Matthew H. Law and Matilda Bradford contributed to this work equally.
Article first published online: 17 JUN 2011
Copyright © 2011 Wiley-Liss, Inc.
American Journal of Medical Genetics Part B: Neuropsychiatric Genetics
Volume 156, Issue 6, pages 709–719, September 2011
How to Cite
Law, M. H., Bradford, M., McNamara, N., Gajda, A. and Wei, J. (2011), No association observed between schizophrenia and non-HLA coeliac disease genes: Integration with the initial MYO9B association with coeliac disease. Am. J. Med. Genet., 156: 709–719. doi: 10.1002/ajmg.b.31213
How to Cite this Article: Law MH, Bradford M, McNamara N, Gajda A, Wei J. 2011. No Association Observed between Schizophrenia and Non-HLA Coeliac Disease Genes: Integration With the Initial MYO9B Association With Coeliac Disease. Am J Med Genet Part B 156:709–719.
- Issue published online: 10 AUG 2011
- Article first published online: 17 JUN 2011
- Manuscript Accepted: 31 MAY 2011
- Manuscript Received: 17 DEC 2010
- myosin IXB;
- association study;
- coeliac disease;
- schizophrenia, Il-2
Schizophrenia is a severe psychotic illness with a heterogeneous presentation and a devastating impact on social and occupational function. Worldwide variations in schizophrenia incidence rates suggest that local conditions may modify disease risk. The human leukocyte antigen (HLA) region has been confirmed to be associated with schizophrenia by genome-wide association studies in populations across the world. While the presence of autoimmune processes in a subgroup of schizophrenia cases is contentious, the immune system could allow environmental exposures to lead to schizophrenia by generating improper immune response. To investigate this topic, we reviewed the current evidence of the relationship between schizophrenia and coeliac disease. Based on this review, we performed genetic analysis of the MYO9B gene and the IL-2/IL-21 locus by genotyping SNPs that have been previously associated with coeliac disease or schizophrenia in 223 families, 108 unrelated individuals with schizophrenia and 120 controls. Finding no evidence for association with these two loci in our study samples, we applied meta-analytic techniques to combine our findings with previous reports. This synthesis, in light of our review of previous reports, suggests a differing developmental trajectory for schizophrenia and coeliac disease. It is possible that these two conditions do not share any functional overlap. © 2011 Wiley-Liss, Inc.
Schizophrenia (SCZD, MIM ID: 181500) is a severe psychotic illness with a heterogeneous presentation. Twin studies suggest significant interactions between genetic and environmental factors in the development of SCZD [Cardno and Gottesman, 2000], which may explain it is varying (7.7–43.0 per 100,000) worldwide incidence [McGrath, 2006]. The human leukocyte antigen (HLA) region has been found to be associated with SCZD in large GWA studies [Purcell et al., 2009; Shi et al., 2009; Stefansson et al., 2009]. Those HLA SNPs that are polymorphic in Chinese populations have since been shown to be associated with SCZD [Li et al., 2010]. While the role of autoimmune processes in developing SCZD is contentious and may be secondary to other pathological processes or medication effects [Strous and Shoenfeld, 2006; Potvin et al., 2008], the illness may result from environmental factors inducing autoimmune responses.
There is a long history suggesting interactions between SCZD and coeliac disease (CD, MIM ID: 212750) [Dohan, 1966; Dohan et al., 1972; Reichelt and Landmark, 1995]. Transglutaminase 2 (TGM2, MIM ID: 190196), an enzyme responsible for post-translational modification of proteins is implicated in the development of CD and we have recently reported that the TGM2 gene is associated with SCZD [Bradford et al., 2009]. TGM2 expression has been found to be increased in post mortem livers of patients with SCZD irrespective of medication [Choi et al., 2009]. CD is typified by an autoimmune response against the small intestine [Fasano et al., 2003; Mäki et al., 2003; Gianfrani et al., 2005]. While the exact processes are unresolved, it appears that in CD native gliadins, TGM2-deamidated gliadin fragments and TGM2–gliadin complexes are presented as antigens by macrophages and dendritic cells [Alaedini and Green, 2005; Briani et al., 2008]. Antigen presentation can trigger both cellular (Th1) and humoral (Th2) immune responses against the intestine via HLA class II molecules [Gianfrani et al., 2005; Briani et al., 2008].
CD diagnosis rates vary by country from 43 to 149 per 10,000 [Rewers, 2005] with up to 115 further cases per 10,000 undiagnosed [Fasano et al., 2003; Mäki et al., 2003; West et al., 2003]. Unequivocal diagnosis of CD requires sero-positivity for anti-gliadin antibodies (AGA), anti-deamidated gliadin protein fragments antibodies (ADGPA), and anti-TGM2 antibodies (anti-TGM2), biopsy confirmation of enteropathy, and improvement on a gluten-free diet [Alaedini and Green, 2005; Briani et al., 2008].
To further explore if there is a commonality between CD and SCZD, we reviewed existing literature on the subject by exploring the following questions: (1) Is there evidence for co-incidence of these disorders? (2) Are the serological measures of CD present in SCZD? (3) Are genetic variants associated with CD also associated with SCZD? Based on this information we identified candidate genes that could be tested in our SCZD samples.
Population-based studies suggest a slight increase in the rate of CD in those with SCZD compared to controls of 5.19 versus 1.35 CD cases per 10,000 (OR = 3.85, 95% CI 1.34–11.02) [Eaton et al., 2006]. When individuals are selected by CD diagnosis the evidence for co incidence of SCZD is less clear. One study reported a non-significant decrease of 25.4 vs. 34.4 SCZD cases per 10,000 (OR = 0.74, 95% CI 0.40–1.35) [West et al., 2006] while others observed a non-significant increase of 10.0 versus 7.4 SCZD cases per 10,000 (OR = 1.36, 95% CI 0.75–2.47) [Ludvigsson et al., 2007]. Ludvigsson et al.  also observed a significant increase in the incidence of non-schizophrenic non-affective psychosis in those previously diagnosed with CD (40 vs. 26 cases per 10,000, OR 1.51, 95% CI 1.12–2.04). Whilst this findings is beyond the scope of our study, it is in keeping with current evidence regarding common heritable risks for both SCZD and bipolar disorder [Craddock et al., 2009].
