Newly developed Laboratory-based Size exclusion chromatography Small-angle x-ray scattering System (La-SSS)

To understand a biological system, it is important to observe structures of biomolecules in the solution where the system is functionalized. Small-Angle X-ray Scattering coupled with Size Exclusion Chromatography (SEC-SAXS) is one of techniques to selectively observe the target molecules in the multi-component system. However, this technique is believed to be available only with a synchrotron-based SAXS instrument due to requirement of high beam intensity and, therefore, the limitation of the beam time was obstacle to satisfy demands from many bio-researchers. We newly developed Laboratory-based Size exclusion chromatography SAXS System (La-SSS) by utilizing a latest laboratory-based SAXS instrument and finely optimization of the balance between flow rate, cell volume, irradiation time and so on. La-SSS succeeded not only decoupling of target protein(s) from non-specific aggregates but also measurement of each concerned component in a multi-component system. In addition, an option: “stopping mode”, which is designed for improving statistics of SAXS profile, realized a high S/N data acquisition for the most interesting protein in a multi-component system. Furthermore, by utilizing a column having small bed volume, the small-scale SEC-SAXS study makes available. Through optimization of instrumental parameters and environments, La-SSS is highly applicable for experimental requirements from various biological samples. It is strongly expected that a La-SSS concept must be a normal option for laboratory-based SAXS in the near future.

SEC-SAXS was considered to be realized only with a synchrotron-based SAXS instrument due to requirement of higher beam intensity by using a flow measuring system. Therefore, the first SEC-SAXS was developed by David and Pérez 1 at the SOLEIL. After their pioneering work, the SEC-SAXS technique has been widely spread over many synchrotron facilities 2-6 around the world. On the other hand, it is still difficult that every biologist performs a SEC-SAXS experiment since the beam time at synchrotron facilities is quite limited. The shortage of beam time should be solved by realization of a laboratory-based SEC-SAXS. Here raises a question, "Is it possible to construct a laboratory-based SEC-SAXS"? even though the most people a priori believe that SEC-SAXS is an only available technique at synchrotron facilities.
Equipment for laboratory-based SAXS, such as a focusing mirror, a collimation system with less parasite scattering pinholes and a photon-counting detector, is in distinguished progress. As a result, a photon flux at sample position in a latest laboratory-based SAXS reaches as much as 1/100 compared to those in a synchrotron-based SAXS instrument 7 . Under this situation, Bucciarelli et al. 8 launched a laboratory-based SEC-SAXS system and succeeded to obtain the SAXS data, which enables the further analysis of a non-aggregated target component for the first time. Here, at least three technical improvements must be considered for a next generation laboratory-based SEC-SAXS. The first one, which is the nuisance problem in SEC-SAXS measurement, is re-mixing between the separated samples. This problem is mainly originated from the retention of samples at a SAXS cell. The proteins with the same sizes are eluted out from a SEC column as groups from larger to smaller ones in the time order. Therefore, the longer retention of the group in the cell has a possibility to make the mixture of the next coming group with the different size. This re-mixing results in the measured SAXS profile that includes   Dark blue and yellow lines correspond to a flow path and an optical fiber, respectively, and the sample flow path was set to 5 mm. Black dotted circle indicates X-ray irradiation zone. The system has another function "stopping mode". When the OD 280 value exceeds the set value, the auto valve automatically switches the path from a SAXS cell to a waste bottle, as shown by green arrow. As a result, the solution exceeding the set OD 280 value, target protein solution, remains in the SAXS cell.
www.nature.com/scientificreports www.nature.com/scientificreports/ the scatterings from the different proteins. The re-mixing is observed as broadening of time evolution of protein concentration at the SAXS cell and, in the worst case, the elution peaks of proteins having different sizes would be overlapped in the SEC-elution chart. We dare to name this re-mixing problem as "broadening problem". The broadening could make it difficult to selectively perform a structural analysis on an interested component in a multi-component complex system. Therefore, the elimination of "broadening problem" in SEC-SAXS is the first technical requirement. The second point is the reduction of injection sample amount, which offers the chance for SEC-SAXS measurements on the sample with a little of available amount. The third point is the amendment of counting statistics of SAXS profiles, contributing to making the quality of structure analysis higher.
