Method to measure strong protein–protein interactions in lipid bilayers using a steric trap

Abstract

Measuring high affinity protein–protein interactions in membranes is extremely challenging because there are limitations to how far the interacting components can be diluted in bilayers. Here we show that a steric trap can be employed for stable membrane interactions. We couple dissociation to a competitive binding event so that dissociation can be driven by increasing the affinity or concentration of the competitor. The steric trap design used here links monovalent streptavidin binding to dissociation of biotinylated partners. Application of the steric trap method to the well-characterized glycophorin A transmembrane helix (GpATM) reveals a dimer that is dramatically stabilized by 4–5 kcal/mol in palmitoyloleoylphosphatidylcholine bilayers compared to detergent. We also find larger effects of mutations at the dimer interface in bilayers compared to detergent suggesting that the dimer is more organized in a membrane environment. The high affinity we measure for GpATM in bilayers indicates that a membrane vesicle many orders of magnitude larger than a bacterial cell would be required to measure the dissociation constant using traditional dilution methods. Thus, steric trapping can open new biological systems to experimental scrutiny in natural bilayer environments.

The formation of protein complexes within or on cellular membranes is often a key event in cellular signaling and the regulation of activity (1, 2). To fully understand how these complexes are constructed, it is essential to measure the affinities of the interacting partners. The typical approach to measuring dissociation constants is to dilute the protein complex until it falls apart (3, 4). This approach is effective for high affinity interactions in solution because it is straightforward to dilute over many orders of magnitude. Dilution can also be effective for low affinity interactions in a bilayer (5, 6), but it can be difficult to dilute sufficiently to dissociate high affinity complexes due to detection sensitivity limits and physical constraints on vesicle size. Substantial stabilization of membrane protein oligomers in bilayers has been observed in the tetrameric M2 channels (7) and the designed Ser-zipper dimers (8) especially with good hydrophobic matching of the peptide and the bilayer. Using a thiol-disulfide exchange method, the dissociation constants of these peptide oligomers could be measured in detergent micelles and thin lipid bilayers such as dilaurylphosphatidylcholine. In palmitoyloleoylphosphatidylcholine (POPC) bilayers, however, whose hydrophobic thickness matches better with those of the peptides, the dissociation constants could not be measured because the oligomers were too stable. In an effort to develop an approach that does not rely on dilution, we test here whether a method we developed to drive soluble protein unfolding (9), called steric trapping, can be extended to the problem of measuring binding equilibria in membrane systems.

We chose to test the method on a high affinity transmembrane helix dimer, glycophorin A (GpATM). GpATM is a single transmembrane helix that forms a strong homodimer in detergent micelles and lipid bilayers (10–12). It has served as an important model system to study the lateral transmembrane helix–helix interactions that occur in the folding of α-helical membrane proteins (13). GpATM was chosen because the thermodynamic stability of the GpATM dimers has been extensively characterized in various detergent micellar systems using sedimentation equilibrium and Förster resonance energy transfer (FRET) methods (14, 15). FRET methods were recently used to measure GpATM association in membrane vesicles, but whether this was a monomer–dimer equilibrium or a nonspecific association between preformed dimers is not clear (16). Several groups have been able to measure GpATM dimerization in natural membranes from living cells. Duong et al. and Finger et al. have used transcription level readouts to measure relative dimer concentrations in live Escherichia coli cells (17, 18). Chen et al. were able to measure GpATM affinities in mammalian membrane blebs using FRET detection (19). These methods are important because they allow for affinity measurements in complex natural membranes, but they do not allow the facile manipulation of the environment required for hypothesis testing and it is impossible to control or measure other competitive equilibria that likely take place. Indeed, as discussed below, we find that the apparent affinity is much higher in purified membrane systems, presenting a significant challenge to measuring GpATM dimerization affinity in model lipid bilayers.

Results

The Steric Trap Approach.

