XDL2011 Workshop 2 Abstracts

Biomolecular Structure from Nanocrystals and Diffuse Scattering
Monday, June 13th - Tuesday, June 14th, 2011

---

Organizers: Ed Lattman (Hauptmann-Woodward Medical Research Inst.), Mavis Agbandje-McKenna (University of Florida), Keith Moffat (University of Chicago), & Sol Gruner (Cornell University)

Workshop Agenda (html)

---

Energy Recovery Linac (ERL) and Ultimate Storage Ring (USR) Properties

Don Bilderback
Cornell University

ERLs and USRs are under consideration for next generation, high-duty cycle (>MHz rep rates), coherent x-rays sources. They both feature extremely high average spectral brightness, diffraction-limited performance and are the response to 3rd generation storage ring users/developers who would like "more x-ray flux, smaller x-ray beam size, more coherent x-ray flux on sample, higher energy resolution probes and/or short pulses for repetitive probing". In most cases, the x-ray beam should minimally impact the sample under study. We review the general features of ERLs and USRs in the context of the Cornell ERL and PEP-X as two well-developed examples from the x-ray community. Additionally, examples of utilization of the advanced properties of these machines will be given. The first example explores using the timing structure to repetitively probe the response of x-ray excited optical luminescence. The second uses the extremely high average brightness to conceptually develop ideas toward a confocal x-ray microscope that is designed to image single atoms based on either Thompson scattering or x-ray fluorescence.

X-ray Detectors: State-of-the-art & Future Possibilities

Sol Gruner
Cornell University

The state-of-the-art of quantitative imaging x-ray detectors is described. We then consider upcoming technologies that may be applied to imaging detectors to advance the state-of-the-art with respect to pixel size and functionality, spatial resolution, time resolution, analog dynamic range, and energy resolution. Specifically, we look into our crystal ball and ask what is likely feasible on a decade time scale, given adequate R&D, given current physical limits of materials and technology.

Nanocrystals, Injectors and Correlations for an ERL

John Spence
Arizona State University & Lawrence Berkeley National Laboratory

It has been shown that, for a model system such as Lysozme, the structure may be solved at Chess using just five nanocrystals, if the efficient Laue method is used together with phasing by molecular replacement (MR)¹. Nanocrystals of Photosystem 1 have also been solved recently² using a continuous liquid jet, and the method of snapshot diffraction, at the LCLS by the methods of Monte-Carlo data merging and MR. Several related issues emerge.

(i) Since an ERL will produce in 10 microsecs the same number of photons as the LCLS does in 10 femotseconds³, I will discuss the design of a new "toothpaste" low velocity containerless lipid cubic phase injector which may allow the more efficient Laue method to be used to solve membrane-protein nanocrystals at ERL, by considering the host buffer viscosity, the rotational diffusion time of the proteins and the time structure of the beam (eg 1 ps pulses at 100 kHz rep rate for ERL, high charge mode).

(ii) Time-resolved nanocrystallography was first attempted at LCLS in June 2010 using our liquid jet particle injector (spraying randomly oriented submicron nanocrystals across the beam), however the results depend on accuracy in merging these partial reflection snapshots, which use a fairly monochromatic and collimated beam, by the Monte-Carlo method⁴. The Laue method is normally used for time-resolved work with a laser trigger, however chirping the LCLS beam will not provide enough energy spread for this to be effective. As an alternative, we plan to use the convergent beam method (CBM)⁵, under these new conditions of full coherence, similar to those used for Ptychography⁶. I will compare coherent and incoherent merging of CBM intensities in order to determine the number of nanocrystals needed in each case, for comparison with Laue mode at ERL.

(iii) The remarkable approach of Z. Kam for ab-inito inversion of SAXs data (without modeling) has recently been shown to work well experimentally for inorganic particles differing in orientation by rotation about a single axis⁷. I will discuss the use of this approach for merging nanocrystal data in the form of "spotty rings", with a few nanocrystals per exposure. I also discuss the dependence of signal-to-noise ratio in snapshot SAXS on number of particles⁸.

