Adaptable SEC‐SAXS data collection for higher quality structure analysis in solution

Abstract

The two major challenges in synchrotron size‐exclusion chromatography coupled in‐line with small‐angle x‐ray scattering (SEC‐SAXS) experiments are the overlapping peaks in the elution profile and the fouling of radiation‐damaged materials on the walls of the sample cell. In recent years, many post‐experimental analyses techniques have been developed and applied to extract scattering profiles from these problematic SEC‐SAXS data. Here, we present three modes of data collection at the BioSAXS Beamline 4–2 of the Stanford Synchrotron Radiation Lightsource (SSRL BL4‐2). The first mode, the High‐Resolution mode, enables SEC‐SAXS data collection with excellent sample separation and virtually no additional peak broadening from the UHPLC UV detector to the x‐ray position by taking advantage of the low system dispersion of the UHPLC. The small bed volume of the analytical SEC column minimizes sample dilution in the column and facilitates data collection at higher sample concentrations with excellent sample economy equal to or even less than that of the conventional equilibrium SAXS method. Radiation damage problems during SEC‐SAXS data collection are evaded by additional cleaning of the sample cell after buffer data collection and avoidance of unnecessary exposures through the use of the x‐ray shutter control options, allowing sample data collection with a clean sample cell. Therefore, accurate background subtraction can be performed at a level equivalent to the conventional equilibrium SAXS method without requiring baseline correction, thereby leading to more reliable downstream structural analysis and quicker access to new science. The two other data collection modes, the High‐Throughput mode and the Co‐Flow mode, add agility to the planning and execution of experiments to efficiently achieve the user's scientific objectives at the SSRL BL4‐2.

1. INTRODUCTION

Small‐angle x‐ray scattering (SAXS) using synchrotron radiation is a very versatile technique applicable to a broad range of biological samples for studying structure and dynamics at different spatial and temporal resolutions. Solution SAXS is one of the principal methods in multidisciplinary structural analysis of biological macromolecules, complementing other methods that provide structural information in solution. In many cases, the SAXS data used for the analysis must be “clean” data in the sense that the scattering signals were obtained from monomodal and monodisperse samples without any problems related to radiation damage, background mismatch, and so forth (Jeffries et al., 2016). Today, SEC‐SAXS (size‐exclusion chromatography coupled in‐line with SAXS) has become a major technique in solution scattering, as it is ideally suited to overcome the problem of sample heterogeneity (Chaudhuri et al., 2017; Lattman et al., ²⁰¹⁸; Svergun, ²⁰¹³). Recent years have seen further significant advancements in the development of chromatography coupled the SAXS data collection, such as the development of combined SEC‐SAXS and ‐WAXS at the Taiwan Photon Source or the combination of other chromatography techniques with SAXS (Hopkins et al., 2017; Hutin et al., ²⁰¹⁶; Konarev et al., ²⁰²²; Meisburger et al., ²⁰²¹; Shih et al., ²⁰²²).

A major experimental challenge for SEC‐SAXS is the sample cell fouling problem: Continuous exposures by the high‐flux synchrotron x‐ray beam during the SEC elution causes the accumulation of radiation‐damaged materials at the beam position on the inner walls of the sample cell. The materials that can cause this problem can be the sample of interest, contaminants in the sample solution, buffer components, or column debris. The fouling makes it challenging to perform proper background subtraction because the fouling causes additional parasitic scattering signals that are prominently detected at low angles.

To minimize these problems, several post‐experimental approaches to correct the baseline have been developed and used recently. The program DELA reduces common subtraction artifacts using Guinier‐optimized buffer subtraction (Malaby et al., 2015). The program US‐SOMO provides two correction methods: a simple linear baseline correction and the integral baseline correction, ideal for instrumental drift and fouling on the sample cell, respectively (Brookes et al., 2016). The program CHROMIXS implements automatic selection of the buffer region from the entire SEC fractions (Panjkovich & Svergun, 2018). The program MOLASS corrects the baseline drift using a low percentile method (Yonezawa et al., 2023). Most of the programs are part of larger biological SAXS software packages, so they are readily available to general users. The wide distribution of these programs reflects the widespread nature of this problem and the lack of a unique approach to correct for the sample cell fouling problem.

Another challenge encountered in SEC‐SAXS experiments originates from overlapping of peaks or insufficient separation between species coming off the size‐exclusion column. It should be noted that the resolution of the sample separation on SEC‐SAXS depends primarily on the quality of the SEC. Therefore, the use of appropriate instruments and columns, as well as optimization of the flow path, and so forth, play a key role in species separation. In cases where overlapping peaks are observed in SEC‐SAXS data sets, the peak decomposition approach can be used to extract individual scattering profiles. The following five decomposition algorithms have been recently developed and used. First, US‐SOMO provides symmetrical and non‐symmetrical Gaussian decomposition tools (Brookes et al., 2016). Second, the DELA allows reconstruction of the sample scattering profile using singular value decomposition (SVD) and Guinier‐optimized linear combination (Malaby et al., 2015). Third, the evolving factor analysis (EFA) is a model‐free decomposition approach employed in the programs EFAMIX and BioXTAS RAW (Hopkins et al., 2017; Konarev et al., ²⁰²²; Meisburger et al., ²⁰¹⁶). Fourth, the program REGALS performs both baseline correction and decomposition using simple expectations as prior knowledge (Meisburger et al., 2021). Fifth and perhaps most rigorous, the program MOLASS performs peak decomposition while taking into account the concentration dependence of the sample, which is another major concern when extracting scattering profiles of biological samples (Yonezawa et al., 2023). Again, the diversity of algorithms for contending with overlapping species reflects that this problem is a common occurrence and has no unique solution.