We firstly constructed the SEC-SAXS system, which is free from the broadening problem even at slow flow rate. Secondly, we introduced an additional technique for enhancing the data quality and, finally, reduced sample injection volume. In this manuscript, we reported the feasibility of our finely optimized Laboratory-based SEC-SAXS System (La-SSS) to several proteins.

Results and Discussion
Detail of setup of La-SSS. We have to take into consideration of the balance between a flow rate and separation resolution, keeping the statistics for data analysis. Namely, the faster flow rate reduces the re-mixing by washing the sample solution out from the SAXS cell quickly, however might deteriorate the separation resolution and the statistics of scattering data due to shortening of an exposure time. Firstly, to figure out the required exposure time experimentally, an ovalbumin (OVA) solution with the concentration of 1.5 mg/mL, which is the nominal protein concentration on a standard SAXS measurement, was measured with the SAXS instrument, NANOPIX (see Materials and methods: SAXS instruments). It was estimated that measurement time of c.a. 20 minutes was at least necessary for the determination of reliable R g value. Next, we optimized the flow rate with Superdex 200 increase 10/300 GL. Under the conventional flow rate of 0.50 mL/min, the full width at half maximum of elution peak of monomeric OVA was found to be 1.0 min. Accordingly, the flow rate for La-SSS should be slower than 0.025 (=0.50/20) mL/min to satisfy the required exposure time. Considering the above experimental requirements, the standard flow rate was selected to 0.020 mL/min for La-SSS. Under this flow rate, the flowing volume corresponds to the SAXS cell (40 μL) in 2 minutes. It implies that the sample at X-ray irradiation area will be thoroughly washed out within 2 minutes. Namely, La-SSS is in principle expected to be free from "broadening problem". In the next section, we will verify the absence of "broadening problem" in more detail based on experimental results.
The detail of components of our La-SSS is shown in Table S1 (refer to supplementary materials) and the schematic view of our La-SSS is also shown in Fig. 1.
Absence of "broadening problem" at SAXS cell. It was reported that the slower flow rate is beneficial for the improvement of separation resolution of SEC 9 . We then experimentally try to verify that La-SSS is free from the broadening problem at several flow rates (including our standard flow rate, 0.020 mL/min) with bovine serum albumin (BSA) solution: the initial concentration and injection volume of 5.4 mg/mL and 500 μl, respectively. Firstly, by changing the flow rates from 0.010 mL/min to 0.20 mL/min, we have compared the time evolution of UV intensity at exit of HPLC (U h (t): t is elution time) to that at just upstream of SAXS-cell (U c (t)) (see Fig. 1(f,h,i), respectively). As shown in Fig. 2, it was confirmed that U c (t) nicely coincided with U h (t) in all the flow rates. Next, we also have compared U h (t) to the time evolution of SAXS intensity integrated from 0.015 Å −1 to 0.030 Å −1 (I int (t)). As shown in Fig. 3, regardless of flow rate, three peaks were observed in U h (t). Their peak positions are highlighted by light green, blue and red arrows from the high molecular weight component to the low one. I int (t) of the peak exhibiting the highest intensity nicely reproduced U h (t) without broadening at all the flow rates investigated. It is then concluded that the SAXS cell in La-SSS is essentially free from "broadening problem" even at the slowest flow rate (0.010 mL/min). It should also be noted that good separation of three peaks is observable from I int (t) at the flow rate of 0.010 mL/min and 0.020 mL/min.

Sec-SAXS measurements: observation of protein with the relatively low molecular weight including the aggregates in solution.