References:

1. Cornaby; S. Acta Cryst D66, 2 (2010)
2. Chapman, H. et al.; Nature 470, 73 (2011)
3. Bilderback, D. et al.; NJP 12, 035011 (2010)
4. Kirian, R. et al.; Acta Cryst. A67, 131 (2011)
5. Ho, J.X. et al.; Acta Cryst. D58, 2087 (2002)
6. Spence, J. and Zuo, J.M.; "Electron Microdiffraction", Plenum (1992)
7. Saldin, D. et al.; Phys Rev Letts. 106, 115501 (2011)
8. Kirian, R. et al.; Phys Rev E (2011) Submitted

The Challenge of Novel, Nanoscale Biological Samples

Alex McPherson
University of California, Irvine

New X-ray sources like the FELS and the ERLs that can provide extremely high X-ray flux densities in micro, and perhaps even nanobeams open a wide range of applications to novel specimens and the attack of challenging biological problems. The conformations of nucleic acids packed within icosahedral viruses, heretofore not seen, is within reach. In situ diffraction of nanofibers, crystals and ordered arrays of macromolecules as exemplified by the hemoglobin fibers that distort sickled blood cells, and the insulin crystals that appear in pancreatic tissue are now possible. Other samples previously to small in physical size to be analyzed with current approaches, such as amyloid fibers associated with dementia, and crystals of both soluble and membrane proteins can similarly be addressed. Along with these new possibilities and challenges come the problems of how to produce or obtain the samples, how to present them to the X-ray source, and how to analyze the diffraction data they produce. Some ideas for obtaining samples, crystalline and fibrous, and exposing then efficiently and in large numbers will be discussed.

Microfluidics to Produce and Manipulate Microcrystals

Seth Fraden
Brandeis University

There is no guarantee that a given protein has a crystalline phase, but if it does then there is a range of equilibrium crystallization conditions, which are found through extensive screening followed by optimization around hits. However, the transformation of a protein solution to a protein crystal is a non-equilibrium process; "nucleation and growth". Consequently, supersaturation kinetics plays an essential role in crystallization. Therefore the optimal crystallization strategy should additionally screen kinetic trajectories involving non-thermodynamic variables that affect supersaturation, such as depth and duration of supersaturation, as well as sample volume.

To address crystallization kinetics we are developing a technology based on emulsion microfluidics in which 1 nl drops of protein solution are encapsulated in oil and stabilized by surfactant. Microfluidic devices process thousands of drops along different kinetic paths simultaneously by varying both temperature and concentration of the protein solution. The objective is to find conditions in which one crystal is grown per drop. As a consequence of the technology, a single crystal will be too small to collect a complete diffraction data set for structure determination. A sample handling device will be built to feed crystals into a synchrotron beam one at a time. This will place extreme demands on the x-ray sample cell, data acquisition, and analysis systems. Proof of principle experiments will be presented.

Data Collection from Nanocrystals with Reduced Radiation Damage

Robert Fischetti
Argonne National Laboratory

The small, intense X-ray beams available at 3rd generation light sources have been exploited by structural biologists to determine the structure of increasing larger or more complex macromolecules. The crystals of many of these macromolecules may have a largest dimension of only 5-10 microns, and may diffract poorly due to lack of internal order. Obtaining data of high signal-to-noise requires exposing the crystal to a beam of high flux density, resulting in increased absorbed dose and radiation damage. Although cryo-cooling of protein crystals significantly reduces X-ray induced radiation damage, it does not eliminate the damage.

