Nanodisc characterization by analytical ultracentrifugation

1 Introduction

In their native environment, membrane proteins are embedded within or attached to the lipid bilayer. They play key roles in many biological pathways, and up to 30% of eukaryotic open reading frames are predicted to encode for membrane proteins [1], [2], [3]. Membranes are not only permeable barriers that maintain the internal environment of cells but also platforms for protein-protein and protein-lipid interactions and modulators of signaling pathways, and membrane proteins represent a class of important drug targets [4], [5]. Lipid context is essential for membrane protein function [6], [7], [8], [9], [10], and the study of membrane proteins requires consideration of the membrane protein-lipid complex as the functional unit, preferably with a native lipid environment. Intact cells are best suited to elucidate the properties of membrane proteins; however, their inherent complexities complicate analysis of the target protein. A variety of methods to reconstitute membrane proteins in detergents, lipid-detergent mixtures, and lipid bilayers mixes have been developed for in vitro studies, including systems such as bicelles, liposomes, planar lipid membranes, nanodiscs, macrodiscs, amphipols, and styrene maleic acid co-polymer lipid particles [11], [12], [13], [14]. Each method has its own advantages and disadvantages, depending on the downstream assay and the membrane protein of interest. In this short review, we focus primarily on the use of nanodiscs and highlight the importance of analytical ultracentrifugation (AUC) for a complete characterization of membrane protein-lipid complexes.

A nanodisc is a discoidal phospholipid bilayer surrounded and stabilized by two molecules of apolipoprotein A-I (ApoA-I) or membrane scaffold protein (MSP) engineered from ApoA-I variants (Figure 1) [17], [18], [19], [20]. The concept is based on the natural lipid-associating properties of ApoA-I, the major component of the high-density lipoprotein (HDL) complex, which is an endogenous nanoparticle [15], [16] used for the removal of excess cholesterol from peripheral tissues and transport to the liver for secretion through reverse cholesterol transport. HDL nanoparticles also transport small molecules such as proteins [21], signal lipids [22], circulating microRNA [23], [24], hormones, and vitamins [21]. These features have led to development of diverse applications for HDL particles as vehicles for site-specific drug delivery [19], [25], [26], [27], [28] and siRNA transport [29]. In vitro, nanodiscs have been proven to be extremely useful for embedded membrane proteins providing stability, solubility, and functional activity in the context of a controllable lipid environment [30], [31], [32], [33], [34], [35], [36], [37], [38]. The basic protocol for the reconstitution of membrane proteins in nanodiscs requires the careful mixing of purified and detergent solubilized membrane protein, MSP, and detergent solubilized lipids in the appropriate ratio and subsequent removal of the solubilizing detergent [39], [40]. As the detergent is removed, the membrane protein associates with the phospholipids, as these assemble into the MSP stabilized discoidal structures. The self-assembly process can be optimized in many ways depending on the protein of interest and specific application – for example, with MSP proteins of different lengths, nanodiscs with diameters of 6 to 17 nm can be prepared [18], [30], [31]. The lipid composition can be controlled based on the initial lipid content. By choosing the appropriate MSP and controlling the initial protein and lipid ratios, it is possible to regulate the stoichiometry of the membrane protein within the nanodisc [41], [42], [43].

Figure 1:

Schematic representation of a membrane protein embedded in a nanodisc. Monomeric rhodopsin (blue, PDB 1U19) is embedded within a central phospholipid bilayer core (gray) stabilized and surrounded by two molecules of a MSP (pink). MSPs are engineered forms of the apolipoprotein A-I, the major protein of HDL. ApoA-I has natural lipid-associating properties stabilizing a discoidal lipid bilayer of ~10-nm diameter [15], [16].

