Abstract for presentation at 11th International Congress of Human Genetics

NRG1 and the Risk of Schizophrenia: Does it Depend on Statistical Epistasis between NRG1 Protein-Interaction Partners ERBb4, CHRNA7, AKT1, DLG4, CAPON and NOS1?

  • Ms Kristin Nicodemus, Genes Cognition and Psychosis Program, CBDB, NIMH, NIH, Bethesda MD, Department of Epidemiology, Johns Hopkins SPH, United States
  • Dr Bhaskar Kolachana, Genes Cognition and Psychosis Program, CBDB, NIMH, NIH, Bethesda MD, United States
  • Dr Radhakrishna Vakkalanka, Genes Cognition and Psychosis Program, CBDB, NIMH, NIH, Bethesda MD, United States
  • Dr Richard Straub, Genes Cognition and Psychosis Program, CBDB, NIMH, NIH, Bethesda MD, United States
  • Dr Michael Egan, Genes Cognition and Psychosis Program, CBDB, NIMH, NIH, Bethesda MD, United States
  • Dr Daniel Weinberger, Genes Cognition and Psychosis Program, CBDB, NIMH, NIH, Bethesda MD, United States

    • Purpose: NRG1 is a candidate gene for schizophrenia that has shown independent replication. Studies have shown that the protein encoded by NRG1 interacts with proteins in other schizophrenia candidate genes. A recent paper has shown evidence for statistical interaction between NRG1 and its receptor ERBb4.
    We sought to comprehensively assess whether variation in NRG1 was associated with schizophrenia by performing an association analysis using a dense map of 48 NRG1 SNPs. We also sought to utilize exploratory statistical methods to test for interaction between SNPs in NRG1 and its protein interaction partners.
    Methods: Using a family sample and a non-independent case control sample, we conducted family-based association tests, unconditional logistic regression, and a score test-based haplotype analysis to assess association between NRG1 single SNPs and haplotypes and risk for schizophrenia. To test for interaction between SNPs in NRG1 and protein-interaction partners, we used a recently proposed Markov chain Monte Carlo boolean logic tree-based algorithm and a CART-based algorithm.
    Results: 3 SNPs in NRG1 were marginally associated with risk of schizophrenia in both family-based and case-control analyses. We observed weak association between the 4 NRG1ICE SNPs in family-based analysis for NRG221132 (p=0.075), and for 3-SNP sliding window haplotypes containing NRG221533 (p=0.086) and SNPs NRG241930 and NRG243177 (p=0.098); the 4 SNP NRG1ICE haplotype was ns in both samples. Both data mining methods created "best" trees using NRG1 rs16879773, rs3802158 and rs11993611, AKT1 rs2494732, CAPON rs4657179 and ERBb4 rs3748960. Focusing solely on these SNPs to reduce multiple testing, we were able to confirm pairwise gene-gene interaction (CAPON*AKT1 p=0.018; NRG1*ERBb4 p=0.033; and NRG1*CAPON p=0.018).
    Conclusion: We were able to confirm weak association between NRG1 and schizophrenia, which was significantly increased by the use of data mining methods to detect interaction with putative partner genes. Our results support the reported interaction between NRG1 and ERBb4. We are currently genotyping an independent case-control sample as a confirmatory set for our data-derived interaction models.

    Conference Organiser - ICMS Pty Ltd