We tested the performance of La-SSS with relatively low molecular weight protein: OVA with M w = 45 kDa. OVA solution was injected to SEC column with the initial concentration and volume of 5.0 mg/mL and 500 μl, respectively. The magnitude of scattering vector: Q is given by 4πsin(θ)/λ, where 2θ and λ correspond to scattering angle and wavelength of X-ray, respectively. Figure 4(a) shows I int (t) www.nature.com/scientificreports www.nature.com/scientificreports/ (yellow circle) and the time evolution of protein concentration (red curve) calculated from U c (t). There are two peaks in the elution chart: one is a clear and extinct peak at 265 min and the other is a weak peak at 155 min. We then averaged the SAXS profiles at two regions; Region 1 (peak at t = 265 min: highlighted by pink dotted rectangle) and Region 2 (peak at t = 155 min: highlighted by light blue dotted rectangle), as shown in Fig. 4(a). The averaged SAXS profiles and their Guinier plots from Regions 1 and 2 are shown in Fig. 4(b,c), respectively. R g from Region 1 is found to be 23.9 ± 0.4 Å, which is well consistent with that of monomeric OVA (R g = 23.9 ± 0.2 Å) measured with synchrotron-based SAXS instrument 4 . Although the statistics of the averaged SAXS profile from Region 2 is not good due to the quite low protein concentration (~0.07 mg/mL), R g value was calculated to be 63.8 ± 28.4 Å. It is considered that Region 2 is dominated by aggregated OVA. In conclusion, monomeric OVA was well separated from the aggregates by La-SSS.
Here, we would like to show the ability of La-SSS to detect the shape difference of protein having similar molecular weight. We selected Fc portion of immunoglobulin G 10 with molecular weight of 50 kDa, which is similar with OVA (45 kDa). Concerning about shape, Fc has a horseshoe-like shape, whereas OVA possesses a barrel-like one. We performed SEC-SAXS measurement of Fc with La-SSS by injecting the concentration and volume of 5.0 mg/mL and 500 μl, respectively. The scattering profiles from monomeric OVA and Fc are shown in Fig. S1(a). Calculated R g value of Fc was 27.4 ± 0.4 Å, which is relatively different from that of OVA (23.9 ± 0.4 Å). Utilizing the same Q range for OVA solution, we also calculated distance distribution function (P(R) function) of Fc. P(R) functions of Fc and OVA are plotted in Fig. S1(b). It can be clearly seen the difference in P(R) functions. R g and P(R) function reflect the difference in shape between two proteins. In conclusion, with La-SSS, we can detect the shape difference as well.

Sec-SAXS measurement: observation of protein with the higher molecular weight including the aggregates in solution.
We also checked the performance of La-SSS with apoferritin (AF) as a relatively higher molecular weight protein (440 kDa). AF solution with the initial concentration and volume of www.nature.com/scientificreports www.nature.com/scientificreports/ 5.1 mg/mL and 500 μl, respectively, was injected to SEC column. Figure 5(a) shows I int (t) (yellow circle) and the time evolution of protein concentration (red curve). Two peaks were observed in I int (t) and the time evolution of concentration. Main and minor peaks were located at 210 min and 150 min, respectively. We then averaged the SAXS profiles at two regions; Region 1 (peak at t = 210 min: highlighted by pink dotted rectangle) and Region 2 (peak at t = 150 min: highlighted by light blue dotted rectangle), as shown in Fig. 5(a). The averaged SAXS profiles and their Guinier plots are shown in Fig. 5(b,c), respectively. A clear upturn at low Q-region is observed in the SAXS profile from Region 2. R g from Region 1 is found to be 53.2 ± 1.6 Å, which is nicely consistent with R g reported by Zabelskii et al. 11 as that of a monomer of AF (R g = 52.9 Å). On the other hand, R g value satisfying the relationship in Q < 1.3/R g was not obtained from Region 2 due to too large R g value. It is again confirmed that La-SSS also contributes to the separation of native protein complex from aggregated one for the protein with higher molecular weight. improvement of statistics: stopping mode. It was confirmed that La-SSS could