Laboratory of Molecular Biophysics
Laboratory Journal 2002
Methods development for protein crystallography
This year we have made progress on our 3 major areas of research, as
detailed in the following 3 sections on A) investigations of radiation damage
in cryocooled crystals, B) structural studies on neuraminidases, and C) elemental
analysis of proteins using a proton microprobe.
Our radiation damage work is supported at the ESRF, Grenoble, by a Long
Term Project (LTP) beamtime allocation held in collaboration with Drs.
Raimond Ravelli (EMBL, Grenoble) and Sean McSweeney (ESRF), and Professor
Martin Caffrey (Ohio State). The LTP amounts to twelve 8 hour shifts every
Since the LTP started, there has been a sharp increase in the profile
of research into radiation damage processes in protein crystals. Many protein
crystallographers are realising that their experiments are being limited
by radiation damage, and there is a keen interest in the topic. Several
factors have contributed to this; not least the burden of experience of
failed MAD structure determinations attributable to radiation damage.
Research efforts to understand damage were fully discussed at the Second
International Workshop on Radiation Damage to Crystalline Biological Samples
held on 1st and 2nd December 2001 at the APS in Chicago. This was organised
jointly by Elspeth Garman, Colin Nave (SRS, Daresbury) and Gerd Rosenbaum
(University of Georgia). James Murray presented our results on scavengers
and unit cell expansion (see next section) at this
Workshop. The availability of carefully collected data on statistically
significant numbers of samples was only possible due to the LTP, which has
given us the beamtime and long term planning potential to carry out such
James Murray also presented a poster on radical scavengers at the 2002
BCA Annual Meeting in Nottingham and won the Biological Structures Group
Poster Prize (the `Blue John' crystal for a year).
As well as the three areas described below, we also undertook some xenon
binding studies on crystals of neuraminidase from Salmonella typhimurium
(STNA) as part of our partner work in EXMAD (Extension of capabilities
for MAD Experiments at synchroton Infrastructures). We are now supplying
STNA purified protein and crystals as a test case to other partners within
EXMAD. We had 2 visitors in connection with this contract: Dr. Maurizio
Polentarutti came from Elettra, Trieste in October 2001 to test his xenon
cell here, and Mr. Ben Hall, a second year Biochemistry undergraduate, did
work experience with us for 2 weeks establishing xenon pressurising conditions
for STNA crystals.
1. Towards an understanding and control of X-ray
Radiation Damage in macromolecular crystals.
James Murray and Elspeth Garman
We have continued our work on radiation damage, investigating both radical
scavengers and possible metrics for radiation damage.
1.1 Metrics for Radiation damage
In order to monitor radiation damage during data collection, it would
be advantageous to have an online metric. Possible candidates include Wilson
B factors, an appropriate merging statistic, the <I>/
(I) in the outer resolution shell and the expansion of the unit cell.
This year we have made a detailed investigation of the expansion of the crystallographic
unit cell with increasing radiation dose. Using beamline ID14-4 at the
ESRF, Grenoble, we monitored this cell expansion on four crystals each
of three different proteins; holoferritin, apoferritin and N9 neuraminidase.
We found that for any individual crystal, the unit cell expansion was approximately
linear with dose, but between crystals of the same protein, the rates of
expansion varied greatly. The results for holoferritin are shown in
Figure 1. We hypothesise that the difference between
crystals may be due to small variations in the cryocooling processes
of the crystals. Whatever the cause, the large variation in the rates of
change of unit cells makes this an unsuitable on-line metric for the monitoring
the extent of radiation damage .
1.2 Free Radical Scavengers
With respect to our study of potential scavengers, by analysis of successive
data sets and resulting electron density maps collected from 3 HEWL crystals
withstyrene and 3 without (matched for size and under as near identical
conditions as possible), we have concluded that styrene has no detectable
effect on lysozyme crystals. We have investigated the effect of ascorbate
on lysozyme (co-crystallisation with 0.5M Sodium Ascorbate) crystals in a
similar way and found it to have a favourable effect on data statistics between
successive datasets, as illustrated in figure 2.
