Resuspending Samples in Analytical Ultracentrifugation

Abstract

In analytical ultracentrifugation it is often very useful to resuspend samples in situ after sedimentation experiments for further investigation. This can be achieved by manually subjecting the entire sample cell assembly to gentle motion that causes the air bubble in the sample compartment to repeatedly move through the solution and thereby cause convection. Here we describe a cell mixing device that can accomplish the same through axial rotation and slow rocking motion. This cell mixer is low-cost, open-source, and can be easily assembled from readily available components. It can efficiently mix multiple sample cells side-by-side and may be used with various centerpiece designs.

Analytical ultracentrifugation (AUC) is based on the real-time observation and analysis of macromolecular sedimentation in a strong centrifugal field [1–3]. It is a classical technique of physical biochemistry. Besides the traditional applications to study synthetic polymers and proteins and their interactions in structural biology, application of AUC is rapidly growing in fields as diverse as vaccine development, archeology, nanoparticles, and gene therapy products [4–11]. AUC has a long history of instrumental improvement and design of customized accessories and protocols. Recent examples include new detectors [12–14], cell alignment tools [15,16], calibration tools [17,18], a cell washer [19], 3D printed centerpieces [20,21], and data acquisition automation [22]. In the present communication we describe an apparatus that facilitates in situ sample mixing so the samples can be utilized for further investigation in AUC or other biophysical methods.

Samples in AUC are filled through filling holes into ‘centerpieces’ where they form narrow, sector-shaped solution columns that are sandwiched between optical windows. A small air volume is left in the centerpiece in order to produce a meniscus that provides a well-defined sample geometry. The entire assembly is fastened in an aluminum barrel. During sedimentation velocity (SV) experiments macromolecules usually sediment away radially from the meniscus to the distal end of the centerpiece (the ‘bottom’), where they are concentrated through the strong gravitational force. For various reasons it can be necessary to resuspend the macromolecules uniformly in solution without disassembly of the sample cell components, and without opening the vacuum-tight filling hole seals.

For example, many laboratories use a protocol where initial adjustment of the optical detection is carried out while the rotor is slowly spinning, followed by resuspension of partially settled material to recreate spatially uniform starting concentrations at the beginning of the SV experiment. While the initial sedimentation can now be accommodated through the recently introduced variable rotor speed analysis [23], sample resuspension is indispensable, for example, when using meniscus-matching centerpieces after the initial liquid transfer between sectors [24], or when rotor acceleration is inadvertently stopped due to malfunctions. More importantly, even after regular sedimentation resuspending the sample can be very useful: After the SV experiment, proteins are usually highly concentrated at the bottom of the centerpiece, and whether they can be completely resuspended or form gels or surface films will provide important clues to the presence of weak attractive interactions and protein phase behavior. Indeed, in our observation many or even most of proteins can be completely resuspended, and can be demonstrated to exhibit unaltered size-distributions and sedimentation behavior in replicate SV experiments. This can offer significant experimental flexibility when working with scarce sample, for example, for carrying out follow-up studies with the same samples at different temperatures, rotor speeds, or using other detection methods. This is a particularly timely consideration with the recently introduced opportunity to study highly concentrated proteins [25]. Bovine serum albumin (BSA) is one well-known protein example that can be completely resuspended many times, and its stability is exploited for verifying correct instrument calibration [26].

To accomplish in situ mixing of the sample it is necessary to manipulate the position of the air bubble such that it repeatedly traverses the entire solution and thereby causes convective mixing (Figure 1). In standard 12 mm pathlength centerpieces we found that inverting the sample cell relative to gravitation causes the bubble to float toward the bottom, after which a gentle slow rocking motion can move it in the direction of the optical path along the bottom, continuously displacing the solution in the highly concentrated region of the cell. However, in shorter optical pathlength centerpieces, such as 1 mm and 3 mm pathlength centerpieces, the bubble occupies virtually the entire thickness between the windows, and a different strategy is required to achieve convective mixing. Here, we found rotation of the cell assembly along the axis of the cylindrical housing barrel will effectively move the bubble up and down, repeatedly squeezing liquid sample past it and causing convection.

Figure 1.

Geometry of sedimentation showing the radially aligned sector-shaped sample cell (A) and strategies for sample mixing by movement of the airbubble (red) in long optical pathlength centerpieces (B) and short optical pathlength centerpieces (C).

Unfortunately, it can be very physically tiresome and time-consuming to manually mix all 7 or 8 samples that may be studied in a single run. This is particularly challenging for centerpieces with shorter pathlength such as 1mm and 3mm. Further, the effectiveness of the mixing is uncertain without a quantitative assessment of the concentration gradient. For this reason, we constructed an apparatus that can automate mixing in both modalities for several cells side-by-side (Figure 2) and explored the remixed samples in subsequent SV experiments to quantitatively measure the homogeneity of the samples, as well as their sedimentation behavior.