The importance of nanodisc technology for the study of membrane proteins is illustrated by many studies, including ones demonstrating that lipids modulate the function of G-protein-coupled receptors (GPCRs) and their signaling pathways [34], [35], [42], [44]. GPCRs play key roles in signal transduction [45] – ligand binding on the extracellular binding site of the receptor induces conformational changes that permit binding to the heterotrimeric G proteins within the cytoplasm, triggering downstream signaling pathways. Termination of GPCR signaling is brought about by phosphorylation of the receptor with GPCR kinases and subsequent arrestin binding, blocking the interaction between activated GPCRs and G proteins [46]. Using a monomeric neurotensin receptor 1 (NTS1) reconstituted into nanodiscs of 10–12 nm diameter, it has been shown that neurotensin binding to NTS1 is not influenced by the local lipid environment. However, the local lipid charge modulates the downstream interaction with Gq-type protein [34]. Furthermore, selective phosphorylation of NTS1 by the GPCR kinases GRK2 and GRK5 requires negatively charged lipids in vitro [47]. In addition to providing a native-like lipid environment, nanodiscs have also been shown to stabilize native protein structures [31]. By using truncated MSPs, smaller nanodiscs with diameters of 6–9 nm have been developed specifically for solution nuclear magnetic resonance (NMR) spectroscopy. This allowed for comparative structural studies of the outer membrane protein X (OmpX) solubilized in detergent or reconstituted in small nanodiscs, showing differences in the structure of the extracellular extensions of the β-sheets proximal to the membrane surface. By measuring the dynamics of these regions in the nanosecond to picosecond timescale, Hagn and co-workers suggested that detergents denature membrane proximal extracellular regions of membrane proteins, thus disrupting biologically relevant interactions. Nanodiscs, on the other hand, maintain a functional and near native environment [31]. Another major advantage of this technology is the ability to assemble, in vitro, multiprotein complexes for structural and functional studies, as exemplified by the resistance nodulation and cell division (RND) efflux system [36]. This superfamily of proteins is an exporter of biological metabolites and antimicrobial compounds that play a prominent role in bacterial drug resistance. The RND pumps located within the inner membrane are driven by the proton motive force working in conjunction with an outer membrane factor (OMF) and a periplasmic membrane fusion protein (MFP) as a part of a tripartite system with the MFP linking the RND component to OMF. Daury and co-workers [36] have reconstructed RND tripartite complexes using the self-assembly properties of nanodiscs during reconstitution. In this approach, both the inner membrane RND pump and outer membrane OMF are embedded within separate nanodiscs, with the tripartite complex assembled by the lipidated MFP bridging the solubilized RND and OMF. Using this technique, the authors demonstrated the intrinsic ability of the native components to self-assemble and thus highlighted the role of the periplasmic MFP as part of the efflux pump exit duct.

These examples demonstrate the importance of nanodisc technology for the study of membrane proteins. An imperative consideration in such studies is the need to characterize the nanodisc preparation in terms of its homogeneity, shape, size, lipid, and protein stoichiometry, as well as the oligomerization status of the membrane protein of interest. Several methods are available for the characterization of membrane proteins embedded in nanodiscs (Figure 2) [18], [34], [38], [41], [42], [43], [44], [48], [49], and a complete characterization may necessitate a combination of these listed methods, with some requiring sample modification or labeling (e.g. fluorescence, NMR). AUC does not require sample modification and provides the most information in terms of sample polydispersity and overall protein and lipid stoichiometry [40].

Figure 2:

Overview of biophysical methods for nanodisc characterization.

2 Nanodisc characterization by AUC

AUC is a method ideally suited for the characterization of macromolecules and macromolecular assemblies, such as nanodiscs [40]. As the name implies, the sample is subjected to a high centrifugal force by spinning it in a rotor at high speed (ultracentrifugation), and the progression of sedimentation is monitored in real time (analytical) using either UV-visible absorbance, Rayleigh interference, or fluorescence emission [50]. The process of sedimentation depends on the sedimentation and diffusion coefficients intrinsic to the material under study, and the method can be used to determine the molar mass, size, and shape of the macromolecule of interest (Figure 3) [52], [55]. In the case of polydisperse systems, information on the size distribution can be obtained [52]. AUC has long been used to study detergent or lipid solubilized membrane proteins; however, the application of the method has undergone a renaissance since the review by Tanford and Reynolds [56]. This is mainly due to the development of computational tools for the analysis of sedimentation velocity (SV) data in terms of diffusion deconvoluted sedimentation coefficient c(s) distributions [53], the ability to combine data from different optical detection systems [57], [58], [59], and the use of density contrast with D₂O or H₂¹⁸O [60] to obtain information on the membrane protein and detergent or lipid stoichiometry. This biophysical methodology is the subject of many excellent reviews [60], [61], and we highlight software tools developed in GUSSI [51] based on the experimental c(s) sedimentation coefficient distributions and protocols developed by Ebel and co-workers [58], [59] to determine the composition of the membrane-detergent/lipid complex by SV-AUC. Membrane protein nanodiscs, unlike detergent solubilized membrane proteins, represent chemically stable, discrete species with a fixed lipid, membrane scaffold, and membrane protein stoichiometry. Their stability does not require the presence of free detergent, simplifying their biophysical characterization and facilitating functional and structural analyses. The absence of free detergent simplifies an analysis by sedimentation equilibrium (SE) AUC (Figure 4) [62], and nanodiscs composed solely of MSPs and lipid (here referred to as empty nanodiscs) can be used to characterize lipid properties such as refractive index increments and partial specific volumes by AUC.