In a study of 1,401 Caucasian-American and African-American individuals diagnosed as having SCZD (n = 1323), schizoaffective disorder (n = 76), and schizophrenifom (n = 2) and 900 control subjects, significantly more cases were positive for circulating IgA AGA and to a lesser extent IgA anti-TGM2 (23.1% vs. 3.1%, OR = 9.33, 95% CI 6.28–13.87, and 5.4% vs. 0.8% OR = 7.17, 95% CI 3.29–15.62, respectively) [Cascella et al., 2011]. Cascella et al.  cautioned that control subjects were at a high risk for CD as their gluten intake and mental illness status data were not available potentially leading to higher than normal antibody scores.
Dickerson et al.  found that CD associated antibody levels were higher in 129 individuals with recent (<2 years) onset psychosis and 191 individuals with multi-episode SCZD (SCZD 96, schizoaffective disorder 87, other SCZD subtype 8) when compared to 151 controls. When grouped by ≥90th percentile of control antibody scores those with recent psychosis had significantly higher IgG and IgA AGA (OR = 5.50, 95% CI 2.65–11.42, and OR = 2.75, 95% CI 1.31–5.75, respectively), while the multi-episode group only differed significantly in IgG AGA (OR = 6.19, 95% CI 2.70–14.16) [Dickerson et al., 2010]. The recent onset group exhibited significantly higher IgG anti-TGM2 levels (OR = 2.29; 95% CI 1.15–4.55) and a small non-significant increases in both groups in IgA anti-TGM2 and ADGPA. Only 1 recent onset SCZD, 1 multi-episode SCZD, and no controls exhibited levels of anti-TGM2 diagnostic of CD; no patients and 1 control exhibited diagnostic levels of ADGPA [Dickerson et al., 2010].
We have recently reported that significantly more Chinese individuals with SCZD were positive for IgA AGA (128/473 SCZD vs. 82/461 controls, OR = 1.72, 95% CI 1.25–2.35). In the AGA positive groups, there were no significant differences in the number of subjects positive for IgA ADGPA (8/128 SCZD vs. 10/82 controls) or TGM2 (1/128 SCZD vs. 1/82 controls) [Jin et al., 2010]. Mean IgA ADGPA and anti-TGM2 were slightly but significantly lower in patients with SCZD compared to controls.
Genes Associated With Both CD and SCZD
HLA alleles represent around 36% of the heritable risk for CD [Petronzelli et al., 1997]. The majority of individuals with CD carry the HLA-DQ2 or HLA-DQ8 variants [Sollid, 2002]; specifically the HLA-DQ2.5 haplotype is carried by ∼90% of UK CD cases compared to ∼25% of controls [van Heel et al., 2007]. A single report observed no differences in DQ2 and DQ8 frequencies between SCZD patients and control subjects [Dickerson et al., 2010]. As the majority of heritable risk for CD comes from non-HLA genes, other genes have been investigated as candidates for SCZD.
Following up their earlier work [Van Belzen et al., 2003], Monsuur et al.  localized a CD linkage signal on the short arm of chromosome 19 (19p13.1) to a 6 SNP haplotype in the Myosin IXB gene (MYO9B, MIM ID: 602129; haplotype AAACTA freq. in CD 38.7% vs. control 30.2%, OR = 1.56, 95% CI 1.27–1.93). The MYO9B association was due to the rs2305764 A allele that confers CD risk (OR = 1.66 for heterozygosity, 95% CI = 1.23–2.13, and OR = 2.27 for homozygozity, 95% CI = 1.56–3.30) [Monsuur et al., 2005].
Gliadin induced immune responses require the gliadin fragments to cross the intestinal permeability barrier or to be taken by dendritic cells for antigen presentation [Ráki et al., 2006]. Genetic variants in MYO9B may facilitate this through altering antigen sensing across the intestine, increasing intestine permeability to antigens, T-cell infiltration, and/or development of anti-gliadin antibodies. MYO9B is actin binding molecular motor protein that can additionally enhance the GTPase ability of the rho-GTPase G protein family members [Bähler et al., 1997]. MYO9B regulates the cytoskeletal remodeling underlying macrophage and monocyte recruitment, leukocyte motility, and phagocytosis [Bement et al., 2006; van den Boom et al., 2007; Hanley et al., 2010]. Through a poorly understood method (for review see Bethune and Khosla ) MYO9B is involved in maintenance of paracellular permeability through its action on apical junction complexes [Bruewer et al., 2004]. Additionally, the MYO9B protein is present in dendritic cells [Hanley et al., 2010], whose dendrites can project into the intestinal lumen through the small intestine epithelium to directly sample gluten-derived antigens [Ráki et al., 2006].
To explore the potential overlap of SCZD and CD, Jungerius et al.  tested the CD-associated MYOB9 haplotype and an additional non-synonymous SNP rs1545620 in a Dutch population with SCZD. While rs2305764 that was strongly associated with CD was not associated with SCZD, the frequency of the haplotype containing the rs2305767 A allele was higher in patients with SCZD (64.0%) than controls (55.9%) with an OR = 1.41 (95% CI 1.18–1.67). rs1545620 also showed a weak association with SCZD with the frequency of rs1545620 C allele42.1% in cases and 35.8% in controls (OR = 1.31, 95% CI 1.10–1.55) [Jungerius et al., 2008]. Accordingly, we sought to build on this work by genotyping rs1545620 and rs2305767 and further characterize local haplotype structure by typing rs8113494 in an alternative linkage disequilibrium (LD) block.
