|Home | About | Journals | Submit | Contact Us | Français|
Genome-wide association studies (GWAS) implicate single nucleotide polymorphisms (SNPs) on chromosome 6p21.3-22.1, the human leukocyte antigen (HLA) region, as common risk factors for schizophrenia (SZ). Other studies implicate viral and protozoan exposure. Our study tests chromosome 6p SNPs for effects on SZ risk with and without exposure. Method: GWAS-significant SNPs and ancestry-informative marker SNPs were analyzed among African American patients with SZ (n = 604) and controls (n = 404). Exposure to herpes simplex virus, type 1 (HSV-1), cytomegalovirus (CMV), and Toxoplasma gondii (TOX) was assayed using specific antibody assays. Results: Five SNPs were nominally associated with SZ, adjusted for population admixture (P < .05, uncorrected for multiple comparisons). These SNPs were next analyzed in relation to infectious exposure. Multivariate analysis indicated significant association between rs3130297 genotype and HSV-1 exposure; the associated allele was different from the SZ risk allele. Conclusions: We propose a model for the genesis of SZ incorporating genomic variation in the HLA region and neurotropic viral exposure for testing in additional, independent African American samples.
Recent genome-wide association studies (GWAS) have detected risk for schizophrenia (SZ) associated with polymorphisms in the chromosome 6p/human leukocyte antigen (HLA) region.1–3 Combined data from independent Caucasian ancestry samples, comprising SZ cases (n = 12 945) and controls (n = 34 591) indicated significant associations corrected for multiple comparisons at 5 SNPs, localized to chromosome 6p21.3-22.1; genomic locations from 27.2Mb to 32.3Mb (National Center for Biotechnology Information map, Build 36).1–3 Follow-up studies using additional Caucasian samples continue to support associations with these and additional SNPs in the HLA region.4,5 Our prior candidate gene studies have also reported associations with different HLA polymorphisms in several ethnic groups.6–11 The risk conferred by individual variants in the HLA region is modest, with most odds ratios (ORs) in the 1.15–1.80 range. The risk due to any one marker could not account for all the associations, suggesting multiple risk loci.2 Further, no functional significance could be ascribed to the associated SNPs, although some of them are in linkage disequilibrium (LD) with HLA markers and other SNPs associated with infectious exposure and autoimmune diseases.2
In this study, we further investigated the HLA associations with SZ as they are related to exposure to infectious agents. HLA polymorphisms are known to influence immune surveillance and there are reports of neurotropic infectious agents as risk factors for SZ.12,13 Although a variety of viral agents have been proposed as putative SZ risk factors, including Toxoplasma gondii (TOX), a protozoan parasite,14,15 many of the studies have not been consistent. It is possible that the lack of consistency stems from the failure to investigate host genetic variations. In support, our prior analyses suggest interactions between host HLA polymorphisms and exposure to herpes simplex virus type 1 (HSV-1) and cytomegalovirus (CMV).10,11 We reported that exposure to CMV is increased among multiplex SZ families versus simplex families (OR 2.47, 95% confidence interval, CI = 1.48–5.33).10 In those earlier studies, we further suggested that CMV exposure increases risk for SZ among Caucasians when considered in conjunction with host genetic variability in the HLA region.10,16 Therefore, in this study variation in the HLA region was analyzed in conjunction with exposure to TOX, as well as HSV-1 and CMV. We investigated cases and controls from an African American multisite collaborative study called the Project Among African Americans to Explore Risks for Schizophrenia (PAARTNERS).17
Our goal was to evaluate published HLA/SZ associations. Using a case-control design and a nominal threshold of statistical significance, we initially evaluated individual SNPs previously reported to be associated with SZ (see online supplementary table 1). The associated SNPs were then individually screened in relation to exposure to 3 putative infectious risk agents for SZ.
Unrelated SZ/schizoaffective disorder (SZA) cases (n = 604) and screened adult controls (n = 404) with self-reported African American ancestry were evaluated through the PAARTNERS study.17,18 Briefly, all participants were interviewed using the Diagnostic Interview for Genetic Studies. Additional clinical information was obtained from medical records and consenting relatives. The detailed information was used to obtain consensus diagnoses based on DSM-IV criteria. All participants provided blood samples.
Venous blood was obtained from participants and genomic DNA extracted using the phenol chloroform method as described.19 Serum was extracted from coagulated blood following centrifugation.
