Extraction and reconstitution of membrane proteins into lipid nanodiscs encased by zwitterionic styrene-maleic amide copolymers

Membrane proteins can be reconstituted in polymer-encased nanodiscs for studies under near-physiological conditions and in the absence of detergents, but traditional styrene-maleic acid copolymers used for this purpose suffer severely from buffer incompatibilities. We have recently introduced zwitterionic styrene-maleic amide copolymers (zSMAs) to overcome this limitation. Here, we compared the extraction and reconstitution of membrane proteins into lipid nanodiscs by a series of zSMAs with different styrene:maleic amide molar ratios, chain sizes, and molecular weight distributions. These copolymers solubilize, stabilize, and support membrane proteins in nanodiscs with different efficiencies depending on both the structure of the copolymers and the membrane proteins.

of pH ≤ 6 and the presence of millimolar concentrations of Ca 2+ , Mg 2+ and polyvalent cations in general 14,17,18 . Precipitation of SMA and SMALPs occurs because of hydrophobic interactions when it becomes neutral at low pH or as a result of association of the polyvalent cations with the negative charges of the carboxyl acids. These incompatibilities preclude experiments that have to be performed at pH ≤ 6 (e.g., studies of H + pumps and H + -gated channels), require Mg 2+ or other divalent cations (e.g., most ATPases), or are regulated by Ca 2+ (e.g., closure of connexin hemichannels). Also, SMA carboxyl groups can interfere with standard His tag-based purification of membrane proteins through coordination of the carboxyl acids of SMA with transition metal ions used in immobilized metal affinity chromatography, or shielding of the affinity purification tag 17,18 . We have solved some of the problems associated with the use of SMAs by designing and synthesizing new styrene-maleic amide copolymers where the maleic acid moieties of traditional SMA were replaced with maleic amide moieties conjugated to zwitterionic phosphatidylcholine (PC) groups 19 . Even though the new copolymers do not contain maleic acid, we decided to refer to them as zSMAs ("z" for zwitterionic). The main reasons for this are the broad use of SMA and SMALP, and the fact that "MA" can be still used to represent the new maleic amide moieties. We showed that these new copolymers are able to solubilize membrane proteins into nanodiscs of controlled sizes 19 . In our previous study, we focused on 1:1 styrene:maleic amide molar ratio (St:MA) zSMAs with molecular weight greater than 10 kDa 19 . In reference to studies on traditional SMAs that solubilize membranes into nanodiscs, oftentimes the commercially-available SMAs with molecular weight smaller than 10 kDa were used [20][21][22][23][24][25][26][27][28] , and the best membrane solubilization was reported with 2:1 St:MA SMAs 23,27,29 . Here, we compared the extraction/ reconstitution of membrane proteins into LNDs by a series of zSMAs with different St:MAs (i.e., 1:1, 2:1, and 3:1), molecular weight, and polydispersity indices (i.e., in-house prepared zSMAs via controlled/"living" polymerization vs zSMAs derived from commercial SMAs). Our results show that the zSMAs solubilize, stabilize, and support membrane proteins in nanodiscs with different efficiency depending on both the structure of the copolymers and the type of the membrane proteins.

Results and Discussion
zSMA copolymers. We compared five different zSMAs in this study. Two of the zSMAs with well-defined sizes and different St:MAs (1:1 and 2:1) were derived from styrene-maleic anhydride random copolymers (P(Stran-MA)) produced by reversible addition-fragmentation chain-transfer (RAFT) polymerization (Fig. 1A). For convenience, we simply refer to these copolymers as zSMAs. It should be noted that for the synthesis of P(Stran-MA) with initial St:MA feeding ratios >1, there is a possibility of a gradient increase of styrene repeating units that could lead to a polystyrene homopolymer segment at the end of individual P(St-ran-MA) chains. To minimize this possibility, we controlled the conversion of styrene in the P(St-ran-MA) to be ~20%. We also prepared three different chain sizes of the 2:1 zSMA to test the effect of molecular weight on membrane solubilization. Three additional zwitterionic copolymers were synthesized using Malvern Lipodisq P[St-ran-MA] copolymers as precursors (Fig. 1B). These are referred to as 1:1, 2:1 and 3:1 M zSMAs (Fig. 1B). As controls, we also prepared conventional SMA copolymers using the same Lipodisq copolymers as precursors (Fig. 1B).