An overlap between CD and SCZD may also be secondary to a general role for autoimmune processes in SCZD, rather than a common pathology with CD. While initially associated with CD [van Heel et al., 2007], meta-analysis of rs6822844, a SNP present between the Interleukin-2 (IL-2, MIM ID: 147680) gene and the interleukin-21 (IL-21, MIM ID: 605384) gene in 4q26-28 [Seigel et al., 1984], suggests that it is likely to be associated with a general risk for autoimmune diseases in Europeans [Maiti et al., 2010]. The location of rs6822844 makes assignment of disease association to one gene or the other problematic. Assignment is further complicated as both genes are likely to be candidates for autoimmune processes. IL-2 is able to both enhance and suppress immune responses. While naïve murine CD4+ T cells do not produce IL-2 [Dooms et al., 2007], activation of CD4+ T cells by antigen recognition leads to expression of IL-2 [Setoguchi et al., 2005] and its high affinity receptor sub-unit CD25 [D'Cruz and Klein, 2005; Fontenot et al., 2005; Dooms et al., 2007]. IL-2 signaling is essential for the survival and homeostasis, but not initial generation, of CD4+ regulatory or memory T cells [D'Cruz and Klein, 2005; Fontenot et al., 2005; Dooms et al., 2007]. IL-2 also assists immune response on re exposure to an antigen by promoting development of CD8+ T memory cells [Williams et al., 2006]. Like IL-2, IL-21 is released by CD4+ T cells, and it has a complex role in homeostasis of lymphocytes [Datta and Sarvetnick, 2008]. IL-21 can induce expression of the genes that enhance inflammatory and Th1 immune responses, including antigen presentation [Strengell et al., 2002]. This study was undertaken to test the hypothesis that the general autoimmune association observed for IL-2/IL-21 is true in SCZD and also to identify a risk haplotype within the IL-2/IL-21 region [van Heel et al., 2007].
MATERIALS AND METHODS
Subjects and Genotyping
Our study utilized a combined British family and case–control sample sets. The family set included 132 trios, comprised of both parents and a SCZD proband as well as 91 single parent and affected offspring duos, in which 159 of the family probands were male and 64 were females. The case control set included 108 unrelated individuals with SCZD (62 male and 46 females) and 120 controls (56 male and 64 female). In total of 331 individuals with SCZD (221 males and 110 females), mean age of 34.1 years (SD = 10.6 years), 120 singleton controls (mean age of 42.1, SD = 12.8 years) and 353 family members were included.
All family samples and the majority of unrelated case–control samples were collected through the Schizophrenia Association of Great Britain, Bangor, UK, in the period between 1990 and 2005. Additional unrelated controls and case samples were collected from 2009 to 2010 from NHS New Craigs Hospital, Inverness. Diagnosis of SCZD was by a clinical interview according to either DSM-III-R or DSM-IV criteria. All subjects gave written informed consent to donating blood samples for the genetic study, which was approved by a research ethics committee and conformed to the requirements of the Declaration of Helsinki and its amendments.
ABI TaqMan® Genotyping Assays for Humans (Applied Biosystems product ID 4351376, individual SNP genotyping kits accessible via their ID, i.e., rs number) were performed on a Techne QUANTICA real time PCR machine. Reactions were performed as per product protocol as 1× TaqMan® Genotyping Assay, 1× ABI genotyping master mix (Product ID 4371357), and 5–20 ng of genomic DNA. PCR cycling: 95°C for 10 min and 45 cycles of 95°C for 15 sec, and 60°C for 1 min.
Literature was initially reviewed by combining the following terms: Schizophrenia, coeliac, celiac, immune system, autoimmune, interleukins 2 and 21 (or IL-2 and IL-21), and MYO9B. Further publications were sourced from the resulting articles' bibliographies.
Genotyping data was analyzed using the Haploview program (version 4.0) [Barrett et al., 2005] to test for Mendelian errors, Hardy–Weinberg Equilibrium (HWE) violations, and to estimate LD. The UNPHASED program (version 3.1.4) was used for analysis of allelic, genotypic, and haplotypic associations using the combined samples [Dudbridge, 2008]. To maximize the data available, UNPHASED imputed missing genotypes [Dudbridge, 2008]; repeating without imputation did not significantly alter results (data not shown). For UNPHASED haplotypic analysis, the zero haplotype frequency was set at 0.01. OR and 95% CI are relative to the allele, genotype, or haplotype containing the major allele(s). P values are two-tailed likelihood ratio X2 tests performed by UNPHASED using the combined transmission and case–control data.
As the IL-2/IL-21 and MYO9B genes were tested independently, study-wise null hypothesis rejection was set at a P value of ≤0.025. To control for the inflation of false positive rates due to testing multiple individual SNPs within the genes for allelic or genotypic associations, UNPHASED performed 10,000 permutations of the test with the lowest P value. Bonferroni correction was applied to post hoc testing of individual haplotypes.
Power analysis for the combined samples as used by UNPHASED is problematic, and an approximation of this was done via SamplePower for SPSS v2.0. Assuming a simple case–control analysis, “case” chromosomes numbers were 2 × (singleton case N + N family probands) = 662 chromosomes, while “controls” was 2 × (singleton controls N + N family probands) = 686 chromosome. Under these assumptions, power to replicate the allelic associations of OR = 1.41 by Jungerius et al.  for MYO9B's rs2350767 gives an adequate power of 80% at P = 0.025. The lower OR observed for rs1545620 leads to a reduction in replication power of 57% at P = 0.025. Assuming that rs6822844 at the IL-2/IL-21 locus has the same role in SCZD as it does in autoimmune diseases with a combined OR of 1.43 our power to replicate is similar at 58%.
