Thermotropic properties of phosphatidylcholine nanodiscs bounded by styrene-maleic acid copolymers

Styrene-maleic acid copolymers (SMA) have been gaining interest in the ﬁ eld of membrane research due to their ability to solubilize membranes into nanodics. The SMA molecules act as an amphipathic belt that surrounds the nanodiscs, whereby the hydrophobic styrene moieties can insert in between the lipid acyl chains. Here we used SMA variants with di ﬀ erent styrene-to-maleic acid ratio (i.e. 2:1, 3:1 and 4:1) to investigate how lipid packing in the nanodiscs is a ﬀ ected by the presence of the polymers and how it depends on polymer composition. This was done by analyzing the thermotropic properties of a series of saturated phosphatidylcholines in nanodiscs using laurdan ﬂ uorescence and di ﬀ erential scanning calorimetry. In all cases it was found that the temperature of the main phase transition (T m ) of the lipids in the nanodiscs is downshifted and that its cooperativity is strongly reduced as compared to the situation in vesicles. These e ﬀ ects were least pronounced for lipids in nanodiscs bounded by SMA 2:1. Unexpected trends were observed for the calorimetric enthalpy of the transition, sug-gesting that the polymer itself contributes, possibly by rearranging around the nanodiscs when the lipids adopt the ﬂ uid phase. Finally, distinct di ﬀ erences in morphology were observed for nanodiscs at relatively high polymer concentrations, depending on the SMA variant used. Overall, the results suggest that the extent of preservation of native thermodynamic properties of the lipids as well as the stability of the nanodiscs at high polymer concentrations is better for SMA 2:1 than for the other SMA variants.


Introduction
Styrene-maleic acid copolymers (SMA) are rapidly gaining interest as tools in membrane protein research due to their capacity of solubilizing biological membranes into nanodiscs without the need for detergents. These so-called "native nanodiscs" thus enclose membrane proteins embedded in a lipid bilayer (for review see ), while retaining native protein-lipid interactions (Dörr et al., 2014;Long et al., 2013;Prabudiansyah et al., 2015;Swainsbury et al., 2014). SMA variants are commercially available differing in average length of the polymer and in average styrene-to-maleic acid ratio. The variants most frequently used in literature are SMA 2:1 (styrene-to-maleic acid ratio of 2:1) (Dörr et al., 2014;Jamshad et al., 2015;Swainsbury et al., 2014) and SMA 3:1 (Cuevas Arenas et al., 2016;Dominguez Pardo et al., 2017;Orwick et al., 2012) of approximately ∼10 kDa. Studies in model membranes showed that the SMA 2:1 variant is slightly more efficient in solubilization than SMA 3:1, while both are significantly more efficient than SMA 4:1 . Similar results were reported for solubilization of proteins from E.coli (Morrison et al., 2016).
A general advantage of using SMA-bounded nanodiscs for membrane protein research is that the bilayer environment reflects the organization of the lipids in the native membrane (Jamshad et al., 2015;Orwick et al., 2012). However, for studies on membrane protein structure and function it is also important to know to what extent the packing of the lipids in the nanodiscs is preserved and resembles that in the native membrane. Previous reports showed that SMA molecules behave as a belt encircling the nanodiscs, where the styrene units intercalate between the lipid acyl chains (Jamshad et al., 2015;Orwick et al., 2012). This intercalation will affect lipid packing and it is thus likely that the properties of the nanodiscs and packing of the lipids will be affected by the composition of the SMA molecules that form the belt. This is supported by the recent observation that SMA composition affects the functional properties of proteins in purified nanodiscs (Morrison et al., 2016).
A convenient way of monitoring the extent to which native lipid packing properties are preserved in nanodiscs is by investigating the thermotropic behavior of the enclosed lipids as compared to that of lipids in vesicles (Denisov et al., 2005;Jamshad et al., 2015;Orwick et al., 2012;Shaw et al., 2004;Tanaka et al., 2015). In order to gain insight into the effect of polymer composition on the properties of the formed nanodiscs and in particular on the extent to which native membrane properties are preserved, we here directly compared the effects of different polymer variants on the thermotropic properties of SMA-bounded nanodiscs. Such a direct comparison is essential, because it is becoming increasingly clear that the properties of nanodiscs are dependent on precise experimental conditions, and can vary for example with SMA concentration (Cuevas Arenas et al., 2016;Oluwole et al., 2017).