Besides such post‐experimental computational approaches, the co‐flow method has been developed at the Australian Synchrotron to physically minimize the buildup of fouling on the sample capillary (Kirby et al., 2016). Under the non‐uniform velocity distribution of laminar‐flow, the sample solution flowing off the SEC column is constrained to the center of the capillary while it is surrounded by a flow of a matched buffer. This approach isolates the sample from the inner walls where the flow velocity is slowed by the laminar flow profile, preventing exacerbation of fouling problems due to the longer detention time at the x‐ray exposure position. Another solution for samples sensitive to radiation damage is the stopped‐flow SEC‐SAXS measurement using a home source instrument with a ~ 10,000 fold less intense x‐ray source (Inoue et al., 2019).

In the following, we first introduce the three SEC‐SAXS data collection modes available at the SSRL BL4‐2, and then we specifically focus on the high‐resolution mode. This high‐resolution mode allows SEC‐SAXS data collection with high sample selectivity delivered by the UHPLC and analytical SEC column. Virtually no further peak broadening was confirmed between the UHPLC UV detector and the x‐ray position. The opening of the x‐ray shutter during the SEC elution can be regulated in two different ways: by indicating the number of images and by setting a threshold of the absorbance value. These options avoid extra fouling on the sample cell by minimizing exposure of unnecessary fractions to the x‐ray beam. Combined with additional cleaning of the sample cell after the buffer data collection, sample data collection can be performed with a clean sample cell, allowing baseline correction‐free analysis. This feature facilitates more reliable downstream data analysis and accelerates new science. The two other data collection modes at SSRL 4–2, the high‐throughput mode and the co‐flow mode, add agility to the design of SAXS experiments to meet the user's scientific objectives. All options described above are also operable in these last two modes, which we intend to discuss in detail in our forthcoming publication. This article should be of interest to molecular biologists who are planning SEC‐SAXS experiments.

2. SEC‐SAXS INSTRUMENTATION

2.1. X‐ray instruments and UHPLC

The beamline 4–2 at the SSRL (SSRL BL4‐2) is dedicated to small‐angle x‐ray scattering and diffraction techniques in structural biology and biophysics (Smolsky et al., 2007). The beamline uses the central part of the radiation fan produced by a 20‐pole, 2 T magnet field wiggler as its source. A rhodium coated mirror at the mid‐point between the source and the detector provides a 1:1 image of the source at the detector plane. A liquid nitrogen cooled Si(111) double‐crystal monochromator provides monochromatic beam in the energy range of 6.5–17 keV. The beamline features a segmented pinhole camera with discretely adjustable sample‐to‐detector distances ranging from approximately 0.25–3.5 m in seven steps. The adjustable distance enables the optimization of the experimental configuration for the specific size of the sample to be measured. The standard configuration at the beamline has 11 keV radiation and a 1.7 m sample‐to‐detector distance that supports measuring samples with an R _g, radius of gyration, range of 20–60 Å. This range corresponds to a sample molecular weight of 30–300 kDa for globular proteins. The x‐ray beam size of 0.3 mm × 0.3 mm (slits collimated) with a photon flux of approximately 2 × 10¹² photons per second at the sample position is part of the standard configuration. The beam size can be further collimated if needed at the expense of a change in flux.

The schematic layout of the SSRL BL4‐2 SEC‐SAXS environment is shown in Figure 1. The SEC‐SAXS system shares both the sample cell assembly and the liquid delivery system of the BL4‐2 SAXS Autosampler used in static SAXS experiments (Martel et al., 2012). The sample cell is composed of a thin‐walled crystallographic quartz capillary (1.0–1.3 mm in inner diameter (ID), Hampton Research Inc., Aliso Viejo, CA) and is sufficiently tolerant to pulsation caused by the UHPLC pump to prevent potential background mismatch due to changes in sample path length during data collection. A dual syringe dispenser is connected to various solutions (water, 2% Hellmanex® III, 50% ethanol, 30% bleach, SEC buffer, etc.) for rigorous cleaning of the sample cell.

FIGURE 1.

Overview of the SEC‐SAXS setup at the SSRL BL4‐2. The exposure of the sample cell is controlled by the x‐ray shutter (bottom center), and the intensities of the scattered x‐rays are measured on the area detector through the vacuum path between the sample position and the detector (top right). The flow path of the UHPLC, through which the sample separated by the SEC flows, lead to the sample cell. The sample cell is also connected to the dual syringe dispenser for cleaning the sample cell. The details of the flow paths for three SEC‐SAXS modes are shown in Figure 2. Synchronization between the x‐ray instrument and the UHPLC allows automated continuous SEC‐SAXS data collection of, currently, up to 50 samples.

TABLE 1.

Summary of SEC‐SAXS data collection.

	Figure 6a	Figure 6b	Figure 6c	Figure 7a	Figure 7b	Figure 7c	Figure 7d	Figure 7e	Figure 7f
Experiment
Sample	BSA	BSA	BSA	Lysozyme	Lysozyme	Lysozyme	Lysozyme	Lysozyme	Lysozyme
SEC‐SAXS mode	HR	HT	HR	HR	HR	HR	HR	HR	CF
Beam energy (keV)	11	11	11	11	11	13	11	11	11
Exposure time (seconds)	1	1	1	2	2	2	2	2	2
Use of x‐ray shutter control	n.a.	n.a.	Applied	n.a.	n.a.	n.a.	Applied	Applied	n.a.
Use of 5 mM DTT	n.a.	n.a.	n.a.	n.a.	Applied	n.a.	n.a.	Applied	n.a.
Sample Conc. (mg/ml)	40	40	40	30	30	30	30	30	60
Sample volume (μl)	10	10	10	10	10	10	10	10	10
Image number used for averaging	343–347	319–323	343–347	435–439	440–444	435–439	455–459	435–439	405–409
Est. Conc. at peak (mg/ml)	3.1	1.9	3.1	4.0	3.9	3.4	4.0	3.9	3.1
Dilution factor	12.9	21.1	12.9	7.5	7.7	8.8	7.5	7.7	19.4
Analysis
q min (Å−1)	0.0072	0.0072	0.0072	0.0221	0.0146	0.0122	0.0144	0.0157	0.0129
q max (Å−1)	0.29	0.28	0.29	0.45	0.45	0.45	0.45	0.45	0.45
Guinier R g (Å)	27.74	28.3	27.78	14.66	14.36	14.51	14.57	14.37	14.39
Guinier I(0) (cm−1)	0.148	0.095	0.144	0.029	0.028	0.028	0.030	0.029	0.023
P(r) R g (Å)	27.74	28.3	27.7	14.66	14.36	14.51	14.57	14.37	14.39
P(r) I(0) (cm−1)	0.148	0.095	0.144	0.029	0.028	0.028	0.030	0.029	0.023
D max (Å)	82.0	86.0	82.0	46.0	43.2	47.0	45.8	44.3	45.6

