X-ray Diffraction at
Synchrotron Light Sources
John R Helliwell, University of Manchester, Manchester, UK
The elucidation of biological structure and function for macromolecules and their
complexes is undertaken extensively using X-ray diffraction. Synchrotron X-radiation is
widely used, due to its exceptional X-ray brilliance and tunability properties, in protein
crystallography, fibre diffraction and solution scattering.
Historical Context and Introduction
Synchrotron radiation (SR) is emitted when charged
particles, usually electrons but also positrons, travel at
speeds close to the speed of light, and are held in circular
orbits by a magnetic field. These machines, accelerators,
were originally developed for particle physics experiments
where the charged particles struck a target or collided with
counter-rotating oppositely charged particles. The emission
of synchrotron light, in a continuum from X-rays to
infrared wavelengths, was a nuisance by-product, an
energy loss that consumed electricity (as radio frequency
power), as far as particle physics was concerned. However,
it was clear from calculations by J. Schwinger in the late
1940s and later by L. Parratt, as well as experimentally,
that this was a potentially very bright, very versatile X-ray
source for the study of the structure of matter by
diffraction, as well as by spectroscopic and imaging
methods. By comparison, the ‘standard’ X-ray tube,
invented by W. Roentgen in 1895, emits rather weak
specific wavelengths(Ka, Kblines), albeitsitting ona white
background of X-rays, but which is exceedingly weak
compared with white SR.
There is a long history to X-ray crystallography, which
commenced in 1912, before SR was first observed in 1948
and the first SR diffraction experiment was performed by
Y. Cauchois in Frascati, Italy in 1963 and the first
biological diffraction from muscle fibres by G. Rosenbaum,
K. C. Holmes and J. Witz (published in 1971 in
Nature) recorded in Hamburg, Germany. The first SR
protein crystal diffraction patterns were recorded at the
Stanford Linear Accelerator Center’s (SLAC) electron
storage ring ‘SPEAR’ by J. C. Phillips and K. O. Hodgson
in 1976. The first dedicated SR X-ray source was the
Synchrotron Radiation Source (‘SRS’) Daresbury, UK
(1981), followed by various others around the world, and
from these ‘second generation’ of SR sources came a steady
flow of innovations in methods. Gradually, large numbers
of protein structure determinations occurred. The study of
noncrystalline samples such as muscle fibres as well as
solutions also progressed well on these sources.
A third generation of SR sources has come on-line in the
1990s. These are much larger than most of the secondgeneration
sources and thus allow ‘insertion device’
periodic magnet sources (wigglers and undulators) to be
incorporated in the ring (Figure 1), and are of exceptional
brilliance. Since biological molecular samples are usually
small (5 0.5 mm, often 50 mm or even smaller) the
brilliance property is very important. Thus micro-samples
of all types, crystals, fibres and solutions, have all become
amenable to study. In another application the intensities
are now such that the electron bunch structure of the
circulating particle beam can be exploited. Even single
bunch sub-nanosecond time-resolved structural studies are
possible on protein crystals. These experiments involve
some perturbation or stimulation of the sample followed,
after a preset time gap, by the recording of the X-ray
diffraction patterns. These polychromatic patterns
(Figure 2) are known as Laue patterns as they involve a
stationary crystal, as in the original 1912 X-ray tube-based
experiment. The largest single category of SR instruments
in theX-ray region comprises thoseused for data collection
on crystals of biological macromolecules or larger biological
macromolecular complexes (Figure 3).
A general consideration for X-ray diffraction structural
studies is that of radiation damageto the sample, whether it
is a crystal, fibre or a solution of biological macromolecules.
The basic effect, and problem, is that X-rays are
absorbed and the energy deposited is converted into
reactive free radicals which can diffuse around the sample
and disrupt the weak molecular bonds that exist between
the molecules and thereby cause a deterioration of the state
of sample order. For a protein crystal, for example, the
resolution deteriorates. In the most brilliant SR X-ray
beams the energy deposited is also enough to heat the
sample by several degrees per second (although the exact
temperature is difficult to measure or calculate exactly).
