Structural biology has developed during the last decade, from an intellectual pursuit of mainly academic interest to a major area of research which has essential implications for progress in molecular and cell biology as well as for biotechnological applications in pharmaceutical drug design and production of modified enzymes used in industrial processes. Synchrotron radiation sources play an increasingly important role in this development as illustrated by a survey of published papers in macromolecular crystallography (Figure 68). Six years ago, during 1990, only 6 of the 23 biological X-ray structures published in "Nature" were based on synchrotron data. In 1995 this figure had increased to 30 of the 41 articles (73 %). Data from the ESRF contributed to six of these articles [1-6]. A similar picture is found in more specialised journals. In the journal "Structure", which is entirely devoted to structural biology, two thirds of the 75 articles on X-ray structures published in 1995 were based on synchrotron data. It is anticipated that the demand for synchrotron data in this field will continue to accelerate since the number of scientists working in this area, as well as their products (X-ray structures) are rapidly growing. 

The ESRF Council and Management have realised this development and structural biology is rapidly becoming one of the major activities at the ESRF. The close collaboration with the EMBL outstation in Grenoble plays an important role for the development of the beamlines and for user support, as well as for providing an excellent biological infrastructure. By the end of 1998 when the construction phase of the ESRF comes to an end, there will be five experimental stations fully devoted to macromolecular crystallography and another three stations partly devoted. By that time we are planning to have all stations equipped with CCD array detectors with a sufficiently fast read-out system to match the exposure time. However, even more important than the quantity of beam time, are the qualitative aspects of both the beam and the beamline equipment at the ESRF, that enables the community to carry out novel and exciting experiments in the front-line of structural biology. We will give here examples of different types of experiments in structural biology, which have been performed at the ESRF and which have important implications for the extended use of synchrotron radiation in the future.

 

 

Large biological assemblies

There are two obvious trends in current structural biology research: 1) to investigate large complex systems involving several different macromolecules that act in a co-ordinated fashion and 2) to obtain a detailed understanding of biological functions of simpler systems. These two aims both require a high brilliance X-ray beam for data collection in order either to obtain data from crystals of very large unit cells which are usually weakly diffracting, or to collect X-ray data of very high resolution giving an accurate and detailed model of a protein structure.

Undulator beams are ideal for this purpose. The High Brilliance beamline ID2, which is equipped with a standard 46 mm ESRF undulator, produces a very small and parallel beam with a high flux through the sample. A number of very demanding data collection projects have now been carried out on the protein crystallography station of this beamline since it became operational in September 1994. The project on the blue-tongue virus by the group of David Stuart, Oxford, is one such example.

The blue-tongue virus (BTV) infects cattle and wild ruminants but not humans. It obtained its name because in some infected animals the tongue becomes swollen, blueish and may protrude from the mouth. The virus particle is large and composed of an outer fuzzy shell and a well-defined inner core. The core is also large, 800 Å in diameter with a molecular weight around 60 million Dalton. It has a stable protein shell, comprising 780 copies of one protein, VP 7, of 39 kDa molecular weight each and 120 copies of a second protein, VP 3, of 103 kDa molecular weight. Inside this shell is the genetic information for making more virus particles in the form of 10 RNA molecules comprising a total of 19,200 base pairs.

Core particles have been crystallised and are the largest biological assemblies that have given well-diffracting crystals. Two different crystal forms have been obtained from two different strains of the virus, BTV 1 and BTV 10. X-ray data have been collected on both forms at the ESRF and Figure 69 shows a diffraction pattern from a BTV 1 crystal. Using these data, D. Stuart's group has obtained a preliminary model of the protein shell of the core particle (Figures 70 and 71). The shell is composed of two layers, an inner scaffold comprising the 120 copies of VP 3 and an outer layer formed by the 780 copies of VP 7.

The protein shell of all spherical viruses is built up using icosahedral symmetry. The 2-fold, 3-fold and 5-fold rotation axes of an icosahedron produces 60 equivalent units, each containing one or several protein subunits. Therefore the number of subunits of a specific protein in a virus shell is a multiple of 60 called a triangulation number, T. Simple geometric consideration has shown that only certain values of T are allowed in order to preserve quasi-equivalence in the packing of the subunits of the shell. VP 7 with 780 copies has T = 13 which is an allowed number and the structure shows that the packing environment of all VP 7 subunits is, at a first approximation, similar or quasi-equivalent. This is not the case, however, for the VP 3 subunits. T = 2 is not an allowed triangulation number, since it is not possible to arrange 120 subunits in a quasi-equivalent way into an icosahedral shell. However, the structure of BTV shows how simply and elegantly nature has overcome this theoretical obstacle. The packing unit is not a single copy of VP 2 but a dimer of two copies (Figure 70). There are 60 dimers, one in each equivalent icosahedral unit forming a perfectly normal icosahedral packing arrangement.

