Achievements and Future Carear Plans:

Solution Scattering
at
ID14-3
Automated Data Collection
and Processing
Adam Round
Contents
ƒ Crystals versus solutions
ƒ What can Solution SAXS tell us
ƒ Solution SAXS data collection
ƒ Automation
ƒ Data Collection
ƒ Automated sample loading robot
ƒ Processing and Analysis
ƒ Integrated data processing pipeline
Crystals versus solution
B
=
X
A
Protein
molecule
Lattice
Fourier
Transformation
Crystal
3-dim information about:
Length of the vector AB
Orientation of vector AB
3-dim protein
structure with
atomic resolution
Scattering
function
I(s) [a.u.]
B
=
X
A
Pair distance
distribution function
p(r)
1
Fourier
Transformation
Radial integration
0.1
2Θ
0.01
0.001
Protein
molecule
Solution
Random
0.05
0.10
0.15
0.20
s [Å]
0
2
4
6
8
10
12
14
16
r [nm]
Maximal dimension of the particle
Mean value
Radius of gyration
Crystals Structure Validation
Based on existing high resolution protein structures from crystallography or NMR
the corresponding SAXS profile can be calculated.
I(s) = A(s)
2
Ω
= A a (s) − ρ s A s (s) + δρ b A b (s)
2
Ω
This allows one to verify high resolution structures
especially in the case of multimeric protein and
larger protein complexes.
Example: The muscle protein titin was found in
different conformations in the crystal.
CRYSOL (X-rays):
Svergun et al. (1995). J. Appl. Cryst. 28, 768
CRYSON (neutrons): Svergun et al. (1998) P.N.A.S. USA, 95, 2267
Refinement of Rigid Domains
Rigid Body Refinement: Moving protein sub-parts (called domains) as
rigid bodies to fit the scattering data
Example: Structural changes upon ligand binding
PX-structure with
ligand
Ligand
SAXS Shape obtained by
GASBOR - unliganded state
Rigid Body refinement
using MASSHA
Adding Missing Linkers
Remodeling of proteins from high resolution fragments/constructs
Program BUNCH (M. Petoukhov; Biophysical J.)
Linkers?!?
A
B
Domains A+B
Domains B+C
Entire Protein (ABC)
C
High resolution protein fragments
from X-ray crystallography
+ sequence data
(TrEMBL/Swissprot)
SAXS Data
of constructs
AB
BC
ABC
Model for the entire protein
including not resolved
linker components
Ab-initio Modelling
Vector of model parameters:
A sphere with diameter Dmax is filled by densely
packed beads of radius r0<< Dmax. A configuration
vector X indicates whether the j-th atom belongs
to the particle or to the solvent.
Solvent
Particle
Position ( j ) = x( j ) =
(phase assignments)
⎧⎪1
⎨
⎪⎩0
if particle
if solvent
The number of model parameters
M ≈ (Dmax / r0)3 ≈ 103 is too large for
conventional minimization methods.
A Monte-Carlo type search starting from a
random X can be employed to find a
configuration that yields the calculated
scattering curve fitting the experimental
data
Chacón, P. et al. (1998) Biophys. J. 74, 27602775.
2r
Dmax
0
Svergun, D.I. (1999) Biophys. J. 76, 2879-2886
Ab-initio Modelling
DAMMIN modelling penalties
Using simulated annealing, finds a compact dummy atoms configuration X that fits
the scattering data by minimizing
f ( X ) = χ 2 [ I exp ( s ), I ( s, X )] + αP ( X )
where χ is the discrepancy between the experimental and calculated curves, P(X)
is the penalty to ensure compactness and connectivity, α>0 its weight.
compact
loose
disconnected
DAMMIN in Action
Ab-initio Can it be Trusted?
Ab initio bead models compared to high
resolution X-ray structures
Solution scattering data collection
Log I(s)
sample – buffer
sample
(subtracted)
buffer
s, nm-1
X-ray Beam
X-ray Scattering
Sample
X-ray detector
Data collection = REPETITION!