evaluate R g from monomeric protein or native protein complex including aggregates. On the other hand, the recent Bio-SAXS analysis requires high statistics data in wide-Q region to build the three-dimensional structural model. To improve the data statistics, especially quality at high Q-region, we installed an optional measurement mode for La-SSS: stopping mode. When the OD 280 value exceeds the set value, the flow path automatically switches to a waste bottle (see Fig. 1(k)) and the solution exceeding the set OD 280 value remains in the SAXS cell. Figure 6 shows the performance of the stopping mode, I int (t)s from the flow mode (red circle) and the stopping mode (blue circle), respectively. The averaged SAXS profiles from the flow and stopping modes are shown in Fig. 6(b,c), respectively: the blue line is the scattering profile calculated with its crystal structure. The evaluated R g values from the flow and stopping modes are 53.2 ± 1.6 Å and 53.1 ± 1.5 Å, respectively. In addition, the statistics around second fringe (Q~0.15 Å −1 ) is clearly improved with the aid of stopping mode. P(R) calculated by GNOM 12 are shown in the www.nature.com/scientificreports www.nature.com/scientificreports/ insets of Fig. 6(b,c), respectively. P(R) from the flow mode was slightly oscillating due to insufficient data quality, as shown by solid arrows in the inset of Fig. 6(b). Contrary to this, P(R) from the stopping mode exhibited smooth curve in the whole R range. It is expected that this stopping mode will contribute to the improvement of statistics at the high Q region even with a lower concentration sample. Finally, it should be kept in mind that SAXS integrated intensity is quite sensitive to the onset of aggregation, hence the presence or absence of aggregation is always monitored by I int (t). In this AF case, no trace of aggregation was observed during the stopping mode for 60 minutes.

Sec-SAXS measurements with multi-component system: BSA.
For further exhibiting the performance of La-SSS, we also utilized La-SSS for more complex system, BSA, which is prone to form several oligomers. Prior to SEC-SAXS measurement, the AUC measurement of BSA solution was performed with the same initial concentration as that of SEC-SAXS. As shown in Fig. S2, there were three peaks in the sedimentation coefficient (S 20 ) distribution (c(S 20 )): one major peak at around S 20 = 4, and two minor peaks around S 20 = 6 and 8. Through the relationship between S 20 value and molecular weight, it was found that the components of S 20 = 4, S 20 = 6 and S 20 = 8 correspond to monomeric dimeric and trimeric BSA oligomers, respectively. With this polydispersed BSA solution, we performed SEC-SAXS measurement. Figure 7(a) shows the time evolution of concentration (red curve) and I int (t) (yellow circle). Three peaks are observable from Fig. 7(a) as expected from the AUC measurement. I(0) and R g as a function of time are sequentially analyzed and the results are summarized in Fig. 7(b). We then averaged the SAXS profiles at three different regions: Region 1 (peak at t = 400 min: highlighted by pink dotted rectangle), Region 2 (peak at t = 300 min: highlighted by light blue dotted rectangle) and Region 3 (peak at t = 240 min: highlighted by black dotted rectangle) as shown in Fig. 7(a). The averaged SAXS profiles from Regions 1, 2 and 3 are shown in Fig. 7(c) and the corresponding Guinier plots are also summarized in Fig. 7(d). The evaluated R g values from Regions 1, 2 and 3 are 27.3 ± 0.3 Å, 39.0 ± 1.9 Å and 47.2 ± 8.0 Å, respectively. The www.nature.com/scientificreports www.nature.com/scientificreports/ reported R g values of monomeric and dimeric BSA were 27.0 ± 0.1 Å and 39.0 ± 0.2 Å 13 , respectively. Hence, our evaluated R g values were surprisingly coincided with those from the synchrotron-based SEC-SAXS studies within the experimental errors. Judging from the AUC result, it is supposed that R g = 47.2 ± 8.0 Å corresponds to trimer of BSA. It is also considered that La-SSS is applicable for the multi-component system.

Sec-SAXS measurements: observation of protein with the low molecular weight: pnGase-pUB.