Abbreviations: CF, Co‐Flow mode; HR, High‐Resolution mode; HT, High‐Throughput mode.

The ThermoFisher UltiMate 3000 UHPLC was employed and positioned adjacent to the BL4‐2 SAXS Autosampler in the hutch. Depending on the sample size, the small analytical SEC column (2.4 mL bed volume) of Superdex 75 Increase, Superdex 200 Increase, or Superose 6 Increase columns (Cytiva, Marlborough, MA) can be chosen for maximal separation of the sample of interest. The small internal volume of the UHPLC minimizes the sample dilution prior to sample injection onto the column, thereby enhancing its sample selectivity. The small bed volume of the column minimizes sample dilution during SEC, enabling SEC‐SAXS data collection at high sample concentrations. The 9 mm screw cap micro vial (<300 μL with 4 μL volume residual) is employed as the sample container (Figure 1). The UHPLC Autosampler is capable of injecting the sample from the vial onto the column in the range of 0.1–100 μL (±0.5% accuracy at 20 μL injection). The UHPLC Autosampler has a carousel that holds three sample racks (3 × 40 vial trays). The racks can store many sample vials at a desired temperature (4–45°C). SEC‐SAXS data collection in a nearly anaerobic environment is feasible through the use of the lined vial cap after the sample solution has been carefully sealed in the vial with the cap in an anaerobic environment. Three different UV flow cells with 0.4, 7 (default), and 10 mm path length are available, depending on the sample, its concentration, and its UV extinction coefficient. For temperature‐sensitive samples, several types of water cooling/heating jackets are available for the column, flow path, and buffer bottle. These cooling/heating lines can extend to the x‐ray sample cell on the SAXS Autosampler. However, the current cooling/heating setup is not suitable for precise temperature control.

Users control the SEC‐SAXS experiment from a work station located outside of the hutch. A graphical user interface, the SECSAXS tab, has been developed for the customized BL4‐2 SAXS variant of the beamline control software Blu‐Ice that is used at many biological crystallography beamlines (McPhillips et al., 2002). The SEC‐SAXS data collection is run with Blu‐Ice while it communicates with the UHPLC program, Chromeleon, running on a separate PC. A DAC (digital to analog converter) board has been installed in the UHPLC UV detector, allowing the Blu‐Ice to receive and process the UV signals. These signals can be monitored on Blu‐Ice SECSAXS tab during data collection and are recorded in a coordinate file (absorbance vs. image number) stored in the same directory as the SEC‐SAXS data. For synchronization between the x‐ray instruments and the UHPLC, communication is established using input, relay signals, or both. Synchronization can be achieved using the appropriate commands in either the Blu‐Ice SECSAXS tab or Chromeleon. A SEC sequence file containing the details of each SEC experiment is first submitted on Chromeleon for synchronization. Up to 50 automated SEC‐SAXS data collection runs can be queued in a single submission on the Blu‐Ice SECSAXS tab.

2.2. Available SEC‐SAXS modes

Three different SEC‐SAXS data collection modes are presently available at the SSRL BL4‐2 (Figure 2). The “High‐Resolution Mode” is the default mode and provides the best data quality in terms of sample separation (Figure 2a). When a large number of samples needs to be measured in a limited beamtime allocation, data collection can be done in the “High‐Throughput Mode” (Figure 2b). This mode enables rapid data collection using two tandem SEC columns. However, the quality of the sample separation is slightly diminished due to the necessary extra flow paths and the valve that are required after the SEC columns. The “Co‐Flow Mode” is designed for samples that are extremely sensitive to radiation damage (Figure 2c).

FIGURE 2.

Operating principles of three different SEC‐SAXS modes: (a) High‐Resolution mode, (b) High‐Throughput mode, and (c) Co‐Flow mode. (a) In the High‐Resolution mode, the data collection is operated with a single SEC column connected to the right pump (blue line). At the beginning of the sample cell cleaning, the column will be switched by the UHPLC 10‐port switching valve to the left pump (green line), which leads to the waste bottle. (b) The High‐Throughput mode allows automated and continuous data collection using two SEC columns (parallel data collection). The data collection is performed with the column connected to the right pump (blue line), while another column connected to the left pump (green line) is washed. At the end of the data collection, the columns are swapped by the UHPLC 10‐port switching valve for the next data collection. While the sample cell is being cleaned, an additional 4‐port switching valve on the SAXS Autosampler allows the eluate to be sent directly to the waste bottle. (c) In the Co‐Flow mode, the eluate is directed to the co‐flow needle, which is aligned to the center of the x‐ray capillary. The SEC buffer is simultaneously pumped by the left pump to the upstream end of the x‐ray capillary so that the flow surrounds the eluate from the co‐flow needle at the x‐ray position located just below the needle tip.

All flow paths of the high‐resolution mode have been finely tuned for the quality of the sample separation. Along with the small internal volume of the UHPLC Autosampler, the adoption of narrow PEEK tubing (ID = 0.005″) from the UHPLC Autosampler to the x‐ray capillary needle (ID = 0.007″) minimizes both additional sample dilution and mixing along the way. No valve, connector, or other instrumentation (e.g., MALS, multi‐angle light scattering, and RI, refractive index, detectors) is added to the path for the same purpose. Note that depending on buffer composition and flow rate, the PEEK tubing after the column may need to be replaced with tubing of a larger ID if the pressure limits of the column are an issue.