Structural characterization of the copolymers. The molecular weights and polydispersity indices (PDIs) of the different P(St-ran-MA) copolymers are presented in Table 1. As expected, the copolymers prepared via RAFT polymerization have much smaller PDIs than those received from Malvern, indicating less variation in their chain size distribution ( Supplementary Fig. 1A vs 1B). Since all the Lipodisq P(St-ran-MA) copolymers have similar sizes (~5-6 kDa), we designed the RAFT P(St-ran-MA) copolymers to have a similar size (~6-7 kDa). For the 2:1 RAFT P(St-ran-MA), we also prepared two other sizes at ~2x and 0.5x of the Malvern copolymer size to examine the effect of chain size on membrane solubilization (Table 1). We used proton nuclear magnetic resonance spectroscopy ( 1 H NMR) to confirm the successful synthesis of each reaction intermediates, as well as to determine the chain size and St:MA of each copolymer product ( Table 1). Examples of 1 H NMR spectra are presented in Supplementary Fig. 2A,B. We also adapted an alternative method based on UV spectroscopy ( Supplementary Fig. 3) to validate the St:MA ratios calculated from the 1 H NMR analysis. In tetrahydrofuran (THF), the styrene moiety has a characteristic UV absorption peak at 245 nm that does not overlap with any other absorption features of the copolymers. We measured the molar absorption coefficient (ɛ 245 ) of our well-defined P(St-ran-MA) copolymers (Supplementary Fig. 3B and Table 1), and compared the values to the expected ɛ 245 of these copolymers calculated on the basis of the St:MA ratios derived from the 1 H NMR analysis using a standard polystyrene sample as reference (PSt 62 , Supplementary Fig. 3A). We found that the measured ɛ 245 of P(St-ran-MA) copolymers agreed nicely ( Supplementary Fig. 3C) with those expected values when the molar fractions of styrene repeating units in these polymers changed from 100% in PSt, to 67% (in P(St-ran-MA) copolymers polymers with St:MA~2:1 as determined by 1 H NMR), and to 50% (in P(St-ran-MA copolymers with St:MA~1:1 as determined by 1 H NMR). Note that we did not measure the ɛ 245 of Malvern Lipodisq P(St-ran-MA) copolymers because of their polydisperse nature.

Solubilization of HR and MsbA in crude membranes by zSMAs.
For the solubilization test we used two very different membrane proteins: Natronomonas pharaonis halorhodopsin (HR) and Salmonella typhimurium MsbA. HR is a photosensitive Cl − pump that belongs to the 7-transmembrane receptor protein family, and MsbA is a twelve-transmembrane helix homodimer that belongs to ATP-binding cassette (ABC) superfamily 6,30 . HR uses the energy of photons absorbed by all-trans retinal for Cl − influx, whereas MsbA hydrolyzes ATP and uses the resulting energy for translocation of lipid A from the inner leaflet to the outer leaflet of the membrane [30][31][32] . A main goal of this study was to identify how variations in the incubation conditions affect solubilization of the recombinant proteins from crude membranes by different zSMA copolymers. For these studies, the total protein concentration of the crude membranes was kept at <3 mg/mL. Recombinant membrane proteins are most frequently expressed in mammalian or insect cells, yeast (P. pastoris or S. cerevisiae) or a variety of E. coli Since increasing the temperature of incubation from room temperature (RT) to 37 °C had small detrimental or no effects on HR solubilization ( Supplementary Fig. 4B), we settled for solubilization of HR at RT. As for MsbA, we decided to use 37 °C because this temperature is well tolerated by MsbA, and increased temperature was reported to improve solubilization by traditional SMA copolymers 26,37 . Data supporting the use of 2-h MsbA solubilization at temperatures higher than 4° are presented in Supplementary Fig. 5. In any case, increasing the temperature from RT to 37 °C has only minor effects on solubilization, and therefore solubilization by zSMA and M zSMA copolymers can be carried out at RT or above without compromising efficiency. Increasing ionic strength is believed to improve solubilization by traditional SMAs due to a decrease of the electrostatic repulsion between the negatively charged membranes and maleic acid groups, which facilitates SMA binding, the first step in membrane solubilization 26,27,29 . Generally, the presence of 100-200 mM NaCl is sufficient, and high salt concentration can be detrimental because the negative charges on traditional SMAs are important to keep the copolymers soluble in aqueous solution 27 . These effects should not be as important for the zSMAs. In fact, we found only minor differences between solubilization in the presence of 150 or 500 mM NaCl for the zwitterionic copolymers ( Supplementary Fig. 6A). Since there were small improvements with 500 mM vs 150 mM NaCl for the 1:1 zSMA and M zSMA (HR), and no effects for the other copolymers in the HR experiments or MsbA, we decided to use the 500 mM NaCl, but