There are no drugs available for the disease caused by the blue-tongue virus. The structural information now obtained by D. Stuart's group will be of crucial importance for the ongoing efforts to develop an efficient and safe drug for this serious disease. Knowledge of the molecular architecture at the atomic level of the virus particle will enable rational design of strongly binding drugs that will prevent infectivity of the virus.

 

 

Small crystals

It is a common experience that only small crystals can be obtained for many proteins even after years of crystallisation trials. Such crystals in general do not give useful X-ray data on home sources and not even from second generation synchrotron sources. However, the high brilliance of undulator beams at the ESRF makes it possible to obtain high resolution diffraction patterns even from very small protein crystals.

The first example of such a project at the ESRF was a cell cycle control protein p13suc1 for which crystals were brought to ID2 by L. Johnson's group, Oxford. To obtain phase information it was essential to collect data for a selenomethionine derivative of the protein. However, the crystals obtained from this derivative were consistently small with maximum dimensions of 20 x 20 x 100 µm3. Nevertheless, high quality data to 2.7 Å resolution could be obtained with an exposure time between 5 and 20 seconds at room temperature. Cell dimensions were a = b = 60.8 Å, c = 265.8 Å. Using these data the structure could be determined [7], which has given important information concerning control of the cell cycle.

High quality data from even smaller crystals were later obtained from another protein called 14-3-3 which is also involved in cell cycle regulation as well as having a critical regulatory role in signal transduction pathways across membranes. The crystals, which were produced by S. Gamblin's group at London, were very thin needles (Figure 72), of size 5 x 9 x 200 µm3. Data were collected to 2.6 Å resolution at ID2 using cryo-cooling and were of sufficiently high quality for the structure to be determined [4].

These results demonstrate that data of high quality can be obtained from very small crystals of biological macromolecules at the ESRF from undulator beamlines. Using a microfocus beam, preliminary experiments have shown that it is possible to use even smaller crystals. Theoretical calculations indicate that crystals as small as 5 x 5 x 5 µm3 of medium-sized protein molecules should be sufficient to obtain useful diffraction patterns.

 

 

MAD-phasing

One must know both the amplitude and phase for each of thousands of diffracted waves from a macromolecular crystal in order to reconstruct an image of the molecule, but only amplitudes can be recovered from standard diffraction measurements. Phase evaluation traditionally has been based on the analysis of isomorphous replacement with heavy atoms but recently, an alternative approach has been developed: the method of multiwavelength anomalous diffraction, MAD.

This method exploits the wavelength-dependent scattering effects of resonance between X-rays and bound atomic orbitals and requires the presence of a few heavier elements as resonance centres. Either intrinsic metal centres, introduced heavy atoms complexed to the protein or incorporation of selenomethionine in place of the natural aminoacid methionine by recombinant DNA-technology are suitable.

MAD-phasing requires small, reproducible and extremely precise wavelength changes and is hence an ideal application for synchrotron radiation, since it can be used to solve the phase problem in an elegant and rapid way.

Two beamlines at the ESRF, BM14 and the CRG-beamline D2AM, have been specifically designed for MAD experiments in structural biology. Both beamlines use the image intensifier CCD detector developed at the ESRF with a current read-out time of 3-5 seconds per image which allows very efficient use of the beam time compared to image plate detectors. More than a dozen MAD-datasets have been collected on these beamlines for different projects and some examples are given here to illustrate different approaches of MAD phasing.

Exonucleases are enzymes that are essential for making and repairing DNA-molecules. D. Suck's group, EMBL Heidelberg [8], was trying to solve the crystal structure of an exonuclease from bacteriophage T5 by isomorphous replacement but failed to obtain isomorphous heavy atom derivatives. They could, however, prepare a modified enzyme by recombinant DNA techniques containing 5 selenomethionine residues per molecule, which gave well-diffracting crystals. Data were collected at BM14 at four different wavelengths (0.9793, 0.9795, 1.035 and 0.87) around the Se absorption edge which provided sufficient phase information to produce an interpretable electron density map and build a model containing 272 amino-acid residues (Figure 73). The model contains an arch built up from two a-helices through which a single-stranded DNA molecule could slide and be cleaved. A mechanism for the catalytic action, which occurs at the base of the arch, has been proposed from this model.