The principle of the steric trap and the specific design for GpATM are illustrated in Fig. 1 A and B. GpATM was expressed as a fusion protein to the water soluble protein staphylococcal nuclease (SNGpATM) as described previously (10), but we added a biotin acceptor peptide (BAP) so that the protein can be labeled with biotin (20). Enzymatic biotinylation by biotin ligase yielded biotin labeling close to 100% (Fig. S1). As illustrated in Fig. 1B, monovalent streptavidin (mSA) (21) can bind freely to one of the biotin tags. If the biotin tags are placed sufficiently close in space, however, a second mSA can bind to the other subunit only if the protein is dissociated. Consequently, the affinity of the first mSA will reflect its intrinsic affinity for biotin, but the affinity of the second mSA will include GpATM dimer dissociation. As a result, the thermodynamic stability of the GpATM dimer can be extracted from a simple mSA binding assay. The advantage of the steric trap method is that dissociation is driven by mSA concentration, not by the concentration of the target complex. The driving force for the reaction can be easily manipulated either by varying the mSA concentration or by altering mSA affinity by mutation (see below).

Fig. 1.

Principles of the steric trap method. (A) Design of the construct used for steric trapping. GpATM (orange) was fused to a biotin acceptor peptide (20) (BAP, green), which can be biotinylated (red dot) on a unique lysine residue using biotin ligase. A unique cysteine was labeled with a pyrene flourescent probe for the detection of binding/dissociation. (B) Reaction scheme describing the dissociation of the SNGpATM dimer coupled to binding of mSA (green shaded sphere). (C) Deconvolution of the mSA-induced fluorescence changes of the pyrene-labeled SNGpATM in 20 mM DM. The dissociation signal (ΔF_d: blue arrow) obtained by adding a 10× molar excess of unlabeled wild-type SNGpATM contributes about 45% to the total change induced by excess wild-type mSA and wild-type unlabeled SNGpATM (ΔF_d + ΔF_b: blue and red arrows, respectively). The change solely induced by a 5× molar excess mSA (ΔF_bd: green arrow) fully accounts for the total fluorescence change (ΔF_d + ΔF_b). ΔF_rd represents the negligible fluorescence change involving any residual dissociation upon the addition of excess unlabeled SNGpATM (yellow arrow). Pyrene was excited at 330 nm. (D) mSA binding curves monitored by pyrene monomer fluorescence. Two micromolars of SNGpATM in 20 mM DM were titrated with wild-type mSA (open circle), mSA-S27R (filled circle), and inactive SA (triangle). The apparent weaker binding of the mSA-S27R (filled circle) relative to its intrinsic biotin binding affinity (K_d,biotin = 6.5 × 10^-6 M) is illustrated by the dotted line, which depicts the expected binding curve in the absence of energetic coupling to dissociation. The attenuated binding phase was fitted to Eq. S1, which yielded the K_d,GpA = 6.0 × 10^-8 M (Fig. S2). The expected fluorescence changes in the first mSA binding (1/2ΔF_b) and the second mSA binding coupled to the dimer dissociation (ΔF_d + 1/2ΔF_b) are also shown. (E) Pyrene fluoresence spectra showing the reversibility in 20-mM DM micelles. The fluorescence increase upon binding with mSA-S27R was abolished upon the addition of free excess biotin.