In the high‐throughput mode, time can be efficiently used for washing and equilibrating the SEC column as well as for cleaning the x‐ray sample cell. The UHPLC 10‐port switching valve allows washing and equilibration of one SEC column while the data are being collected with another SEC column. The eluate from the UHPLC UV detector leads to the 2‐port switching valve mounted on the SAXS Autosampler. After the last image acquisition is completed, the 2‐port valve is switched to direct the eluate to the waste bottle in order to start sample cell cleaning using a syringe dispenser. This valve position is also used when conventional high‐throughput equilibrium (i.e., static or batch) SAXS data collection by the SAXS Autosampler is in service (Martel et al., 2012). For high‐throughput SEC‐SAXS data collection, the 6‐port buffer selection valve can be added at the UHPLC pump upon request (Figure 1). A total of eight solvents can be connected to the pump.

The co‐flow (sheath flow) method (Kirby et al., 2016) is available in the co‐flow mode. It provides comparable sample separation quality to that of the high‐resolution mode. The reduction in signal‐to‐noise ratio per image depends on the actual sample path length (sheath width) in the x‐ray path, which can be adjusted by changing both the sample and buffer flow rates in the sample cell.

2.3. SEC‐SAXS analytical software

The automated real‐time SEC‐SAXS data processing and analysis pipeline, SECPipe, was developed for the SSRL BL4‐2 SEC‐SAXS data analysis (https://www-ssrl.slac.stanford.edu/smb-saxs/content/documentation-software-secpipe). It implements the programs SASTOOL (https://www‐ssrl.slac.stanford.edu/smb‐saxs/content/documentation/sastool) and the ATSAS AUTORG (Petoukhov et al., 2007). The program SASTOOL normalizes individual two‐dimensional images using the transmission intensity value recorded by a photodiode mounted in the beamstop. Individual SAXS curves were then generated by azimuthal integration of each detector frame. Once an average buffer (blank) scattering curve has been generated (typically from the first 100 images), the background‐subtracted curves of subsequent images are generated in real time, followed by automated Guinier analyses performed by the program AUTORG. The results of the Guinier analysis are immediately plotted in the Blu‐Ice SECSAXS tab (Figure 3), which allows the user to closely monitor the ongoing experiment. A scaling factor to convert the intensity values to the absolute intensity scale (in unit cm⁻¹) is determined by a measurement of water scattering (Orthaber et al., 2000). This should be done by the beamline staff during the experimental setup and written in the input file of SASTOOL (integ.mpp file) in the analysis directory.

FIGURE 3.

The SECSAXS tab in the Blu‐Ice. The real‐time UV and SEC‐SAXS Guinier plots (SEC‐SAXS profile) are monitored during data collection. Experimental parameters and options are specified for each sample (enlarged view). They correspond to Figure 4.

As mentioned above, the UV absorbance from the UHPLC (2 outputs) is displayed in the UV plot (mAU vs. image number) on the Blu‐Ice SECSAXS tab. The time delay between the UHPLC UV detector and the x‐ray position is taken into account in the image number of the UV plot in all output files. The default UV wavelengths are set to 280 and 260 nm. By specifying the mass UV extinction coefficient of the sample in the Blu‐Ice SECSAXS tab, the unit of UV absorbance can be converted to sample concentration (in mg/ml). The UV plot also includes the value converted using the following Warburg‐Christian equation (Warburg & Christian, 1941):

$$\text{[Protein} (\text{mg}⁄\text{ml)}]=(1.31\times A_{280})\text{–(}0.57\times A_{260})$$

$$\text{Protein}\left[\mathrm{mg}/\mathrm{ml}\right]=1.31\times \left[{\mathrm{A}}_{280}\right]–0.57\times \left[{\mathrm{A}}_{260}\right]$$

where A ₂₈₀ and A ₂₆₀ are the blank‐subtracted absorbance at the 1 cm light path.

The Python script, secdata.py, and the data processing program for the BL4‐2 data, bl42‐dat, which has a Graphical User Interface (GUI), are available at the SSRL BL4‐2 for manual refinement of the SECPipe output data. The secdata.py can also be used when data sets need to be re‐processed after redefining the buffer region or the q _L position. For immediate quality check, the SECPipe and secdata.py generate a variety of plots in the PNG format. The actual q _L value can be confirmed on these plots. More details about the q _L value and the I(q _L) plot are described in the next section. The outputs also include the plots of automatic Gaussian curve fitting on the UV profile to facilitate the selection of scattering profiles from regions where peaks partially overlap.

3. AVAILABLE EXPERIMENTAL OPTIONS FOR STRATEGIC SEC‐SAXS DATA COLLECTION

Outlined below are available experimental options to overcome the sample cell fouling problems. As a prerequisite, each SEC‐SAXS data collection must start with a perfectly clean sample cell. Rigorous cleaning procedures have been designed and implemented for each SEC‐SAXS mode to prevent carryover of the fouling to the next data collection. For example, the washing procedure of the high‐resolution mode is set by default to do rigorous cleaning for ~13 min. The clean capillary enables acquisition of clean buffer data if the buffer region is specified at the beginning of the fractions. However, it is possible for the sample cell to become fouled by long‐term irradiation during a data collection run. The features of SSRL BL4‐2 described below have been designed to overcome the problems observed in conventional SEC‐SAXS data collection experiments.