Lectins are proteins that recognise and bind specific carbohydrates. They play an important role in cell-cell recognition and are essential for such diverse functions as inflammatory response in humans and nitrogen fixation in plants. The structure of a new family of lectins, maclura pomifera agglutinin, has recently been determined from MAD data at BM14 [9]. The crystals contained both mercury and lead and MAD data were collected at six different wavelengths around both the Hg and Pb absorption edges. Crystals were of spacegroup C 222 with unit cell dimensions a = 58.8 Å, b = 120 Å, c = 155 Å, and data were recorded on the image intensifier CCD detector to 2.9 Å resolution with exposure times of 1 minute per frame. The data quality was excellent with Rsym of around 2 %, high redundancy [4-6] and more than 95 % completeness. The resulting electron density map is interpretable (Figure 74) and a model of the protein is being built.

The cutting edge of applying the MAD method is to extend it to larger and larger protein structures. Janet Smith, Purdue University (USA), has explored the use of ESRF beamlines for MAD phasing of large structures. A selenomethionine experiment was done on the bending magnet beamline BM14 for the enzyme glutamine PRPP amidotransferase, a protein of 110,000 molecular weight. However, MAD experiments from proteins of this size benefit most from the high brilliance of an undulator beam. A MAD experiment was therefore done for the enzyme GMP synthetase, a structure of 220,000 molecular weight, on the high brilliance undulator source ID2. In both cases, high quality electron density maps were obtained (Figure 75) from the phase information due only to selenomethionine.

These examples demonstrate that MAD phasing has now matured as a method and provides a rapid and efficient method for solving biological structures especially if CCD detectors are used. The high quality of the diffraction patterns obtained by an undulator beam makes the method feasible even for large structures using only the signal from selenomethionine.

 

 

Time-resolved studies in structural biology

In the Highlights of last year we described feasibility studies on ID9 from nanosecond time-resolved experiments of release of carbon monoxide, CO, from crystals of the protein myoglobin, using Laue diffraction from a polychromatic single bunch [10].

These studies have now been extended [11], the data have been analysed and the electron densities obtained show significant features of movement inside the protein on nano- and microsecond time scales.

Data sets were collected at different time delays (4 ns, 1 µs, 7.5 µs, 50 µs, 350 µs and 1.9 µs) after the laser pulse by exposures of very short polychromatic X-ray pulses (150 ps) to allow the determination of the time evolution of the atomic motions involved and hence to deduce the role played by the different constituents of the protein molecule. The difference maps clearly show the release of the CO molecule, the displacement of the Fe and some further, minor rearrangements.

For the first time point (Figure 76) there is also a weak density in a position identical to the docking site determined by low-temperature crystallographic studies. The CO molecule is assumed to occupy this site after release from the heam before it escapes out from the heme interior. The last time points confirm the reversibility of the reaction which is of importance for the possibility of collecting redundant and accurate data. The results obtained establish the feasibility of acquiring data with sub-nanosecond exposure times and to deduce structural information from such data. However, to complete the story, further time points of improved quality, if possible, are needed. For this purpose a subsequent experiment has been undertaken with several technical improvements and the data analysis is underway. The experiment was made possible by careful experimental design and newly developed equipment including a fast shutter, a low-noise detector, X-ray and optical monitoring systems.

A feasibility study has also been done to determine the accuracy of structural information that can be obtained using Laue diffraction data from a single bunch at ID9. Two insertion devices were used in series and at 15 mA up to 1.8 x 107 photons / 0.1 % bandwidth can be delivered at 15 keV on a 200 x 200 µm2 sample in a single 150 ps X-ray pulse. Crystals of cutinase were used which is a fungal 22 kDa enzyme whose structure is known at 1.0 Å resolution. From a single bunch Laue data set with a total exposure time of 8.5 nanoseconds, the structure of native cutinase could be determined at 1.5 Å. In order to mimic a fast time-resolved experiment, we used as a starting model for refinement the known structure of a mutant where residue 196 has been changed from arginine to glutamic acid. This mutant crystallises in a different space group. A satisfactory model of native cutinase was obtained, with Rcryst = 19.3 % (Rfree = 24.2 %). Discrepancies between this model and the assumed «true» structure of cutinase (obtained from monochromatic data collected to 1.0 Å resolution, Rcryst = 9.5 %) were minor. The wild-type Arg196 could be readily positioned in the electron density (Figure 77), and significant main and side-chain displacements due to packing constraints were successfully retrieved with the Laue data [12]. The electron density maps showed good connectivity and were of sufficient quality to solve unambiguously the conformational changes between wild type and mutant. This feasibility study shows that single-bunch Laue diffraction is a powerful tool to perform time-resolved studies down to the 150 ps timescale with suitable macromolecular crystals.