To allow comprehensive and systematic comparison, we studied a series of saturated phosphatidylcholine lipids in SMA-nanodiscs by laurdan fluorescence and DSC analysis. Briefly, we found that lipids in SMA 2:1-nanodiscs retain to a higher extent the thermodynamic properties of the native membrane as compared to lipids in SMA 3:1-nanodiscs under a variety of experimental conditions. Moreoever, upon varying the SMA concentration both the thermodynamic properties and the morphology of the nanodiscs were best retained in SMA 2:1-nanodiscs. These results are discussed in terms of the membrane interactions of the styrene units in SMA and how they may depend on the phase state of the lipids.

Preparation of multilamellar vesicles (MLVs)
Phospholipid stock solutions were prepared in chloroform/methanol (9:1 v/v) in concentrations of either 5 mM or 20 mM (for DSC measurements), based on the analysis of total phosphate (Rouser et al., 1970). A 5-mM laurdan stock solution was prepared in ethanol/ chloroform (1:1 v/v). Aliquots from the phospholipid stock solutions and from the laurdan stock solutions, if required, were taken and the solvent was removed under a stream of N 2 . The resulting lipid film was dried in a desiccator under vacuum for at least 1 h. Multilamellar vesicles (MLVs) were obtained by hydrating the lipid films with buffer (50 mM Tris-HCl, 150 mM NaCl, pH 8.0) to a final concentration of 0.5 mM or 20 mM (for DSC measurements). The samples were then subjected to 10 freeze-thaw cycles, each consisting of 3 min freezing in liquid N 2 (-196°C) and 3 min thawing in a water bath above T m of the lipids (Marsh, 2013).

Preparation of nanodiscs
Dispersions of MLVs containing either 0.5 mM or 20 mM lipid were incubated with SMA (SMA-to-lipid mass ratios used for solubilization are detailed in the legends of each figure) for 1 h at the phase transition temperature (T m ) of the lipid components (Lewis et al., 1987;Marsh, 2013). After incubation with SMA, the samples were placed in an Optima Max ultracentrifuge (Beckman-Coulter, Brea, CA). Traces of nonsolubilized material were removed by centrifugation at 115,000 × g for 1 h at 4°C.

Laurdan fluorescence
Laurdan fluoresence was measured using a Varian Cary Eclipse fluorescence spectrophotometer (Santa Clara, CA). 1-mL aliquots of ∼0.5 mM phosphatidylcholine nanodisc solutions (lipid-to-laurdan molar ratio of 200:1) were placed in a 10-mm quartz cuvette and excited at 340 nm. The excitation and emission slits were adjusted to 5 nm. Emission spectra were recorded in the range of 400-550 nm at a speed of 60 nm/min. The temperature was controlled with a Peltier cuvette holder (Santa Clara, CA) in the range of 5-70°C. Exact temperatures of the samples were obtained from a thermosensor dipped inside the cuvette. Generalized polarization parameters defined as GP = (I 440nm -I 490nm )/(I 440nm + I 490nm ) (Parasassi et al., 1991) were obtained as function of temperature. Phase transition midpoints were determined by nonlinear least-squares fitting using a sigmoidal fitting function (Kemmer and Keller, 2010). GP values were obtained from a single heating scan. For selected samples additional heating cycles were performed as a control and negligible differences in the GP curves were observed.

Differential scanning calorimetry
DSC measurements were performed using a Discovery DSC (TA Instruments, Newcastle, DE) calorimeter. 10-μL aliquots of nanodisc solutions or MLV dispersions containing ∼20 mM lipids were placed in hermetic Tzero pans (TA Instruments, Newcastle, DE). Heating curves were recorded in the ranges of 5-70°C at a scan rate of 5°C/min at least 3 times. Calorimetric enthalpies (ΔH cal ) were determined from the area under the peak of the main phase transition using Trios software (TA Instruments, Newcastle, DE). The calorimetric enthalpies reported represent the average obtained from the 2nd and the 3rd heat scan from 2 independent samples. Error bars reported for both T m and ΔH cal values correspond to the average from the 2nd and 3rd heating cycles from 2 independent samples.The full width at half height of the peaks (ΔT 1/2 ) were determined manually using Adobe Illustrator software (San José, CA). For determination of the calorimetric enthalpy, the total lipid material contained in the calorimetric pans was recovered by perforating the pans in a glass tube containing 1-mL of water. The tubes were vortexed vigorously in order to assure that all lipids were extracted from the pans. Next, the total lipid was quantified in triplicate by applying a phosphate analysis (Rouser et al., 1970) on 200-μL aliquots of the resulting nanodisc solutions or MLV dispersions.