Articles that investigated rs2305764 in the context of CD in as of November 30, 2010 (n = 9) were included in the meta analysis if they (a) represented unique study sets, (b) studied European Caucasian populations and, (c) confirmed diagnosis of CD by biopsy and serum antibody test, (d) provided sufficient information to calculate rs2305764 allele frequencies and OR. At the time of submission, a single study [Wolters et al., 2007] could not be accessed nor could the data be retrieved from the author, leaving n = 8. To combine case–control and family data sets, meta-analyses were performed using the R package case–control and TDT meta-analysis package, catmap Version 1.6 [Nicodemus, 2008]. The R software, version 2.12 and the current catmap package are accessible via http://www.r-project.org. A number of studies reported on both TDT and case control sets [Amundsen et al., 2006; Giordano et al., 2006; Núñez et al., 2006; Latiano et al., 2008]; because meta-analysis assumes independent studies, only the larger case control set were included excepting Giordano et al.'s  TDT set. For random effects, weighting was by the DerSimonian Laird method while, for fixed effects, inverse variance was used. Heterogeneity (Q test P value is <0.05) suggests that a fixed-effect model is inappropriate as the assumption that the only between study difference is statistical power is likely violated [Huedo-Medina et al., 2006]. Forest plots were generated using GraphPad Prism v5 (GraphPad Software, San Diego, CA, www.graphpad.com).
Data analysis by Haploveiw showed no HWE deviations in genotyped SNPs or Mendelian errors in family sets. Analysis of our samples using UNPHASED found that no SNPs (Tables I and II) or haplotypes (Table III) were significantly associated with SCZD. The SNPs genotyped for MYO9B represent two LD blocks (Fig. 1), with rs2305767 and rs1545620 associated previously with SCZD [Jungerius et al., 2008]. As shown in Table IV, 2-SNP haplotypic analysis (excluding rs8113494) revealed an uncorrected significant association due to over-transmission of the major allele haplotype (P = 0.0476, OR = 1.332, 95% CI 0.997–1.778); correction for multiple testing renders this non-significant (threshold P ≤ 0.00625).
|Myo9B||TDT||CC||Combined effects||Genotype||TDT||CC||Combined effects|
|SNP||Allele||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P|
|rs8113494||A||297 (0.672)||300.6 (0.674)||150 (0.694)||176 (0.733)||1.000||0.875||AA||98 (0.443)||102.6 (0.46)||53 (0.491)||66 (0.55)||1.000||0.923|
|T||145 (0.328)||145.4 (0.326)||66 (0.306)||64 (0.267)||0.982 (0.786–1.227)||AT||101 (0.457)||95.3 (0.427)||44 (0.407)||44 (0.367)||1.017 (0.759–1.361)|
|TT||22 (0.1)||25.2 (0.113)||11 (0.102)||10 (0.083)||0.923 (0.560–1.521)|
|rs2035767||A||264 (0.597)||258.9 (0.58)||129 (0.597)||146 (0.613)||1.000||0.642||AA||77 (0.348)||81.3 (0.365)||42 (0.389)||44 (0.37)||1.000||0.724|
|G||178 (0.403)||187.1 (0.42)||87 (0.403)||92 (0.387)||0.951 (0.769–1.176)||AG||110 (0.498)||95.9 (0.43)||45 (0.417)||58 (0.487)||1.026 (0.751–1.402)|
|GG||34 (0.154)||45.8 (0.205)||21 (0.194)||17 (0.143)||0.881 (0.574–1.351)|
|rs1545620||A||270 (0.614)||250.3 (0.564)||130 (0.602)||132 (0.555)||1.000||0.168||AA||79 (0.359)||79.4 (0.358)||42 (0.389)||40 (0.336)||1.000||0.214|
|C||170 (0.386)||193.7 (0.436)||86 (0.398)||106 (0.445)||0.861 (0.696–1.066)||AC||112 (0.509)||91.7 (0.413)||46 (0.426)||52 (0.437)||0.984 (0.727–1.331)|
|CC||29 (0.132)||50.9 (0.229)||20 (0.185)||27 (0.227)||0.7000 (0.45–1.091)|
|IL-2/IL-21||TDT||CC||Combined effects||Genotype||TDT||CC||Combined effects|
|SNP||Allele||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P|
|rs6840978||C||96 (0.217)||88.7 (0.199)||46 (0.213)||45 (0.188)||Ref||0.483||CC||9 (0.041)||9.4 (0.042)||4 (0.037)||5 (0.042)||Ref||0.669|
|T||346 (0.783)||357.3 (0.801)||170 (0.787)||195 (0.813)||0.911 (0.703–1.181)||CT||78 (0.353)||70 (0.314)||38 (0.352)||35 (0.292)||1.094 (0.525–2.281)|
|TT||134 (0.606)||143.6 (0.644)||66 (0.611)||80 (0.667)||0.954 (0.450–2.021)|
|rs6822844||G||80 (0.182)||75.8 (0.171)||39 (0.181)||37 (0.154)||Ref||0.586||GG||6 (0.027)||5.6 (0.025)||3 (0.028)||3 (0.025)||Ref||0.828|
|T||360 (0.818)||368.3 (0.829)||177 (0.819)||203 (0.846)||0.927 (0.705–1.218)||GT||68 (0.309)||64 (0.289)||33 (0.306)||31 (0.258)||0.850 (0.348–2.073)|
|TT||146 (0.664)||152.4 (0.687)||72 (0.667)||86 (0.717)||0.792 (0.321–1.955)|
|rs8113494||rs2035767||rs1545620||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P|
|A||A||A||14.8 (0.034)||8.9 (0.020)||8.1 (0.038)||15.4 (0.064)||Ref||0.226|
|A||A||C||128.6 (0.292)||140.9 (0.317)||71.6 (0.332)||80.2 (0.334)||0.732 (0.389–1.380)||0.791|
|A||G||A||150.4 (0.342)||147.4 (0.332)||70.1 (0.325)||75.2 (0.313)||0.735 (0.39–1.387)||0.949|
|A||G||C||1.2 (0.003)||3.7 (0.008)||0.1 (0.001)||5.3 (0.022)||0.176 (0.019–1.588)||0.155|
|T||A||A||80.3 (0.183)||60 (0.135)||36.2 (0.168)||33.1 (0.138)||0.902 (0.463–1.756)||0.104|
|T||A||C||39.3 (0.089)||49.6 (0.112)||13.1 (0.06)||18.9 (0.079)||0.535 (0.262–1.091)||0.073|