Time-resolved studies have also been made on single muscle fibres using monochromatic radiation from the High Brilliance beamline ID2. Generation of force and shortening in a contracting muscle are attributed to the interaction of myosin cross-bridges extending from the thick filament and the thin actin filament. Structural changes in intact single fibres from frog skeletal muscles have been studied in a joint project between an Italian and a British group [13].

X-ray data were collected in 10 ms time frames with a 2D gas-filled detector during the rise of the tetanic tension and the subsequent steady shortening at a velocity that gives nearly the maximum power output. Changes in intensity and spacing of the third-order myosin meridional reflection, arising from the axial mass projection of the myosin heads, could be precisely measured by signal averaging from only one fibre. Fibres were mounted between a force and a length transducer by means of aluminium foil clips to minimise the series compliance. Tension development (half-time 33 ms, at 4 °C and 22 mm sarcomere length) is accompanied by an increase in axial spacing of the reflection from 14.34 nm (rest value) to 14.57 nm (tetanus plateau value), which has been attributed to myosin heads attaching to actin. The reflection intensity decreases to 50 % of the resting value within 30 ms, due to the loss of resting crystallographic order (Figure 78), then increases with a time course that almost superimposes tension development. Steady shortening decreases the tension and the intensity of the reflection to 60 % and 50 % of their respective isometric values, but does not affect the spacing of the reflection by more than 0.3 % indicating that the intensity signal arises from myosin cross-bridges interaction with actin. Consistent with previous work, these results suggest that attached myosin heads developing isometric force have their long axis nearly perpendicular to the filament axis while, during the execution of the working stroke that drives shortening, they undergo a change which broadens their axial mass projection.

 

 

Single fibre diffraction

Dragline spider silk has aroused considerable interest due to its excellent structural and mechanical properties. In order to relate macroscopic properties to microscopic models,

X-ray structural data should preferably be obtained from single fibres. This allows differences in crystallinity and orientation between individual fibres to be avoided. X-ray structural data are, however, very limited and have always been obtained on fibre bundles. It has now been shown for the first time that structural data on a single spider fibre of < 10 µm diameter can be obtained in a few seconds using the focused beam of a low-ß undulator [14].

X-ray diffraction data were collected on dragline fibres from two different spider species, Nephilia clavipes and Kuculeania hibernalis. Experiments were performed on the Microfocus beamline at the ESRF using a 7 µm beam (full width at half maximum) at a wavelength of 1.488 Å and a flux of 1011 photons/sec. Individual fibres were fixed on electron microscopy apertures. The aperture was placed on a goniometer head and the fibre aligned by a long-distance microscope on a kappa-goniometer so that the fibre axis was oriented vertically.

An image intensified video camera with converter screen and on-line digitiser was used for data collection with an integration time per frame of 4 sec. Experiments were performed at room temperature.

Figure 79 shows a scanning electron microscopy image of a fibre of Nephila clavipes and Figure 80 the corresponding diffraction pattern. The equatorial and first layer line reflections have been fitted with Gaussian profiles. Structural features disappeared in less than 30 sec which suggests the use of cooling techniques in order to get a more complete data set.

Preliminary experiments have shown that liquid nitrogen cooling gives a lifetime improvement of more than one order of magnitude as well as a sharper diffraction pattern.

These experiments indicate that using third generation synchrotron sources fibre diffraction studies may enter a new era where individual domains of single fibres of a few micrometer sizes may be studied.

 

 

Publications

[1] L. Shapiro (a), A.M. Fannon (b), P.D. Kwong (a), A. Thompson (c,e), M.S. Lehmann (d,e), G. Grübel (e), J.F. Legrand (e), J. Als-Nielsen (e), D.R. Colman (b), W.A. Hendrickson (a), Nature 374, 327 (1995)

(a) Columbia University, New York (USA)

(b) Mount Sinai School of Medicine, New York (USA)

(c) EMBL, Grenoble (France)

(d) ILL, Grenoble (France)

(e) ESRF

[2] P.L. Roach (a), I.J. Clifton (a), U. Fülöp (a), K. Harlos

(a), G.J. Barton (a), J. Hajdu (a), I. Andersson (a,b), C.J. Schofield (a), J.E. Baldwin (a), Nature 375, 700 (1995)

(a) Univ. of Oxford, (UK)

(b) Biomedical Centre, Uppsala (Sweden)

[3] T. Bizebard (a), B. Gigant (a), P. Rigolet (a), B. Rasmussen (b,c), O. Diat (c), P. Bösecke (c), S. Wharton (d), J. Skehel (d), M. Knossow (a), Nature 376, 92 (1995).