From the SEC-SAXS studies on OVA, Fc, AF and BSA, we could successfully exhibit the performance of La-SSS. In order to extend the accessible range of M w of target protein, we also tested the performance of La-SSS with low molecular weight protein by changing sample to detector distance. The target protein is PNGase-PUB 14 with M w = 13.4 kDa. For this SEC-SAXS study, Superdex 75 increase 10/300 GL was utilized for SEC column. PNGase-PUB solution was injected to SEC column with the initial concentration and volume of 2.7 mg/mL and 300 μl, respectively. In order to enhance S/N of SAXS data, the flow rate of 0.010 mL/min was selected for this study. Figure 8(a) shows I int (t) integrated from 0.035 Å −1 to 0.050 Å −1 (yellow circle) and the time evolution of protein concentration (red curve) calculated from U c (t). We then averaged the SAXS profile at t = 860 min: highlighted by blue dotted rectangle. The averaged SAXS profile and its Guinier plot are shown in Fig. 8(b,c), respectively. R g was calculated to 15.3 ± 0.5 Å, which is consistent with that calculated from its crystal structure (=15.8 Å) within the experimental error. Present result surely certificates the applicability of La-SSS to low molecular weight protein as well.
further challenge: Approach for small-scale Sec-SAXS system. It is required for c.a. 2.5 mg (=5.1 mg/mL × 500 μl) proteins to maximize the concentration of a target protein in our La-SSS (a normal scale system). However, the preparation of such huge amount is still drawback for performing a SEC-SAXS www.nature.com/scientificreports www.nature.com/scientificreports/ experiment. The reduction of sample amount opens the accessibility of SEC-SAXS studies, hence we also build a small-scale SEC-SAXS system: The SEC column is Superdex 200 increase 3.2/300 of which injection volume can be reduced to 50 μl and the flow rate is adopted at 0.010 mL/min. We tested this small-scale SEC-SAXS system using an AF solution with the initial concentration of 5.1 mg/mL. Figure 9(a) shows the time evolution of protein concentration (red curve) and I int (t) integrated from 0.015 Å −1 to 0.030 Å −1 (green circle). In the small scale SEC-SAXS system, the contribution of non-natively aggregated AF was not well decoupled from that of native AF. It is expected that the resolution of separation of Superdex 200 increase 3.2/300 is not superior as that of Superdex 200 increase 10/300 GL. With the result of the small-scale SEC-SAXS system, to eliminate the contribution of non-natively aggregated AF, we firstly fitted the I int (t) with two Gaussians. Each Gaussian component is depicted by dotted light blue line (component 1) and dotted light pink line (component 2), respectively, in Fig. 9(a). Then, the region excluding the tail of component 1 was chosen for averaging region, as shown by blue dotted rectangle in Fig. 9(a). The averaged SAXS profile and its Guinier plot are shown in Fig. 9(b,c), respectively. R g value was 55.2 ± 3.8 Å same as that obtained from the normal-scale SEC-SAXS system within experimental error (Fig. 5(c) and more detailed information should be referred to Table 1).
Although the error of R g from the small-scale system is still larger than that from the normal-scale system, it is considered that the scattering profiles having similar S/N can be obtained from this small column through further optimization of experimental conditions and/or utilization of "stopping mode". It is also expected that optimum selection of carrier of SEC column would further improve the performance of SEC-SAXS with small column volume. Finally, it will open further opportunity for La-SSS studies on small sample volume.
Summary. In this paper, we have extended the feasibility of laboratory-based SEC-SAXS system (La-SSS) by optimizing instrumental parameters and environments. We could successfully obtain the scattering profiles and evaluate R g of target proteins without aggregation: The relationship between data quality and flow rate is given in Fig. S3. In addition, dimensionless Kratky plot from all the monomeric proteins exhibited the bell-shaped curves having the peaks at QR g = 1.7. These results strongly support that all the monomeric protein possesses well-folded globular structure with La-SSS (refer to Fig. S4). In addition to decoupling the target monomeric or native protein complex from non-specific aggregation, selective structural analysis of aggregates was practically realized. It is