|T||G||A||24.4 (0.056)||33.4 (0.075)||15.5 (0.072)||9.9 (0.041)||0.674 (0.322–1.412)||0.610|
|T||G||C||0.9 (0.002)||0.142 (0.0003)||1.2 (0.006)||2.16 (0.009)||1.824 (0.214–15.53)||1.000|
|rs6840978||rs6822844||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P|
|C||G||80 (0.182)||73.1 (0.165)||37 (0.171)||37 (0.154)||Ref||0.578|
|C||T||14 (0.032)||16.2 (0.036)||9 (0.042)||8 (0.033)||0.964 (0.522–1.78)||0.927|
|G||G||0.032 (0.0001)||2.3 (0.005)||2.03 (0.0094)||0.024 (0.0001)||1.090 (0.158–7.54)||1.000|
|G||T||346 (0.786)||352.4 (0.794)||168 (0.778)||195 (0.812)||0.923 (0.700–1.217)||0.561|
|rs2035767||rs1545620||T (freq.)||U (freq.)||SCZD (freq.)||Control (freq.)||OR (95% CI)||P|
|A||A||95.2 (0.216)||69.6 (0.157)||44.3 (0.205)||48.1 (0.201)||1.332 (0.997–1.778)||0.0476|
|A||C||167.8 (0.381)||190 (0.428)||84.7 (0.392)||99.2 (0.413)||Ref||0.2197|
|G||A||174.8 (0.397)||180.7 (0.407)||85.7 (0.397)||85.2 (0.355)||1.067 (0.845–1.348)||0.8404|
|G||C||2.22 (0.005)||3.68 (0.008)||1.28 (0.006)||7.54 (0.031)||0.621 (0.152–2.535)||1|
As we were not able to replicate the findings of Jungerius et al. , it may be that the strength of the MYO9B association with SCZD is lower than expected. To test this, we explored the literature regarding association between CD and MYO9B. Replication of the initial rs2305764 association with CD has been variable, a problem explored by Latiano et al.  using a meta-analysis method that does not directly combine case–control and family datasets. To address this, we utilized the R package “catmap: Case–control And TDT Meta-Analysis” [Nicodemus, 2008] (Table V and Fig. 2). This also allowed us to include more recent studies and the fuller transmission data of Giordano et al. . Where exact allele counts were not provided, they were estimated from published frequencies [Monsuur et al., 2005; Amundsen et al., 2006; Giordano et al., 2006; Hunt et al., 2006; Núñez et al., 2006; Cirillo et al., 2007; Sánchez et al., 2007; Latiano et al., 2008]. As the Q P-value is <0.05 (Table V), a random effect model is the most appropriate means of synthesizing the overall effect size. Inclusion of more recent findings and utilizing catmap supports the conclusion by Latiano et al.  that the initial associations of MYO9B with CD are either population specific or overestimate the strength of the association (fixed P = 0.0348, OR = 1.071, 95% CI 1.006–1.167, and random P = 0.288, OR = 1.071, 95% CI 0.944–1.216).
|Pooled estimates rs2305764||Values (95% CI)|
|Inverse variance fixed-effects||OR 1.083 (1.006–1.167)|
|Inverse variance fixed-effects||X2 4.46, P = 0.0348|
|Q statistic (heterogeneity)||X2 19.64, P-value 0.00640|
|DerSimonian & Laird random-effects||OR 1.071 (0.944–1.216)|
|DerSimonian & Laird random-effects||X2 1.13, P = 0.288|
|Study||ORs (95% CI)||Study weights|
|Monsuur et al. 2005||1.424 (1.203–1.686)||134.5|
|Núñez et al. 2006||1.068 (0.880–1.297)||101.8|
|Hunt et al. 2006||0.861 (0.700–1.060)||88.94|
|Giordano et al. 2006||1.023 (0.880–1.303)||60.74|
|Amundsen et al. 2006||1.071 (0.880–1.303)||99.98|
|Sánchez et al. 2007||1.260 (0.903–1.758)||34.61|
|Cirillo et al. 2007||0.870 (0.697–1.085)||78.58|
|Latiano et al. 2008||1.109 (0.903–1.362)||90.79|
The relatively modest findings from even the largest GWA studies suggest that other avenues need to be considered if we are to fully understand the pathology of SCZD. The immune system protects an organism against foreign invaders or exogenous compounds yet its dysfunction in response to these insults can lead to deleterious autoimmune processes. Altered immune processes have been observed in SCZD; in vivo alterations in the levels of IL-1RA, sIL-2R, and IL-6, as well as in vitro IL-2, have been replicated across numerous studies of SCZD [Potvin et al., 2008]. IL-2 and IL-21 are intriguing candidates to consider when exploring the potential role of autoimmune processes in SCZD yet the low minor allele frequency of rs682244 reduces study power. Given the modest size of our study population, we can only report that we could find no evidence of a strong genetic association with SCZD. However, publication of these data makes them available for future researchers.
While our power to replicate the most strongly associated MYO9B SNP rs2305767 was adequate (80%), we did not observe a statistically significant association between MYO9B and SCZD. We found a non-significant over transmission of the A–A haplotype (OR = 1.332, 95% CI 0.997–1.778) previously reported as associated with SCZD. However, this is questionable as the alternative A–C haplotype was most strongly associated with SCZD by Jungerius et al.  rather than A–A. The frequency of the A allele at rs2305767 in controls is similar in our study and Jungerius et al. , and in turn close to the reported HapMap database frequency for Europeans. Thus, while Jungerius et al.  used a portion of the controls from the original CD and MYO9B association, it was unlikely to be due to an abnormally low control A allele frequency.