3.1. Optimizable basic parameters and setup

The beam energy, sample‐to‐detector distance, and beam size can be decided during setup. As tested below, using a higher energy beam may help alleviate the problems associated with radiation damage. The exposure time per image, the interval between exposures, and the total number of exposures (number of images) can be changed for each data collection in the SECSAXS tab of Blu‐Ice. Note that the x‐ray shutter operates in a repetitive open‐close cycle during the SEC elution that is dictated by the specified parameters (Figure 3), and that the shutter does not remain open throughout the data collection. The default buffer region, where the averaged background (blank) profile will be calculated, is defined by the first image number and the total number of images (set by beamline staff during the setup). The region will be used for the automatic data processing by SECPipe. The procedure of rigorous sample cell cleaning following the last image acquisition is highly customizable (selection of cleaning solutions, number of injections or rinses per solution, their volume, number of oscillation cycles, etc.). Some of these changes require adjustments to the Chromeleon SEC sequence file for synchronization.

3.2. Multiple x‐ray positions on the sample cell

The multiple x‐ray beam positions on the sample cell can be stored and used during the SEC‐SAXS data collection (Figure 1). The SAXS autosampler with the sample cell is mounted on a motorized sample stage, which can be moved horizontally and vertically. The different sample cell positions relative to the x‐ray beam are determined and stored by the beamline staff during the setup as per request (maximum 5 positions). The positions are typically selected within <2.5 mm vertically from the tip of the x‐ray capillary needle, and no significant sample dilutions (peak broadening) due to positional differences were observed during experiments. The position can be selected and used per data set on the Blu‐Ice SECSAXS tab (Figure 3). This option distributes the x‐ray irradiations of the sample cell for more efficient and longer use; this iterative use of multiple x‐ray positions on the sample cell allows each position to undergo rigorously cleaning multiple times while other positions are in use.

3.3. X‐ray shutter control

The irradiation of fractions that are not of interest can lead to unnecessary additional sample cell fouling, which can be circumvented with the following options during the data collection. The x‐ray shutter, mounted upstream of the sample cell (Figure 1), can be manipulated using the “Shutter Closed” parameters and the “UV‐Control” options. Both options can be configured on the Blu‐Ice SECSAXS tab (Figure 3). They are functional in all fractions after the buffer region (typically after the 100th image). The “Shutter Closed” parameters consist of three image numbers: “Start/End/Again.” The x‐ray shutter in the range between “Start” and “End” is forcibly closed (yet blank images without x‐ray irradiation are still collected and stored for bookkeeping purposes). The opening of the x‐ray shutter recommences at the next image of “End,” and then it is closed again after the image at “Again.” Furthermore, the x‐ray shutter actuation can be gated by the value of the UV absorbance. This option permits the x‐ray shutter to re‐open when the UV absorbance of the fraction exceeds the threshold value. This shutter control avoids unnecessary irradiations of the sample cell where the sample concentration is too low to obtain analyzable scattering signals. The threshold value is set by the beamline staff during setup (changeable on request). Five typical cases using these options are shown in Figure 4. Regardless of the use of these options during data collection, it is always possible to manually open or close the x‐ray shutter by clicking on the buttons at the bottom of the Blu‐Ice SECSAXS tab (Figure 3). Even in the case of unexpected sample elution, such manual interventions allow SEC‐SAXS data collection to be resumed.

FIGURE 4.

X‐ray shutter control options for the SEC‐SAXS data collection. After buffer data collection, the x‐ray shutter can be closed, re‐opened, and then closed again using the parameters of “Start/End/Again.” It can also be automatically re‐opened when the UV absorbance of a fraction exceeds the UV threshold value (“UV‐Ctrl” checkbox in Figure 3). Additional sample cell cleaning can be inserted after the buffer data collection. (a–f) depict five typical usages of these options: (a) Scattering data of all fractions are measured (conventional method). No option is applied. (b) Unnecessary data collection are avoided. The x‐ray shutter is closed during the void volume after the buffer data collection as well as after the peak of interest has been passed. The parameters of “Start/End/Again” are only applied. (c) Precise x‐ray shutter control is applied to the fraction of interest. Together with additional sample cell cleaning, this can provide the highest quality data while avoiding radiation damage problems. A test run would be required for optimizing the shutter control parameters. (d) The shutter control by the UV threshold is applied solely. This can avoid unnecessary irradiations to the fractions having low sample concentration that does not provide enough scattering, but it may exacerbate the fouling problem. All fractions above the UV threshold are measured. This is ideal for an initial run if the elution profile of the sample is unknown. (e) The combined use of the two options. This can be used to avoid known regions of aggregates. This is also useful when optimizing the injection volume and/or sample concentration.

3.4. Additional sample cell cleaning after buffer data collection

SAXS is a contrast method where the scattering of the buffer is subtracted from the scattering of the sample solution to isolate the scattering from the biomacromolecule. The difference data are sensitive to buffer mismatches and is a common source of failure of SAXS experiments by novice users. The fouling problem on the sample cell can even occur during the buffer data collection (Kirby et al., 2016). Numerous buffer data are generally collected in order to improve the signal‐to‐noise level of the averaged buffer profile and such long irradiations of the sample cell can lead to the accumulation of damaged buffer components, debris from the SEC column, or both. An additional sample cell cleaning operation can be inserted immediately after the buffer data collection (Figure 4). The sample cell can be rinsed with the SEC buffer at the end of the cleaning if the buffer is set to the syringe dispenser (Figure 1), and all cleaning processes can be completed within the void volume of the SEC column. This cleaning option can be selected on the Blu‐Ice SECSAXS tab (Figure 3). The parameters of this additional cleaning step can be customized.