Several approaches have helped to get around these
Article Contents
Introductory article
. Historical Context and Introduction
. Structure Elucidation of Large Biomolecules at or Near
to Atomic Resolution
. Studies of Crystals with Large Unit Cell Dimensions
. Multiple Wavelength Anomalous Dispersion (MAD)
Technique
. Time-resolved Structural Analysis and Polychromatic
X-ray Diffraction
. Noncrystalline Diffraction
. Other Probes of Structure; Neutron and Electron
Diffraction and NMR Spectroscopy
. An Overview of Achievements of SR Diffraction in the
Life Sciences
1ENCYCLOPEDIA OF LIFE SCIENCES © 2001, John Wiley & Sons, Ltd. www.els.net
chemically or thermally damaging effects. The use of short
wavelengths reduces the absorption and since it varies as
wavelength cubed versus the diffraction signal varying as
wavelength squared there is a benefit of the short
wavelength. The faster exposure time at SR sources is also
valuable since the diffusion time of the free radicals is
significant. The most effective approach of all is to cool the
sample to cryo temperature (100 K), thus preventing the
diffusion of free radicals altogether. However, the sample
lifetime is not infinite. The deposited energy will break
bondsand eventually cause physicalbreakup of thesample
order. While a big benefit, there are subtle structural
changes upon cryo-cooling which increase the number of
multiple conformations of amino acid side-chains and
affect the vibrational signatures of atoms. Neutron
spectroscopy techniques show that there is a glass-like
transition at around 180 K and a loss of protein function
occurs below such a temperature. Naturally, there is a
continued interest in room temperature measurements.
R. F. cavity
Injection magnet
Quadrupole magnets
Dipole magnet
Insertion devices
Synchroton radiation
Beam from injector
Figure 1 A schematic diagram of the layout of a synchrotron radiation (SR) source. In practice, SR storage rings are much larger, having many more
machine‘cells’thanthe fourshownhere.AttheEuropeanSR Facility(ESRF)inGrenoble,forexample,thethird-generationhigh-brilliancesourceis840min
circumference and has 32 cells.
Figure 2 A polychromatic Laue diffraction pattern from a protein crystal
with spots colour-coded for wavelength. Short wavelengths in blue and
then through the rainbow colours to red at the longest wavelengths. Actual
wavelength bandpass used 0.5 A˚ to 2.5 A˚. Reproduced with the permission
of Acta Crystallographica.
X-ray Diffraction at Synchrotron Light Sources
2
Structure Elucidation of Large
Biomolecules at or Near to Atomic
Resolution
Use is made of the high intensity and collimation of SR for
recording the weak, high-angle outer parts of the diffraction
pattern which allows a high-resolution protein crystal
structure refined model to be achieved.
The resolution achievable at the synchrotron is now
often reaching ‘atomic resolution’ (i.e. where the electron
density images of the protein crystal unit cell resolve
individual atoms). Indeed the data available now can reach
a sufficient number to allow molecular model refinement
where the individual motions of atoms are described by a
full anisotropic model (ellipsoids) and single versus double
bonds are clearly resolved as is done routinely in chemical
(smaller molecule) structure refinement.
The technical advance of cryo-cooling the protein
crystal sample (typically at 100 K) to prevent X-ray
irradiation damage has also widened the number of such
studies. This method was introduced in the study of
ribosome crystals, pioneered by A. Yonath and H. Hope
and co-workers, and is now used for almost all protein
crystal samples especially at high-brilliance third-generation
SR sources. At such sources beam damage even of
cryo-cooled crystals is observed on the insertion device
beamlines.
