Ted Baker School of Biological Sciences University of Auckland New Zealand

1. New Drug Targets from Mycobacterium tuberculosis:Strategies, Progress and Pitfalls from a Structural Genomics Enterprise Ted Baker School of Biological Sciences University of Auckland New Zealand On behalf of TB Structural Genomics Consortium

2. The challenge posed by complete genome sequences

3. The Mycobacterium tuberculosis genome • Approx. 3900 open reading frames (ORFs) • ~60% of gene products have an inferred function (mostly by homology) • ~25% are “conserved hypotheticals” • ~15% are “unknowns” • ~30% can be related to proteins of known 3D structure - but only ~25 TB protein structures • Many metabolic pathways appear incomplete

4. Function from structure? Relationships that are hidden at the sequence level SpeB – virulence factor from S. pyogenes Actinidin – plant cysteine protease - < 10% sequence identity

5. Structural Genomics • The use of genomic information to guide protein structure discovery - and its inverse • The use of protein structure analysis to add value to genomic sequence data – to deduce function - • Reversal of the ‘traditional’ direction of structural analysis • Many targets – whole genomes, pathways, functional classes, folds

6. Beginnings…~1998A pilot pilot programme – Pyrobaculum aerophilum • Using laboratory-scale approaches - PCR cloning - Expression in E. coli, cleavable affinity tags - Variation of expression temperature - Purification by affinity chromatography and gel filtration • Genomic approach – most tractable first

7. Results – P. aerophilum • Cloned 25 (274) • Expressed 20 (168) • Soluble 12 (80) • Purified 12 (43) • Crystallized 6 (24) • Structures 4 (11) Main bottlenecks - solubility - crystallization

8. HisF (imidazoleglycerol phosphate synthase) Banfield et al.Acta Cryst. D (2001) Pa_989 (TB homologue)

9. Pa_2307 (unknown) • ‘Ancient conserved domain’ found in bacteria and archaea. No functional annotation • Reproducible crystals with Li2SO4 - but twinned • Two crystals grown from PEG/phosphate • 1.5 A native data from one, SAD data from Pt(NO2)4 deriv of the other (used gel shift) • Structure solved: SAD/Solve/Resolve/ARP

10. Pa_2307

12. The next phase – larger enterprises • Publicly funded - NIH Protein Structure Initiative (USA) - Initiatives in Japan, Germany, UK, France, Canada • Biotech companies - Structural Genomix, Syrrx

13. NIH Protein Structure Initiative • 10 groups (consortia) funded • Aim to develop methods and tools for “high throughput” structure determination • Goals primarily structural - representative structures for all protein sequence families - discover novel folds (cover “fold space”) - estimate 10,000 structures needed But evolving

14. Mycobacterium tuberculosis Causative agent of TB One-third of world’s population affected - approximately 3 million deaths annually Five front-line drugs (isoniazid, pyrazinamide, ethambutol, rifampin, streptomycin) but… - effective only against actively-growing bacteria - very long treatment regime (6-9 months) - resistance rising - need for new drugs

15. Peculiarities of the organism • Very slow-growing Gram-positive organism • Complex waxy cell wall – outer layer rich in unusual lipids, glycolipids, polysaccharides • Novel biosynthetic pathways • Complex lifestyle - persistence - enters dormant state within active macrophages - survives through switches in metabolism - can be reactivated years later

16. Led in United States by: - Tom Terwilliger (Los Alamos NL) - David Eisenberg (UCLA) - Jim Sacchettini (Texas A&M) - Bill Jacobs (Albert Einstein Coll. of Med.) - Tom Alber (UC Berkeley)….. and many others • Aims are focused on function: - understanding TB biology - discovery and structural analysis of novel drug targets http://www.doe-mbi.ucla.edu/TB/

17. Philosophy and policies • Open participation - to all with an interest in TB • Operates as a wider consortium of >30 participating labs in 13 countries worldwide • Collaboration between structural biologists TB biologists, chemists…. • Commitment to common policies - collaboration and cooperation - shared database for logging progress - sharing of data and materials - structures to be placed in public domain

18. Operational aspects • Central facilities for - bioinformatic analysis and data storage - protein expression and evolution - crystallization - synchrotron data collection - gene knockouts • Technologies and facilities available to all • Individuals choose their own targets according to their own interests – and assign priorities • Targeting scores determine priorities of facilities • Parallel efforts in individual labs

19. Progress to date • Most of structural results to date come as a result of efforts in individual labs • But - availability of high-throughput facilities gives flexible options for individual labs and for efforts in the facilities • Within facilities – 688 genes cloned (out of 720 targeted to date) • First phase – concentrate on soluble proteins • Next phase – the insoluble proteins

20. Folding Reporter - GFP • Function of R (GFP) depends on solubility of X-L-R. • Solubility of X-L-R depends on X. Express fusion protein X-L-R L N C X R Non-functional R Detect function R Insoluble Soluble Dealing with insoluble proteins GFP fusions as reporter of solubility– G. Waldo

21. Cell Colonies X-L-GFP FUSION FLUORESCENCE In Vitro Transcription + Translation Soluble Fraction SDS-PAGE X (Non-Fusion) Pellet Fraction

22. Insoluble Protein Mutate Gene FORWARD EVOLUTION Recombine Optima Clone Select Recombine Optima & Wild type Clone BACKCROSSING Select Soluble Protein Using GFP-fusions to engineer proteins for solubility G.Waldo

23. Solubilisation by evolutionRv2002 – Se Won Suh • Putative ketoacyl ACP reductase • Rendered soluble by 3 random mutations • I6T and T69K mutations are on the molecular surface • V47M mutation enhances a semi- exposed hydrophobic contact

