Transcript Shotgun crystallization of the T. maritima proteome

Shotgun crystallization of the
Thermotoga maritima proteome
Protein properties and crystallization
conditions that correlate with
crystallization success
Rebecca Page
The Scripps Research Institute
3.30.2004 – PSI, NIH
Data mining for faster structure
determination
Crystallization
Conditions
Protein
Properties
Data mining for faster structure
determination
Crystallization
Conditions
Protein
Properties
Data mining for faster structure
determination
Crystallization
Conditions
Protein
Properties
Data mining for faster structure
determination
• Minimize
initial
crystallization
screens
Crystallization
Conditions
Protein
Properties
Data mining for faster structure
determination
• Minimize
initial
crystallization
screens
• Improve
target
selection
Crystallization
Conditions
Protein
Properties
Experimental design
• Process all T. maritima
proteins through the JCSG
structure determination
pipeline
• Targets are not prefiltered
Thermotoga maritima
1877 ORFs
• Targets are processed
using identical
experimental methods
Lesley, et al. (PNAS, 2002)
Experimental design
A more complete, less
biased crystallization
dataset for data mining
Thermotoga maritima
1877 ORFs
Lesley, et al. (PNAS, 2002)
The Numbers
Targets
1791
1376
539
539
• 258720 crystallization
experiments
• 465 of 539 (86%)
proteins crystallized
• 472 of 480 (98%)
conditions produced
crystals
• 5546 total crystal hits
Data mining crystallization
conditions
Minimize initial crystallization screens
Data mining crystallization
conditions
Minimize initial crystallization screens
Many proteins crystallized in 5 or more
of the original 480 conditions
51 or
21; 3.9%
more
32; 5.9%
26 to 50
19; 3.5%
073; 13.5%
21 to 25
24; 4.5%
16 to 20
47; 8.7%
11 to 15
73; 13.5%
6 to 10
249; 46.2%
1 to
5
00
11-5to 5
66-10
to 10
11-15
11
to 15
16-20
16
to 20
21-25
21
to 25
26-50
26
to 50
51 or
51
ormore
more
Identify minimal crystallization screens
MINCOV
Iterative selection algorithm that
identifies minimal screens, subsets of
the original 480 conditions that would
have crystallized all 465 proteins
Repeat 472 times (each condition)
• 472 minimal
screens
• Each
contained 108116 conditions
• Intersection =
Core Screen
Slawomir Grzechnik
Core Screen
Best 96 conditions crystallize 448 proteins
Core Screen
180
180
Screen
OriginalCore
Screen
All Conditions
Core Screen
160
140
140
120
• 67 conditions (14%)
• All precipitants
• 392 proteins crystallized (84%)
100
100
80
60
60
Expanded Core Screen
40
20
20
• 96 conditions (20%)
0
High MW
Low MW
High
MW Low
MW
PEG
PEG
PEG
PEG
Salts
Salts
Polyalcohols
Poly-
alcohols
Organics
Organics
• 448 proteins crystallized (96%)
Page, et al. (Acta Cryst D, 2003)
Data mining protein properties
Improve target selection
Data mining protein properties
Improve target selection
Better target selection for JCSG pipeline
Frequency
20
Gravy Index
- hydrophilic
+ hydrophobic
15
10
5
-1.0
0.0
Gravy Index
1.0
Identify upper and
lower bounds of
crystallized proteins
and use these limits
in future target
selection
Proteins with 40 or more SEG
residues rarely crystallize
Low-complexity segments
TPPTMPPPPTT
GGGSSSSHS
PNGLPHPTPPPP
QQQGRQQQQQLK
• SEG: Filtering to
identify low complexity
segments
• Long SEG segments
can be unstructured
Proteins with 40 or more SEG
residues rarely crystallize
• SEG: Filtering to
identify low complexity
segments
• Long SEG segments
can be unstructured
% crystallized
30
20
10
0
0
20
40
60
80
100
Number of SEG residues
New target selection
Characteristic
Proteins
Eliminated
Crystals
Eliminated
Protein
Total
(1877)
Crystal
Total
(465)
120
1
1757
464
188
0
1602
464
204
0
1562
464
pI
57
0
1538
464
TMHMM;
SignalP
538
15
1245
448
Coiled-Coil
72
4
1213
445
SEG
144
6
1187
Length
Charged AA
Gravy
63%
439
94%
Goal: more structures!
Crystallization
Conditions
Protein
Properties
Goal: more structures!
Crystallization
Conditions
Protein
Properties
Goal: more structures!
Crystallization
Conditions
Protein
Properties
Goal: more structures!
Crystallization
Conditions
Protein
Properties
GNF / TSRI - CC
Ray Stevens
Scott Lesley
Rebecca Page
Carina Grittini
Jeff Velasquez
Kin Moy
Eric Sims
Bernard Collins
Tom Clayton
Angela Walker
Heath Klock
Eric Koesema
Eric Hampton
Jamison Campbell
Mike Hornsby
Tanya Biorac
Dan McMullan
Kevin Rodrigues
Mike DiDonato
Andreas Kreusch
Glen Spraggon
Marianne Patch
Xiaoping Dai
Terry Cross
Kevin Rodrigues
Polat Abdubek
Eileen Ambing
SSRL - SDC
Keith Hodgson
Ashley Deacon
Mitchell Miller
Henry van den Bedem
Guenter Wolf
S. Michael Soltis
R. Paul Phizackerley
Irimpan Mathews
Qingping Xu
Amanda Prado
John Kovarik
Hsiu-Ju Chiu
Ross Floyd
Inna Levin
Ronald Reyes
Fred Rezazadeh
UCSD - BIC
TSRI - AC
John Wooley
Ian Wilson
Adam Godzik
Peter Kuhn
Susan Taylor
Marc Elsliger
Slawomir Grzechnik
Frank von Delft
Bill West
Vandana Sridhar
Andrew Morse
Dan Taillac
Jie Quyang
Xianhong Wang
Jaume Canaves
Lukasz Jaroszewski
Robert Schwarzenbacher
Ray Bean, Josie Alaoen
Exploratory Projects
Kurt Wüthrich, TSRI
Linda Columbus
Touraj Etezady
Margaret Johnson
Wolfgang Peti
Virgil Wood, UCSD
Phillip Bourne
Barbara Cottrell
Raymond Deems
Jack Kim
Dennis Pantazatos
Geoffrey Chang, TSRI