SNPs were genotyped primarily using iPLEX, a multiplexed single base extension method using the MassArray MALDI-TOF MS detection platform (SequenomInc.) (http://www.sequenom.com/getdoc/197b98fa-93f7-40e8-9deb-a8dcfecf899e/iPLEX-brochure_web/). SNPs unsuitable for iPLEX were assayed using the multiplexed SNaPshot platform (Applied Biosystems, New Jersey) (http://www3.appliedbiosystems.com/cms/groups/mcb_support/ documents/generaldocuments/cms_041203.pdf). One SNP (rs2517614) was genotyped by Sanger sequencing. All assays included Centre d’Etude du Polymorphisme Humain (CEPH) samples with known genotypes, as well as blind duplicates and negative samples. Genotypes were read blind to case/control status. Assays were repeated for ambiguous genotypes.
SNPs for which satisfactory genotypes could not be generated were replaced with tag SNPs in LD (r2 > 0.9). The relative locations of chromosome 6p SNPs and their LD patterns are provided in figure 1 and online supplementary figure 1, respectively.
The titers of IgG antibodies to HSV-1, CMV, and TOX were estimated using solid phase enzyme immunoassay kits (obtained from KMI Diagnostics Inc, Minneapolis).21,22 Using cutoff values based on internal controls and the manufacturer’s recommendations, individuals were classified as exposed (raised titers) or unexposed to the appropriate infectious agent.16
The study was approved by the Institutional Review Boards (IRB) at the participating collaborative sites. Written informed consent was obtained from all participants in accordance with IRB guidelines.
Associations between individual SNPs and SZ risk were initially tested among the cases and controls using logistic regression analysis. Case-control status was the outcome, with SNP minor allele dosage as the predictor variable, co-varying for admixture proportion. Logistic regression analyses were also used to evaluate interactions between SNPs and exposure variables in relation to SZ risk using case/control status as the outcome and individual SNP genotype, exposure variable, admixture proportion, and demographic variables as covariates. Corrections for multiple comparisons were not applied.
To test for associations with viral and TOX exposure, separate logistical regression analyses were conducted for each infectious agent with serological status (exposed/unexposed) as the outcome. Minor allele dose for each SZ-associated SNP, age, gender, group status (case/control), and admixture proportion were used as covariates. These analyses were conducted using participants with available serological data (n = 749). Population admixture was estimated from the AIMs using LAMP software,23 assuming 2 populations. Ancestral allele frequencies were estimated using HAPMAP CEU and YRI genotypes.
Cases were significantly younger than controls. There were proportionately more men among the cases. Cases and controls did not differ significantly with respect to exposure rates for CMV, TOX, or HSV-1, following correction for age and gender. There were no significant case-control differences with respect to the individual estimates for admixture (table 1).
Nominally, significant associations with SZ were noted for 5 SNPs (see table 2; rs12214031—BTN3A2_3UTR, P = .004; rs9393709—BTN3A2, P = .015; rs12199613—BTN3A2, P = .016, rs6932590, P = .007; rs3130297, P = .0007; uncorrected for multiple comparisons).
To evaluate associations between infectious agent exposure and SNPs associated with SZ from table 2, logistic regression analysis was used for each SNP that was associated with SZ, with exposure status as the outcome. HSV-1 exposure was nominally associated with rs3130297, one of the SNPs associated with SZ (P < .05, table 2). At this SNP, allele A (minor allele) was associated with HSV-1, while the major allele (G) was associated with SZ risk. None of the other SNPs were significantly associated with HSV-1, CMV, or TOX exposure.
Our goal was to evaluate previously reported GWAS results in the HLA region. We detected nominally significant associations between SZ and 5 SNPs. The associated alleles are consistent with the published GWAS reports, although an earlier GWAS study of African Americans did not detect genome-wide significant associations in the HLA region.1 rs3130297, one of the SZ-associated SNPs is also associated with exposure to HSV-1 but the risk alleles differ. The allelic differences are reminiscent of a Caucasian ancestry sample in which we reported that the alleles of an exonic SNP at the MICB locus in the HLA region were associated with SZ or with CMV exposure.11. The basis for such associations is uncertain as there is no known functional effect of the sequence variation at rs3130297. They could indicate an epistatic effect at rs3130297 or a SNP in LD with it. Published studies indicate that exposure to HSV-1 is associated with impairment in specific cognitive domains among SZ patients and community-based control individuals,16,24–27 although an association between HSV-1 exposure and SZ risk per se has not been convincingly demonstrated.28 Nevertheless, our results provide a testable model summarized in figure 2.