either concentration should work equally well. We also explored the effects of solution pH on the copolymers solubilization efficiency, testing pH 6.5, 7.5 and 8.3. We observed a decrease in solubilization by 2:1 SMA at the more acid pH ( Supplementary Fig. 6B), which is not surprising given the tendency of traditional SMA to precipitate at low pH 14,[17][18][19] . There were significant increases in HR solubilization by 1:1 M zSMA at pH 8.3 vs 6.5 (P < 0.005) and 2:1 M zSMA at pH 7.5 vs 6.5 (P < 0.01), and in MsbA solubilization by 1:1 zSMA at pH 7.5 vs 6.5 (P < 0.01). Since there were no statistical differences between pH 7.5 and 8.3, we settled for the use of pH 7.5, but any pH ≥7 should work well for solubilization by zSMAs and M zSMAs ( Supplementary Fig. 6B). Based on the effects of copolymer concentration, incubation temperature and time, as well as salt concentration and pH, our basic solubilization conditions consisted of 1% copolymer, 500 mM NaCl and pH 7.5, with incubation for 2 h at RT temperature for HR and 37 °C for MsbA. However, our comparative studies also suggest that solubilization of membrane proteins by zSMAs can be accomplished under a broad set of conditions without compromising the efficiency significantly. Figure 2 shows the efficiency of solubilization of HR and MsbA under our basic solubilization conditions. Individual experimental results obtained over the course of ~2 years are shown in Supplementary  Fig. 7. For HR ( Fig. 2A), which is extracted very well from crude membranes by detergent (1.5% n-dodecyl-β-D-maltopyranoside; DDM), the solubilization by copolymers was less efficient, but still >50% for the zSMA and M zSMA copolymers in contrast to <40% for 2:1 SMA. Taken together, both sets of zwitterionic copolymers with different St:MAs displayed a good extraction efficiency for HR. For MsbA (Fig. 2B), which is not extracted very efficiently by detergent (2% DDM/0.04% sodium cholate), the copolymers did better than the detergent, in particular the zSMA copolymers. Of these, the 2:1 zSMA showed the best performance, with solubilization of ~60% of the MsbA in crude membranes, ~3.5-folds the extraction efficiency by the detergent mix. An increase in solubilization efficiency by SMAs with St:MA molar ratios 2:1 vs 1:1 has been reported 23,27,29 , and our data suggest that this improved efficiency depends, at least in part, on the target protein (Fig. 2). We also noticed that zSMAs outperformed M zSMAs by ~2-folds under otherwise similar conditions. This significant difference for MsbA solubilization between zSMAs and M zSMAs could result from their different microstructures and polydisperse indices (Supplementary Fig. 1 and Table 1) due to the different synthetic routes, as previous studies suggested that molecular weight distribution could play a role in the solubilization/reconstitution 21,25 . However, we are not clear why HR extraction is less sensitive to this difference ( Fig. 2A) than MsbA, and we cannot rule out other possibilities. A relatively long styrene-rich hydrophobic tail of SMA copolymers at St:MAs higher than 1 has been associated with reduced solubilization efficiency 25,28,38 . However, HR and MsbA solubilization efficiencies by 2:1 zSMA were very similar (60-65%), and better than solubilization by 2:1 SMA tested under the same conditions for both proteins (Fig. 2), and comparable to those for extraction of a variety of proteins from membranes by SMAs 23,39 . This suggests that the early termination of the polymerization reaction in our synthesis helps minimize the potential styrene-rich hydrophobic tail and its associated adverse effect on solubilization. Overall, these results showed that our zSMA and M zSMA copolymers are practical alternatives to detergents for the solubilization of membrane proteins, have the distinct advantage of the direct reconstitution in a lipid bilayer membrane, and avoid the precipitation that occurs with the use of SMAs at pH ≤ 6 or in the presence of polycations 19 .  Supplementary Fig. 2). Figure 3 shows the effects of 2:1 zSMA copolymers of different molecular weights on solubilization, and a representative Western blot is shown in Supplementary Fig. 8. The solubilization conditions were those of the basic protocol used for the experiments in Fig. 2. Here we tested small-, medium-, and large-size zSMAs derived from RAFT P(St-ran-MA) copolymers with molecular weights of 3.1, 6.4 and 12.0 kDa, respectively. The solubilization by the small-size zSMA was reduced at 2.5 vs 1% copolymer concentration for both HR (P < 0.05) and MsbA (P < 0.01), whereas no differences were detected for the solubilization between the three copolymers at the preferred concentration of 1%. These results suggest that zSMA copolymers perform essentially similarly in the molecular weight range tested, and P(St-ran-MA) copolymers in the 3-to 12-kDa size range can be safely used as a starting point for further development of zSMAs.