ƒ Enter Hutch
ƒ Clean cell (there can be no cross contamination)
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
Rinse with water
Flush with cleaning solution
Rinse with water
Dry
Load new solution (making sure cell filled properly)
Leave / Interlock hutch
Measure
ƒ Repeat for n samples with buffers at multiple concentrations
ƒ
ƒ
ƒ
ƒ
2n+1 (if all samples are in the same buffer)
3n if different buffers
one individual construct at 3 concentrations = 7 measurements
Data collections at X-33 in Hamburg usually between 100 to 200
measurements per day
ƒ At ID14-3 we hope for more
Automation
ƒ Implementation of automated data collection
ƒ Design to be based on the first generation
prototype developed at EMBL-Hamburg in
collaboration with the Fraunhofer Institute of
Manufacturing Engineering and Automation
ƒ To progress in stages
ƒ Manual operation
ƒ Automated cleaning
ƒ Pump to load samples and buffers into correct position
ƒ XYZ stage for loading of multiple solutions
ƒ In observed mode
ƒ Fully automated unattended data collection
Prototype in use at X33
(Image courtesy of Fraunhofer Institute Stuttgart)
(Image courtesy of D. Svergun )
Current Status of the System
ƒ In user operation since September 2007
ƒ Since then over 4500 SAXS images have been collected at X-33
ƒ Which means over 2000 protein samples of various concentrations
have been loaded by the automatic sample changer for measurement
by over 50 separate international user groups
ƒ Tested failure rate using an automatic loading script was less
than 0.5 %
ƒ Operational efficiency has been greatly increased
ƒ loading is more reliable
ƒ Time not spent changing the samples can be used for further sample
preparation or data analysis
ƒ Data reliability improved as cross contamination reduced
Automated Data Processing
Pipeline
Data
AUTOSUB
AUTOPilatus
Automatic data
transformation of 2D raw
data into 1D scattering curves
Automatic samplebuffer subtraction
with automatic
estimation of radius of
gyration (AUTORG)
AUTOGNOM
Automatic estimation of
distance distribution
function
Results:
Verification there are no radiation or
concentration effects in data
Sample Size: Rg, Mw, Dmax and excluded
volume
Oligomeric state
Low Resolution particle shape
AUTODAMMIF
Automatically start the abinitio shape determination
program and fitting with
simple geometrical bodies
Summary
What can we learn from solution SAXS
ƒ Complete high resolution structure known: validation of crystal structure in
solution under physiological conditions
ƒ Ligand binding reactions: Internal structure of multicomponent particles
and large macromolecular complexes
ƒ High resolution structure of domains/subunits known: quaternary structure
using docking/rigid body refinement
ƒ Incomplete high resolution structure known: probable configuration of
missing portions
ƒ Nothing known: ab initio low resolution structure
Summary
ƒ Aim is to provide a world leading facility to compliment
the current research at the ESRF, ILL and EMBL
ƒ Optimisation of ID14-3 and implementation of Sample
environment first priority
ƒ First users expected in Autumn 2008
ƒ Manual data collection with partial automated analysis
ƒ Development of automated systems in collaboration
with EMBL-Hamburg
ƒ Sample loading robot to be developed in stages to enable
thorough testing
ƒ Data analysis software to be implemented for user operation
after validation by developers at EMBL-Hamburg
ƒ Beta testing and feedback to aid in future developments
Acknowledgments
ƒ EMBL-Hamburg
SAXS Group
ƒ D. Svergun
ƒ M. Roessle
ƒ M. Petoukhov
ƒ D. Franke
ƒ A. Kikhney
ƒ P. Konarev
Instrumentation Group
ƒ R. Klaring
ƒ U. Ristau
ƒ EMBL-Grenoble
ƒ ESRF MX Group
ƒ Fraunhofer Institute of
Manufacturing Engineering
and Automation
ƒ
ƒ
ƒ
S. Moritz
R. Huchler
D. Malhtan
EU FP6 Design study SAXIER Grant number RIDS 011934
Contact
Until June 2008
around@embl-hamburg.de