Studies of Crystals with Large Unit Cell
Dimensions
Large macromolecular complexes, i.e. of proteins with
proteins or proteins with nucleic acids, in cases such as
spherical or cylindrical viruses or the ribosome, have been
crystallized and their structures determined. These are at
resolutions typically of 3 A˚ , occasionally 2.5 A˚ . Nevertheless,
for the size of these structures these are major
achievements of structural biological crystallography. The
first animal virus determined was the common cold virus
(rhinovirus) by M. G. Rossmann using the Cornell
‘CHESS’ Synchrotron, USA in 1985, with preliminary
work undertaken at Daresbury, UK and Hamburg,
Germany. Figure 4 shows one of the diffraction patterns
measured from a rhinovirus crystal at CHESS. Approximately
160 different crystals, each yielding one such
pattern, were used in the analysis. The crystal structure
of the large subunit of the ribosome has been determined
by P. D. Moore and T. Steitz and colleagues at Yale
University in 2000 using SR data from the National
Synchrotron Light Source at Brookhaven National
Laboratory, USA. A variety of ribosomal subunit
structural studies are now being reported in the literature.
The ribosome is now known to be a ribozyme.
Figure 3 (a) Electron density map image of a protein crystal. (b) Ribbon
diagram of the same protein crystal to the same scale and the same view.
The protein is the tetramer of concanavalin A from jack beans with
glucoside in atom-by-atom format and metal atoms also shown as
individual spheres. Concanavalin A–sugar complexes are typical of the
proteins studied to high resolution at a second-generation SR source (in this
case DaresburySRS), i.e.up to 100 kDain theasymmetricunit ofthe crystal.
X-ray Diffraction at Synchrotron Light Sources
3
These crystals make optimal use of the high intensity of
the SR beam as they are weakly diffracting, as one can
readily imagine because there are rather few unit cells in the
crystal (e.g. 108
rather than 1014
). Moreover the diffraction
spots are very close together in angle and so the fine
collimation (small divergence) of the SR beam is needed.
SR diffraction pattern recording for crystal unit cells in
excess of 1000 A˚ edge length is now quite feasible.
Multiple Wavelength Anomalous
Dispersion (MAD) Technique
The tunability of synchrotron radiation has made this
method of structure elucidation feasible, and it is growing
in importance. The core problem in solving a protein
crystal structure, which this method addresses in a direct
way, is the determination of the phase angle of each and
every X-ray reflected beam from a protein crystal. From
the 1950s through to the 1980s the method used involved
heavy atom soaking of protein crystals with measurements
being made from each crystal type, using a method called
multiple isomorphous replacement (MIR). However,
instead, it is possible to utilize the wavelength-dependent
changes (‘anomalous dispersion’) to the X-ray scattering of
metal atoms that are in a protein naturally or are
introduced via derivatization or, more frequently today,
by deliberately producing selenomethionine-substituted
proteins. Exploitation of these effects (at maximum
approximately only 10–15% of the wavelength-independent
scattering of an atom) required the improvements
available from SR X-ray intensity data collection to
reliably measure the effects, which are typically at the 1–
2% change in intensity level. The technical requirements of
the SR source and beamline necessitated rather exacting
developments of source and beamline optics stability, i.e.
exact wavelength and wavelength-band setting, as well as
improved diffraction detectors (i.e. image plates and
charge coupled device detectors). Today the MAD method
is allowing a much higher throughput of protein crystal
structure determinations since the recording of the data,
the determination of the positions of the metal sites and the
calculation of an interpretable electron density image is
achievable on a third-generation source undulator beamline
in hours. The bottlenecks that remain are highthroughput
expression and purification of pure protein as
well as high-throughput crystallization. A new field of
‘structural genomics’ has thus opened up.
Time-resolved Structural Analysis and
Polychromatic X-ray Diffraction
The purpose of structure elucidation is to understand the
biological function. A direct way to probe function is to
stimulate the functional cycle of a molecule or molecular
biological system and then to probe the structural
intermediate state in real time using X-ray diffraction
measurements at given time points after the stimulation.
This is a relatively new area of SR diffraction research and
development, and to date only 10 successful biochemical
crystallographic real-time studies have been published.
The methods of stimulation include diffusion of substrate
(for enzymes), pH jump, light flash or temperature jump.