As mentioned in the Introduction, one of the uses of SMAs and zSMAs is in the solubilization of purified membrane proteins reconstituted in liposomes of defined composition. Such experiments can be useful to determine the effects of the bilayer composition on membrane proteins. In spite of the versatility of zSMAs and M zSMAs as solubilization agents, our data in membranes with similar lipid composition and under the same  www.nature.com/scientificreports www.nature.com/scientificreports/ experimental conditions show significant differences in the solubilization of MsbA from BL21 E. coli membranes and liposomes formed by E. coli polar lipids. This is consistent with previous reports showing that the type and state of the membrane affects solubilization by the copolymers 20,35-37 . In general, solubilization of HR from proteoliposomes was better, and the same was true for MsbA solubilization by 2:1 SMA, with the differences being more marked for the latter (Fig. 4A). In contrast, MsbA solubilization by 2:1 zSMA from crude membranes was more efficient than from proteoliposomes (Fig. 4A). The results do not support a uniform role of the protein:lipid ratios between the membranes and proteoliposomes (higher in the membranes) that explains the observed differences in solubilization, since the membrane protein and polymer are clearly significant factors. More studies will be needed to explain the differences presented in Fig. 4A. Nevertheless, the differences in the efficiency of MsbA extraction from crude membranes and proteoliposomes by the zSMAs were not major, and the results show that solubilization of both membrane proteins from proteoliposomes by zSMAs and M zSMAs is straightforward and feasible (Fig. 4A). Figure 4B shows size distributions of the copolymer nanodiscs formed by solubilization of HR or MsbA reconstituted in proteoliposomes. As shown in Fig. 4B,C, the 1:1 and 2:1 zSMALPs have a very focused size distribution with an average diameter of 10-11 nm. In contrast, the 2:1 SMALPs and M zSMALPs, both of which are encased by copolymers derived from Malvern Lipodisq P[St-ran-MA] precursors, have a relatively broader size distribution with a peak diameter of ~7 nm. Identical DLS analysis yielded diameters of 6.0 ± 0.2 nm for detergent-solubilized MsbA (n = 21) and 9.7 ± 0.3 nm for MsbA in traditional lipid nanodiscs (n = 5). Although we do not have a definitive explanation for the differences in apparent sizes, the similarity between the diameters calculated from DLS for MsbA in traditional LNDs and zSMALPs suggests that the latter form similar nanodisc-like structures, whereas the MsbA-loaded SMALPs and M zSMALPs may consist of a mixture of nanodiscs and smaller www.nature.com/scientificreports www.nature.com/scientificreports/ polymer-lipid nanoparticles due to the polydisperse nature of the commercial P[St-ran-MA] precursors. Overall, our results are consistent with the formation of nanodisc-size structures following solubilization by 2:1 SMA, zSMAs and M zSMAs.
One potential advantage of the reconstitution in zSMALPs vs solubilization in detergent is membrane protein stabilization 36,[39][40][41][42] . Figure 5 shows that stabilization does indeed occur following zSMALP formation (see Supplementary Fig. 9 for a representative experiment). After a heat shock (15 min at 65 °C) essentially no MsbA in detergent remained in solution (7 ± 2%; P < 0.001 vs copolymers). In contrast, the protein that remained in solution after the heat shock averaged ~45 and ~70% for MsbA in 1:1 and 2:1 zSMALPs, respectively (P < 0.002 for 1:1 zSMA vs 2:1 zSMA). These results are not particularly surprising in that they confirmed the general finding that membrane proteins reconstituted in lipid bilayers are more stable than those in detergent micelles. However, the improved thermal stabilization in 2:1 zSMALPs vs 1:1 zSMALPs was somewhat surprising. It can result from a stronger interaction of the more hydrophobic 2:1 zSMA copolymer with the lipid bilayer, but understanding the bases for membrane protein stabilization will need additional studies.