Figure 3:

Sedimentation velocity of MSP1E3D1 nanodiscs reconstituted with POPC. (A) Absorbance sedimentation velocity scans for a sample of MSP1E3D1/POPC empty nanodiscs collected at 42,000 rpm, 20°C and 280 nm. (B) Interference sedimentation velocity scans for the same sample of MSP1E3D1/POPC empty nanodiscs collected at 655 nm during the same SV-AUC run. In both cases, radial scans were collected at approximately 3.7-min intervals, and for the purposes of clarity, only every third scan and third data point is shown. Solid lines show the best fits from a continuous c(s) distribution of sedimenting species; the corresponding residuals are shown in the plot below. Data were plotted in GUSSI [51]. The best fit absorbance (C) and interference (D) c(s) distributions show the presence of a major species at 3.17 S having a frictional ratio of 1.2 and an estimated molar mass of 265 kDa. Based on the sedimentation coefficient, a hydrodynamic diameter of 10.1 nm is determined for the empty nanodiscs. The absorbance signal for this species, obtained by integration of the 3.17 S c(s) species observed, reports on the nanodisc protein content, whereas the corresponding interference signal reports on both the protein and lipid content. Using the consensus lipid refractive index increment in Table 1, a lipid content of 200 POPC per nanodisc is determined. The c(s) distributions support a nanodisc loading concentration of 17 μM, along with evidence for traces (~4% based on the interference signal) of 550 kDa and larger species. The absence of detergent used for nanodisc reconstitution is noted by the absence of slower sedimenting species in the interference SV-AUC data and corresponding c(s) distribution. In SV-AUC, a homogenous solution of the sample of interest is analyzed at the highest possible rotor speed. The process of sedimentation is governed by the Lamm equation c(r,t)=−1r∂∂r(csω2r2−Dr∂c∂r) that relates the progression of the radial r- and time t-dependent concentration c(r, t) of the material with a sedimentation coefficient s and a diffusion coefficient D, where ω is the angular rotor speed [52], [53], [54]. In a continuous c(s) distribution analysis, experimental SV scans are numerically fitted in terms of a distribution of Lamm equation solutions spanning a range of sedimentation coefficients. The frictional ratioff0 is used as a fitting parameter to describe the diffusion coefficient D(s), and the molar mass M(s) of each resolved species can be determined based on the Svedberg equation M(s)=sRTD(s)(1−v¯ρ), where R is the gas constant, T the absolute temperature, ν̅ the partial specific volume of the species of interest, and ρ the solvent density [40], [52], [53]. Experiments are carried out following standard protocols, as described in Refs. [40], [50], [54] and references cited therein. Briefly, 400 μl of a nanodisc sample and 400 μl of a matching reference buffer are loaded into standard two channel sector shaped AUC cells and temperature equilibrated at 0 rpm under vacuum. The rotor is accelerated to 42,000 rpm, and SV-AUC absorbance and interference data are collected until all of the material has sedimented to the bottom of the cell [54]. Data are subsequently analyzed in SEDFIT in terms of a c(s) distribution using experimentally determined or calculated values for the solvent density ρ, solvent viscosity η, and nanodisc partial specific volume ν̅ [52], [53]. Even though SV-AUC experiments presented here were carried out at 20°C, we routinely carry out such experiments at 10°C (Figure 5) due, in part, to concerns regarding sample stability and the fact that lipid partial specific volumes were determined at 10°C (Figure 4). Within the error of the method, we obtain identical sedimentation coefficients, molar masses, and lipid stoichiometries for both the empty and NTS1 nanodiscs irrespective of whether SV data were collected at 10°C or 20°C.

