ÐÓ°ÉÔ­´´

If we could watch enzymes in action, we might discover the secrets of
some of the most important biochemical reactions. Enzymes are proteins that
speed up chemical reactions in living systems and to understand how they
do this we need to know their atomic structures. Researchers worked out
the structure of the first enzyme in 1965. This was a static model. At the
time it was impossible to capture a protein molecule in action. About seven
years ago, protein crystallographers teamed up with physicists to produce
a technique that combines some of the brightest X-rays in the world with
a revival of an old method of recording information about crystal structures.
This new technology should allow us to watch nature’s biological catalysts
at work.

The technique of study-ing crystals by X-ray diffraction goes back to
1912, when two students of the physicist Max Theodor Felix von Laue, W.
Friedrich and P. Knipping, recorded the first X-ray photo of a crystal –
copper sulphate – using the broad spectrum radiation emitted by an X-ray
tube. No small part of their achievement was to keep the tube working for
the 20 hours it took to record the diffraction pat-tern, a series of spots
imprinted on a photographic film.

A year later, the physicist Lawrence Bragg provided the connection between
spots and atoms, with the law that bears his name. This law relates the
distances between planes of atoms in a crystal to the way they diffract
X-rays of a particular wavelength. Bragg’s law made it possible to decode
diffraction patterns into an atomic structure. The process is akin to the
formation of an image in an optical microscope, where light rays are scattered
by an object and then recombined using a system of lenses to form an image.
In X-ray crystallography the X-rays are scattered by the crystal and the
pattern recorded. There is no lens capable of refocusing the X-rays. Instead,
the diffracted beams are ‘recombined’ into an image of the atoms with a
computer using a mathematical technique called Fourier transforms.

Bragg solved the structure of zinc blende, sodium chloride and other
simple crystals. By the 1940s, biochemists realised that the same method
might be a useful tool for studying protein crystals, and in 1959, John
Kendrew and his team at the Cambridge Medical Research Council Laboratories
solved the crystal structure of the first protein molecule, myoglobin. This
molecule, which is similar to a subunit of haemoglobin, binds oxygen in
muscle, but does not catalyse a chemical reaction.

The study of enzymes as crystals is not straightforward. Most enzyme-catalysed
reactions take place in solution, so biochemists have to be sure that the
structure and activity of the enzyme is not altered by crystallisation.
Until recently there was no way of proving this, but researchers could accumulate
enough circumstantial evidence, by crystallising under different conditions
and by comparing structures, to be fairly certain that crystallisation did
not alter the biological structure. Fortunately, they also found that many
enzymes are active in the crystal as well as in solution.

Twenty-six years ago, David Phillips and his team at the Royal Institution
in London succeeded in working out the first enzyme crystal structure –
a small but important enzyme called lysozyme. This discovery was the culmination
of several tens of thousands of tedious X-ray measurements. From its structure,
it was possible to propose a detailed mechanism for how the enzyme worked.
These proposals caused immediate excitement among biochemists, who realised
that this could be the beginning of a new understanding of chemical reactions
in living systems.

Lysozyme is a good example of why enzymes are important. It destroys
bacterial cell walls by means of a hydrolysis (water splitting) reaction
that breaks a carbon-oxygen bond in the large polysaccharide molecules of
the cell wall. Chemically, the same reaction can be speeded up only by strong,
boiling acid. How can the enzyme do what chemistry cannot? The answer lies
in its structure – more precisely, in certain carboxyl groups which are
poised in exactly the right position to bring about the hydrolysis.

There were several problems with extending X-ray crystallography to
other enzymes. Many enzymes are available in very small quantities and are
difficult to crystallise. Once crystallised, the crystals are fragile and
are susceptible to damage, not only by X-rays but also by variations in
temperature and in the type of solvent. More serious still is their size:
proteins are large, complex molecules. The technique of X-ray crystallography
works for crystals such as sodium chloride because these are made up of
fairly small repeating units, spaced roughly the same distance apart as
the wavelength of the X-rays. In proteins the repeating units are much larger
and as much as half of their volume can be water, so they diffract X-rays
poorly.