Initial reports can overestimate the strength of a SNP association in a process termed Winner's Curse [Zollner and Pritchard, 2007]; it may be that the role of MYO9B in SCZD is more modest than estimated for our power analysis. Arguably, this may be because the association between CD and MYO9B may itself over estimated (Table V) and the role of MYO9B in gluten-related immune responses is small. Weighting accounted for differences in quality and as all retrievable studies met the inclusion criteria, and the SE funnel plot suggested no systemic publication bias (data not shown) this finding is not the result of biased study inclusion. However, meta-analysis cannot itself prove that effect size pooling is justified or meaningful. Heterogeneity (differences in effect sizes and confidence intervals between studies) should not be ignored, rather examined to understand why an outcome might differ between populations [Egger et al., 1997]. Differing LD structures, founder effects or multiple risk haplotypes of a relatively recent origin would render meta-analysis of rs2305764 across European populations inappropriate. Thus, we can be confident that across the entirety of Europe rs2305764 represents a small, if any, risk for CD development but we cannot rule out a higher risk within subpopulations.
Recent work has shown that high anti-gliadin antibodies are present in a large proportion of SCZD cases in a range of different populations [Reichelt and Landmark, 1995; Dickerson et al., 2003; Dickerson et al., 2010; Jin et al., 2010; Li et al., 2010]. Yet the CD diagnostic anti-TGM2 or ADGPA antibodies while slightly increased are not ubiquitous with AGA in SCZD. The majority of individuals with CD carry the HLA-DQ2 or HLA-DQ8 variants [Sollid, 2002], yet an initial report of HLA-DQ variant frequencies in SCZD found no significant differences from controls [Dickerson et al., 2010]. Finally the concordance rates between CD and SCZD are low or possibly non-absent [Eaton et al., 2006; West et al., 2006; Ludvigsson et al., 2007]. Together with the work we have reported here, this suggests a differing developmental trajectory for SCZD and CD beyond gliadin antibody generation. That is while individuals SCZD exhibits certain autoimmune characteristics and may be more likely to mount an immune response against gluten, SCZD itself may not share any functional overlap with CD. However, the high frequency of CD-related HLA risk alleles in the healthy population indicates non-HLA genes are involved in CD risk. We reasoned that investigation of non-HLA CD risk genes in SCZD may shed light on both any overlap between these illnesses and the role of autoimmune factors in SCZD. The candidates we investigated, MYO9B and IL-2/IL-21 did hot, however, exhibit significant association with SCZD.
This work does not exhaust all the potential non-HLA CD risk loci. Zhong et al.  identified a CD linkage region spanning 6p23–6p22.3, linkage being strongest at the 6p23 microsatellite D6S259 [Zhong et al., 1996]. However, others have suggested that non-HLA chromosome 6 linkage with CD may be due to association signals extending from the HLA region itself [Houlston et al., 1997; Forabosco et al., 2009]. Separately Straub et al.  identified a broad region spanning 6p24-21 to be in linkage with SCZD; D6S259 itself marginally so (P = 0.05). Follow up by the same group refined the SCZD association to two peaks in 6p24.3 and 6p22.3 [Straub et al., 2002]. The strongest CD-associated markers fall between these peaks [Zhong et al., 1996] though the SCZD peak linkage signal in the 6p22.3 region includes the CD-linked D6S260 [Straub et al., 2002]. Further fine mapping indicated that the 6p22.3 SCZD linkage is likely due to haplotypes in the Dysbindin gene and these do not extend into 6p23 [Straub et al., 2002]. Separate work suggests that the 6p24 SCZD linkage may be due to a gene involved in cognition [Hallmayer et al., 2005], possibly BMP6 or TXNDC5 [Lin et al., 2009]. This suggests that while close to one another the discussed SCZD and CD linkage peaks in 6p may not overlap, and identification of further 6p CD candidate genes would require more detailed fine mapping beyond the scope of this work. As we are interested in testing CD candidate genes in a population with SCZD rather than re-testing existing genes for SCZD this region was not pursued any further in this study.
An intriguing possibility is that MYO9B represents a general risk for autoimmune diseases, as suggested for IL-2/IL-21. With the caution of our expansion of Latiano et al.'s  meta-analysis, MYO9B has been found to be associated with a range of immune conditions besides CD, including rheumatoid arthritis [Sánchez et al., 2007], type 1 diabetes [Santiago et al., 2008], ulcerative colitis and Chron's disease [van Bodegraven et al., 2006; Latiano et al., 2008]. It is worth noting that there is an inverse relationship between SCZD and rheumatoid arthritis [Mors et al., 1999]. Other studies have observed both positive and negative rates of co-incidence of type 1 diabetes and schizophrenia, a situation that is further confounded as maternal type-1 diabetes may itself be a risk factor for SCZD [Wright et al., 1996; Mors et al., 1999; Eaton et al., 2006; Juvonen et al., 2007]. Different combinations of risk variants leasing to different autoimmune phenotypes may explain this complex interplay between SCZD and autoimmune diseases. If MYO9B is a general autoimmune risk factor, it may have a relatively low effect size as observed for IL-2/IL-21, meaning that only large samples are powered to define the role of these genes in SCZD. However, it is unusual to find a genetic analysis paper that does not end with a call for replication in a larger population. Proper exploration of the role of autoimmune responses in SCZD requires combination of genetic testing with functional measurements, for example, by identifying dietary antibodies or auto-antibodies positive sub-populations.
The authors thank the patients and healthy volunteers for their support and participation. They also thank their colleagues at the Department of Diabetes & Cardiovascular Science, UHI Millennium Institute, for their supportive work. Specific thanks go to Professor Ian Megson for proofreading of this article. This study was supported by the Schizophrenia Association of Great Britain, Bangor, UK, and NHS Highlands R&D office, Inverness, UK.
- 2005. Narrative review: Celiac disease: Understanding a complex autoimmune disorder. Ann Intern Med 142(4): 289–298. , .
- 2006. Association analysis of MYO9B gene polymorphisms with celiac disease in a Swedish/Norwegian cohort. Hum Immunol 67(4–5): 341–345. , , , , , , , , .
- 1997. Physical mapping of human myosin-IXB (MYO9B), the human orthologue of the rat myosin myr 5, to chromosome 19p13.1. Genomics 43(1): 107–109. , , , , .
