwan indicated increased prevalence of auto-immune diseases including Graves' disease, psoriasis, celiac disease, pernicious anemia, and hypersensitivity vasculitis [Chen et al., 2012]. An increased prevalence of auto-immune disorders has also been observed among non-psychotic relatives of patients with SZ [Eaton et al., 2010; Benros et al., 2012]. On the other hand, patients with SZ have reduced prevalence of rheumatoid arthritis, another auto-immune disease [Chen et al., 2012]. Others have also reported a reduced prevalence of type 1 diabetes mellitus among patients with SZ in a Finnish national registry [Juvonen et al., 2007] in contrast to the Danish studies [Benros et al., 2011].
Several mechanisms may explain the epidemiologic data [Strous and Shoenfeld, 2006]. Individuals with auto-immune diseases have elevated prevalence of antibodies directed against brain proteins or may produce antibodies that cross-react with brain proteins [Irani and Lang, 2008]; individuals with SZ can also produce antibodies against proteins in the frontal cortex [Henneberg et al., 1994], cingulate gyrus [Ganguli et al., 1987; Kelly et al., 1987; Henneberg et al., 1994], hippocampus [Ganguli et al., 1987] and against glutamate receptors [Tsutsui et al., 2012]. Individuals with autoimmune disorders also develop further CNS dysfunction in conjunction with microbial infections [Benros et al., 2011]. The inflammatory reactions to these insults could conceivably compromise the blood–brain barrier, permitting the transport of noxious agents [Eaton et al., 2006; Dalman et al., 2008; Dantzer et al., 2008]; similar mechanisms are plausible in SZ. It has even been proposed that maternal infection provokes immunological reactions and producing autoantibodies that disrupt neural development, thus elevating the risk for SZ [Kirch, 1993]. Finally, there may be shared genetic or environmental risk factors (e.g., stress) for specific autoimmune disease and SZ.
HLA variants could serve as shared genetic risk factors because there are well known associations for several autoimmune diseases [Todd et al., 2007; van Heel et al., 2007; Genetic Analysis of Psoriasis Consortium et al., 2010; Zhernakova et al., 2011]. Particular HLA molecular configurations have been identified as prominent risk factors for individual auto-immune diseases, such as ulcerative colitis and type I diabetes mellitus [Trucco, 1992; Achkar et al., 2012]. To see if HLA variants from GWAS of auto-immune disease also alter risk for SZ, we conducted a simplified survey of published data (http://gwas.nih.gov/). Of the most significantly associated HLA region SNPs for auto-immune disease, none overlap with the significantly associated SNP list in the current SZ mega analysis [Ripke et al., 2011]. However, rs3131296 an HLA region SNP reported in an earlier SZ GWAS [Stefansson et al., 2009] is in significant LD (r2 > 0.73) with SNPs that are significantly associated with celiac disease [van Heel et al., 2007], type-1 diabetes mellitus [Todd et al., 2007] and systemic lupus erythematosus [Harley et al., 2008]. More sophisticated and precise analysis based on HLA variants may help understand whether shared HLA risk variants explain the increased (or decreased) prevalence of individual auto-immune diseases among persons with SZ.
In summary, the co-occurrence of auto-immune disease among persons with SZ and their relatives could be explained by several mechanisms. Shared etiology is an appealing possibility that could be explored precisely through HLA associations.
Role for SZ Associated SNPs in Non-Immune Related Functions
A substantial proportion of genes on chromosome 6p do not have immune-related functions, so it is plausible that the GWAS associations in this region also indicate abnormalities unrelated to immune dysfunction. Candidate gene studies have indicated nominal associations with genes with no obvious role in immune function, for example, NOTCH4 [Wei and Hemmings, 2000], HSPA1B [Pae et al., 2005], and HSPA1L [Kim et al., 2008b]. Indeed, a GWAS from Iceland indicates genome-wide significant results at this locus [Stefansson et al., 2009]. However, NOTCH4 SNPs are not significantly associated with SZ following corrections for multiple comparisons in the GWAS mega analyses. In summary, the SZ associations on chromosome 6p could be plausibly related to non-immune related functions, as illustrated by the reported associations at NOTCH4 [Shayevitz et al., 2012].