24. Potential new TB drug targets Early results from the TB Structural Genomics Consortium

25. Target ORF Selection in Mycobacterium tuberculosis • Selection of ORFs: (a) potential drug targets and (b) to understand TB biology • Biosynthetic enzymes for essential amino acids, cofactors, lipids, polysaccharides • Secreted proteins • Proteins implicated in antibiotic resistance or response • Proteins implicated in persistence

26. Cell wall biosynthesis- mycolic acids (Sacchettini lab) • Long chain branched lipids - form dense waxy outer layer of the mycobacterial cell wall • Contribute to its impenetrability • Implicated in both virulence and persistence • Either covalently attached to cell wall or released as trehalose dimycolate (“cord factor”) • Modification of mycolic acids, eg. cyclopropanation – varies between pathogenic and non-pathogenic species

27. Cyclopropanation of mycolic acid chains • Cyclopropane groups introduced by methylation

28. Three cyclopropane synthases(C. Smith, J. Sacchettini – Texas A&M) CmaA1 CmaA2 PcaA

29. Secreted proteins(Eisenberg lab) Secreted proteins attractive drug targets for M. tuberculosis because: • Often determinants of virulence or persistence - involved in cell wall modification - role in survival in macrophages • M. tuberculosis secretes large number of proteins • Cell wall is impermeable to many anti- bacterial agents

30. C N Secreted proteins(C. Goulding, D. Anderson, H. Gill, D. Eisenberg – UCLA) Rv1886c Antigen 85B Mycolyl transferase Rv2220 Glutamine synthetase - Synthesis of poly-(L-Glu-L-Gln) for cell wall Rv1926c Unknown, resembles cell surface binding proteins (invasin, adaptin, arrestin)

31. 3. Targets against persistence(Sacchettini lab) • Persistence within activated macrophages facilitated by switch in metabolism • Glycolysis downregulated – instead glyoxalate shunt allows use of C2 substrates generated by b-oxidation of fatty acids • Enzymes isocitrate lyase and malate synthase are drug targets for persistent bacteria

32. Glyoxalate shunt enzymes(V. Sharma, J. Sacchettini - Texas A&M) Rv0867 Isocitrate lyase Rv1837c Malate synthase

33. DNA microarray analysis of TB ORFs upregulated by exposure to isoniazid Some code for proteins ofknown function – cell wall biosynthesis Others represent ‘unknowns’ The proteins encoded bythese ORFs may represent the bacterial response to thetoxic effects of the antibiotic 4. Antibiotic resistance- Isoniazid response genes Wilson et al.,PNAS96:12833-12838 (1999)

34. Rv0340 Rv0341 Rv0342 Rv0343 Putative INH response operon • Four ORFs appear to make up part of a putative operon in the TB genome: Rv0340, Rv0341, Rv0342, Rv0343. • None of the four ORFs have detectable sequence homologues in other organisms. • Rv0340 and Rv0341 are paralogues, as are Rv0342 and Rv0343 • Same genes also upregulated by ethambutol.

35. Isoniazid response – Rv0340Moyra Komen, Vic Arcus, Shaun Lott • Crystallization attempts • NMR – shows only partially folded • Limited proteolysis – gives N-terminal fragment with excellent NMR spectrum Oil Spherulites

36. NMR spectrum – Rv0340(residues 1-131) • Indicates helical bundle with flexible tail • Possible homology with acyl carrier protein • Gives putative role in cell wall biosynthesis

37. Problems of partial or incorrect functional annotation • Widespread in bacteria, but not eukaryotes • No clearly indicated function - closest sequence homologs: malonyl CoA decarboxylase siderophore biosynthesis aminoglycoside acetyltransferase • No structure prediction Rv1347c

38. Rv1347c structure - Graeme Card Rv1347cAcetyl-CoA dependent aminoglycoside acetyltransferase (11% identity) Aminoglycoside N-acetyl transferase (GCN5 family) ~ 11% sequence identity Rv1347c

39. Problem of partial or incorrect functional annotations • Putative SAM-dependent methyltransferase catalysing final step in menaquinone biosynthesis • Potential drug target – menaquinone pathway is essential and is not present in humans • Genome also includes ubiE (Rv0558) - catalyses this step in both menaquinone and ubiquinone biosynthesis (menG is specific for menaquinone) • Expressed, refolded, crystallized, solved to 1.9Å by SIRAS Rv3853 - “menG”

40. Common methyltransferase fold

41. MenG structure – Jodie Johnston • Structure does not look like a methyltransferase • Resembles a phosphate transfer domain? • Incorrect annotation

42. Challenges for the future • Membrane proteins • Solubility of expressed proteins • Hetero-oligomeric proteins • Protein-protein interactions • Assignment of function to “unknowns” • Cellular pathways - metabolic pathways - signalling pathways

43. Conclusions • Structural biology is being transformed by new technologies – some driven by genomics • Less effort in solving initial structures – more emphasis on “downstream” studies • TB structural genomics consortium – a different model for large scale structure determination - access to centralised facilities - international effort on a common goal - collaboration rather than competition - opportunities for smaller labs

44. Thanks • Mycobacterium tuberculosis structural genomics consortium • Members of Auckland Structural Biology Laboratory – Vic Arcus, Kristina Backbro, Mark Banfield, Heather Baker, Graeme Card, Jodie Johnston, Rainer Knijff, Moyra Komen, Shaun Lott, Andrew McCarthy, Clyde Smith • Marsden Fund Health Research Council New Economy Research Fund