The predominant mechanism of interaction of an X-ray with a low-Z atom in the crystal is the emission of a photoelectron, which carries away most of the energy of the incident X-ray. When the emitted photoelectron scatters off another atom, it loses energy to the atom resulting in local damage. As the photoelectron energy decreases, the probability of interacting with yet another atom increases causing more frequent interactions until finally the photoelectron is recaptured. Thus, if the X-ray beam size is small compared to the distance the photoelectron travels from its point of emission, then deposition of photoelectron energy outside the beam footprint may reduce radiation damage inside the beam footprint. Monte-Carlo simulations predict that a photoelectron of typical energy could travel 4 - 5 microns from the point of emission before being absorbed. We studied radiation damage to lysozyme crystals by monitoring the diffracted intensity of 18.5-keV X-rays as a function of dose and beam size (0.86 - 15.6 micron) at beamline 23-ID-B at the Advanced Photon Source. We observed a 3-fold reduction of damage per dose within the footprint of the smallest compared to the largest beam. In addition, the spatial extent of radiation damage was mapped using both 15.1- and 18.5-keV X-rays and a ~1-micron beam. The damage profiles displayed spatial anisotropy with greater damage occurring along the direction of the X-ray polarization, as expected. The spatial extent of the damage was limited to about 4 microns. How we can exploit this reduced damage to collect data from "nanocrystals" will be discussed.

Predicting and Processing Nanocrystal Diffraction Data

James Holton
Lawrence Berkeley National Laboratory

The use of "coherent" beams from X-ray lasers and very small crystals starts to raise questions about which of the various assumptions made in traditional crystal diffraction theory are no longer valid. For example, at what point does a "nanocrystal" become so small that it diffracts like a single molecule? Does the diffraction pattern with a "coherent" x-ray beam look any different than it does with a "regular" x-ray beam? Is the intensity diffracted from a perfect nanocrystal really brighter than what we observe with larger, less perfect crystals? Where, exactly, does one "mosaic domain" end and the next one begin? To answer these questions, a first-principles total scattering simulator called nearBragg was created. This program simply calculates the linear distance from one or more points in the "source" to each point "atom" in the sample and then on to the center of a pixel on the detector, and adds up the sine waves that result.

No assumptions are made about unit cells, crystallinity, "coherence lengths", or even Bragg's law, and indeed even near-field geometries may be investigated with nearBragg. The "coherence" of the source is implemented simply by a user-selectable switch to add up either the squared amplitude of the sine waves on each pixel or to add all the waves as complex numbers. Using this formalism, the answers to the above questions appear to be: there is no practical difference between crystal diffraction patterns with "coherent" and "regular" beams. Nanocrystals have the potential for better signal/noise from partial reflections, but not fulls. Small crystals experience Scherrer broadening of their spots as the number of unit cells approaches one, and a "mosaic domain" is identical to the "coherence length" of the camera: ~3-10 microns, depending on the detector distance. Drawing on these findings, the new fastBragg program implements much quicker routes to the spot shapes obtained using nearBragg, fast enough perhaps to obtain structure factors from "stills" by directly fitting a total scattering model to the entire diffraction image.

G Protein Coupled Receptor Structure Determination Enabled by Microdiffraction Technology

Roger Sunahara
University of Michigan

For decades efforts to obtain structural information of membrane proteins have been fraught with problems with crystallogenesis and G protein-coupled receptors (GPCRs) are no exception. However, the application of a combination of protein engineering, lipidic cubic phase technology and microcrystallography have had led to significant advances in the last three or four years. Here we present the latest major advance using similar technologies to solve the structure of a GPCR in a complex with a heterotrimeric G protein. An agonist-bound β2-adrenergic receptor was isolated in a complex with its cognate heterotrimeric G protein, the stimulatory G protein, Gs, in its nucleotide-free form. The crystal structure, together with data from single particle reconstructions from electron microscopy and deuterium exchange mass spectroscopy analysis, reveal significant changes in the G protein α-subunit and aid in the elucidating the mechanism for receptor-mediated nucleotide exchange.

Into the Future - a structure biologist's dreams

Ilme Schlichting
Max Planck Institute, Heidelberg

Electron microscopy and X-ray crystallography provide unique insight into the architecture of cells, molecular assemblies and (macro)molecules. New X-ray sources such as ERLs and FELs promise to enable a plethora of new experiments, for example by allowing analysis of nanocrystals as has been shown recently in a proof of principle experiment. The high coherence of the X-ray beam also allows analysis of single particles by imaging or correlation analysis. Some potential applications in studying self assembly reactions and cellular imaging and the associated challenges will be presented.

