1. Insights Into Complex Disease Genetics (?) David M Evans

2. Outline Gene-gene Interaction (Epistasis) (Allelic) Heterogeneity Rare Protective Mutations (Controversial!) Genetic Testing in complex disease 2

3. Ankylosing Spondylitis Auto-immune arthritis resulting in fusion of vertebrae Often associated with psoriasis, IBD and uveitis Prevalence of 0.4% in Caucasians. More common in men. Model for complex disease?!

4. Gene-gene Interaction (Epistasis)

5. Gene-Gene Interaction Most of the heritability of complex traits and diseases is yet to be explained Gene-gene interaction is a possible explanation, but thus far no strong examples have been demonstrated. The idea that the whole is more than the sum of the parts 5 Maher et al. (2009)

6. Gene-gene Interaction Against: – Twin Studies, Genetic relationship matrix (Visscher) For: – Model organisms (Brem, Storey, Kruglyak) – Multifactor Dimensionality Reduction (Ritchie, Moore) Reality? 6

7. 7 Evans et al. (submitted) CARD9 TBKBP1 ANTXR2 PTGER4

8. 8 Evans et al. (submitted)

9. (Functional) Implications ERAP1 could predispose to AS through it’s action on peptide presentation or cytokine receptor cleavage The restriction of the ERAP1 association to HLA-B27 positive cases implies functional interaction between the two and that ERAP1 exerts its effect through a mechanism involving peptide presentation 9 NS Evans et al. (submitted)

10. Implications HLA interactions the first of many epistatic interactions to be discovered? 10

11. (Allelic) Heterogeneity

12. 12 Evans et al. (submitted) Odds of disease in double recessives ~ 4.3 x lower! (epistasis again?!)

13. Functional Activity of ERAP1 Variants in vitro 13 (rs30187) (rs10050860) (rs17482078) Evans et al. (submitted)

14. (Functional) Implications Could allelic heterogeneity be the rule rather than the exception and consequently explain a significant portion of the missing heritability in common diseases? Could loss of function variants which are risk protective be a running theme in the genetics of auto-immune diseases? 14

15. Rare Protective Mutations

16. GENESNPRISK ALLELE ANCESTRAL ALLELE? IL23Rrs11209026GYES IL23Rrs11209032ANO KIF21Brs2297909GYES Chr2prs10865331ANO ERAP1rs30187TYES ERAP1rs10050860CYES HLA-Brs4349859ANO Chr22rs378108GNO RUNX3rs11249215ANO IL12Brs6556416CYES LTBRrs11616188ANO ANTXR2rs4389526AYES PTGER4rs10440635AYES CARD9rs10781500TNO TBKBP1rs8070463CNO Relatively high proportion of ancestral variants are risk conferring Loss of function variants protect against an over active immune system?

17. AAAGGG Cases Controls61591164 AAAGGG Cases Controls133232337 AAAGGG Cases21722072 TASC: WTCCC2: Replication: rs11209026 0 74 990 0 161 1625 Rare genotypes just uncommon? Functioning receptor necessary for developing AS? Block receptor -> stop people developing AS? G -> A is rare (6 – 7%) SIFT predicts G -> A at rs11209026 is detrimental

18. Risk Prediction

19. (from Yang et al. 2003 AJHG) (from Janssens et al. 2004 AJHG) Risk Prediction

20. 20 Evans et al. (submitted) CARD9 TBKBP1 ANTXR2 PTGER4

21. BCEAFG Class I Region MICB MICA rs4349859 21

22. Is rs4349859 evidence of the ‘linked gene’? B27 SUBTYPEETHNIC GROUPAS ASSOCIATED? TAGGED BY RS4349859? *02 WHITE EUROPEAN YES *03 AFRICAN YESNO *04 EAST ASIAN YESNO *05 WHITE EUROPEAN YES *07 ASIAN YESNO *08 WHITE EUROPEAN YES *09 SARDINIAN NOYES *10 WHITE EUROPEAN YES 22

23. Criteria for a Useful Diagnostic Test Criteria Disease should be serious or potentially so√ Disease should be relatively prevalent in population√~0.4% in Europeans Disease should be treatable√NSAIDs, TNFα blockers, physio Treatment should be available to those who test positive√NSAIDs, TNFα blockers, physio The test should not harm the individual√Blood test The test should accurately classify diseased and non- diseased individuals ? (Based upon an extract from “The Statistical Evaluation of Medical Tests for Classification”, Pepe (2004))

24. Diagnostic Testing in AS Radiographs, clinical symptoms, HLA-B27 testing: Notoriously difficult early in disease HLA-B27 testing is expensive! Testing in first degree relatives?

25. Conclusions Epistatic interactions contribute to risk of Ankylosing Spondylitis and Psoriasis. Could gene-gene interaction also contribute to other complex diseases? Allelic Heterogeneity is present at the ERAP1 locus. In general allelic heterogeneity may explain significant portions of the missing heritability problem 25

26. Conclusions RS11209026 is an example of a rare protective mutation. It’s full significance is yet to be determined, but rare protective mutations may be more enlightening about disease etiology than rare predisposing ones Ankylosing Spondylitis is the first condition where the results from a GWAS can be translated directly into clinical practice 26

27. BrisbaneWTCCC2 Matthew Brown Adrian CortesTony KennaPeter Donnelly Gethin Thomas Linda BradburyChris Spencer Karena PryceJohanna HadlerGil McVean Katie CreminPatrick Danoymany others.. Oxford BathHoustonShanghai TASC Paul Wordsworth Millicent StoneJohn ReveilleHuji Xu Michael Weisman Jenny PointonRui JinLei JiangMichael Ward Claire FarrarXiaodong ZhouYu Liumany others.. Louise Appleton The AzoresSardiniaLeeds Jacome Bruges-Armas Rosa SorrentinoAnn Morgan Helena Marzo-Ortega Funding: NIAMS, ARC (UK), Wellcome Trust, NHMRC (Australia), ACRF, Arthritis Australia, Rebecca Cooper Foundation and others. Acknowledgements 27