By the late 1960s, researchers had worked out the structures of several
complex proteins, including haemoglobin, but it was a painstaking task.
Apart from the thousands of calculations needed to interpret the diffraction
patterns, recording the diffraction pattern was laborious, because the crystal
had to be rotated gradually, through a couple of degrees for each set of
measurements. This process had to be repeated for the next rotation and
so on, until a complete three-dimensional diffraction pattern was obtained.

For enzymes there was the additional difficulty that however accurate
X-ray analysis of a static structure is, it can tell only part of the story.
The catalytic process is a complex cycle with several distinct steps. First,
the enzyme must recognise the starting material (substrate) for the reaction
and bind to it. Then it must bring about the chemical changes of bond-breaking
or bond-making, often by stabilising an ‘in between’ molecule (the transition
state) that is neither substrate nor product. The next step is to bring
about the conversion of the intermediate to a product molecule. And it must
still recognise this product molecule – almost all enzyme reactions are
reversible, so they can work either forwards or backwards (see Box 1 for
details).

To do all this, enzymes have the ability to respond in a different way
to each part of their task by subtle changes in their shape. So to understand
enzyme catalysis we need to know not only the structure of a single stage,
but the structure of the enzyme as it passes through all the different stages.
Could biochemists make a moving picture of an enzyme at work? Until only
a few years ago recording such a sequence was impossible. In the race to
record what was going on, the enzymes won hands down, even when researchers
used the fastest X-ray methods. The reaction was over before experimenters
could record the crucial step: the structure of the complex formed between
the enzyme and the substrate.

As a result, most of what we know about enzymes has come indirectly
from the study of transition state analogues. These are molecules of known
structure which organic chemists believe resemble the state of affairs at
the point of highest energy on the path from substrate to product. In practice
this transition state is so unstable that it is almost impossible to isolate.
Researchers have tried to study enzyme-substrate complexes in different
ways. Some studied the enzyme at an unfavourable pH or at a low temperature,
which slowed the reaction enough to allow time for X-ray measurements. But
such studies are complicated because at these low temperatures water in
the protein crystal freezes, and crystals are not so easy to grow from other
solvents. Researchers have to compare many experiments to check that these
changes have not altered the way the enzyme works.

A turning point came in the early 1980s with two key developments. One
was the application to biological crystals of the method Laue used earlier
this century for the first X-ray studies of crystals. The second, and most
important, was the exploitation of synchrotron radiation. This is an intense
form of electromagnetic radiation that spans all wavelengths, including
X-rays (see Box 2 for more details). Synchrotron radiation provided researchers
with X-rays at least 100 times as intense as those produced by conventional
sources. Pioneering experiments on synchrotron radiation were carried out
in the late 1970s, at accelerators in Paris and Hamburg designed for the
study of elementary particle physics. Such accelerators emit synchrotron
radiation as a by-product. Then in 1981 a source designed solely to produce
this radiation was commissioned at Daresbury in Cheshire. By combining synchrotron
radiation with conventional methods of data collection (single-wavelength
X-rays), researchers have been able to determine the structures of large
numbers of complex proteins and viruses, which would otherwise have been
impossible. This application has been the most successful and productive
application of synchrotron radiation in protein crystallography but the
Laue method has great potential for the future.

In the Laue method the crystal is irradiated with X-rays of all wavelengths,
instead of a single wavelength used in conventional X-ray crystallography
. More radiation hits the crystal, so it can be exposed to the X-rays for
a much shorter time, although the greater intensity of the X-rays also increases
radiation damage. Since each set of planes of atoms can find the correct
wavelength for their diffraction, the Laue method removes the need to rotate
the crystal. Laue diffraction patterns are beautiful but complex. Each spot
is generated by reflection from a different wavelength, and the spots are
close together, so converting the spots on the film to accurate measurements
of their intensity requires great care.

