1. Repetitive Beta Folds Form, Function, and Properties

2. Overview of Presentation Introduction to traditional beta-helix fold –Structural and functional properties –Structure prediction by BetaWrap program Introduction to trimeric viral attachment fibers –Structural comparison to traditional beta-helices –Current computational approaches

3. Basic Parallel Beta-Helix Beta-helix is an all beta fold Three strands per rung Right-handed helix (RHBH) Also left-handed helices (LHBH) Function: Sugar cleavage Mainly occurs in bacterial pathogens King lab studies RHBH Tailspike Three faces form a prism

4. Analyzing Beta-Helices Solved structures –RHBH: 5 SCOP SuperFamilies –LHBH: 2 SCOP SuperFamilies –48 solved structures in PDB –8 HSSP representatives Predicting novel beta-helices –Homology modeling, threading, and HMMs do not successfully predict occurrence in cross-validation BetaWrap (King, Berger et al. 2001) successfully predicts RHBH

5. Lessons from BetaWrap Joint sequence-structure analysis important –Discovered conserved hairpin turn –Discovered internally packed asparagines Beta-strand packing interactions are important –BetaWrap energy function uses strand-to-strand packing probabilities Prediction is not enough –BetaWrap does not predict active site, etc. –Other methods (rotamer libraries etc.) may supplement initial prediction

6. Trimeric Viral Attachment Fiber Proteins

7. Trimeric Viral Attachment Fibers King’s interest in beta helix led to interest in two new folds 1.Triple beta-helix (TBH) 2.Triple beta-spiral (TBS) These two folds are our current research area –Consist of three identical interacting chains –TBH is structurally similar to beta-helix –TBS is structurally distinct Both folds characterized by unusual stability to heat, protease, and detergent

8. Triple Beta-Helix Described by van Raaij et al. in JMB (2001) HomoTrimeric (consists of three identical chains) Two solved structures –Portion of T4 short tail fibre SwissProt: P10390 –Cell puncturing device of T4 SwissProt: P16009

9. Triple Beta-Spiral Described by van Raaij et al. in Nature (1999) HomoTrimeric (three identical chains) Two solved structures –Human Adenovirus 2 Fibre SwissProt: P03275 –Reovirus Attachment Fibre SwissProt: P03528 Characterized by regular repeat pattern in literature

10. Preliminary Analysis TBS more regular than TBH –TBS characterized by sequence repeat –Can use standard regex techniques (like PROSITE) to find many putative TBSs –See http://www.baobob.net/cgi-bin/repeat/stored-queries.pl http://www.baobob.net/cgi-bin/repeat/stored-queries.pl TBH has so far defied basic characterization –Only two solved structures –The quasi-repeat is less regular than the TBS

11. Current Research What are we trying to do with TBH and TBS? –There are too few for rigorous prediction tool –Right now we are just “characterizing” them –Searching for sequence-structure patterns –Searching for unique properties –Searching for repetitive sequence motifs Regular Expression is first attempt Search with PSSM sequence profile

12. Repetitive Sequence Motif Search Existing Methods for repetitive motif search –RADAR (Heger & Holm) and others attempt this –Existing methods do not find the adeno repeat –TBH repeat is not regular enough to search Our approaches (tried so far…) –Basic regular expression (more in supplemental) –PSSM characterizing repeat (in progress)

13. Thank you Peter Weigele (pweigele@mit.edu)pweigele@mit.edu and Eben Scanlon (eben@mit.edu)eben@mit.edu

14. Supplemental Slides

15. TBS Information –Human Adenovirus 2 Fiber and Reovirus Attachment Protein σ1 have 27% sequence identity, 52% sequence similarity –Searching SwissProt for the Adenovirus repeat (regex) pattern with more than 6 occurrences finds 3158 matches –Searching SwissProt for the Reovirus repeat (regex) pattern finds 37578 matches –PDB ID’s are 1kke, 1qiu

16. TBH Information –The T4 Short tail fiber TBH and the T4 cell puncturing TBH have 32% sequence identity and 61% sequence similarity –There is no clear repeat pattern in TBH –Tried PSSM and HMM models with alignments derived from known repeat strands in TBH Have not yet figured out a way to restrict to matches with a large number of recurrent repeats Also may want to add a high non-affine gap penalty beyond a certain extension –PDB ID’s are 1H6W, 1K28 –Need to use PQS (http://pqs.ebi.ac.uk) to get trimer imagehttp://pqs.ebi.ac.uk