Figure 4:

Sedimentation equilibrium profiles for MSP1E3D1/POPC empty nanodiscs at 10°C plotted as a distribution of the absorbance at 280 nm versus radius. Absorbance sedimentation data were collected at 10,000 (red), 15,000 (blue), and 20,000 (green) rpm and a loading concentration corresponding to 0.68 A₂₈₀ in a 1.2 cm pathlength cell. In SE, the opposing transport forces of sedimentation and diffusion balance out, resulting in no net flux of material, and as described below, equilibrium is usually reached only after a sufficiently long time has passed, typically of the order of days. As shown, data are typically collected at three rotor speeds chosen such that the lowest rotor speed results in a shallow concentration gradient with the ratio of the signal at the cell bottom (high radius) to the signal at the meniscus (low radius) of approximately 3–4. The highest rotor speed is chosen such that meniscus depletion is observed, along with a steep concentration gradient. In this manner, the method can report on all of the species present in solution and test for sample monodispersity [40], [54], [55]. At sedimentation equilibrium, an exponential concentration gradient c(r)=c(r0)exp[M(1−v¯ρ)ω2RT(r2−r02)] is observed for a single non-interacting species, where R is the gas constant, T is the absolute temperature, and r, r₀, and ω are the radial position, a reference radial position, and the angular rotor speed, respectively. This concentration distribution is a special solution to the Lamm equation for a single species. Experiments are carried out following standard protocols, as described in Refs. [40], [50], [54], [56] and references cited therein. Briefly, 160 μl of a nanodisc sample and 160 μl of a matching reference buffer are loaded into standard two channel sector shaped AUC cells. Data are initially collected at 10,000 rpm and 6-h intervals until equilibrium is reached, as determined by pairwise comparison of scans using the equilibrium tool functions in SEDFIT, GUSSI [51], or WinMATCH. Once equilibrium is reached at 10,000 rpm, the rotor speed is increased to 15,000 rpm and the process repeated at this and the subsequent rotor speed. Seventy-two hours was required to reach equilibrium at the initial rotor speed, and 48 h, for the higher rotor speeds. Because of the long times for SE-AUC experiments, we routinely collect data at 10°C. Data were sorted in GUSSI [51] and analyzed globally in SEDPHAT (http://www.analyticalultracentrifugation.com) in terms of a single non-interacting species to obtain a value for the buoyant molar mass M(1−v¯ρ), where M is the molar mass, ν̅ is the partial specific volume, and ρ is the solvent density. The excellent fits observed confirm that the nanodisc preparation is monodisperse and homogenous. As the buoyant molar mass is the sum of the buoyant molar masses of the MSP1E3D1 and POPC, the best-fit value of the buoyant molar mass allows for a determination of the lipid partial specific volume. The solid lines show the best fit obtained, with the corresponding residuals in the plot below. Data are plotted in GUSSI [51], and for clarity, only every other experimental point is shown.

Nanodiscs, with or without an embedded membrane protein, have been characterized using a variety of biophysical methods that report on their overall size and shape, as well as the properties of the membrane protein. Size exclusion chromatography (SEC) is widely employed as the final purification step, and in combination with the analysis of a set of calibration proteins, SEC provides an estimate of the hydrodynamic radius [30]. The method is readily accessible; however, it is not universally applicable as we have found that the nanodisc embedded NTS1 is not stable following SEC, even though other nanodisc embedded membrane proteins have been purified and characterized this way [63], [64], [65], [66]. Another method for the characterization of nanodiscs is dynamic light scattering (DLS). We have previously discussed the use of DLS [40]; however, when compared to SV-AUC, the method exhibits poor hydrodynamic resolution and requires pure material to report the correct hydrodynamic radius. Sample purity is critical as DLS measurements report on all of the species present in solution. As the measurement is based on the intensity of scattered light, DLS is highly sensitive to the presence of larger sized impurities. The intensity of scattered light scales as the sixth power of the hydrodynamic radius, and in the case of paucidisperse systems, DLS reports a z-average hydrodynamic radius [67]. Static light scattering can provide information on the nanodisc molar mass and radius of gyration, and the combination of SEC with multi-angle light scattering (SEC-MALS) has been used to show that the active forms of nanodisc embedded Staphylococcus aureus accessory gene regulator protein-A and -C are dimers [68], [69]. Related scattering methods, specifically small-angle X-ray (SAXS) and small-angle neutron (SANS) scattering, have been used to characterize both empty and membrane protein nanodiscs. SAXS was used to characterize empty nanodiscs and show that their radius of gyration depended on the MSP used [30]; these differences reflect a change in nanodisc radius as the scattering data support a constant nanodisc thickness. Similar studies on bacteriorhodopsin embedded nanodiscs confirmed that the bacteriorhodopsin trimer spans the phospholipid bilayer [41]. The use of SAXS to confirm proper membrane protein insertion for Agrobacterium curdlan synthase [70], human cytochrome P450 [71], into nanodiscs has also been reported. Empty [72], [73], [74], [75] and membrane protein nanodiscs [66] have been characterized by SANS and neutron reflectivity. Both SEC and DLS allow for a rapid characterization of the nanodisc preparation and can be quite useful when optimizing the conditions for membrane protein reconstitution; however, due to their poor hydrodynamic resolution, the methods may not necessarily identify the presence of small amounts of impurities. In addition, these methods do not provide definitive information on the lipid and membrane protein stoichiometry, which may be important for downstream applications.