Transmission electron microscopy
Size characterization of di-16:0 PC and di-18:0 PC nanodiscs was performed by transmission electron microscopy (TEM). 1-mL aliquots of 0.5 mM nanodisc solutions were adsorbed on carbon-coated mica following the carbon flotation technique and stained with a staining solution containing 2% (w/v) sodium silicotungstate as detailed before . Images were taken under low dose conditions at a nominal magnification of 49,000 with a T12 electron microscope (FEI, Hillsboro, OR) at an operating voltage of 120 kV using an ORIUS SC1000 camera (Gatan, Inc., Pleasanton, CA). The average size of the nanodiscs was estimated from at least 25 well-defined individual particles. The maximum diameter was determined using Adobe Illustrator software (San Jose, CA).

Results
3.1. Native thermotropic lipid properties are better retained in SMA 2:1 nanodiscs than in SMA 3:1 nanodiscs Shifts in phospholipid phase behavior can be conveniently followed by incorporating amphipathic solvatochromic dyes, such as laurdan, into lipid membranes. Laurdan probes are extremely sensitive to solvent relaxation and can be used to track changes in membrane fluidity based on a shift in emission maximum. Briefly, when the temperature is raised above the main gel-to-fluid phase transition temperature (T m ) of the lipid constituents of the membrane, a spectral shift of the emission maximum takes place from 440 nm to 490 nm (Parasassi et al., 1991). This is illustrated in Fig. 1A where laurdan is incorporated into di-16:0 PC MLVs, which have a T m of around 41°C (Lewis et al., 1987). Below T m , at 26°C and 35°C, the emission maximum is at 440 nm. At 42°C, the intensity of the emitted fluorescence is decreased at 440 nm but is increased at 490 nm. At higher temperatures of 46°C and 51°C, the emission maximum is completely shifted to ∼490 nm.
Insight into the relative amount of gel and fluid phase present can be obtained by calculating the generalized polarization (GP) values from the maximum intensities of the two emission peaks (Parasassi et al., 1991). As shown in Fig. 2 (green line) for di-16:0 PC MLVs the GP values slightly decrease as the temperature is raised, with a sharp inflection around T m , indicating that phase interconversion is taking place with a high degree of cooperativity (Parasassi et al., 1994). By applying non-linear regression analysis T m was found to be ∼41°C.
The same experimental approach was used for nanodiscs obtained after addition of SMA to di-16:0-PC MLVs. We inititally focussed on SMA 2:1 and SMA 3:1, as they are the commonly used SMA variants in literature. As shown in Fig. 1B, for SMA 2:1 a decrease in the emission at 440 nm is observed that now begins already at lower temperatures in the range of 27°C to 36°C. In addition, there is no clear peak appearing at 490 nm upon further increasing the temperature. Similar but more pronounced effects are observed for di-16:0 PC SMA 3:1 nanodiscs (Fig. 1C). These results are attributed to a blue shift of the emission wavelength that is particularly pronounced in the fluid phase (Parasassi et al., 1993) and that can be ascribed to a decrease in i) the motility of laurdan molecules or ii) the polarity of the solvent (Parasassi et al., 1994;Parasassi et al., 1993). Here it is possible that both factors are involved as a result of binding of SMA molecules to the surface of the nanodiscs, which may impede the motility of the lipids (Cuevas Arenas et al., 2016) while at the same time insertion of styrene moieties will decrease the polarity.
Analysis of the GP curves observed for laurdan in the nanodiscs reveals that for SMA 2:1 nanodiscs the phase transition occurs at ∼38°C and for SMA 3:1 nanodiscs at ∼27°C (Fig. 2). The decreased steepness of the slope of the inflection curve around T m suggests a decrease in the cooperativity of the transition. This effect is more pronounced for SMA 3:1 nanodiscs than for SMA 2:1 nanodiscs. Furthermore, the higher GP values of SMA 2:1 nanodiscs and particularly SMA 3:1 nanodiscs in the fluid phase suggest a less hydrated or more ordered environment than for MLVs.