We have previously demonstrated that, in contrast to membrane proteins in SMALPs, proteins reconstituted in zSMALPs can be studied at low pH (proteorhodopsin, a H + pump) and in the presence of divalent cations (MsbA) 19 . Here, we compared the activity of HR and MsbA reconstituted in 1:1 and 2:1 zSMALPs. For these studies, purified HR or MsbA were first reconstituted in liposomes and then solubilized with zSMAs. The HR experienced the typical spectral blue shift between the Cl − -free and Cl − -bound states ( Fig. 6A and Supplementary  Fig. 10) 43,44 . The shift in wavelength maxima between the Cl − -free and Cl − -bound states was similar for HR in detergent, proteoliposomes, zSMALPs and 2:1 M zSMALP (20 ± 1 nm; n = 18; Fig. 6B). Unexpectedly the change was still present but was smaller for HR in 2:1 SMALPs (10 ± 1 nm; P < 0.001; n = 3). We presently do not have an explanation for the detrimental effect of reconstituting HR in SMALPs while HR function is well maintained in detergent, liposomes or zSMALPs. In contrast to the observations with HR, there were noticeable differences between MsbA in 1:1 and 2:1 zSMALPs. Consistent with our previous observations, the ATPase activity of MsbA (measured in 5 mM MgATP) was higher in 1:1 zSMALPs than in detergent (P < 0.001), and the protein has no activity in 2:1 SMALPs (Fig. 6C) because Mg 2+ is required for MsbA ATPase activity and this cation causes aggregation/precipitation of SMALPs 19 . Figure 6C also shows that MsbA activity in 1:1 zSMALPs was similar to that of MsbA in liposomes. The ATPase activity data in proteoliposomes and nanodiscs are comparable since we have previously shown that MsbA is reconstituted inside-out in proteoliposomes, with the ATP-binding sites accessible to the bulk solution 6,45 . Unexpectedly, although MsbA in 2:1 zSMALPs was active, its ATPase activity was similar to that in detergent, and significantly lower than that of MsbA in 1:1 zSMALPs or proteoliposomes (P < 0.001). Even more surprising was the absence of ATPase activity in 2:1 M zSMALPs. Although we do not have a definitive explanation, it may be related to the small size of the M zSMALPs with the resulting restriction in MsbA conformational changes (Fig. 4C). The inverse relationship between MsbA thermal stability and function (Figs. 5 and 6C) in 1:1 and 2:1 zSMALPs points to a delicate balance between the copolymers structure and membrane protein stability and function that may shift depending on the different interactions among the membrane protein, Figure 5. Thermal stability of solubilized MsbA. Purified MsbA was studied in detergent (Det) and zSMALPs (solubilized from proteoliposomes). For these experiments the samples were heated to 65 °C for 15 min and MsbA in the supernatant, after centrifugation at 100,000 g for 45 min, was quantified on Western blots probed with an anti-His antibody. The 1:1 and 2:1 zSMALPs were prepared using the corresponding zSMAs derived from RAFT P(St-ran-MA) copolymers with molecular weights of 6.7 and 6.4 kDa, respectively. *Denotes P < 0.001 vs both copolymers; † Denotes P < 0.002 vs 2:1 zSMA. Data are means ± SEM (n = 4 per condition).
lipids, and copolymers when the nanodiscs are encased within the copolymers. Although protein-polymer charge interactions are important in the functional reconstitution of membrane proteins 46 , that is not a likely factor for the difference in ATPase activity between MsbA reconstituted in 1:1 and 2:1 zSMAs. The molecular bases for the difference are unclear, but the results suggest that, at least for some membrane proteins, 1:1 zSMAs may be a better choice for functional studies even when the yield and stability of zSMALPs for 2:1 zSMAs are higher. Therefore, it will be advisable to test both copolymers since 2:1 zSMA will be preferable when increased stability is critical, whereas there may be differences in activity for a given membrane protein depending on the St:MAs. Understanding the effect on stability and function by different copolymers will need additional research. Such studies will be important to rationally design new and improved copolymers for specific applications.
conclusions SMALPs have gained popularity as membrane protein platforms because SMA copolymers can solubilize membranes directly, bypassing the use of detergent, and produce nanodiscs where the membrane patch includes the native lipids [14][15][16] . SMAs can also be used on purified membrane proteins reconstituted in liposomes of predetermined composition for studies of the effects of lipids on membrane protein structure and function [14][15][16] . Buffer incompatibilities of SMAs and SMALPs constitute a major problem when polyvalent cations at mM concentration or low pH are needed 14,[17][18][19] . Newer copolymers that offer distinct advantages over the traditional SMA copolymers have been studied recently 19,[47][48][49][50] . Diisobutylene/maleic acid copolymer (DIBMA), poly(styrene-co-maleimide) (SMI), styrene maleimide quaternary ammonium (SMA-QA) and our zSMAs are compatible with high concentrations of divalent cations 19,[47][48][49] . SMI can solubilize membrane proteins into nanoparticles efficiently at pH < 7.5, but not at more alkaline pHs, and the smaller diameter of the "SMILPS" could www.nature.com/scientificreports www.nature.com/scientificreports/ limit its application to "small" proteins 48 . In contrast, our zSMAs are stable at acidic and alkaline pH, and zSMAs with molecular weights ranging from ~10,000 to 45,000 can produce zSMALPs of increasing sizes 19 .
