1. Structure prediction with FAMS for proteinsscreened critically to autoimmune diseasesbased upon bioimformaticsShigeharu Ishida (Dept. Phys., Chuo Uinv.)Hideaki Umeyama (Dept. Bio. Sci., Chuo Univ.)Mitsuo Iwadate (Dept. Bio. Sci., Chuo Univ.)Y-h. Taguchi (Dept. Phys., Chuo Uinv.)

2. 1. IntroductionAutoimmune disease:Autoimmune disease arise from aninappropriate immune response of the bodyagainst substances and tissues normallypresent in the body.Ex.Rheumatoid Arthritis (RA; Alopecia areata 3. Previous findings B. M. Javierre et al (2010) Genome Res. 20[2] pp 170-179Target:Systemic lupus erythematosus(SLE; )RADermatomyositis (DM; )Seek common promoter methylationSLEOnly 4. Y-h. Taguchi, DMSM2010 (2010)http://www.ai-gakkai.or.jp/jsai/sig/dmsm/012/Successfully picked up 33 promoters commonlymethylated for RA, SLE, and DM.But, no biological validations ..Data base : , Blast : In this paper, we show that protein structureprediction by FAMS is useful tool for functional annotation of proteins and also possibly usefulfor drug discovery... 5. 3. FAMS (Full Automatic Modeling System)The computer program FAMS performshomology modeling of protein structures bymeans of an algorithm consisting of databasesearches and simulated annealing.Umeyama & Iwadate (2004)Curr Protoc Bioinformatics. 6. 4. ResultsReferenceModelgenesymbol PDBIDP-valuegene symbolAIM22OQ0_B4.00E-92 GAMMA-INTERFERON-INDUCIBLE PROTEIN IFI-16CARD153CIY_B7.00E-64 TOLL-LIKE RECEPTOR 4, VARIABLE LYMPHOCYTE (TLR4)CD822BG9_A4.60E-01 ACETYLCHOLINE RECEPTOR PROTEIN, ALPHA CHAINCSF1R 3B43_A5.00E-83 TITINCSF3 1GNC_A 2.00E-66 GRANULOCYTE COLONY-STIMULATING FACTORCSF3R3DMK_A 1.00E-71 DOWN SYNDROME CELL ADHESION MOLECULE (DSCAM)DHCR24 2Q4W_A 1.00E-115 CYTOKININ DEHYDROGENASE 7 (CKO7)ERCC3 2W74_D1.00E-152 TYPE I RESTRICTION ENZYME ECOR124IIR PROTEIN (HSDR)GRB73HK0_B2.00E-73 GROWTH FACTOR RECEPTOR-BOUNDP ROTEIN10 (GRB10)HGF 2F83_A1.00E-111 COAGULATION FACTOR XIHOXB2 2D5V_A9.00E-24 HEPATOCYTE NUCLEAR FACTOR 6 (HNF-6)IFNGR21FNF_A1.00E-37 FIBRONECTINLCN21X71_A1.00E-51 NEUTROPHIL GELATINASE-ASSOCIATED LIPOCALIN (NGAL)LMO22XJY_A2.00E-33 RHOMBOTIN-2LTB4R 2KS9_A2.00E-83 SUBSTANCE-P RECEPTORMMP14 1SU3_B1.00E-160 INTERSTITIAL COLLAGENASE (MMP-1)MMP81SU3_B1.00E-171 INTERSTITIAL COLLAGENASE (MMP-1)MPL 3L5H_A4.00E-63 INTERLEUKIN-6 RECEPTOR SUBUNIT BETA (IL6RB)PAD142DEW_ X 0.00E+00 PROTEIN-ARGININE DEIMINASE TYPE IVPECAM1 3DMK_A 1.00E-104 DOWN SYNDROME CELL ADHESION MOLECULE (DSCAM)PI31TWP_A 2.00E-19 WHEY ACIDIC PROTEIN (WAP)RARA3DZY_A4.00E-95 RETINOIC ACID RECEPTOR RXR-ALPHAS100A22RGI_A4.00E-19 PROTEIN S100-A2SEPT9 3FTQ_A1.00E-137 SEPTIN-2SLC22A181PW4_A1.00E-108 GLYCEROL-3-PHOSPHATE TRANSPORTERSPI11GVJ_B1.00E-21 C-ETS-1 PROTEIN (ETS1)SPP11D2T_A3.00E-14 ACID PHOSPHATASE (ACP)STAT5A 1Y1U_ A 0.00E+00 SIGNAL TRANSDUCER AND ACTIVATOR OF TRANSCRIPTION (STAT5ASYK 2OZO_A1.00E-168 TYROSINE-PROTEIN KINASE ZAP-70TIE1 3DMK_A 2.00E-84 DOWN SYNDROME CELL ADHESION MOLECULE (DSCAM)TM7SF3 [1AR1_A] 6.00E-88 CYTOCHROME C OXIDASETRIP6 1B8T_A2.00E-32 CYSTEINE-RICH PROTEIN1 (CRP1)VAMP82KOG_A 1.00E-21 VESICLE-ASSOCIATED MEMBRANE PROTEIN 2 (VAMP2) 7. Almost all proteins are modeled with verysmall P-values (10E-20) (with highsignificance)Obtained proteins have functional annotations(because they are listed in PDB)Protein structure prediction by FAMS turnsout to be also useful for functional annotation,because model protein registered in PDB is oftenwell studied and annotated. (Otherwise nonehope to decide 3D structure with much effort andmoney!) 8. Almost all functional annotations arerelated to immunology/autoimmune selected promoters turned out to bebiologically meaningful e.g., AIM2 modeled by IFI-16 Structures of the HIN Domain: DNA Complexes Reveal Ligand Binding and Activation Mechanisms of the AIM2 Inflammasome and IFI16 Receptor Jin et al (2012/4/20) Immunity. 9. 5. To Drug Discovery...5.1 Ligand Searchmmp8/14 are modeled by mmp1. Ligand designedfor mmp1 possibly can bind to mmp8/14, too. mmp8mmp14 10. 5.2 Search for protein complexProtein AProtein Complex Protein Bin PDBProtein Complex Candidate 11. Huge number of protein complex candidates werefound.... CSF1RPECAM1 410 12. CSF1R vs PECAM1 modeled by 2ZJS 13. Why protein complex?If we can terminate protein complexformation by finding molecules to bindinterface between two proteins, they can bedrug. 14. 6. ConclusionFAMS predicts protein structure of geneswhose promoter are commonly de/methylatedfor autoimmune.Protein structure prediction is useful forfunctional annotationIt also helps ligand search FAMS is also useful protein complexinference which may lead to drug discovery