Toward Rational Crystallization for Structure-Function Studies of Membrane Proteins

Martin Caffrey
Trinity College, Ireland

One of the primary impasses on the route that eventually leads to membrane protein structure through to activity and function is found at the crystal production stage. Diffraction quality crystals, with which structure is determined, are particularly difficult to prepare currently when a membrane source is used. The reason for this is our limited ability to manipulate proteins with hydrophobic/amphipathic surfaces that are usually enveloped with membrane lipid. More often than not, the protein gets trapped as an intractable aggregate in its watery course from membrane to crystal. As a result, access to the structure and thus function of tens of thousands of membrane proteins is limited. In contrast, a veritable cornucopia of soluble proteins have offered up their structure and valuable insight into function, reflecting the relative ease with which they are crystallized. There exists therefore an enormous need for new ways of producing structure-quality crystals of membrane proteins. One such approach makes use of lipidic liquid crystalline phases (mesophases)¹. In my presentation, I will describe the method, its successes and how the Energy Recovery Linac and Ultimate Storage Ring might be used to establish its molecular underpinnings for a more rational approach to crystallogenesis and high resolution structures where interactions that are integral to human health are revealed.

References:

1. Caffrey, M.; "Crystallizing Membrane Proteins for Structure Determination. Use of Lipidic Mesophases", Annual Review of Biophysics 38:29-51 (2009)

This work was supported by grants from the Science Foundation Ireland and the National Institutes of Health.

Membrane Proteins and Membrane Potentials

Doug Rees
California Institute of Technology

By providing a boundary that separates the cellular interior from the external environment, membranes serve to establish voltage and concentration gradients across the bilayer that may be used to regulate the activities of membrane proteins. While the pace of structure determination is increasing, an understanding of how membrane proteins respond to imposed gradients in terms of structure and particularly dynamics has been more challenging to acquire, since the architectures of two- and three-dimensional crystals are generally incompatible with gradient formation. Time resolved structural studies characterizing the response of membrane proteins to changes in transmembrane gradients would be quite informative for establishing the mechanistic details of this fundamental event in cellular signaling and regulation. The preparation of polyhedral proteoliposomes, together with diffraction based analyses of transmembrane electrical and concentration gradients, provide potential approaches to such studies that would greatly benefit from the next generation of light sources.

Non-Bragg Scattering from Protein Crystals

George Phillips
University of Wisconsin, Madison

Proteins are designed to be flexible and dynamic as a part of their biological functions. Even when cooled to cryogenic temperatures, they are generally trapped in minima that are parts of broad conformational landscapes of similar potential energies. These ensembles of conformations are sampled by crystallization, which affects the nature of the conformations. Coarse-grained modeling (less detailed than full atomic treatment) of the dynamics of proteins in crystals can be calibrated to match not only the experimentally refined crystallographic B-factors, but also to the diffuse scattering from protein crystals. Thus reasonable density of states can also be obtained from the models. These models should be useful in understanding and interpreting diffraction patterns from nanocrystals and single particles, where the samples are not well approximated by infinite lattices.

Next Generation Solution Scattering

Lee Makowski
Northeastern University

What advantages will next generation sources bring to solution scattering from biological macromolecules? SAXS and WAXS are currently undergoing rapid development due to the high quality of data generated by current sources and the increasing impact of computation that allows production of low resolution molecular envelopes and the testing of atomic coordinate sets at higher resolutions. The sensitivity of solution scattering to molecular fluctuations is increasingly being used to estimate the extent of intramolecular motions and the effect of mutations, ligand binding and environmental factors on that motion. Anomalous scattering is providing the radial position of individual atoms. Robotic sample handling is providing high throughput and use of much smaller volumes of material. Very short, intense x-ray pulses promise to provide time resolution adequate to observe angular correlations in scattering, greatly enhancing the information content of data sets.