The limitations of these methods highlight the need for an alternative, like AUC, and we have previously demonstrated the utility of AUC for the characterization of nanodisc embedded membrane proteins [40]. Specifically, we have utilized SV-AUC to determine the purity and polydispersity of the nanodisc preparations, and determine the sedimentation coefficient, molar mass, and hydrodynamic radius of the resolved major species (Figures 3 and 5). By simultaneously collecting absorbance and interference sedimentation data, we were able to independently estimate the protein-to-lipid ratio for each of the resolved species. We have also utilized SE-AUC to determine the buoyant molar mass of empty nanodiscs and thus the partial specific volumes of the lipids (Figure 4). As both the lipid partial specific volume and interference refractive index increment are key parameters for nanodisc characterization, we briefly highlight their importance and refer the readers to Refs. [40], [54] for a more complete description of current AUC methodology. When compared to other biophysical methods, AUC has a number of advantages, which allow for a more complete characterization. Unlike diffusion-based methods where the resolution scales as the inverse of the particle radius, SV-AUC resolves particles based on their radius squared [50] allowing for unsurpassed resolution (Figures 3 and 5). This dependency arises from the value of the sedimentation coefficient (s) that scales proportionally to the molar mass (M) and inversely to the frictional coefficient (f):

Figure 5:

Sedimentation velocity of MSP1D1 nanodiscs reconstituted with POPC (red curves) or POPG (blue curves). (A) Absorbance c(s) distributions for NTS1 reconstituted into nanodiscs. The integral of the distribution reports on the scaffold and membrane protein concentrations. (B) The corresponding interference c(s) distributions for the same sample of NTS1 reconstituted into nanodiscs. Absorbance and interference data were collected in the same SV experiment. The integral of the interference distribution reports on the contributions from scaffold and membrane proteins, as well as the lipids, allowing for an estimate of the lipid stoichiometry in the sedimentation region of interest. The major species observed at 6.4–6.9 S represent the MSP1D1 lipid nanodisc containing a single copy of NTS1. (C) Interference c(s) distributions for empty MSP1D1 nanodiscs highlighting the differences in density between POPC (red curve) and POPG (blue curve) and confirming the nature of the lipid reconstitute. Sedimentation coefficient differences for NTS1 reconstituted nanodiscs are less pronounced due to the smaller proportion of lipid. See reference [40] for more details. Sedimentation velocity experiments were carried out at 40,000 rpm and 10°C. Based on the experimental sedimentation coefficients and calculated molar masses, we obtained hydrodynamic radii of 4.2 nm and 3.6 nm for the empty MSP1D1/POPC and MSP1D1/POPG nanodiscs, respectively [40]. Using the reported diameter (2R_disk) and thickness (L_disk) for MSP1D1/POPC nanodiscs, based on SAXS data [30], to calculate the parameter α=Ldisk/2Rdisk and equation (3) in [76], the experimental hydrodynamic radius for the MSP1D1/POPC nanodiscs corresponds to a disk radius of 4.5 nm. Hydrodynamic radii of 5.4 nm and 5.2 nm are determined for NTS1/MSP1D1/POPC and NTS1/MSP1D1/POPG nanodiscs, respectively [40].