To monitor SNGpATM dissociation, we attached a fluorescent pyrene probe at a unique cysteine, Cys-72. We did not observe excimer formation at this position, but we did see significant self-quenching of pyrene fluorescence in the dimer that was relieved upon dissociation and mSA binding (Fig. 1C, ΔF_d + ΔF_b). A large enhancement of fluorescence was seen when the wild-type mSA was added in excess (Fig. 1C, ΔF_bd). This enhancement can be broken into two components: First, decreased self-quenching occurs upon dissociation of the dimer, which is observed by adding excess, unlabeled SNGpATM competitor (Fig. 1C, ΔF_d). Second, an enhancement of fluorescence is observed when mSA binds to the biotin site (Fig. 1C, ΔF_b). The SNGpATM dissociation signal (ΔF_d) contributes to approximately 45% of the total fluorescence increase, whereas the mSA binding signal (ΔF_b) contributes the remaining 55%. Thus, when mSA binds to the first site, we expect a change of about 1/2ΔF_b because only half the potential mSA binding sites become occupied. But when mSA binds to the second biotin we expect a change of 1/2ΔF_b + ΔF_d. These results suggest that, of the total change in fluorescence, about 30% will correspond to binding of mSA to the first biotin in the context of the dimer and 70% will correspond to the binding of the second mSA and concomitant dissociation of the dimer. Thus, we expect that titration of the SNGpATM steric trap construct will yield a binding curve with two phases, and this is essentially what is observed.

A typical binding curve is shown in Fig. 1D (filled circles). Two phases are clearly visible with roughly the expected amplitudes. The first phase reflects high affinity binding of the first streptavidin, and the second phase monitors dissociation of the SNGpATM dimer (also see below). The addition of excess free biotin leads to restoration of the quenched fluorescence signal (Fig. 1E), indicating that the mSA-induced dissociation of the SNGpATM dimer is completely reversible. The mSA-induced dissociation and free biotin-induced reassociation of the dimer were also confirmed by SDS-PAGE experiments (Fig. S1). From the binding curves in Fig. 1D, we can extract dissociation constants as described in Fig. S2 and Eq. S1.

Tuning mSA Affinity for a Wide Range of Dimer Affinities.

Because the steric trap method does not rely on dilution to achieve dissociation, a high sensitivity detection of the protein association/dissociation reaction is not necessary, as long as the mSA affinity is properly tuned. This is illustrated in Fig. 1D, which compares a binding assay employing wild-type mSA [intrinsic K_d,biotin ∼ 10^-14 M (21)] with a binding assay employing a mutant mSA, mSA-S27R (intrinsic K_d,biotin = 6.5 × 10^-6 M). With wild-type mSA, a single phase is observed reflecting stoichiometric binding. It is only with the lower affinity mSA-S27R that the two phases can be observed. The appropriate affinity of the mSA mutant will depend on the affinity of GpATM dimerization and the concentration of SNGpATM (Eq. S1). We, therefore, built a library of mSA variants that have intrinsic affinities for monomeric biotinylated SNGpATM (K_d,biotin) ranging from 2.1 × 10^-9 M to 6.5 × 10^-6 M (Fig. S3) so that the appropriate biotin binding affinity can be selected.

Testing in Detergent Micelles.

For comparison to more traditional approaches, we first tested the steric trap method in detergent micelles. We chose the detergent decylmaltoside (DM) because the protein is well behaved in this detergent and the affinity of GpATM in DM was previously measured using a FRET assay (22). As discussed previously, GpATM dimer stability depends on its concentration in the detergent micelle, not its bulk concentration (23). Thus, increasing detergent concentration at a fixed bulk protein concentration has the effect of diluting and destabilizing the GpATM dimer. In the steric trap method, we would expect that mSA binding should become easier as the SNGpATM dimer is destabilized by dilution and this is exactly what we observe. As shown in Fig. 2A, increasing the DM micellar concentration at a fixed concentration of SNGpATM shifted the second phase of the mSA binding curves to lower mSA concentrations.

Fig. 2.