- 2005. Haploview: Analysis and visualization of LD and haplotype maps. Bioinformatics 21(2): 263–265. , , , .
- 2006. Rho GTPase activity zones and transient contractile arrays. Bioessays 28(10): 983–993. , , .
- 2008. Parallels between pathogens and gluten peptides in celiac sprue. PLoS Pathog 4(2): e34. , .
- 2009. The TGM2 gene is associated with schizophrenia in a British population. Am J Med Genet B Neuropsychiatr Genet 150B(3): 335–340. , , , , , .
- 2008. Celiac disease: From gluten to autoimmunity. Autoimmun Rev 7(8): 644–650. , , .
- 2004. RhoA, Rac1, and Cdc42 exert distinct effects on epithelial barrier via selective structural and biochemical modulation of junctional proteins and F-actin. Am J Physiol Cell Physiol 287(2): C327–C335. , , , , .
- 2000. Twin studies of schizophrenia: From bow-and-arrow concordances to star wars Mx and functional genomics. Am J Med Genet 97(1): 12–17. , .
- 2011. Prevalence of celiac disease and gluten sensitivity in the United States clinical antipsychotic trials of intervention effectiveness study population. Schizophr Bull 37(1): 94–100. , , , , , , , .
- 2009. Effects of typical and atypical antipsychotic drugs on gene expression profiles in the liver of schizophrenia subjects. BMC Psychiatr 9: 57. , , , , , , , .
- 2007. Do MYO9B genetic variants predispose to coeliac disease? An association study in a cohort of South Italian children. Dig Liver Dis 39(3): 228–231. , , , , , , .
- 2009. Psychosis genetics: Modeling the relationship between schizophrenia, bipolar disorder, and mixed (or “schizoaffective”) psychoses. Schizophr Bull 35(3): 482–490. , , .
- 2005. Development and function of agonist-induced CD25+Foxp3+ regulatory T cells in the absence of interleukin 2 signaling. Nat Immunol 6(11): 1152–1159. , .
- 2008. IL-21 limits peripheral lymphocyte numbers through T cell homeostatic mechanisms. PLoS One 3(9): e3118. , .
- 2003. Association of serum antibodies to herpes simplex virus 1 with cognitive deficits in individuals with schizophrenia. Arch Gen Psychiatr 60(5): 466–472. , , , , , .
- 2010. Markers of gluten sensitivity and Celiac disease in recent-onset psychosis and multi-episode schizophrenia. Biol Psychiatr 68(1): 100–104. , , , , , , , , , .
- 1966. Wartime changes in hospital admissions for schizophrenia. A comparison of admission for schizophrenia and other psychoses in six countries during World War II. Acta Psychiatr Scand 42(1): 1–23. .
- 1972. Antibodies to wheat gliadin in blood of psychiatric patients: Possible role of emotional factors. Biol Psychiatr 5(2): 127–137. , , , , .
- 2007. Interleukin-2 enhances CD4+ T cell memory by promoting the generation of IL-7R alpha-expressing cells. J Exp Med 204(3): 547–557. , , , .
- 2008. Likelihood-based association analysis for nuclear families and unrelated subjects with missing genotype data. Hum Hered 66(2): 87–98. .
- 2006. Association of schizophrenia and autoimmune diseases: Linkage of Danish national registers. Am J Psychiatr 163(3): 521–528. , , , , , , .
- 1997. Meta-analysis: Principles and procedures. Br Med J 315(7121): 1533–1537. , , .
- 2003. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: A large multicenter study. Arch Intern Med 163(3): 286–292. , , , , , , , , , , , , , , , , , .
- 2005. A function for interleukin 2 in Foxp3-expressing regulatory T cells. Nat Immunol 6(11): 1142–1151. , , , .
- 2009. Meta-analysis of genome-wide linkage studies in celiac disease. Hum Hered 68(4): 223–230. , , , , , , , , , .
- 2005. Adaptive and innate immune responses in celiac disease. Immunol Lett 99(2): 141–145. , , .
- 2006. A family-based study does not confirm the association of MYO9B with celiac disease in the Italian population. Genes Immun 7(7): 606–608. , , , , , , , .
- 2005. Genetic evidence for a distinct subtype of schizophrenia characterized by pervasive cognitive deficit. Am J Hum Genet 77(3): 468–476. , , , , , , , , , , , .
- 2010. Motorized RhoGAP myosin IXb (Myo9b) controls cell shape and motility. Proc Natl Acad Sci USA 107(27): 12145–12150. , , , , , , , , .
- 1997. Linkage analysis of candidate regions for coeliac disease genes. Hum Mol Gen 6(8): 1335–1339. , , , , , , , , , , , , , , , , , , , , , .
- 2006. Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychol Methods 11(2): 193–206. , , , .
- 2006. Lack of association of MYO9B genetic variants with coeliac disease in a British cohort. Gut 55(7): 969–972. , , , , , , , , , , .
- 2010. A study of circulating gliadin antibodies in schizophrenia among a Chinese population. Schizophr Bull. (Published online September 30, 2010) DOI: 10.1093/schbul/sbq111. , , , , , , .
- 2008. Is MYO9B the missing link between schizophrenia and celiac disease? Am J Med Genet B Neuropsychiatr Genet 147(3): 351–355. , , , , , .
- 2007. Incidence of schizophrenia in a nationwide cohort of patients with type 1 diabetes mellitus. Arch Gen Psychiatry 64(8): 894–899. , , , , , , , .
- 2008. The association of MYO9B gene in Italian patients with inflammatory bowel diseases. Aliment Pharmacol Ther 27(3): 241–248. , , , , , , , , , .
- 2010. Common variants in major histocompatibility complex region and TCF4 gene are significantly associated with schizophrenia in Han Chinese. Biol Psychiatry 68(7): 671–673. , , , , , , , , , , , , .