The most prompt method of stimulation is light flash and is
applicable to naturally light-sensitive proteins (e.g. bacteriorhodopsin,
phospho yellow protein, carbon monoxide
myoglobin to name three cases that have been studied this
way). Also amenable to this approach are cases where a
‘caged’ substrate is pre-equilibrated in the crystal sample
(e.g. by diffusion or co-crystallization) and the natural
substrate is released from the cage at the protein active site
upon light flash.Proteincrystals might bethought tobe too
rigid an environment to allow protein structure conformational
change. Of course this can indeed be the case but the
fact that there is a large fraction of the crystal that is open
solvent, even up to 80% in some cases, allows some
conformational change to occur, unless the active site is
truly blocked by molecule-to-molecule crystal packing.
Obviously large multidomain rearrangements are not
Figure 4 A monochromatic rotating crystal diffraction pattern recorded
from a human rhinovirus crystal at the Cornell High Energy SR Source
(CHESS), USA. Rhinovirus was the first mammalian virus structure to be
solved (in 1985). Reproduced with the permission of Professor M. G.
Rossmann, Purdue University, USA and Acta Crystallographica.
X-ray Diffraction at Synchrotron Light Sources
4
feasible in the crystalline state without break up of the
crystal, and these cases can only be investigated in solution.
A different but related approach, at least for slower
structural processes, is to freeze the protein crystal to trap
structural intermediates. Since the intrinsic time to freeze
the crystal (of around 5 0.5 mm size) is in the sub-second
range this sets a limit to the time resolution achievable. A
faster approach involving freeze trapping is to use very thin
samples and to use electron diffraction techniques; then the
freezing time is a matter of milliseconds. Electrons scatter
more strongly than X-rays but are more damaging. They
are used as preference, however, where three-dimensional
crystals cannot be obtained. In all freeze-trapping methods
the precise moment to trap the intermediate structure has
to be determined and this decision can be guided by
spectral measurements on the protein crystal or if there is
not a spectral signal associated with the structural change
then it can be done via repeated recording of a small subset
of the diffraction spots to look for intensity changes.
For faster structural changes only the real time approach
is viable. Using polychromatic SR diffraction, whereby a
large bandpass of wavelengths is used, parallel acquisition
of the diffraction X-ray reflections is possible (Figure 2).
This is known as the Laue method as it is the same
diffraction geometry used in the original 1912 experiment,
but of course using a hugely powerful X-ray source by
comparison! Exposure times in the sub-nanosecond range
have been realized on third-generation SR sources. Clearly
on such time scales the rotation of the crystal, as employed
with the standard monochromatic diffraction geometry, is
not viable. The Laue method uses a stationary crystal for
each exposure and to record a full angular range of the data
requires a number of separate exposures with the crystal
orientation reset between each. Substantial methods
development has been required to harness this technique
for accurate structural analysis, in particular for wavelength
normalization, i.e. to bring all the spot intensities
onto a common scale recorded from the various wavelengths,
and also involving a thorough re-examination of
the Laue diffraction geometry. Instrumentation and software
engineering development has also been a major effort.
The main SR centres for these developments and initiatives
have been Cornell (CHESS synchrotron), Daresbury
(SRS), University of Chicago and Grenoble (European
SR Facility, ESRF). An interesting spin-off has been the
application to neutron protein crystallography where
neutron reactor ‘steady state’ sources produce a white
spectrum of neutrons.
The longer term promises femtosecond X-ray time
resolution capability via the X-ray free electron laser
installation proposed for the ‘DESY’ accelerator physics
laboratory in Hamburg (scheduled for 2010 for first beam,
if approved for funding). A similar project is proposed at
SLAC in Stanford.
Noncrystalline Diffraction
This embraces all other states of order, namely solutions,
fibres or amorphous solids. The case of polycrystalline
powders is not included in this category. Powders are
finding increased importance in smaller molecule ‘chemical
crystallography’ but their use is not widespread in the life
sciences where the powder lines would be exceedingly close
together, i.e. overlapping. (One test case, for example, is
that at ESRF, bacteriorhodopsin.)
Fibres are very important in the life sciences since many
naturally occurring fibres exist, such as DNA, muscle,
collagen, silk and filamentous viruses for example. Helical
structures produce a very distinctive signature of a crisscross
diffraction pattern. The high SR intensity is needed
for single fibres and/or fast time resolution. In such cases
conventional X-ray sources are too weak. Of special
significance are muscle fibres and especially time-resolved
muscle diffraction under induced electrical twitch. Indeed,
the first example of biological diffraction with SR in 1971
involved muscle fibre diffraction as mentioned in the
introduction.