s=M(1−ν¯ρ)Nf

where N is Avogadro’s number, ν̅ is the partial specific volume, and ρ is the solution density. Both the value of the sedimentation coefficient and best-fit estimate of the corresponding molar mass can be used to determine the loading stoichiometry of the membrane protein onto the nanodisc. The determination of the molar mass requires the use of the proper value of the partial specific volume of the nanodisc [52]. It contains contributions from MSP, the embedded membrane protein, and constituent lipids, as noted in equation 3.11 of Inagaki et al. [40]. Protein partial specific volumes can be reliably calculated based on the amino acid sequence in SEDNTERP (http://sednterp.unh.edu [77]) or the utility calculator functions in SEDFIT (http://www.analyticalultracentrifugation.com). Limited data are currently available for lipid partial specific volumes [78], [79], [80], [81]. We therefore determined these values experimentally by SE-AUC of the monodisperse empty nanodiscs and reported partial specific volumes of 0.981 cm³/g for 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) in MSP1D1 and MSP1E3D1 nanodiscs (Figure 4) and 0.968 cm³/g for 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (POPG) in MSP1D1 nanodiscs at 10°C [40], [47]. As noted in Table 1, MSP1D1 and MSP1E3D1 are derivatives of human ApoA-I. MSP1D1 is used to prepare POPC containing nanodiscs with diameters of approximately 9.7 nm, whereas the use of MSP1E3D1 results in larger nanodiscs with diameters of approximately 12.1 nm and a corresponding larger proportion of lipid [18], [30]. The difference in the lipid partial specific volumes further manifests itself in slight differences of the measured sedimentation coefficient with lipid composition, particularly for the empty nanodiscs where the lipid content is highest (Figure 5C). This further highlights the additional resolution of SV-AUC, provided through differences in the density of the sedimenting nanodisc as a function of lipid composition. It is interesting to note that partial specific volumes of 0.986 cm³/g and 0.938 cm³/g are calculated for POPC and the anionic form of POPG at 25°C based on their chemical composition [82]. While these calculations confirm the experimentally observed trends in the dependence of the overall partial specific volume as a function of the lipid head group, they do not account for the actual differences that contain contributions from temperature, differential lipid head group hydration, and counterion binding [83].

Determining the partial specific volume for the sedimenting particle is important, as both the sedimentation coefficient in SV-AUC and the experimental concentration gradient in SE-AUC (Figure 4) depend on the buoyant molar mass M(1−v¯ρ). To a first approximation, the partial specific volume is the inverse of the particle density, and the buoyancy correction factor −M(v¯ρ) accounts for the mass of displaced solvent reflecting the basic fact that sedimentation experiments are essentially exercises in density contrast. As described in Ref. [50], the sedimenting particle also incorporates contributions from hydration and bound counterions – hydration contributions are minimal in buffers with a density close to that of water and the effects of bound counterions are usually incorporated into an experimentally determined effective partial specific volume. SV c(s) distributions allow for a determination of molar masses of the resolved non-interacting species, based on the best-fit frictional ratio (Figures 3 and 5). Molar mass estimates depend critically on the use of the correct partial specific volume [52], and in the case of nanodiscs, we have found that the precision of the mass estimate appears to correlate inversely with lipid content (e.g. see Table 3.1 of [40] and correction in [84]). This more than likely arises from the fact that the lipid partial specific volumes are nearly matched by the buffer density, leading to large uncertainties in the lipid contribution. Independent information on the lipid contribution can be obtained from a comparison of the absorbance and interference signals for selected portions of the c(s) distribution. The absorbance signal accounts for the MSPs stabilizing the nanodisc and the inserted membrane protein, whereas the interference signal also has contributions from the lipid.