3.5. Real‐time quality check for SEC‐SAXS data collection

In addition to the real‐time SEC‐SAXS Guinier plot (also termed the SEC‐SAXS profile: I(0) and R _g versus image number, [see Figure 3]), on the Blu‐Ice SECSAXS tab, the I(q _L) plot has been introduced to provide an immediate assessment of the ongoing data collection (formerly called the I(q _min) plot). The I(q _L) value corresponds to the actual intensity value at q = q _L, which by default is set to the 6th lowest measured q value (Figure 5a). The sixth q‐value was selected because it is of high enough q to avoid the sometimes noisy area in the immediate vicinity of the beamstop that can often lead to the exclusion of the lowest q data points from Guinier analysis. Assuming that the experimental configuration sufficiently covers the Guinier region of the sample of interest, the I(q _L) value of good data must conform to the Guinier law and exhibit a good agreement with the I(0) value estimated by the Guinier analysis (Figure 5b). The scattering intensities at the lower q, including I(q _L), can serve as a sensitive indicator of many types of problems, including contamination of aggregates, interparticle interactions (concentration‐dependent interactions), fouling of the sample cell, buffer mismatch, and so forth. When such problems are moderate but not severe, the Guinier analysis is still feasible by using the other points at slightly higher q values and one can estimate reasonable I(0) and R _g values, resulting in the discrepancy between I(0) and I(q _L) (Figure 5c, d). Thanks to the program AUTORG that automatically discards unreliable data points during analysis, the comparison between the I(0) and I(q _L) provides supplementary information about the data quality. A strong agreement between the values is absolutely indispensable. The position of the q _L can be changed by the secdata.py when re‐analyzing the data. For instance, such a change would be necessary when the number of data points in the Guinier region is limited by samples with elongated shapes.

FIGURE 5.

Introducing the I(q _L) plot for rapid quality assessment of the SEC‐SAXS data collection. (a) I(q _L) is defined as the experimental intensity at q = q _L, where the q _L is the 6th lowest measured q value (by default). A comparison between I(0) and I(q _L) values can serve as an indicator of data quality. (b) Schematic representation of the I(0) and I(q _L) plots for high‐quality data. It exhibits a good agreement between I(0) and I(q _L) during the data collection. (c) Schematic representation of the plots when aggregates were eluted right before the peak. (d) Schematic representation of the plots for low‐quality data where I(0) and I(q _L) disagree. This type of disagreement can be often observed when moderate fouling occurs on the sample cell.

4. TEST DATA

4.1. High‐Resolution mode versus High‐Throughput mode

Figure 6 shows SEC‐SAXS results collected with two different SEC‐SAXS modes at the SSRL BL4‐2. They were performed with the standard setup (11 keV and approximate 1.7 m sample‐to‐detector distance). For the high‐resolution mode (Figure 6a, c), a 10 μL volume of 40 mg/mL BSA was injected into the Superdex 200 Increase column at a flow rate of 0.05 mL/min, and images were collected with 1 s exposure every 5 s. On the other hand, for the high‐throughput mode (Figure 6b), images were collected with 1 s exposure every 3.3 s at a flow rate of 0.08 mL/min. The x‐ray shutter was closed only in the void volume after the buffer data collection (Image #: 102–199), but all eluted samples after the void volume were measured for demonstration purposes. Neither the buffer nor sample solution contained DTT (1,4‐dithiothreitol). The I(0) plot of the high‐resolution mode (Figure 6a) exhibits a good agreement with the UV absorbance plot measured by the upstream UHPLC UV detector, signifying that further dilution and mixing did not occur between the detector and the x‐ray position (internal volume is ~20 μL). In fact, no peak broadening was discerned at the BSA monomer peak, and automatic Guinier analysis did not work between the dimer and monomer peaks due to low sample concentrations. In contrast, the I(0) plot of the high‐throughput mode (Figure 6b) showed peak broadening compared to the UV plot due to the less‐optimized flow path design necessary for the higher‐throughput mode compared to the high‐resolution mode. Upon comparing both results, the high‐resolution mode maintains ~1.5 times higher sample concentration (3.1 vs. 1.9 mg/mL at the BSA monomer peak position); however, the BSA monomer of the high‐throughput mode exhibited an extended plateau on the R _g plot. No severe radiation damage problem was detected in either data set, as indicated by the good agreement between the I(0) and I(q _L) plots.

FIGURE 6.

SEC‐SAXS profiles collected at (a) High‐Resolution and (b) High‐Throughput modes. A 10 μL of 40 mg/mL BSA was injected into a Superdex 200 Increase column (3.2/300). (c) An exemplar of strategic SEC‐SAXS data collection at the high‐resolution mode. The scattering profiles (i.e., SAXS curves) of BSA monomer were collected using the x‐ray shutter control option. (d) An averaged scattering profile of the BSA monomer obtained in (c) (averaged 5 scattering profiles at the peak position). Guinier plot and P(r) function are also shown in the insets. The theoretical curves from the BSA crystal structure (green, χ ² = 1.88) and the SAXS model (black, χ ² = 1.28) well fit with the experimental curve. The program CORAL was used for reconstructing disordered 26 residues at the N‐terminal (colored black in the ribbon model). The averaged curve is purely background‐subtracted data, devoid of any post‐experimental analyses such as peak decomposition or baseline correction. Table 1 summarizes the results of the SAXS data collection and analysis.

The results of an exemplar SEC‐SAXS data collection experiment are shown in Figure 6c, d. The x‐ray shutter was re‐opened using the UV threshold of 1000 mAU (Figure 4d). The averaged scattering curve corresponding to the BSA monomer exhibited good agreement with the theoretical curves from the BSA the crystal structure (PDB ID: 3V03) and the SAXS model in which 26 residues in the N‐terminal region were reconstructed by the program CORAL exhibited even better agreement (Figure 6d). In instances where samples have encountered radiation damage problems, such selective SEC‐SAXS data collection yields higher‐quality data in most cases at the SSRL BL4‐2.

