Functional discoveries from a structural genomics approach

to TBTed Baker

Palmerston North - centre of the Universe

Guy born and grew up here

Palmerston North

Oxford York

Auckland

M. tuberculosis genome sequence

From the whole genome sequence: Mycobacterium tuberculosis

• Approx. 3900 genes

• Most functions assigned by homology (not direct experiment)

• ~ 60% of unknown or uncertain function 15% found only in this organism

• Many metabolic and other pathways appear incomplete or are not yet known at all

• Can we use structure to discover function?

Our programme at University of Auckland

• Began ~ 1999 - pilot project focused on “unknown” proteins from Pyrobaculum aerophilum

- a hyperthermophile

• Switched in 2000 to larger project focused on Mycobacterium tuberculosis – cause of TB

• As part of a larger worldwide network

TB Structural Genomics Consortium

- some central facilities in US, funded by NIH - primarily a focus for collaboration, coordination

Target selection• P. aerophilum:

- Proteins of unknown function, homologous with proteins from M. tuberculosis

• M. tuberculosis:

1. Biosynthetic enzymes - potential drug targets

2. “Unknowns” important in TB biology

eg genes upregulated in response to

- hypoxia

- antibiotics

Progress to date• 120 ORFS cloned

• 106 proteins expressed• 42 soluble• 25 crystallized• 15 structures solved (10 TB, 5 P. aerophilum)

Bottlenecks - soluble expression - crystallization

Not really high throughput

1. Biosynthetic enzymes

LeuA - leucine MenB - menaquinone MshB - mycothiol

MbtI - mycobactinMenG - menaquinoneTrpD - tryptophan

CH3 C C COO-

CH3

H O

CH3 C C

COO-CH3

H OH

CH2 COO-

CoAAcetyl-CoA

αα-isopropylmalate-isopropylmalateαα-ketoisovalerate-ketoisovalerate

αα-isopropylmalate-isopropylmalatesynthasesynthase

• Essential for growth of Mycobacterium tuberculosis - and for survival in lung (inside granulomas)

• No equivalent enzyme in humans

• Good potential drug target

LeuA

Catalyses first committed step in leucine biosynthesis

Nayden Koon, Chris Squire

LeuA structure• Solved by MAD from SeMet protein – refined at 2.0 Å• In the first electron density maps we found:

Bound -KIV substrate moleculeZn ion – verified by fluorescence

Overall structure

Dimer 2 x 70 kDa

N-terminal catalytic domain

C-terminal domain

2 linker domains

Function of C-terminal domain?

• Activity lost when domain removed

• Novel fold – duplicated modules

• Unusual juxtaposition of helix N-termini

• Mutations that abolish feedback inhibition map here

• Hypothesis: Site of leucine inhibition

Leucine feedback inhibition

Now shown

crystallographically

Hinge ?

2. “Unknowns”

Rv2874 (DsbD) – EM domain Rv1347c – acyltransferase

Pa2754 = Rv1720c Pa2307 = Rv3735 Pa1218 = Rv0820

Rv2238c – AhpE

Pa_2754 and its homologues

• Small protein (16 kDa) of unknown function

• Multiple homologues found in certain archaea P. aerophilum: Pa_0151, Pa_0285, Pa_0337, Pa_2754

• and in Mycobacteria M. tuberculosis: Rv0065, Rv0549, Rv0960, Rv1720c

• Classified in Pfam database as PIN domains

• More than 300 other examples… including > 30 examples in the M. tuberculosis genome

• No fold prediction

Vic Arcus, Kristina Backbro, Annette Roos

Structure determination – Pa_2754

• Soluble expression, crystallized easily BUT

• Tetramer in solution, 2-3 tetramers in the crystal – no interpretable heavy atom derivative

• Engineered 2 Met residues (L M mutations)

• Structure solved by MAD phasing at 3.5 Å - found 15/16 Se (SOLVE)

• Autobuilt and refined at 2.8 Å

How to find function?

• Conserved Asp and Glu residues

1. Multiple sequence alignment

Conserved residues in Pa_2754

• Suggest conserved metal binding site

Pa_2754 dimer

How to find function?

• DALI search – a few very weak “hits”, none with Z > 3.0 - new fold?

• Alexei Murzin similar to T4 phage RNase H - folds appear different, but central -sheet and some helices are in common

AND – conserved acidic residues match!

• Hypothesis: protein is a nuclease

• Proved by experiment! Cleaves single-stranded overhangs from double-stranded DNA

• Common function for PIN domains?

2. Search of Protein Data Bank for proteins with similar fold

Why so many PIN domain proteins in M. tuberculosis?

• Many PIN domains now discovered to be the toxincomponent in toxin:antitoxin (TA) pairs- occur in tandem (cassettes) on plasmids and bacterial chromosomes

• Roles in plasmid maintenance, cell cycle arrest,

and dormancy - Science (2003) 301, 1496-1499

• M. tuberculosis has at least 30 of these

toxin:antitoxin pairs (all involving PIN domains)

- role in dormancy?

TA

Acknowledgements

Shaun Lott, Vic Arcus, Kristina Backbro, Heather Baker, Graeme Card, David Goldstone, Anthony Harrison, Jodie Johnston, Nayden Koon, Simon Li, Andrew McCarthy, Neil Peterson, Miriam Sharpe

Marsden Fund of NZ Health Research Council of NZ New Economy Research Fund

TB Structural Genomics Consortium

Guy & Eleanor