The first Laue photographs of protein crystals were recorded in 1984
by Keith Moffat at the Cornell High Energy Synchrotron Source (CHESS) in
the US and by John Helliwell at the Synchrotron Radiation Source (SRS) at
Daresbury. By 1987, these and other researchers had developed methods for
processing the films. During the 1980s, Janos Hajdu and our team from the
University of Oxford used the intense X-rays from the SRS to study single
crystals of phosphorylase, a large enzyme (subunit molecular weight about
100 000) which controls the conversion of insoluble glycogen to soluble
glucose-1-phosphate to provide energy for muscle contraction. By combining
fast X-rays from the synchrotron with the Laue method we were able to reduce
the time it took to record a diffraction pattern of the enzyme from hours
to seconds or even milliseconds. We used these data to show how an oligosaccharide
molecule binds to phosphorylase.

Timing in right

But we had to wait a little longer for the first real enzyme movie.
To study an enzyme-catalysed reaction in progress, it was necessary to enlist
the help of chemists. Their contribution was to devise a ‘caged’ substrate.
This is a substrate molecule such as phosphate, which is made biologically
inactive (it is protected from the enzyme) by the addition of a blocking
group. The group is photolabile – that is, it can be removed quickly and
easily by illuminating it with a laser or a xenon flash lamp. With release
of the cage made to coincide with the opening of the X-ray shutter, researchers
can synchronise the start of the reaction in the crystal with the start
of the X-ray measurements. This is important for pinpointing the ‘zero time’
for both reaction and measurements.

One of the most outstanding applications of the Laue method to the study
of enzyme-catalysed reactions has been recent work with a protein known
to be associated with uncontrolled growth in cancer tumours. This is the
relatively small protein called Ha-Ras p21 (molecular weight 21 000). Precisely
what this protein does has not been established but from analogies with
other systems, it is thought to be in an ‘active’ state when bound with
guanosine triphosphate (GTP) and in an ‘inactive’ state when bound with
guanosine diphosphate (GDP). GDP is formed by hydrolysis of GTP.

The p21 protein has slow GTP hydrolysis activity and so once bound to
GTP it can switch itself off by converting GTP to GDP. This reaction is
greatly enhanced by binding the p21 protein to another protein, a GTP hydrolysis-activating
protein, GAP. In about one in 10 of human cancer tumours the p21 protein
is mutated at a single amino acid. In these so-called transforming mutants,
the interaction between p21 and GAP is defective. Hence, the p21 is only
able to hydrolyse the GTP slowly, the p21-GTP complex exists for a longer
time and the signal for growth is prolonged.

Two groups of scientists, one led by Sung-Hou Kim at the University
of California at Berkeley and the other by Emil Pai at the University of
Heidelberg, working independently, had determined the structure of p21 between
1985 and 1989. The location and most of the features of the enzyme’s catalytic
site were known from conventional X-ray studies of the GDP-p21 complex.
But the all-important complex between the enzyme and GTP was relatively
short-lived – its half-life was 40 minutes – so the structure could not
be trapped by conventional crystallography. The trick Pai and his team used
was to cage the GTP and crystallise it at the active site of the enzyme.
After liberation of the cage they then recorded Laue photographs that represent
two successive stages in the reaction: the GTP complex (after 4 minutes),
and a GTP-GDP mixture (after 14 minutes). From these photographs they identified
structural differences in two loop regions of the enzyme between the GTP
complex and the GDP complex, and it seems likely that these regions are
involved in the recognition of GAP.

The same techniques can also be applied to viruses which, like proteins,
are difficult to study by X-ray methods because they are so large. Since
Phillips’s pioneering work on lysozyme, over 600 protein structures have
been solved by X-ray methods. Several of these are the structures of the
same protein from different species or in another crystal form and the number
of unique structures is estimated to be about 140. For many of them this
information has been extended to working out the biochemical mechanism by
which they catalyse reactions. Given the recent developments in synchrotron
radiation and Laue diffraction, some of the most exciting enzyme movies
may be coming soon.

Louise N. Johnson FRS is David Phillips Professor of Molecular Biophysics
at the University of Oxford.