4.2. Overcoming radiation damage problems by strategic SEC‐SAXS data collection

Lysozyme at acidic pH is empirically known to be extremely sensitive to x‐ray irradiation at the SSRL BL4‐2. Long‐term irradiation of lysozyme leads to severe fouling problems on the sample cell made of quartz, so it is often used to optimize and generalize the experimental protocols at the SSRL BL4‐2. To evaluate the experimental options, SEC‐SAXS data collection using lysozyme from chicken egg white were performed in the high‐resolution mode (Figure 7). For these demonstrations, the sample‐to‐detector distance of approximate 1.7 m (standard setup), which is typically longer than optimal for samples with small R _g like lysozyme, was used for the data collection in order to collect more data points at low q. A 10 μL volume of 30 mg/mL Lysozyme was injected into the Superdex 75 Increase column at a flow rate of 0.05 mL/min. The use of higher sample concentrations and/or injection volumes was avoided to eliminate potential concentration‐dependent problems that might complicate the evaluation of the fouling problem. Images were collected with a longer exposure time (2 s per every 5 s) to induce severe fouling problems, as opposed to standard data SEC collection (1 s per every 5 s). The 50 images in early fractions were used for buffer data set. Due to higher extinction coefficient of lysozyme, the UV flow cell of 0.4 mm flow path was used for the data collection. Severe fouling problems were observed in Figure 7a where the x‐ray shutter was open during the entire elution (conventional method). This can be confirmed by the fact that I(q _L) values increased around the peak as a function of the number of images, that automatic Guinier analysis did not work in the late part of the peak, and that the R _g values surged right before the Guinier analysis stopped. No severe fouling problem before the peak position (i.e., buffer radiation damage) was observed during this time. Figure 7b, c show that using DTT or a higher energy beam (changed from 11 to 13 keV) were effective in alleviating the fouling problem. A better agreement between I(0) and I(q _L) values can be confirmed even in the late part of the peak, although estimates of R _g were noisier due to low sample concentrations. The x‐ray shutter option was applied in the absence and presence of DTT (Figure 7d, e). Additional sample cell cleaning was performed after buffer data collection, and the x‐ray shutter was then re‐opened at the peak position. Clean scattering curves can be obtained in both data sets without any signs of fouling, even in the data set without DTT. The result that scattering curves without DTT (Figure 7d) are even comparable to those with DTT (Figure 7e) proves the effectiveness of these experimental options for strategic data collection. We suggest optimizing the timing of the shutter re‐opening based on preliminary measurements so that the fractions right after re‐opening can be used for further analysis (e.g., fractions after the top of the peak).

FIGURE 7.

SEC‐SAXS data collection of lysozyme by different methods. A 10 μL of 30 mg/mL Lysozyme was injected to the column. (a) A series of 500 images was collected at 11 keV in conventional SEC‐SAXS. (b) The 5 mM DTT was added in both buffer and sample solution as a radical scavenger. (c) The x‐ray beam was switched from 11 to 13 keV. No DTT was included. (d) The data were collected at 11 keV with the x‐ray shutter control. No DTT was used. (e) The data were collected at 11 keV with the x‐ray shutter control. The 5 mM DTT was used. Additional sample cell cleaning after the buffer data collection was performed on (d) and (e). (f) The data measured by the co‐flow mode. A 10 μL of 60 mg/mL lysozyme was injected to the column. Neither DTT nor the x‐ray shutter control was used. (g) An averaged scattering profile of the BSA monomer obtained in (e) (averaged five scattering profiles at the peak position). The theoretical curve from the lysozyme crystal structure (black) well fit with the experimental curve. Guinier plot and P(r) function are also shown in the inlets. A Guinier plot of (a), colored in green, is also shown, along with a black line whose slope corresponds to the same R _g of (ed) for comparison. As with the BSA data in Figure 6, the averaged curve is purely background‐subtracted data, devoid of any post‐experimental analyses such as peak decomposition or baseline correction. Table 1 summarizes the results of the SAXS data collection and analysis.

Further data collection was also demonstrated on the co‐flow mode (Figure 7f). A volume of 10 ul of 60 mg/mL lysozyme without DTT was injected instead of 30 mg/mL. The flow rates for the sample and buffer flow lines within the sample cell were set to 0.05 mL/min and 0.10 mL/min, respectively. The clean scattering data were obtained, although the recorded scattering signals were weaker because of the sheath flow in the sample cell.

The averaged scattering curves derived from Figure 7e are presented in Figure 7g, as well as Guinier plot and P(r) function. The experimental curve showed a good agreement with the theoretical curve from the Lysozyme crystal structure (PDB ID: 2vb1).

5. DISCUSSION

Taking advantage of the small internal volume of the UHPLC as well as the small analytical SEC column, the high‐resolution mode maintained a great selectivity with sample economy. As demonstrated in Figures 6 and 7, the amount of sample required for a SEC‐SAXS data collection is typically equal to or less than that required for a concentration series for equilibrium (conventional) SAXS experiments using the SAXS Autosampler. Because no peak broadening was observed at the BSA monomer peak (Figure 6a), it is clear that the SAXS instrumentation following the UV detector to the x‐ray position cannot cause appreciable peak broadening. Therefore, the quality of the SEC‐SAXS data, in terms of sample separation, is contingent upon how the SEC is performed, including factors such as column selection, injection volume, and sample concentration. The MALS and RI detectors specifically designed for UHPLC are available and can be inserted into the flow line between the UHPLC and SAXS instruments. However, separate use of these detectors is recommended to obtain better quality of the SEC‐SAXS data. These detectors can be used in parallel with a 2nd UHPLC instrument at the beamline. Alternatively, they can be connected and used before or after the beamtime.

The small analytical SEC column with 2.4 mL bed volume is capable of separating the sample at high concentration compared to typical larger columns, providing the SEC‐SAXS data with the higher signal‐to‐noise ratios. The higher samples concentrations are also ideal for the complex samples that tend to dissociate at low sample concentrations (i.e., those with larger dissociation constants). However, we acknowledge two limitations with the small column. First, the buffer exchange during the SEC is difficult to complete. For example, if the sample solution contains a certain amount of glycerol (e.g., for freezing purposes) that is not in the SEC buffer, background mismatch can be observed. The sample solution must be exactly replaced with the SEC buffer beforehand. Second, due to higher sample concentration at the x‐ray position, the sample may exhibit concentration dependencies similar to those in the equilibrium SAXS data collection experiments. Therefore, a few data collection runs with different injection volumes (or different sample concentrations) are strongly recommended to check for the presence of concentration dependence.