Here, we showed the formation of polymer-encased nanodiscs by new zSMA and M zSMA copolymers under a variety of conditions. We found significant effects on solubilization by temperature, incubation time, polymer concentration, polymer chain size and microstructure, salt concentration, and pH. We also found that the membrane and membrane protein are additional players, but systematic studies will be needed to define their precise roles. Based on the variables explored, we suggest the use of 1% copolymer, intermediate ionic strength (150-500 mM NaCl), pH in the 7.5-8.3 range, incubation at RT or above, and an incubation time of ~2 h. The differences among different copolymers appeared to depend on the copolymer structure, membrane structure/ composition, and/or membrane protein. In general, zSMAs performed better than M zSMAs under similar conditions, with remarkable differences in MsbA activity between the zSMALPs encased by the two different families of copolymers, suggesting an important role of the polymer chain size distribution and microstructure in nanodisc formation and support of membrane protein function. Among all the zwitterionic copolymers, those with a 2:1 St:MA ratio showed better solubilization efficiency than those with a 1:1 St:MA ratio, and a higher stability of the corresponding zSMALPs, but improved efficiency and stability do not translate into a higher membrane protein activity, as clearly showcased in the example of MsbA. As for the polymer chain sizes, the results showed that P(St-ran-MA) copolymers in the 3-to 12-kDa size range can be safely used as a starting point for the preparation of nanodiscs.
The overall conclusion is that solubilization by the new 1:1 and 2:1 zSMA, and 1:1, 2:1 and 3:1 M zSMAs to a large extent, is quite forgiving and better than that by 2:1 SMA, a benchmark for the SMALP studies due to the recognition that the solubilization efficiency by the 2:1 SMAs is better than that of the 1:1 SMAs 23,27,29 . This is important because, in principle, solubilization conditions already established for a membrane protein using SMAs or detergents can be used with few or no changes to solubilize them into zSMALPs with reasonable results, minimizing the need for extensive trials. In summary, the new zwitterionic copolymers developed and presented here provide new tools for membrane-protein studies in polymer-encased nanodiscs, and solid bases for further development of improved zSMA copolymers.

Synthesis of zSMA via reversible addition−fragmentation chain-transfer (RAFT) polymer-
ization. The reaction scheme used to prepare zSMAs is shown in Fig. 1A. We first prepared the random copolymers of styrene-maleic anhydride, P(St-ran-MA), with different St:MA ratios by controlling the stoichiometry of the two monomers 19 . In a typical reaction to prepare 2:1 P(St-ran-MA) with a molecular weight of ~6 kDa, styrene (2.08 g, 20 mmol), maleic anhydride (0.98 g, 10 mmol), DATC (36.4 mg, 0.10 mmol) and AIBN (3.3 mg, 0.02 mmol) were dissolved in 3.0 g THF in a 10-mL Schlenk flask, stirred, and degassed by three freeze-pump-thaw cycles. The flask was sealed and immersed in a 65 °C-oil bath, and after a predetermined time of ~50 min the reaction was quenched by liquid nitrogen, such that the conversion of styrene was controlled to ~20%. The reaction time was determined by the kinetics study commonly used in controlled/"living" polymerization 52 . The mixture was then diluted with THF and precipitated three times in a mixture of ether/chloroform (3/1 v/v). The molecular weights and polydispersity indices of different zSMA copolymers are summarized in Table 1.
Finally, P(St-ran-MA) was modified with the cysteamine-PC to prepare the zSMAs. In a typical run, cysteamine-PC (0.36 g, 0.97 mmol) and dicyclohexylcarbodiimide (DCC) (0.20 g, 0.97 mmol) were dissolved in 10 mL of DMSO. SMA (0.093 g, 0.32 mmol maleic anhydride) dissolved in 10 mL of DMSO was added dropwise into the mixture. After stirring for 2 days at RT the mixture was dialyzed against Millipore water. After two days, the undissolved white solid was filtered away and the solution was dried in a lyophilizer.

Synthesis of zSMA using the Lipodisq P[St-ran-MA] copolymers from Malvern Cosmeceutics.