We may anticipate that probing the details of scattered intensity - that are currently averaged in time and space - will raise solution scattering to a new level of utility in characterization of the structural and dynamical basis of biomolecular function.

Time-resolved Scattering of Proteins in Solution: new opportunities for an ERL

Philip Anfinrud
National Institutes of Health

To generate a deeper understanding into the relations between protein structure, dynamics, and function, it is crucial to study protein structural dynamics in solution. We have developed time-resolved small- and wide-angle X-ray scattering (SAXS/WAXS) methods capable of probing changes in protein structure on time scales as short as ~100 ps. This infrastructure was first developed on the ID09B time-resolved beamline at the ESRF, and more recently on the ID14B BioCARS beamline at the APS. In these studies, a picosecond laser pulse first photoexcites a protein and triggers a structural change, then a suitably delayed picosecond X-ray pulse passes through the protein. Scattered X-rays are imaged on an integrating 2D detector, which is read out after 1100 repeated pump/probe measurements. The q-dependent intensity of these images responds to changes in the size, shape, and structure of the protein. These time-resolved scattering "fingerprints" provide stringent constraints for putative models of conformational states and structural transitions between them. Might an ERL prove a superior source for these investigations? If so, how might we best take advantage of its unique properties? These questions will be discussed in the context of current state-of-the-art time-resolved SAXS/WAXS capabilities at the APS.

This research was supported in part by the Intramural Research Program of the NIH, NIDDK.

Small-angle Scattering from Biological Solutions: potential of the ERL/USR sources

Dmitri Svergun
European Molecular Biology Laboratory

Small-angle X-ray scattering (SAXS) experiences a renaissance in the studies of macromolecular solutions allowing one to study low resolution structure of native particles and to rapidly analyze structural changes in response to variations in external conditions. Novel data analysis methods significantly enhanced resolution and reliability of structural models provided by the technique. Emerging automation of the experiment, data processing and interpretation make solution SAXS a streamline tool for large scale structural studies in molecular biology. The new possibilities of biological solution SAXS due to the advent of Energy Recovery Linacs (ERLs) and Ultimate Storage Rings (USRs) will be discussed. The highly coherent and extremely intense beams of nanometer size and femtoseconds-scale time structure allow for the experiments, which are not yet feasible on the third generation sources.

The new experiments include, in particular, cryo-SAXS analysis on flash-frozen solutions. Rapid freezing yielding vitreous ice probes allows one to significantly diminish the radiation damage and also to study the time course of biological and chemical reactions by stalling them at the defined time points. The problems of the use of flash-frozen samples will be discussed in light of the opportunities provided by ERLs and USRs.

The ERLs/USRs are expected also to have a major impact on the use of protein purification methods such as size exclusion chromatograpy (SEC) inline with the SAXS experiment. SEC purification becomes increasingly popular now allowing one to study highly purified solutions and transient complexes. The major problem of the approach is extremely low amounts and concentrations of the solutes coming out of the columns, such that reliable signals are often difficult to record with conventional sources. The intense and small ERLs/USRs beams may allow means to overcome the problem e.g. by employing microfluidic lab-on-a-chip devices with the micron-sized channels for electrodynamic focusing of the protein flow.

Biological Opportunities with Solution Scattering

Brian Crane
Cornell University

Small-angle x-ray Scattering (SAXS) is a proven technique for defining the shapes, oligomeric states and globular properties of macromolecules. We have found SAXS to be particularly useful for defining the component juxtapositions and overall symmetries of multi-domain proteins and higher order complexes. In addition, time-resolved measurements can provide structural information on transient protein complexes. With the potential for short, bright and coherent x-ray pulses delivered by an Energy Recovery Linac (ERL) we encounter new possibilities for characterizing molecular structure in solution. This talk will explore potential applications of bright, coherent x-rays for solution scattering of macromolecules in partially ordered samples, with an eye toward molecular alignment techniques and the opportunities they may bring.