(a) CNRS-Université Paris-Sud, Gif-sur-Yvette (France)

(b) EMBL, Grenoble (France)

(c) ESRF

(d) MRC National Institute for Medical Research, Mill Hill, London (UK)

[4] B. Xiao (a), S.J. Smerdon (a), D.H. Jones (a), G.G. Dodson (a,b), Y. Soneji (a), A. Aitken (a), S.J. Gamblin (a), Nature 376, 188 (1995)

(a) Nat. Inst. for Medical Res., Mill Hill, London (UK)

(b) Univ. of York (UK)

[5] S. Iwata (a), C. Ostermeier (a), B. Ludwig (b), H. Michel (a), Nature 376, 660 (1995)

(a) MPI für Biophysik, Frankfurt (Germany)

(b) Johann-Wolfgang-Goethe Universität, Frankfurt (Germany)

[6] C.R. Kissinger, H.E. Parge, D.R. Knighton, C.T. Lewis, L.A. Pelletier, A. Tempczyk, V.J. Kalish, K.D. Tucker, R.E. Showalter, E.W. Moomaw, L.N. Gastinel, N. Habuka, X. Chen, F. Maldonado, J.E. Barker, B. Bacquet, J.E. Villafranca, Nature 378, 641 (1995)

Auguron Pharmaceuticals Inc., San Diego (USA)

[7] J.A. Endicott (a), M.E. Noble (a), E.F. Garman (a), N. Brown (a), B. Rasmussen (b), P. Nurse (c), L. Johnson (a), EMBO J. 14, 1004 (1995)

(a) Univ. of Oxford (UK)

(b) EMBL and ESRF, Grenoble (France)

(c) ICRF Laboratories, London (UK)

[8] T.A. Ceska (a), J.R. Sayers (b), G. Stier (a), D. Suck (a), Nature 382, 90 (1996)

(a) EMBL, Heidelberg (Germany)

(b) Univ. of Sheffield (UK)

[9] M. Young (a), R. Johnstone (a), X. Lee (b), J. Biesterfeldt (b), H. Ton-That (b), A. Thompson (c), to be published

(a) National Research Council of Canada, Ottawa (Canada)

(b) Cleveland Clinic Research Institute, Cleveland (USA)

(c) EMBL and ESRF, Grenoble (France)

[10] D. Bourgeois (a,b), T. Ursby (a), M. Wulff (a), C. Pradervand (c), A. Legrand (c), W. Schildkamp (c), S. Labouré (a), V. Srajer (c), T.Y. Teng (c), M. Roth (b), K. Moffat (c), J. Synchr. Rad. 3, 68 (1996)

(a) ESRF

(b) IBS, Grenoble (France)

(c) Univ. of Chicago (USA)

[11] V. Srajer (a), T. Teng (a), T. Ursby (b), C. Pradervand (a), Z. Ren (a), S. Adachi (c), W. Schildkamp (a), D. Bourgeois (b,d), M. Wulff (b), K. Moffat (a),

accepted by Science

(a) Univ. of Chicago (USA)

(b) ESRF

(c) Inst. of Physical and Chemical Res., Saitama (Japan)

(d) IBS, Grenoble (France)

[12] D. Bourgeois (a,b), S. Longhi (c), M. Wulff (a), C. Cambilleau (c), J. Appl. Cryst., in print

(a) ESRF

(b) IBS, Grenoble (France)

(c) CNRS, Marseille (France)

[13] I. Dobbie (a), G. Piazzesi (b), M. Reconditit (b), P. Bösecke (c), O. Diat (c), M. Iroing (a), V. Lombardi (b), J. Muscle Res. Cell Motility 17, 163 (1996)

(a) King's College, London (UK)

(b) Università degli Studi di Firenze, Florence (Italy)

(c) ESRF

[14] A. Bram (a), C.-I. Brändén (a), C. Craig (b), I. Snigireva (a), C. Riekel (a), J. Appl. Cryst., in print

(a) ESRF

(b) Harvard Univ., Cambridge (USA)