- 2009. Clustering by neurocognition for fine mapping of the schizophrenia susceptibility loci on chromosome 6p. Gene Brain Behav 8(8): 785–794. , , , , , , , , , , , , , , , , , , , , .
- 2007. Coeliac disease and risk of schizophrenia and other psychosis: A general population cohort study. Scand J Gastroenterol 42(2): 179–185. , , , .
- 2010. Confirmation of an association between rs6822844 at the Il2-Il21 region and multiple autoimmune diseases: Evidence of a general susceptibility locus. Arthritis Rheum 62(2): 323–329. , , , , , , , , , , , , , .
- 2003. Prevalence of Celiac disease among children in Finland. N Engl J Med 348(25): 2517–2524. , , , , , , , , , , , .
- 2006. Variations in the incidence of schizophrenia: Data versus dogma. Schizophr Bull 32(1): 195–197. .
- 2005. Myosin IXB variant increases the risk of celiac disease and points toward a primary intestinal barrier defect. Nat Genet 37(12): 1341–1344. , , , , , , , , , , , , , , , , , , , .
- 1999. A population-based register study of the association between schizophrenia and rheumatoid arthritis. Schizophr Res 40(1): 67–74. , , .
- 2008. Catmap: Case-control and TDT meta-analysis package. BMC Bioinformatics 9: 130. .
- 2006. No evidence of association of the MYO9B polymorphisms with celiac disease in the Spanish population. Tissue Antigens 68(6): 489–492. , , , , , , , , .
- 1997. Genetic contribution of the HLA region to the familial clustering of coeliac disease. Ann Hum Genet 61(Pt 4): 307–317. , , , , , , , , .
- 2008. Inflammatory cytokine alterations in schizophrenia: A systematic quantitative review. Biol Psychiatr 63(8): 801–808. , , , , , .
- 2009. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460(7256): 748–752. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , .
- 2006. A unique dendritic cell subset accumulates in the celiac lesion and efficiently activates gluten-reactive T cells. Gastroenterology 131(2): 428–438. , , , , , .
- 1995. Specific IgA antibody increases in schizophrenia. Biol Psychiatr 37(6): 410–413. , .
- 2005. Epidemiology of celiac disease: What are the prevalence, incidence, and progression of celiac disease? Gastroenterology 128(4 Suppl 1): S47–S51. .
- 2007. MYO9B gene polymorphisms are associated with autoimmune diseases in Spanish population. Hum Immunol 68(7): 610–615. , , , , , , , , , , .
- 2008. Association of MYO9B haplotype with type 1 diabetes. Hum Immunol 69(2): 112–115. , , , , , , .
- 1984. Gene for T-cell growth factor: Location on human chromosome 4q and feline chromosome B1. Science 223(4632): 175–178. , , , , , .
- 2005. Homeostatic maintenance of natural Foxp3(+) CD25(+) CD4(+) regulatory T cells by interleukin (IL)-2 and induction of autoimmune disease by IL-2 neutralization. J Exp Med 201(5): 723–735. , , , .
- 2009. Common variants on chromosome 6p22.1 are associated with schizophrenia. Nature 460(7256): 753–757. , , , , , , , , , , , , , , , , , , , , , , .
- 2002. Coeliac disease: Dissecting a complex inflammatory disorder. Nat Rev Immunol 2(9): 647–655. .
- Genetic Risk and Outcome in Psychosis (Group), , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , . 2009. Common variants conferring risk of schizophrenia. Nature 460(7256): 744–747. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
- 2002. Genetic variation in the 6p22.3 gene DTNBP1, the human ortholog of the mouse dysbindin gene, is associated with schizophrenia. Am J Hum Genet 71(2): 337–348. , , , , , , , , , , , , , , , .
- 1995. A potential vulnerability locus for schizophrenia on chromosome 6p24-22: Evidence for genetic heterogeneity. Nat Genet 11(3): 287–293. , , , , , , , , , .
- 2002. IL-21 up-regulates the expression of genes associated with innate immunity and Th1 response. J Immunol 169(7): 3600–3605. , , , , .
- 2006. Schizophrenia, autoimmunity and immune system dysregulation: A comprehensive model updated and revisited. J Autoimmun 27(2): 71–80. , .
- 2003. A major non-HLA locus in celiac disease maps to chromosome 19. Gastroenterology 125(4): 1032–1041. , , , , , , , .
- 2006. Genetic variation in myosin IXB is associated with ulcerative colitis. Gastroenterology 131(6): 1768–1774. , , , , , , , , , , , , , , , , , , , , , , .
- 2007. The Myosin IXb motor activity targets the myosin IXb RhoGAP domain as cargo to sites of actin polymerization. Mol Biol Cell 18(4): 1507–1518. , , , , .
- 2007. A genome-wide association study for celiac disease identifies risk variants in the region harboring IL2 and IL21. Nat Genet 39(7): 827–829. , , , , , , , , , , , , , , , , , , , , , , , , , , , .
- 2003. Seroprevalence, correlates, and characteristics of undetected coeliac disease in England. Gut 52(7): 960–965. , , , , , , , , .
- 2006. Risk of schizophrenia in people with coeliac disease, ulcerative colitis and Crohn's disease: A general population-based study. Aliment Pharmacol Ther 23(1): 71–74. , , , .
- 2006. Interleukin-2 signals during priming are required for secondary expansion of CD8+ memory T cells. Nature 441(7095): 890–893. , , .
- 2007. The MYO9B gene is a strong risk factor for developing refractory celiac disease. Clin Gastroenterol Hepatol 5(12): 1399–1405 1405.e1391–e1392. , , , , , , , , .
- 1996. Autoimmune diseases in the pedigrees of schizophrenic and control subjects. Schizophr Res 20(3): 261–267. , , , , , , .
- 1996. An autosomal screen for genes that predispose to celiac disease in the western counties of Ireland. Nat Genet 14(3): 329–333. , , , , , , .
- 2007. Overcoming the winner's curse: Estimating penetrance parameters from case-control data. Am J Hum Genet 80(4): 605–615. , .