We adopted a similar approach to prepare M zSMAs, as shown in Fig. 1B, except that the P(St-ran-MA) copolymers with 1:1, 2:1 and 3:1 St:MA were obtained from Malvern Cosmeceutics. The molecular weights and polydispersity indices of the different Lipodisq P(St-ran-MA) copolymers are summarized in Table 1. As controls, we also prepared SMAs www.nature.com/scientificreports www.nature.com/scientificreports/ used with dimethylformamide (DMF) with 0.01 M ammonium acetate as eluent for SMAs and THF for PSt, at a flow rate of 0.5 mL/min. 1 H NMR was performed on a 400-MHz JEOL liquid-state NMR spectrometer.

Expression, purification and reconstitution of halorhodopsin. A synthetic gene (Genscript,
Piscataway, NJ) coding for Natronomonas pharaonis halorhodopsin (HR) fused at the C-terminal end to a 6-His tag followed by a Flag tag ( Supplementary Fig. 11) was cloned into the NcoI/BamHI sites of the expression vector pET19b (Novagen, Madison, WI). HR was overexpressed in the E. coli strain BL21 (DE3) (Agilent Technologies, Santa Clara, CA) transformed with the pET19-HR plasmid. The cells were grown at 37 °C in 2YT medium with 200 µg/mL ampicillin, and 10 µM all-trans retinal was added at the time of induction (at OD 600 ~1) with 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG). The cells were harvested 4 h later, and all subsequent procedures were performed at 4 °C unless specified otherwise. Cell pellets were resuspended in 50 mM Tris/HCl and 5 mM MgCl 2 , pH 8, with 10 µg/mL lysozyme, 10 µg/mL DNAse I, 1 mM phenylmethanesulfonyl fluoride (PMSF) and a protease inhibitor cocktail (1 tablet/100 mL of buffer; complete EDTA-free, Roche), and lysed on a microfluidizer. Crude membranes were prepared by centrifugation at 135,000 g for 1.5 h, and were solubilized for 2 h at 4 °C in a buffer containing 300 mM NaCl, 50 mM MES (2-(N-morpholino)ethanesulfonic acid), 5 mM imidazole, 1 mM PMSF and 1.5% DDM (Inalco Pharmaceuticals, San Luis Obispo, CA), pH 6.5, at a total protein concentration <3 mg/mL. The DDM-solubilized lysate was centrifuged at 100,000 g for 30 min, and the supernatant was incubated overnight with Talon Co 2+ beads (Talon Superflow, Clontech, Mountain View, CA). The resin was washed first with 10 column volumes of a buffer containing 300 mM NaCl, 50 mM MES, pH 6.5, 0.1% DDM and 45 mM imidazole, and then with 10 column volumes of the same buffer, but with NaCl increased to 1 M. The last wash was with 4 column volumes of 300 mM NaCl, 50 mM MES, 0.1% DDM and 65 mM imidazole, pH 6.5. Imidazole was added to a concentration of 200 mM for elution. After elution, the imidazole was removed by exchanging the buffer to 50 mM sodium phosphate and 50 mM sodium citrate, with 0.1% DDM, pH 7. The purity of the preparation was assessed by staining gels (16% SDS PAGE) with Instant Blue (Expedeon, San Diego, CA) and by UV-Vis spectroscopy. The HR samples were concentrated to ~1 mg/mL and were stored at −80 °C until use. For reconstitution into liposomes, purified HR solubilized in DDM was mixed with E. coli polar lipids (Avanti Polar Lipids, Alabaster, AL) at a 1:3 protein:lipid ratio (w/w), and proteoliposomes were formed by detergent removal by gel filtration on columns (Zeba columns, Thermo Fisher Scientific, Waltham, MA) pre-equilibrated with 200 mM NaSO 4 and 50 mM sodium phosphate, pH 7. After reconstitution, the samples were extruded through a 200-nm polycarbonate filter (Mini-Extruder, Avanti Polar Lipids). Fig. 11) was expressed and purified as previously described 6,45 . Briefly, MsbA expression in BL21 DE3-RILP E. coli cells (Agilent Technologies) was induced for 4 h at 30 °C with 1 mM IPTG at an OD 600 of ~1. Crude membranes were prepared as described above for HR, and were solubilized for 1 h at RT with 2% DDM and 0.04% sodium cholate in a buffer containing 100 mM NaCl, 20 mM Tris/HCl, with 15% glycerol, 0.5 mM Tris (2-carboxyethyl) phosphine hydrochloride (TCEP) and 1 mM PMSF, pH 8. Solubilized MsbA was purified by metal affinity chromatography (Talon Superflow; Clontech) followed by size-exclusion chromatography using a Bio-Scale Mini Bio-Gel P-6 Desalting Cartridge (Bio-Rad, Hercules, CA) equilibrated with storage buffer: 100 mM NaCl, 20 mM Tris/HCl, 0.065% DDM, 0.04% sodium cholate, with 15% glycerol and 0.2 mM TCEP, pH 7.5. MsbA was stored at −80 °C until use. Protein concentration was determined by absorbance at 280 nm and purity was estimated at >95% from SDS-PAGE gels stained with Instant Blue (Expedeon). Reconstitution of purified MsbA into liposomes formed by E. coli polar lipids was performed by gel filtration and extrusion, as described for HR, but using a 1:10 protein:lipid ratio (w/w) in 100 mM NaCl, 20 mM Tris/HCl, with 0.1 mM TCEP, pH 7.4.