Amorphous samples in the life sciences occur naturally,
such as connective tissue, bone and tendon. Because these
tend to form large samples, high intensity is generally
unnecessary, nor is fast recording needed for time-resolved
diffraction studies. However, recording of diffraction from
tiny areas of a large sample in a scanned matrix of points
allows local variations in order and structure to be probed
as well as normal and pathological tissues to be compared
(e.g. young versus old or normal versus cancerous tissue).
Microfocus SR beams on third generation sources have
had a big impact here. In one instance a sample of hair from
Napoleon (the diameter of a hair is around only 20 mm) was
examined to determine the arsenic content (it is thought
that he was systematically poisoned by his British captors
on St Helena).
Solutions of dissolved proteins are used to determine the
radius of gyration and molecular weight of the macromolecule
in question from the ‘small angle X-ray scattering
pattern’ produced. Synchrotron radiation is used where
time-induced conformational re-arrangements are made.
Examples include virus assembly or disassembly, enzyme
domain re-arrangements, etc. A closely related technique is
‘small angle neutron scattering’ (SANS). With the SANS
technique, by altering the H2O/D2O mix or the H/D
balance, the contrast of macromolecules against solvent or
protein versus protein or protein versus nucleic acid in
complexes can be varied. The SANS technique allows then
the dissection by structure analysis of complex molecular
systems including molecular machines, e.g. the ribosome,
chaperonin complex, viruses, etc.
X-ray Diffraction at Synchrotron Light Sources
5
Other Probes of Structure; Neutron and
Electron Diffraction and NMR
Spectroscopy
It is worth emphasizing that some 40% of a given genome
involves membrane-bound proteins and many of these
(most?) may not be amenable to crystallization. Electron
diffraction of two-dimensional arrays allows for a different
approach. Also it is possible to record electron microscope
images of large numbers of individual molecules in many
different orientations on an em sample grid to reconstruct a
three-dimensional image of the whole molecule in the
computer. This has been particularly successful for
obtaining medium resolution (10 A˚ ) of ribosomal particles
already; and improvements towards 5-A˚ resolution are
taking place. Nuclear magnetic resonance (NMR) is a
completely different approach which uses the radio
frequency resonance signals of different isotopes to
elucidate structure. The use of NMR for structure
determination for proteins with molecular weights up to
20 kDa is routine, and 30 kDa is feasible. The big
advantage is that the protein does not need to be
crystallized as a solution is used. The precision of the
atomic positions is not as good as from a protein crystal
structure analysis but if a crystal cannot be obtained the
NMR structure is invaluable. In another category of study
by NMR of protein ligand binding, weakly interacting
complexes can be studied whereas protein crystallography
requires at least approximately 30% occupancy of binding
to pick up the structural details in the difference electron
density image that is determined. In rational drug design
programmes, i.e. involving computer design, weak ligand
binding may result and so NMR studies can be used to
guide the design process through this stage.
An Overview of Achievements of SR
Diffraction in the Life Sciences
In the 20 years of wide SR usage in structural studies
involving diffraction techniques there have been spectacular
successes, especially for huge multi-macromolecular
complexes such as viruses and ribosomes at secondgeneration
and now also third-generation SR sources (i.e.
for yet larger complexes). Sub-nanosecond time-resolved
protein crystallography has been realized at third-generation
SR sources using the Laue method. Tiny crystals and
single fibres can now yield usable diffraction data at thirdgeneration
SR sources. High-throughput protein crystal
structure determination, ‘structural genomics’, is becoming
a reality where the step of optimizing the crystal size can
be relaxed and MAD phasing is applied. As more and more
dedicated SR facilities become available and more easily
accessible geographically, i.e. on a regional as well as a
national or an international basis, the number, range and
type of application will be much more diverse and the
numbers of structures available will increase dramatically.
X-ray Diffraction at Synchrotron Light Sources
6