Similar results were obtained when the lipid acyl chain length was varied from 15 to 18 C-atoms (see GP curves in Fig. S1). Overall, the Laurdan data show a downshift in T m of ∼3-5°C for saturated phosphatidylcholines organized in SMA 2:1 nanodiscs as compared to MLVs (Table 1, left column), while the same lipids assembled in SMA 3:1 nanodiscs show a downshift of ∼14-17°C. In addition, the SMA 3:1 nanodiscs exhibit a further increased broadening of the phase transition and a less hydrated environment in the fluid phase as compared to SMA 2:1 nanodiscs for all lipids tested.
The laurdan fluorescence data were complemented with differential scanning calorimetry (DSC) analysis. DSC has the advantage of being a Fig. 1. Normalized fluorescence emission of laurdan molecules incorporated in di-16:0 PC self-assemblies: A) MLVs, B) SMA 2:1 nanodiscs and C) SMA 3:1 nanodiscs. Spectra were recorded using an excitation wavelength of 340 nm. The data shown correspond to the spectra obtained during the first heating cycle. Nanodiscs were obtained at a SMA-tolipid mass ratio of ∼ 1.7. high precision technique which does not require the addition of fluorophores that may perturb the phase equilibrium of the membrane. As shown in Fig. 3, DSC thermograms of di-16:0 PC MLVs display a sharp peak with a maximum corresponding to T m at ∼41°C.SMA 2:1 nanodiscs and SMA 3:1 nanodiscs both exhibit a lowering of T m with the downshift being more pronounced in SMA 3:1 nanodiscs, as was also observed for PC lipids with shorter and longer chain lengths (Fig. S2). The T m values show downshifts of approximately ∼2-4°C in T m of lipids in SMA 2:1 nanodiscs and of ∼11-13°C in SMA 3:1 nanodiscs (Table 1, middle column). These shifts are slightly less than those observed with laurdan fluorescence, most likely due to differences in experimental conditions, but nevertheless the results are qualitatively similar.
DSC analysis also provides insight into the cooperativity of the gel-to-fluid phase transition. The narrow peak width at half maximum (ΔT 1/2 ) observed for di-16:0 PC phospholipids in MLVs indicates a high degree of cooperativity (Biltonen and Lichtenberg, 1993), while the heat curves of both nanodisc samples show a clear broadening, suggesting a notable loss in the cooperativity (Denisov et al., 2005;Orwick et al., 2012;Shaw et al., 2004). The broadening is more pronounced in SMA 3:1 nanodiscs as compared to SMA 2:1 nanodiscs, as was also observed for other lipids (Fig. S2). These results, as quantified in Table 1, are fully consistent with the observed effects of the polymers on the steepness of the inflection points in the laurdan GP-curves ( Fig. 2 and S1).

Calorimetric enthalpy values of the gel-to-fluid phase transition may be determined by multiple processes
Another characteristic property of the gel-to-fluid phase transition is the molar calorimetric enthalpy (ΔH cal ) associated with the melting process. As displayed in Table 2, the ΔH cal values for MLVs of lipids increase with lipid chain length, in line with previously reported data (Goto et al., 2009;Lewis et al., 1987;Marsh, 2013). For all lipids, a dramatic loss in calorimetric enthalpy is observed when present in SMA-bounded nanodiscs. For SMA 2:1 nanodiscs ΔH cal is approximately 20% of ΔH cal found in MLVs (Table 2). Rather surprisingly, the loss of ΔH cal is less dramatic in SMA 3:1-nanodiscs, where ∼45-50% of the value is retained.
To obtain further insight into the origin of this somewhat puzzling observation we decided to extend the systematic approach by going to system of lipids with an even longer chain length. As shown in Fig. 4, di-20:0 PC MLVs exhibit a narrow, highly cooperative gel-to-fluid phase transition while the SMA 2:1-nanodiscs again show a broadening and a small downshift in T m . Interestingly, a remarkable feature now occurs in the thermogram obtained for SMA 3:1-nanodiscs, where the main phase transition appears to be split into two peaks. Identical thermograms were obtained upon repeated heating showing complete reversibility of this transition. A similar, reversible effect of apparent multiple transitions was detected upon extending the data-set to include nanodiscs of SMA 4:1 copolymers, as shown for di-18:0 PC nanodiscs (Fig. S3). These results suggest that interpretation of effects on ΔH cal in these nanodiscs is not straightforward, and that possibly SMA molecules themselves contribute to the enthalpy of the transition. We will further elaborate on these effects in the discussion.