The interference signal increment, analogous to the absorption extinction coefficient, can be determined from the refractive index increment dn/dc, and in our previous work [40], we have used refractive index increment values of 0.185 cm³/g for proteins [85] and 0.164 cm³/g for lipids [86]. Using absorbance and interference SV data collected for preparations of MSP1D1 and MSP1E3D1 empty nanodiscs, we have re-evaluated the value of dn/dc for lipids within nanodiscs. We have determined MSP extinction coefficients based on their composition using the method of Pace et al. [87] using the calculator in SEDNTERP [77]. Refractive index and interference signal increments were determined based on their composition using the calculator in SEDFIT [88] and a complement of 110 and 240 lipids for the empty MSP1D1 and MSP1E3D1 nanodiscs, respectively [30]. Table 1 summarizes our results with an average lipid refractive index increment of 0.146 cm³/g, a value significantly different from the literature value of 0.164 cm³/g [86]. However, we note that the latter is based on a lipid volume, determined using a cross-sectional area from X-ray diffraction and a monolayer lipid thickness, and a lipid refractive index of 1.45. Conversely, the value we report depends critically on the assumed lipid:MSP stoichiometry – the values we use reflect the lipid:MSP stoichiometry used for reconstitution, which is essentially based on SAXS dimensions [30] and theoretical values for the lipid surface area [39]. The refractive index increment determined also depends on the values used for the MSP absorption extinction coefficient and interference signal increment. The issue of average lipid stoichiometry is clearly important, and based on the small data set presented in Table 1, it would appear as though lipid dn/dc values for the smaller MSP1D1 nanodiscs are slightly higher than those determined for the larger MSP1E3D1 nanodiscs. This may reflect errors in the assumed stoichiometry. Much like the density increment ∂ρ∂c≡(1−v¯ρ), the refractive index increment determined in this manner will also contain hydration and bound counterion contributions. We further note that the value presented here is similar to values of 0.138 cm³/g and 0.146 cm³/g determined experimentally for the related 1,2-dioleoyl-sn-glycero-3-phosphocholine (DPPC) and 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DOPC), respectively [89].

In addition to our work on NTS1 [34], [40], [47], AUC has been used by other groups to characterize nanodisc preparations. Alami and co-workers used SV c(s) profile to characterize their nanodisc SecYEG complex [90], whereas Xu and co-workers combined SV absorbance and interference signals to determine the sedimentation coefficients, molar masses, and lipid to protein stoichiometries of both a KcsA-Kv1.3 chimeric ion channel nanodisc and the corresponding empty MSP1D1/DMPC nanodiscs [91]. As the MSP was labeled with an Alexa Fluor 488 tag, Xu and co-workers also carried out SV experiments by monitoring the absorbance at 496 nm, along with separate experiments utilizing a fluorescence detection system [91]. Altogether, these highlight the usefulness of AUC for the characterization of nanodisc preparations – a single multisignal SV experiment collecting both absorbance and interference data can report on the monodispersity of the nanodisc preparation with high hydrodynamic resolution, as well as the lipid and protein stoichiometries. The method has the potential to study the interaction of the nanodisc solubilized membrane protein with physiological partners in vitro [92], with access to the study of high affinity interactions when using the fluorescence detection system [93].

3 Current state of the art

Nanodiscs, originally developed as a method to deliver drugs [26], [94] or solubilize membrane proteins [38] for structural and functional studies, have now found applications in the proteomic and biomedical fields. Roy and co-workers [95] have used nanodiscs to create a membrane protein library for a proteomic analysis of detergent-extracted membrane proteins obtained from mammalian cells, improving upon similar protocols developed for bacteria [96] and yeast [97]. These protocols led to the identification of low-abundance membrane proteins not identified by standard proteomic methods; in addition, by adjusting the lipid composition of the reconstitute, the authors demonstrated the differential partitioning of membrane proteins, again highlighting the importance of the local lipid environment. Importantly, the membrane proteins extracted using this method retain their functional activity in vitro, also exemplified by recent studies reporting on the preparation of a synaptic membrane protein nanodisc library to identify targets that bind β-amyloid oligomers [98]. Interestingly, Smith and co-workers have demonstrated that a recombinant Factor VIII lipid nanodisc complex, when administered retro-orbitally to mice with hemophilia A, has a higher pro-coagulant effect compared to either the recombinant Factor VIII or the lipid nanodisc alone [99]. Because the nanodisc represents a reconstructed HDL-like particle, the particle is not expected to be immunogenic and the authors demonstrate similar immunogenicity for the administration of the recombinant Factor VIII alone or as a nanodisc complex. Empty nanodiscs were also used as a template for ⁶⁴Cu incorporation and applied as contrast agent for in vivo imaging of mice [100]. In addition to demonstrating the accumulation of the contrast agent in tumors, liver, and kidney, the authors noted minimal accumulation in the spleen, indicative of low immunogenicity.

The potential use of nanodiscs for biomedical applications and drug delivery will require a thorough characterization of the preparation and formulation. There is a vast body of work on the biophysical characterization of biopharmaceuticals, and we highlight the importance of AUC [101] for the accurate quantitation of aggregates [102], [103], the characterization of high concentration formulations [104], and the possibility to utilize fluorescence detection AUC to characterize a fluorescently labeled nanodisc drug or contrast agent delivery system in the serum following delivery [105].