Ultrafast Biophysical Studies of Biomolecules at the APS

National Institute Of Diabetes And Digestive And Kidney Diseases

Investigators

Linked publications & trials

Abstract

Due to the Covid-19 pandemic, the Advanced Photon Source (APS) excluded external users for over a year, and our first opportunity to resume time-resolved x-ray studies of biomolecules arose in June 2021. This hiatus allowed us to focus on the analysis of data already acquired and work on a major overhaul of the software developed by our group to operate the BioCARS beamline. In this report, we focus on time-resolved SAXS/WAXS studies of biomolecules in solution using the pump-probe method. Briefly, a laser pulse (pump) photoactivates or thermally excites a biomolecule, after which a suitably delayed X-ray pulse (probe) passes through the sample and records a scattering pattern on a 2D detector. It is well known that the SAXS region of the scattering pattern reports on the size and shape of the biomolecule, while the WAXS region is sensitive to secondary and tertiary structure. Time-resolved SAXS/WAXS scattering patterns therefore provide 'fingerprints' that can be correlated with protein structure via molecular models and can assess which models best describe reaction pathways in solution. However, the full potential of these measurements has not yet been realized as scattering in the WAXS region is much weaker than in the SAXS region, and has required the development of experimental methodologies that not only maximize the signal-to-noise ratio of these measurements, but also identify sources of systematic error and correct for them. Our ongoing efforts to improve data analysis methodologies are quite technical, as follows. The $1.6 M Rayonix X-ray detector employed on the BioCARS beamline at the APS consists of a 4x4 tiled array of CCD chips, each of which is illuminated via a fiber optic taper whose front surface is in contact with a thin phosphor film that absorbs x-rays and emits visible photons. The fiber taper consists of a hexagonal array of light pipes that deliver visible photons from the phosphor to the CCD chip. Due to aliasing arising from registration of the hexagonal light pipes with the rectilinear pixel array, there is substantial variation in the coupling efficiency across the pixel array. To correct for coupling efficiency variation and image distortion arising from the tapered fiber bundles, Rayonix developed a proprietary method to map real pixels in the mosaic CCD array (7680x7680) to virtual pixels (3840x3840) and then apply appropriate digital gain settings to achieve uniform responsivity across the detector. The virtual images generated by the detector are invariably contaminated by so-called zingers, in which clusters of pixels randomly report signal levels well above the background, even up to saturation, and arise from radioactive decay or cosmic rays interacting with the phosphor, fiber taper, or CCD chip itself. By characterizing in detail the dark and light statistical variance for every pixel in the 3840x3840 (15 Mpixel) virtual image, we not only establish statistical criteria for identifying and rejecting zingers, but also for calculating statistically-weighted averages of measured scattering intensities, thereby providing the most accurate measurement possible with this detector. Our statistical analysis found significant variation in the pixel-to-pixel and CCD-to-CCD chip responsivity, which distort the WAXS curves and leads to artifacts in the pair-distribution functions needed to generate accurate fingerprints of secondary and tertiary structure elements. To address this problem, we developed an approach to characterize the detector non-uniformity pixel-by-pixel and use this information to normalize our scattering images prior to performing the radial integration needed to characterize the q-dependent scattering intensity. The magnitude of the uniformity correction is strongly dependent on the x-ray energy, and appears to be near a minimum around 14 keV. Starting in June 2021, when the APS relaxed its pandemic-related restrictions and allowed us to return and continue our time-resolved x-ray studies of biomolecules, we started acquiring all our data at 14 keV, instead of the 12 keV energy used in prior studies. With this new, higher energy, we acquire scattering data over a range of q spanning from 0.025 to 6.25 -1, which implies spatial resolution down to 1. One of the remaining challenges is to fully extract the structural information embedded in our time- and temperature-dependent SAXS/WAXS curves. Much effort has focused on developing methods to quantitatively account for all sources of scattering and accurately isolate the contributions that arise from the biomolecule of interest. In a recent review article entitled Time-resolved X-ray scattering studies of proteins, which was published in Current Opinion in Structural Biology in 2021, we pointed out that scattering studies published thus far have not yet achieved this goal. This effort is complicated by the fact that water accounts for most of the scattering intensity and requires accurate subtraction to isolate the far weaker contribution from the biomolecule of interest. Even when properly subtracted, the remaining scattering signal arises from both the biomolecule and its hydration shell. As pointed out in last year's annual report, time- and temperature-resolved changes in the hydration shell scattering can generate difference signals larger than those that arise from conformational changes in the biomolecule. Our efforts to develop novel methods that allow us to isolate the biomolecule scattering contribution from its hydration shell with high accuracy is ongoing. We are also developing a new, Methods-based approach for acquiring time-resolved x-ray datasets on the BioCARS beamline. Dr. Friedrich Schotte, a Staff Scientist in our research group, has developed code that operates our home-built Field-Programmable-Gate-Array (FPGA) timing system and implements a novel piano-player mode of operation. Data acquisition protocols invariably make measurements as a function of one or more dependent variables and then repeat those measurements until sufficient statistical accuracy is achieved. Our home-built FPGA timing system controls all time-critical aspects of the data acquisition sequence, and with metronome precision, repeats the sequence as datasets are being accumulated. Dr. Schotte has also written the Python code used to operate the beamline, which entails controlling all motorized stages and detectors used to properly align, focus, and deliver x-ray and laser pulses to the sample at the correct times and with properly assigned operating parameters. Our Methods-based approach seeks to markedly enhance the efficiency of beamline operations: to acquire a dataset, one simply selects a Method from a pull-down menu and then clicks a button to start the acquisition. Each Method represents an entry in a table that fully specifies all parameters required to acquire a dataset in the desired sequence. When the order of execution requires a pause to change a dependent variable, the FPGA switches from Aquire to Idle mode, and when properly set up, switches back to Acquire mode and resumes data collection. This effort has required a new software framework in which every component on the beamline that plays a role in data acquisition is controlled by a 'soft' IOC (input-output controller) that communicates across the network via the channel-access paradigm. We hope to have this new framework fully operational and commissioned for general user beamline operation by the end of the year. As our time-resolved SAXS/WAXS methodology becomes more precise and easier to use, we expect it to become an ever more important complement to time-resolved Laue studies and time-resolved optical spectroscopy studies of biomolecules, and will help provide a structural basis for understanding how biomolecules function.