Ultrafast Biophysical Studies of Biomolecules at the APS

National Institute Of Diabetes And Digestive And Kidney Diseases

Investigators

Linked publications, trials & patents

Abstract

Our time-resolved x-ray studies of biomolecules have been pursued on the BioCARS beamline at the Advanced Photon Source (APS), which shut down operations in May 2023 to implement a major upgrade (APS-U) that aims to reduce significantly the horizontal emittance of electrons circulating in the synchrotron. This effort will allow x-rays generated from those electrons to be focused to a smaller spot and will boost the coherence of those focused x-rays. However, the fill patterns required to achieve this goal contain significantly fewer electrons in each bunch. For example, the APS-U will no longer support their hybrid mode of operation, which will reduce by about a factor of four the flux available in a single x-ray pulse. This reduction in the single-shot signal level achievable in time-resolved studies will require longer integration times to generate the same quality data. On the plus side, the spectral bandwidth of the x-rays is expected to be modestly narrower than before, which will improve somewhat the precision of the SAXS scattering pattern. In 2025, the APS resumed limited operations, and we expect to resume our time-resolved x-ray scattering studies on the BioCARS beamline in late 2025. In the meantime, we have been focusing on x-ray detectors, how to extract the highest quality scattering data possible from them, and beamline control software. The x-ray detector is a critical component in our time-resolved SAXS/WAXS studies and we have put significant effort into image processing approaches that properly flag and eliminate data distortion from zingers, properly account for all sources of scaling corrections, including a uniformity correction to compensate for pixel-to-pixel scaling errors, and finally, when converting the 2D x-ray image into a one-dimensional scattering curve, produce a statistically-weighted average of the measured scattering intensities from all pixels assigned to each q-bin. The current workhorse on the BioCARS beamline is a Rayonix MX340-HS x-ray detector, a technological wonder that has in the past generated 4-5 TB of image data per week of x-ray beamtime. Thanks to its large size, we can acquire time-resolved SAXS/WAXS data over a range of q spanning 0.02 to 5.25 Å-1, covering both SAXS and WAXS regions with a single detector. Unfortunately, its point spread function includes a weak, broad pedestal that contributes to scaling errors near edges and requires correction to generate accurate scattering amplitudes at very low q. Moreover, the image consists of virtual pixels whose effective gain varies substantially from pixel-to-pixel across each CCD chip, which we characterize in terms of variance per count (VPC), a detector-specific statistic. Knowing VPC, we can properly flag and correct for zingers that contaminate the images. We also correct for scaling errors using a uniformity correction (UC) generated from experimental datasets. When corrected for zingers and properly rescaled with UC, the signal-to-noise ratio achieved with this detector is quite impressive. For example, when averaging 160 bursts of a 500 ns duration x-ray pulse train, all within 1 second, the signal-to-noise ratio (S/N) of the one-dimensional scattering curve, which represents a statistically weighted average of all pixels assigned to each q-bin, is well over 1000:1 in the WAXS region. Since proteins at 20 mg/ml occupy approximately 1.4% of the solution volume, most of the measured scattering intensity in the WAXS region comes from water and must be subtracted to isolate the biomolecule WAXS scattering signal. Hence, the ability to achieve such high S/N in the WAXS region is crucial to extract from the WAXS scattering signature information about a biomolecule’s secondary structure. We are finalizing a manuscript describing methods we developed to characterize the detector properties pixel by pixel, how to flag zingers, properly scale the data to achieve shot-noise-limited detection sensitivity, and how to put our scattering data on an absolute scale to achieve accurate comparison and/or differencing of datasets acquired at different times. The manuscript is entitled ‘Absolute scaling of small and wide-angle x-ray scattering images recorded with short duration x-ray pulses on a large area fiber-taper x-ray detector’, Hyun Sun Cho, Friedrich Schotte, and Philip A. Anfinrud, and we intend to submit it to the Journal of Synchrotron Radiation. BioCARS recently acquired an ePix10k 2M x-ray detector from SLAC that was developed specifically for x-ray free-electron laser applications. Being a direct detector, its point-spread function is pixel-size limited and smaller than that achieved with the Rayonix detector. Moreover, rapid readout following each trigger pulse makes this detector virtually zinger free. However, the ePix detector dimension is approximately half that of the Rayonix detector. Is it possible to use this relatively small detector and still retain access to the WAXS region? The short answer is, yes. We designed and fabricated a kinematic detector support that allows the detector to be locked down in a conventional on-axis fashion with a single thumb screw. However, it can also be locked down in an alternative (unconventional), 45-deg rotated and offset configuration, with the x-ray beam centered near the middle of the bottom module. With this configuration, the scattering range achieved spans from 0.02 to over 5 Å-1. Are there any negatives? The ePix detector is not a commercial detector, and the software provided to operate it was found to be quite kludgy. Hence, we are developing our own software to make this detector function similar to an integrating detector by performing on-the-fly averaging of background-subtracted images, writing files in a conventional image format, and provide live access to images as they are acquired. These innovations should make this detector quite useful for our time-resolved x-ray scattering studies and also those of other outside users of BioCARS. Based on preliminary tests performed with this detector, it is likely to become the detector of choice in future time-resolved x-ray scattering studies of biomolecules. Polyphony, a new software suite developed by Dr. Schotte to operate the time-resolved BioCARS beamline at the APS is nearing completion. Polyphony aims to control all time-critical components needed for time-resolved studies in a harmonious fashion. Polyphony employs what we call a ‘Methods’ based approach for operating the beamline, with the user-selected Method completely specifying all control parameters required to acquire a dataset. User specified sequencing of the data acquisition is accomplished in the form of nested loops in which the innermost loop specifies the fast variable, such as the time delay between laser and x-ray pulses. The number of nested loops is dependent on the number of variables the user wishes to vary during data acquisition, such as sample temperature or phi orientation, with the outermost loop being a Repeat loop that executes all inner loops a user-specified number of times, thereby facilitating signal averaging. Key to this approach is writing code that controls individual hardware devices in a consistent, event-driven fashion with standards that ensure properly timed operation. This Methods based approach is extensible. Provided software written to control new hardware conforms to a set of standardized Polyphony ‘rules’, control of new hardware can be achieved by simply adding a new column in the Methods table and specifying appropriate control parameters as a string entry in that table. Hence, integrating new hardware into Polyphony can be accomplished solely with configuration parameters, i.e., no new coding is required. This new approach is currently being tested by our group with an aim to deploy it for all BioCARS users in the coming year. The hardware and software we have developed for use on the BioCARS beamline not only benefits our research, but that of other outsider users of that facility, many of whom are NIH-supported investigators. As our time-resolved methodology becomes more precise and easier to use, it will become an ever more important and powerful complement to other experimental methods and should help provide a structural basis for understanding how biomolecules function from a mechanistic point of view. This fundamental understanding of biomolecule function should help promote a more rational approach to drug development with significant benefits to human health.