3.3. Thermotropic properties are more sensitive to polymer concentration in nanodiscs bounded by SMA 3:1 or SMA 4:1 than in nanodiscs bounded by SMA 2:1 Next it was investigated to what extent the SMA-to-lipid ratio at which the nanodiscs are obtained may affect the thermotropic properties of the lipids in the nanodiscs. As shown in Fig. 5A, T m of di-16:0 PC is shifted to lower temperatures as the SMA-to-lipid ratio increases. Importantly, this concentration effect seems to be less pronounced in SMA 2:1-nanodiscs than in SMA 3:1-nanodiscs and SMA 4:1-nanodiscs.
For all samples also the calorimetric enthalpy values of the gel-tofluid transition were determined and these are shown in Fig. 5B. The results show that ΔH cal corresponding to di-16:0 PC lipids bounded by Table 1 T m values and peak widths at half maximum (ΔT ½ ) of saturated phosphatidylcholine self-assemblies measured by laurdan fluorescence and DSC. DSC data are given as averages of the 2nd and 3rd heating cycle from 2 independent samples, with errors representing the standard deviation.   3.2 ± 0.3 3.8 ± 0.1 4.8 ± 0.6 4.5 ± 0.1 SMA 2:1 is low and rather constant, independent of the SMA concentration used for solubilization., When either the SMA 3:1 or the SMA 4:1 variant are used, the values of ΔH cal are larger and fluctuate. This fluctuation of the ΔH cal values may indicate that i) other factors may be contributing energetically to the phase transition of lipids or ii) the morphology of the nanodiscs is affected by the concentration of SMA.
3.4. Morphological integrity is better preserved in nanodiscs bounded by SMA 2:1 than in nanodiscs bounded by SMA 3:1 or SMA 4:1 Finally it was investigated how the SMA variants affect the size and morphology of the nanodiscs and whether this depends on the concentration of the polymer. As illustrated in Fig. 6, nanodiscs bounded by SMA 2:1 form well-defined nanodiscs at all polymer concentrations with a somewhat inhomogeneous size distribution of approximately d ∼ 9-18 nm. A remarkable feature here is the formation of phospholipid stacks. These are also known as "rouleaux" stacks and have been observed before in nanodiscs bounded by SMA molecules and MSP-like proteins Zhang et al., 2011), where they were ascribed to an artefact resulting from the interaction between positively-charged choline head groups with the negativelycharged inorganic crystals from the staining solution. These stacks are also observed in nanodiscs bounded by SMA 3:1, where they can be seen most clearly at lower polymer concentrations (SMA-to-lipid mass ratio of 0.75 and 1.5). Here the nanodiscs are still well-defined and a size distribution is observed of 9-16 nm, similar as for SMA 2:1 nanodiscs. At higher concentrations of SMA 3:1 (SMA-to-lipid mass ratio > 1.5) the morphological integrity of the nanodiscs is affected, as shown by the formation of nanoscopic aggregates. Remarkably, EM micrographs of nanodiscs bounded by SMA 4:1 revealed a much higher degree of polydispersity as compared to the other SMA analogues and a higher tendency to aggregate. The presence of circular structures is ascribed to artifacts during the preparation of the grids. Stacks of nanodiscs are not observed in micrographs of SMA 4:1 nanodiscs. Together these data suggest that SMA 3:1 and in particular SMA 4:1 are less prone than SMA 2:1 to form well defined circular-shaped nanodiscs and that SMA 2:1-nanodiscs "allow" a high amount of SMA in solution without affecting their morphological integrity.
Similar differences in behavior between the SMA analogues were found using dynamic light scattering. These experiments (Figure S 4) revealed rather homogeneous hydrodynamic diameter distributions for nanodiscs bounded by either SMA 2:1 and SMA 3:1 (d ∼ 9-12 nm), and a heterogeneous size distribution for SMA 4:1 nanodiscs at the various SMA concentrations.