Here, we developed some experimental options to overcome difficult situations caused by radiation damage problems during the SEC‐SAXS data collection. As demonstrated in Figure 7b, the use of radical scavenger is still primarily encouraged. Previous severe radiation damage tests using Lysozyme revealed that more than 2 mM DTT appeared to be most effective to prevent fouling of the sample cell (data not shown). This result is consistent with the previous report by Brooks‐Bartlett, et al. (Brooks‐Bartlett et al., 2017). In consideration of the degradation of DTT, a slightly higher concentration (e.g., 3–5 mM DTT) is recommended especially for long SEC‐SAXS data collection beamtime. In light of the cost‐effectiveness of DTT and the volume of SEC buffer required, we continue to recommend the use of DTT, even if the buffer is marginally outside the optimal pH range for DTT. In fact, the use of 5 mM DTT in Lysozyme buffer (pH = 4.8) effectively prevented fouling problems at the SSRL BL4‐2. However, the use of TCEP (Tris(2‐carboxyethyl)phosphine) is recommended for highly acidic environments due to its superior reactivity and stability compared to DTT.

Like Figure 7e, in most cases, the combined use of the x‐ray shutter control options, 3–5 mM DTT, and additional sample cell cleaning after buffer data collection reduced fouling on the sample to an undetectable level (Bush et al., 2019; Chen et al., ²⁰¹⁹; Shi et al., ²⁰²³; Yang et al., ²⁰¹⁹). However, DTT cannot be used for some samples with exposed disulfide bond(s) that could be reduced, thereby causing structural changes. Even in this case, the combined use of the shutter control options and additional sample cell cleaning was very helpful in obtaining clean SEC‐SAXS data (Figure 7d) (Cogan et al., 2020; Garcia et al., ²⁰²²; Horikoshi et al., ²⁰²¹; Panigrahi et al., ²⁰¹⁸). As also demonstrated in Figure 7d, even when the radical scavenger cannot be used, this method eliminates the problems caused by carryover fouling from the sample data measured immediately after re‐opening the x‐ray shutter, making data quality competitive with that measured by the static SAXS method. In addition, this method allows increases in exposure times, which can compensate for weak scattering from samples that cannot be sufficiently concentrated. In contrast, this is a challenge for conventional SEC‐SAXS data collection because longer exposure times per image exacerbate the fouling problem. Prior to image acquisitions at the fractions of interest, the sample cell becomes dirtier by superfluous irradiation during the unnecessary earlier fractions. As demonstrated in Figure 7c, increasing the x‐ray beam energy is another option to effectively mitigate radiation damage problems. When using higher x‐ray energy, the accessible q‐range will change so optimization of the sample‐to‐detector distance may be necessary depending on the q range required for the analysis.

The co‐flow mode offers another approach to mitigate radiation damage problems. However, it involves a traded‐off between data quality and sample consumption. In the present studies, a quartz capillary of approximately the same size was used for all data collection, as shown in Figure 7. Nevertheless, it is recommended to use a capillary with a slightly larger internal diameter (ID) in the co‐flow mode to enhance sample dilution. Further development and application of the co‐flow mode, such as the use of a high‐flux multilayer beam, are currently underway.

The Gaussian peak fitting on the UV profile is automatically performed by the SECPipe. Unlike the SAXS intensity, the UV absorbance is not influenced by inter‐particle interactions (concentration‐dependence) and aggregates. Assuming no peak broadening from the UHPLC UV detector to the x‐ray position, the use of these fitting results to select the scattering profiles is legitimate. In instances where severe effect by inter‐particle interactions are detected or peak overlap is observed, post‐experimental SEC‐SAXS decomposition programs can be applied. Nevertheless, the methods presented herein can help to start the analysis with cleaner data with minimal fouling and separation problems.

The SSRL is a user facility and open for all scientists from academia, industry and other types of research organizations. Access to the SSRL is based on the score assigned to the user's proposal during peer review. Further updates and information about BL4‐2 SEC‐SAXS and workshops will be posted on our website (https://www-ssrl.slac.stanford.edu/smb-saxs).

6. MATERIALS AND METHODS

Lyophilized powders of albumin from bovine serum (BSA) and lysozyme from chicken egg white were purchased from Millipore Sigma (St. Louis, MO). 40 mg/mL BSA and 30 mg/mL lysozyme solutions were prepared in the BSA buffer (50 mM Tris–HCl at pH = 7.0) and lysozyme buffer (50 mM Sodium Acetate at pH = 4.8 and 150 mM sodium chloride), respectively. Both solutions were centrifuged at 14,100 × g for 10 min and then filtered using 0.22 μm pore size filter.

The curve fitting was performed for the BSA and Lysozyme curves using the program CRYSOL (Svergun et al., 1995). The crystal structures of the BSA (PDB ID: 3 V03) and the lysozyme (PDB ID: 2vb1) were employed for the fitting analyses. The N‐terminal disordered region (26 residues) of the BSA crystal structure was reconstructed using the program CORAL (Petoukhov et al., 2012). The program GNOM was used for the indirect Fourier transform to estimate the distance distribution function P(r) (Svergun, 1992).

AUTHOR CONTRIBUTIONS

Tsutomu Matsui: Conceptualization; writing – review and editing; methodology; writing – original draft; investigation; data curation. Ivan Rajkovic: Software; writing – review and editing; data curation; investigation. Blaine H. M. Mooers: Writing – review and editing. Ping Liu: Software; writing – review and editing; investigation; data curation. Thomas M. Weiss: Supervision; writing – review and editing; investigation; methodology.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Matsui T, Rajkovic I, Mooers BHM, Liu P, Weiss TM. Adaptable SEC‐SAXS data collection for higher quality structure analysis in solution. Protein Science. 2024;33(4):e4946. 10.1002/pro.4946