It has been suggested that transition metal compounds could be used
as radical scavengers in crystals. However calculations on the effect of
metal ions on the primary absorption of the crystal suggests that it would
be increased too much to be a beneficial additive, as illustrated in figure 3. We are thus not planning experimental studies
on these scavengers.
In addition to the collection of crystallographic data from crystals,
we have used an offline microspectrophotometer to measure the absorption
spectra of cryocooled crystals before and after exposure to an X-ray beam.
For native lysozyme crystals we have seen an absorption maximum at ~400nm
which we believe corresponds to a disulfide radical species, since this
absorbance maximum is not present in irradiated crystals without disulfide
bonds. The ascorbate containing crystals showed no absorption maximum at
400nm when exposed to X-rays, suggesting that ascorbate is preventing this
radical from accumulating in the crystal, (see figures
4a and b). As already mentioned, lysozyme crystals containing 0.5 M
sodium ascorbate showed better response to incident radiation than native
crystals (see Figure 2). We
thus believe that ascorbate is preventing damage to the crystals both from
crystallographic evidence and spectrographic evidence.
We have secured a Royal Society Equipment Grant to assemble an online
microspectrophotometer on beamline qID14-4 at the ESRF to measure the changes
in absorption in parallel with the collection of crystallographic data,
and this should be installed during 2003.
1. Investigation of free-radical scavengers amd metrics for radiation
damage in protein cryocrystallography. Murray J.W. and Garman E.F. J.
Synch. Radiation (2002) 9, 347-354
2. Bacterial Neuraminidases.
Mutational Investigation of the active site of Salmonella Typhimurium
James Murray and Elspeth Garman in collaboration with Eric Vimr,
University of Illinois at Urbana-Champaign, USA.
Exceptionally well-diffracting crystals of the sialidase from Salmonella
typhimurium LT2 (STNA) can be grown (~0.9Å resolution). Thus it
is possible to undertake ultra-high resolution structure-function studies.
The core active site of the exo-alpha sialidases (EC 126.96.36.199) is very
well conserved. The influenza enzyme has a cavity into which the guanidino
group of the anti-influenza drug Relenza (GG167) binds. The Salmonella
enzyme lacks this cavity and so Relenza will not bind to it. The crystals
of the influenza N9 neuraminidase subtype have been studied to their resolution
limit (~1.4 Å).
We have collected atomic resolution data at SSRL, Stanford from D100S
STNA variant protein with and without the inhibitor DANA bound. Atomic
refinement of these structures with SHELXL is ongoing. Typical electron
density is shown in figure 5. We intend to collect
atomic resolution data from crystals of the set of variants described in
last year's report, to carry out a detailed structural comparison. At SSRL
we also collected data from wild-type STNA with DANA bound and from N6 viral
neuraminidase with Relenza bound.
As part of the EU-funded EXMAD project, we have shown that STNA can
be derivatised by xenon under pressure (see Figure 6).
Our collaborators, Dr. Christoph Krattky and his group in Graz, Austria,
have successfully used STNA crystals supplied by us in their own experiments
on xenon derivatisation.
3. Analysis of proteins by microPIXE (Proton Induced X-ray Emission)
Elspeth Garman in collaboration with Geoff Grime of the Department of
Materials, University of Oxford.
The microPIXE technique  involves bombarding samples of dried liquid
protein or protein crystals with a microbeam (1µm in diameter) of
3 MeV protons to induce X-ray emission from the elements present in the
This year, the proton microprobe has continued to be used for a range
of projects, many involving measurements for colleagues or collaborators
from outside Oxford. This year more liquid than crystalline samples were
A lithium drifted silicon X-ray detector with high energy resolution
enables these characteristic X-rays to be identified as coming from specific
elements. The proton beam is scanned in X and Y, and the X-rays of interest
have software windows set around them. These events are then sorted into
separate 2-D X-Y plots for each element, to build up 2-D `contour maps' of
the individual elements in the sample.