Discussion
In this study we analyzed the thermotropic properties of saturated phosphatidylcholines in nanodiscs that are bounded by SMA with a styrene-to-maleic acid ratio of 2:1, 3:1 or 4:1. Laurdan fluorescence as well as DSC results for lipids with varying acyl chain length showed that the native main phase transition temperature (T m ) of the lipids is downshifted in the SMA-bounded nanodiscs as compared to the situation in large multilamellar vesicles, and that the cooperativity of the gel-to-fluid phase transition is highly reduced. Nevertheless, native-like properties seemed best preserved in SMA 2:1 nanodiscs, where in all lipid systems the effects on T m and broadening were least profound. In theory, a likely explanation would be that SMA 2:1 nanodiscs have a somewhat larger size than SMA 3:1 or SMA 4:1 nanodiscs. However, EM and DLS analyses in the present study gave no indication for this. Even though the size distribution in all samples was somewhat heterogeneous, it is unlikely that the differences in effect of the polymers can  be ascribed to a difference in size of the nanodiscs. We speculate that these effects are due to differences in polymer composition. SMA 3:1 and SMA 4:1 have a more inhomogeneous distribution of styrene and maleic acid units along their sequence as compared to SMA 2:1 . This would result in a more inhomogeneous membrane partitioning and thus might explain the larger broadening and lower T m for SMA 3:1 and SMA 4:1 nanodiscs as compared to SMA 2:1 nanodiscs.
Further DSC analysis resulted in a remarkable observation that SMA 2:1 nanodiscs have the lowest calorimetric enthalpy, as was observed for all phosphatidylcholines tested when compared with SMA 3:1 and for di-16:0 PC lipids at different concentrations of SMA 2:1, SMA 3:1 or SMA 4:1. This effect would be consistent with a smaller size of the SMA 2:1 nanodiscs, in analogy to effects observed for MSP nanodiscs, where it was found that the calorimetric enthalpy in nanodiscs systematically increases with increasing size of the nanodiscs (Denisov et al., 2005). We speculate that also these effects are due to differences in polymer composition. Using nanodiscs bounded by SMA 3:1 it has been shown that insertion of styrenes into gel phase lipids is thermodynamically more favorable than insertion into fluid phase lipids, demonstrating that there are differences in polymer-lipid interactions as function of the phase state of the lipids in the nanodiscs (Cuevas Arenas et al., 2016). Indeed, insertion of the styrenes inbetween the lipid acyl chains can be expected to be less favorable for lipids in a fluid phase than for lipids in a gel phase, because in the fluid phase the chains will occupy more space causing steric hindrance. This effect will become stronger when the lipids are longer and when there is a higher density of styrenes. We therefore propose that the relatively large enthalpy for SMA 3:1 and SMA 4:1 nanodiscs is due a change in organization of the polymer around the rim of the nanodisc, which accompanies phase interconversion and which energetically contributes to the transition. This might also be the basis for the complex thermograms obtained for di-20:0 PC SMA 3:1-nanodiscs and for di-18:0 PC SMA 4:1-nanodiscs, where the main peak of the thermogram is split into two peaks. If there indeed is such a contribution, then this would imply that calculations of the cooperative unit as number of lipids participating in the phase transition (Oluwole et al., 2017;Orwick et al., 2012) are not useful in SMA-nanodiscs, because they are based on broadening and enthalpy values of a lipid melting transition only. It must be noted that this energetic contribution can not be interpreted as a gel-to-ripple phase pretransition since such ripples exclusively occur in pure extended lipid bilayer systems (i.e MLVs or supported lipid bilayers) and they are easily abolished by the presence of membrane-interacting molecules (Heimburg 2000). Furthermore, the pretransition can be considered as a macroscopic reorganization of lipids and since the periodicity of the ripples is at least ∼5 times larger than the size of nanodiscs (Czajkowsky et al., 1995;Heimburg 2000) their presence in nanodiscs is highly improbable.
Another relevant finding in this manuscript is that the morphological integrity of the nanodiscs can be affected by increasing concentrations of SMA, depending on the hydrophobicity of the polymer. For SMA 2:1 nanodiscs no notable effect of varying the SMA concentration was observed, but for SMA 3:1 at higher SMA concentration regular discs could no longer be observed. Based on both EM and DLS analysis, SMA 4:1 nanodiscs do not appear to self-assemble in regular circular nanodiscs at any SMA concentration, but rather tend to form aggregates. Thus, the term nanodiscs may be incorrect for solubilized SMA 4:1 particles.
In conclusion, the data presented here suggest that the SMA 2:1 polymer is able to preserve native lipid packing properties in the nanodiscs to a higher extent than the SMA 3:1 polymer. Together with recent studies showing that SMA 2:1 is also more efficient in solubilizing membranes (Morrison et al., 2016;Scheidelaar et al., 2016), this supports the notion that SMA 2:1 is more suitable as tool for characterization of membrane proteins in native nanodiscs.