* * *

1: Nature’s answer to slow chemistry

Enzymes are biological catalysts that speed up nearly all chemical reactions
in living systems. They are protein molecules which are precisely folded
into compact structures of a particular size and shape. The sequence of
amino acids that make up these proteins is important. This determines the
three-dimensional structure which in turn holds the key to how enzymes work.

In the first stage of an enzyme-catalysed reaction the enzyme recognises
the molecule that is to undergo chemical change – the substrate – and binds
it relatively loosely to form an enzyme-substrate complex. As the reaction
proceeds the substrate goes through a high-energy structure known as the
transition state, which is an intermediate between bond-making and bond-breaking
on the chemical pathway from substrate to product.

Almost all enzymes have as part of their structure groups of atoms which
help to stabilise this enzyme-transition state complex. This has the effect
of lowering the energy barrier to the reaction, which is speeded up by as
much as 10 orders of magnitude compared with the same reaction without an
enzyme. Once the substrate has been converted to a product molecule, the
enzyme-product complex splits up, and the enzyme returns to its original
shape.

Many enzyme-catalysed reactions are far more complex than this. They
may go through several steps, each having an intermediate that is stable
for a short time, or involve two or more substrates or products. The order
of events – binding of substrates or release of products, for example –
is also important. To follow an enzyme-catalysed reaction biochemists ideally
need to identify the structures of the intermediate stages. But transition
states are by their nature unstable, and researchers have to rely on studies
of transition state analogues – compounds of known structure that resemble
the transition state, and which bind tightly to the enzyme.

* * *

2: How X-rays from electrons make Laue patterns

Synchrotron radiation is electromagnetic radiation produced when charged
particles are accelerated at velocities close to the speed of light and
forced to move along a curved path by a magnetic field. The radiation extends
from X-rays to the wavelengths of visible light.

In a synchrotron, such as the SRS at Daresbury, the charged particles
are electrons. These are injected from a linear accelerator into a booster
synchrotron and from there into a doughnut-shaped storage ring run at energies
of between 1 and 20 gigaelectronvolts, where they form currents of up to
several hundred milliamps.

In the storage ring the electrons are usually grouped in small bunches
a few millimetres long. These produce short (between 10 and 500 picosecond)
intense bursts of synchrotron radiation whenever they pass through the fields
of the bending magnets positioned at intervals around the circumference
of the ring. The radiation is confined to a narrow cone centred around a
tangent to the orbiting electrons.

Filters select the particular range of wavelengths needed for particular
experiments, for example infrared or X-rays. For protein crystallography
the useful wavelength range is in the X-ray region, usually between 0.2
and 2.0 angstroms.

By inserting a device called a wiggler magnet in the ring it is possible
to increase the bending radius of the electron path over short distances.
Such ‘wiggles’ in the electron beam shift the range of the radiation towards
shorter wavelengths and provide more intense X-rays.

In conventional X-ray diffraction the crystal is irradiated with a monochromatic
(single wavelength) beam of X-rays. Planes of atoms in the crystal with
spacing d produce diffraction spots at an angle q for an X-ray wavelength
l, where the position of the spots is governed by Bragg’s Law: 2d sin theta
= n lambda where n is the order of diffraction (n=1, 2, 3 . . . ). The crystal
must be rotated to put different planes in the right position to diffract.

In Laue diffraction, the crystal is illuminated with a polychromatic
beam of X-rays. All (or nearly all) the planes in the crystal can find a
particular wavelength to satisfy Bragg’s Law. This means that a single exposure
to X-rays produces many reflections simultaneously in a short time. The
wider the range of wavelengths, the greater the number of reflections that
can be recorded. In practice, the wavelength range is between 0.25 and
2.0 angstroms.

For highly symmetrical crystals, such as the cubic crystals of some
viruses, a single snapshot is enough to record almost all the reflections.
For less symmetrical crystals such as the tetragonal crystals of glycogen
phosphorylase, about three photographs are needed. The intensity of each
diffraction spot is dependent on wavelength, so to obtain a self-consistent
set of data, the measurements must be adjusted so that they are all refer
to a particular wavelength. Usually this is done by comparing the intensities
of reflections which we know to be related in a predictable way by the symmetry
of the crystal.