Figure 2.

AUC cell mixer with rotating beams on a rocking stage (A). The bottom pictures show a sample cell assembly containing blue dextran pelleted after SV at 50,000 rpm prior to mixing (B), during operation with an air bubble moving through the solution column and causing turbulent convective mixing (C), and after mixing (D). Movies of the operation can be found in SI Movie 1 (rotation movement) and SI Movie 2 (rotation and rocking).

The mixing device was assembled from LEGO® bricks (Billund, Denmark) to enable low-cost and open source fabrication from widely available standardized components. In the present work we used standard motor functions and gear systems (Figure 2) to construct a series of beams, formed by long axles of the gear system, on which multiple AUC cell assemblies can be placed. AUC cell assemblies are rotated through friction on the rotating supporting beams. The array of axles is mounted on top of a slowly rocking platform to cause liquid mixing both laterally and radially. A list of parts and a LEGO digital designer (lxf) file that clarifies assembly details is provided in the Supplementary Material.

We tested the performance of the mixer using Blue dextran (MW: 2,000 kDa), BSA, and thyroglobulin, diluted in phosphate buffered saline to concentrations between 0.5 mg/ml and 20 mg/ml and placed in centerpieces with 1 mm, 3 mm, and 12 mm pathlength. After complete sedimentation at 50,000 rpm, as recorded by Rayleigh interference optical detection, we placed the cell assemblies on the mixer for different length of times. For the sample with blue dextran the turbulent mixing can be visually followed (Supplementary Material SI Movie 1 and SI Movie 2). After mixing we repeated the SV experiment. By analyzing the sedimentation profiles acquired initially and after resuspending the material, it is possible to assess spatial uniformity of the macromolecules after mixing and any loss of concentration of the sedimenting sample.

Figure 3 shows an example for sedimentation profiles of a mixture of BSA and thyroglobulin in a 12 mm pathlength centerpiece before (Panel A) and after mixing (Panel B), and a superposition of the obtained sedimentation coefficient distributions in 5 consecutive SV/mixing cycles (Panel C). Quantitatively, the total sedimenting signal was constant within better than 0.2%, and the reproducibility of the major sedimentation coefficients for BSA and thyroglobulin was 0.21% and 0.32%, which is consistent with expected repeatability of SV experiments. Similar results were obtained for a sample of 20 mg/ml of BSA in 3D printed 1 mm pathlength centerpiece, exhibiting slightly lower reproducibility of the total material (0.6%) and monomer and dimer s-values (0.3% and 0.6%), as may be expected from the increased number of fitting parameters in the nonideal c(s) sedimentation model. Variation of sedimentation time at 50,000 rpm between 5 hours and 18 hours did not show any impact on mixing efficiency indicating that highly concentrated protein after long time centrifugation can be resuspended. Mixing times of 10 min or 20 min provided indistinguishable results.

Figure 3.

Sedimentation profiles of 1 mg/ml BSA with 0.5 mg/ml thyroglobulin in phosphate buffered saline in a 12 mm pathlength centerpiece. (Panel A) Original sedimentation data from the first sedimentation at 50,000 rpm. Color temperature indicates progression of time (blue to red). (Panel B) Sedimentation data after remixing the sample from (A) recorded during by a second sedimentation phase at 50,000 rpm. (Panel C) Superposition of sedimentation coefficient distributions c(s) from the initial run in (A) (black), the second run in (B) (blue), and from four more consecutive mixing/sedimentation cycles (cyan, green, red, orange).

In summary, we describe an automated, low-cost, and open source cell mixer for analytical ultracentrifugation that can efficiently resuspend pelleted macromolecules after SV experiments. This can support the work-flow of carrying SV experiments in practice. There could be variation of the effectiveness of mixing for solutions with different surface tension and viscosity, and for narrower centerpieces, as these factors impact the movement of the bubble on which the mixing relies. Thus, it is important to closely inspect the movement of the bubbles in the sample sectors when initiating the mixing process, as additional tapping or shaking may need to be applied to initially dislodge the bubbles. In this study, we have established that the mixing device performs well for conventional dilute or semi-dilute protein solutions in aqueous buffer. It is important to emphasize that whether or not a given protein can be resuspended will largely depend on its macromolecular properties, including weak associations and phase behavior. The application of the mixer allows us to more conveniently and efficiently assess the propensity for irreversible aggregation or gel formation at high concentration by comparison of the original and resuspended sample, and if the behavior is identical, to extend the experiments.

• Resuspending samples in analytical ultracentrifugation is important but tedious

• It is part of the workflow and can reveal protein properties at high concentration

• A new low-cost open-source cell mixer is described and tested