For proteins, the sulphur in the cysteine and methionine residues provides
an X-ray signal which can be used as an internal calibration for the number
atoms of the element of interest per protein molecule. Absolute measurements
are unnecessary and an accuracy of ±6-10% in the elemental
composition for elements heavier than neon can usually be obtained.
The Rutherford Backscattered protons are detected in a silicon surface
barrier detector inside the sample chamber (which is under vacuum) and the
resulting proton spectrum can be fitted  to extract the sample thickness.
This thickness is then used in the analysis of the X-ray spectrum to correct
for self absorption of X-rays in the sample.
Liquid protein samples were analysed for Christoph Mueller and Serge
Cohen (EMBL, Grenoble)[zinc], Ehmke Pohl (EMBL, Hamburg)[iron and zinc],Wolfram
Meyer-Klaucke (EMBL, Hamburg)[zinc for 2 samples], and Randy Reid and Sharon
Miller (Cambridge Institute for Medical Research)[Zn, Ca]. These measurements,
performed for a range of protein sizes, concentrations and sulphur content,
have allowed us to refine our plot of the liquid protein concentration
we require to perform a measurement. This updated plot is shown in Figure 7.
Several of the liquid samples sent to us have contained HEPES buffer
or DTT. These compounds both contain sulphur and thus interfere with the
internal normalisation in calculating the number of atoms of the element
of interest per protein molecule. For instance, there is 0.32mg/ml in 10mM
HEPES, whereas for a 1.6mg/ml 15.5kD protein with 9 sulphur containing amino
acids, the sulphur concentration is 0.031mg/ml. The sulphur signal from the
protein is thus 10 times smaller than that from the HEPES, and quantitative
measurements are not possible. NaCl buffer is also problematic, as
the chlorine in it gives an intense signal next to the sulphur. Although
the peaks are easily resolved, if the chlorine peak is large, it produces
a tail under the sulphur peak giving a large error in the final result.
We have had success in exchanging NaCl for NaBr buffer in these cases.
Crystal samples were analysed for Martin Noble and John Sinclair (LMB)
[P, S], Stephen Cusack and Paul Backe (EMBL, Grenoble) [phosphorous from
putative bound DNA]. This latter was a particularly elegant application of
the technique, as the result in terms of the number of protein molecules bound
to each single stranded DNA oligomer (extracted from the P/S ratio, knowledge
of the protein sequence and the DNA oligomer length) was so unexpected that
the group looked at the molecular replacement self rotation functions anew,
and were eventually able to solve the structure, helped by the microPIXE information.
Enquiries about possible use of the technique were received from
9 other groups round the world, and the University of Leipzig's proton microprobe,
usually used for analysing geological samples, has started to make measurements
on protein samples in collaboration with researchers from EMBL Hamburg.
The Proton Scanning Microprobe in the Department of Materials at Oxford
University, designed, constructed and run by Dr. Geoff Grime, moved with
him to the Ion Beam Centre at the University of Surrey on 1st September
2002. We hope to continue our collaboration with him in his new location.
2. Leaving no element of doubt: analysis of proteins using microPIXE.
Elspeth Garman. Structure (1999) 7, R291-299.
3. The `Q factor' method: quantitative micoPIXE analysis using RBS normalisation.
G.W. Grime Nucl. Inst. Meth. in Phys. Research B (1996) 109/110,
4. Architecture of a protein central to iron homeostatis: crystal structure
and spectroscopic analysis of the ferric uptake regulator. Ehmke Phl, Jon
C. Haller, Ana Mijovilovich, Wolfram Meyer-Klaucke, Elspeth Garman and Michael
Vasil. Molecular Microbiology (2003), 47 in press.
Last updated: 15-MAY-2003 15:09