Expression, purification and reconstitution of MsbA. A fully-active MsbA mutant (Supplementary
Extraction of HR and MsbA from crude membranes and proteoliposomes with copolymers. For the basic protocol, crude membranes from E. coli expressing HR or MsbA were resuspended at a final concentration of 30 mg/mL (wet membrane mass; total protein <3 mg/mL) in 500 mM NaCl and 50 mM Tris/HCl, with 10% glycerol, pH 7.5. Copolymers were added to a final concentration of 1% (w/v) and the samples were incubated with gentle rotation for 2 h at RT for HR or at 37 °C for MsbA. Insoluble material was removed by centrifugation at 100,000 g for 45 min at 4 °C, and the supernatant containing HR-or MsbA-loaded zSMALPs was analyzed by Western blotting using an antibody against the His tag that is conjugated to Alexa Fluor 647 (HisTag Antibody [iFluor 647], GenScript). The signal was visualized using an Odyssey Infrared Imager (Li-Cor Biosciences). The efficiency of protein solubilization was quantified from the Western blots by densitometry (UN-SCAN-IT, Silk Scientific). In some experiments, copolymer concentration was increased to 2.5 or 5%, NaCl concentration was decreased to 150 mM, or pH was changed to 6.5 or 8.3, as indicated in the text. For extraction by copolymers of HR or MsbA reconstituted into liposomes we used the conditions of the corresponding basic protocols described above.
HR and MsbA activity assessment. We used the blue shift in absorption spectra elicited by NaCl as an indicator of HR response to Cl − binding 43,44 . Absorption spectra of HR in DDM, liposomes or polymer-encased nanodiscs were collected on a spectrophotometer (Jasco V-630, Easton, MD) at 22 °C in absence of Cl − and after addition of 250 mM NaCl. The ATPase activity of MsbA was measured using a variant of the ATPase linked assay 45,53 . In this system, regeneration of the ATP consumed by MsbA is tied to the conversion of NADH to NAD + , and hydrolysis activity is determined by the decline in NADH measured by the absorbance at 340 nm. Measurements were performed at 37 °C under conditions approaching V max , using 0.5 to 1 µg of MsbA in 200 μL of linked enzyme cocktail containing: 100 mM NaCl, 20 mM Tris/HCl pH 7.5, 12 mM MgSO 4 , 5 mM ATP, 0.1 mM EGTA, 3 mM phosphoenolpyruvate, 0.1 mM TCEP, 25 μg/mL lactate dehydrogenase, 50 μg/mL pyruvate kinase and ~0.7 mM NADH. For measurements in detergent the reaction also contained 0.065% DDM and 0.04% sodium cholate. For the measurements, MsbA-loaded nanodiscs were enriched based on the affinity of the His-tagged MsbA for Ni 2+ . The supernatant from the solubilization of proteoliposomes was mixed with Ni-NTA beads (Thermo Fisher Scientific) at a ratio of 100 µL resin/mL solubilized protein, and after incubation at 4 °C overnight with gentle rotation the samples were transferred to a gravity flow column. The resin was washed with 10 column volumes of 100 mM NaCl and 20 mM Tris/HCl, with 0.1 mM TCEP and 20 mM imidazole, pH 7.4, and elution was achieved by increasing imidazole to 200 mM. Eluted fractions were analyzed on gels (16% SDS-PAGE) stained with Instant Blue (Expedeon) and used for the ATPase measurements.

Estimation of nanodiscs size by dynamic light scattering (DLS).
Measurements were performed at 22 °C on a Zetasizer Nano ZSP (Malvern Instruments, Westborough, MA) using 40-µL disposable microcuvettes. Size-number distributions were generated using the Zetasizer software version 7.11 and were analyzed using the protein analysis distribution.
Statistics. Statistical comparisons were performed by the Student's t test for paired or unpaired data, or one-way analysis of variance, as appropriate. P < 0.05 in a two-tailed analysis was considered significant. The number of experiments given in the main text and figure legends corresponds to independent measurements.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding authors on reasonable request.