Dilution effects on the stability of the SNGpATM dimer in DM micelles and POPC lipid bilayers. (A) Dependence of the second mSA binding phase on the concentration of DM. Only the second binding phase is shown. Two micromolars of SNGpATM in the indicated concentrations of DM were titrated with mSA-S27R (K_d,biotin = 6.5 × 10^-6 M) and incubated overnight. The dashed line represents the binding curve expected based on the intrinsic biotin binding affinity of mSA-S27R with no contribution from dimer dissociation. (B) Comparison of the apparent dissociation free energies [ (M)] obtained by the steric trap method (filled square, open square) and FRET [solid line from Fisher et al. (22)]. Data were obtained by the titration measurements in the absence (filled square) and the presence (open square) of the 1-μM wild-type mSA (Fig. S2). (C) The second binding phases of SNGpATM titrated with mSA-E44Q/S45A (K_d,biotin = 9.1 × 10^-9 M) at an increasing lipid-to-protein molar ratio. Only the second phases of the binding curves are shown. The dashed line represents the binding curve expected based on the intrinsic biotin binding affinity of mSA-E44Q/S45A if there was no contribution from dimerization. (D) Comparison of (X) in detergent micelles [DM and C₈E₅ (23)] and POPC bilayers as a function of mol fraction of SNGpATM. In A and C, pyrene was excited at 330 nm, and the pyrene monomer fluorescence change was monitored at 390 nm.

Fig. 2B shows a comparison of the dissociation free energies obtained by the steric trap method to the results obtained previously by Fisher et al. using a FRET assay (22). The apparent dissociation free energies, (M), were calculated in terms of bulk GpATM concentrations with a standard state of 1 M. With bulk concentration units, the apparent free energy of dissociation changes with detergent concentration. Fig. 2B shows a plot of (M) obtained with the steric trap method as a function of DM concentration. The line shown in the plot is the best fit to the data from Fisher et al. (22). Our data agree very well with the previously determined line, indicating that the steric trap method can be reliably used to extract dissociation constants of protein oligomers.

Extension to Bilayers.

We next tested whether the steric trap method could be extended to POPC bilayers. We first had to determine if the biotin tags were accessible in the proteoliposomes. We employed an avidin binding assay and found that 60–80% of biotin tags were exposed outside of the vesicles and accessible for trapping (Fig. S4). Because the dimer affinity of GpATM is not known in model lipid bilayers, various mSA mutants were tested to select an adequate K_d,biotin for probing the stability of the wild-type GpATM dimer (Fig. S5A). For the wild-type GpATM under the conditions used, we found that the mSA-E44Q/S45A mutant (K_d,biotin = 9.1 × 10^-9 M) showed the characteristic biphasic binding curve. The addition of excess free biotin reversed the fluorescence changes, which implies that the system is reversible (Fig. S5B). A similar SNGpATM dimer stability was obtained when the next weaker biotin binding mSA-W79M (K_d,biotin = 4.0 × 10^-8 M, Fig. S3) was used (Fig. S7). Finally, we found that the second phase of the biphasic curves depended on the SNGpATM concentration in bilayers, indicating that the second phase of the binding curves report the binding/dissociation reaction (Fig. 2C). Thus, we believe the steric trap method works in POPC liposomes as it does in detergent micelles.

Comparison of Affinity in Detergent and Bilayers.

The steric trap method now allows us to compare the stability of SNGpATM in detergent and POPC bilayers. To make a direct comparison, we express the dissociation constants in mole fraction units to obtain free energies of dissociation, (X) (23). The results are plotted in Fig. 2D. Ideally, the (X) should be independent of concentration, but this is not the case, indicating nonideal behavior (23). Nevertheless, at comparable concentrations, the (X) in POPC bilayers is ∼4–5 kcal/mol higher than in detergent. Thus, bilayers impart a dramatic stabilization to the SNGpATM dimer.