There are some shortcomings in our analyses. We did not correct the initial genetic association analyses for multiple comparisons, as the type of correction necessary for previously associated GWAS SNPs is uncertain. Further evaluations of our results are therefore necessary in other independent samples, preferably with African American ancestry. The replicate samples would necessarily require available DNA and serum samples. Another concern is that exposure to infectious agents was indexed indirectly using antibody titers in the serum because demonstration of the viruses in the host target tissues is difficult.29,30 The serological assays clearly indicate infectious exposure, but do not reveal when it occurred. The timing of the exposure may be a critical determinant of viral effects on neurodevelopment thought to be critical for SZ pathogenesis.
In conclusion, our analyses suggest a complex relationship between individual genomic variability, exposure to infectious agents, and SZ risk. The associations suggest a testable model of SZ genesis.
National Institute of Mental Health at the National Institute of Health (R01 grant numbers: MH66006 [LDB], R01 MH66278 [BD], R01 MH066049 [NE], R01 MH66181-03 [RCPG], R01 MH66121 [REG], R01 MH066005 [JK], R01 MH66050 [JPM], R01 MH66263 [VLN] and R01 MH66004 [AS], and K08 MH79364 [MC]; The Stanley Medical Research Institute (07-R1721[VLN]).
Supplementary material is available at http://schizo phreniabulletin.oxfordjournals.org.
The authors thank the study participants and research faculty and staff for their time and effort, and they thank the Western Mental Health Center and Dr Thomas Hobbs for their support and contribution to recruitment in Ensley and Birmingham, AL. The following research faculty and staff contributed at the 8 sites: University of Alabama at Birmingham (central administrative site)—Roberta May; Charlie Swanson, Jr, MD; Laura Montgomery-Barefield, MD; Tolulope Aduroja, MD; Ryan Coleman; Rakesha Garner; Lee Prichett, RN; Thomas Kelley, RN; Marguerite Ryan Dickson, PhD; Muktar Aliyu, MD, DrPH; Adrienne Lahti, MD; Duke—Linda Blalock, RN; University of Mississippi—Karen Richardson, MS; Morehouse School of Medicine—Deirdre Evans-Cosby, MD; George W. Woods, MD; Kendaly Meadows, RN; Sandra Cummings, MSW; Cara Stephens, LCSW; and Kent Baker; Medical University of South Carolina—Shirley Hendrix, Cynthia Gilliard, Wanda Smalls-Smith, and Steven McLeod-Bryant; University of Pennsylvania—Felipe Da Silva; Alexandra Duncan-Ramos, MS; Jarrod Gutman; CarlaAnn Henry; Paul Hughett, PhD; Farzin Irani, PhD; Jennifer Jackson Greene, MS; Stephen J. Kanes, MD, PhD; Christian Kohler, MD; David Rice; Devon Seward; Steven Siegel, MD, PhD; Bruce Turetsky, MD; and Robert Witalec; University of Pittsburgh—Mary Miller, LPN; and Frank Fleischer, MBA; University of Tennessee—Kristin Beizai, MD; Marie Tobin, MD; Alyssa English, MD; Richard Sanders, BS; Shelia A. Dempsey, ADN; Martha Velez, CPS; Marianne Smith, BS, MA; Martha Garriott, MS, NCC; Nancy Fowler; Derrick W. Allen, MSSW; Phyllis Meyer, BS, PA; and Lynn Heustess, BS. Drs Allen, Bradford, Calkins, Devlin, Edwards, Go, McEvoy, Nimgaonkar, Santos, Weiner, Wood report no competing interests. Dr Kwentus receives grant support through collaborations of the University of Mississippi with Eli Lilly, Astra Zeneca, Pfizer, Bristol Meyers, Johnson & Johnson, and Takeda. Dr Raquel Gur receives grant support through collaborations of the University of Pennsylvania with AstraZeneca and Pfizer. Dr Henry Nasrallah has received research grants from Forest, Lilly, Roche/Genentech, Otsuka, Shire and honoraria for consulting or speaking at Genentech, Grunenthal, Janssen, Lundbeck, Merck, Novartis, Sunovion, and Boehringer-Inglheim. The other authors did not report any competing interests.