To examine how the bilayer environment might alter the energetic contributions of individual side chains to dimerization, we compared the effects of mutations on the stability of SNGpATM in C₈E₅ micelles, previously measured by sedimentation equilibrium (14), with DM micelles and POPC bilayers, measured using the steric trap method. The results are summarized in Fig. 3. Ile76 in the GpATM is involved in side-chain packing in the dimer. The Ile76Ala mutation destabilized the dimer in detergent ( in C₈E₅ and 1.2 ± 0.3 kcal/mol in DM), and we find a similar destabilization in lipid bilayers (). For the rest of the mutants tested, however, the mutational effects in lipid bilayers generally surpassed the effects in detergents. The severe destabilization by the Gly83Ile mutation in the GxxxG motif observed in detergent micelles ( in C₈E₅) was even greater in lipid bilayers (). The mutation of Val80Ala led to a significantly larger destabilization () in lipid bilayers than in detergent micelles ( in C₈E₅ and 0.4 ± 0.1 kcal/mol in DM). Thr87 is of particular interest because it is known to form a hydrogen bond between the hydroxylic group of Thr87 and the carbonyl group of Val84 in lipid bilayers (24), but not in detergent (25) (Fig. 3). Indeed, the effect of Thr87Ala mutation was more pronounced in lipid bilayers () compared to the detergents ( both in C₈E₅ and DM), which is consistent with the existence of the hydrogen bonding interactions in lipid bilayers. The magnitude of the destabilization energy of Thr87Ala mutation corresponds to ∼1.0 kcal/mol per residue. Although relatively modest, it is similar to the contributions of putative hydrogen bonding interactions measured in other systems (26–28). These results indicate that the enhancement in stability seen in POPC bilayers is not only a consequence of reduced order in the dissociated state, but also a better organized dimer. The destabilizing effects of the mutants demonstrate that we are measuring dissociation of the structurally characterized dimer.

Fig. 3.

Comparison of the mutational effects on the stability of the SNGpATM dimer in DM micelles and POPC lipid bilayers. The structure of the GpATM dimer solved by solution NMR in DPC micelles (25) is shown at Left. The modeled structure of the GpATM dimer in lipid bilayers using the constraints obtained by solid-state NMR (30) is shown at Right. The closest distances for the potential intermolecular hydrogen bonding pair involving Thr87 are shown for each structure. The side chains of the substituted residues are shown in space filling representation and colored as follows: Ile76 in green, Gly79 and Gly80 in orange, and Thr87 in blue. The of wild type and each Ala mutant were measured by the steric trap method at 20 mM DM. The difference free energies () for each mutant in C₈E₅ (14), DM and POPC bilayers are shown, colored according the specific residue as above. For of each Ala mutant in POPC bilayers, was measured at several concentrations and the results interpolated to the same mole fraction as the wild type (Fig. 2D). The [POPC]/[GpATM] for each mutant varied from 1,400 to 2,600 (Fig. S4).

Discussion

The remarkably enhanced stability of SNGpATM in bilayers can be explained by fewer degrees of freedom in the dissociated state (29), but there are clearly also alterations in the dimer structure itself that are reflected in the energetics of association. In particular, solid-state NMR experiments by Smith et al. indicate that the orientation of the GpATM helices is different in bilayers than in detergent (30) and a new hydrogen bonding interaction is observed across the dimer (24). We have now been able to demonstrate that the new hydrogen bond significantly enhances stability in bilayers. Moreover, other destabilizing mutations have larger energetic effects in bilayers than in detergent, suggesting better organization of the dimer itself.

The stability we observe in POPC vesicles is much higher than was observed by Chen et al. in extruded mammalian membrane vesicles (19) and, we believe, also must be higher than in the bacterial membrane system employed by Duong et al. and Finger et al. (17, 18). Chen et al. report a of 3.9 kcal/mol, using a standard state concentration of 1/nm², which corresponds to a K_d of 1.5 × 10^-3 nm^-2 at 25 °C. Although Duong et al. and Finger et al. cannot determine an absolute K_d in bacterial membranes using the TOXCAT and GALLEX systems, respectively, they both argue that considerable SNGpATM monomer must exist under the conditions used. If we roughly approximate an E. coli cell as a cylinder of 1-μm radius and 2-µm height, the total surface area is about 10⁷ nm². Thus, at the lowest possible concentration with one molecule per bacterial cell, the concentration would be 10^-7 nm^-2. So for there to be a substantial monomer population, the K_d in E. coli cells must be much higher than 10^-7 nm^-2. Converting our results in POPC bilayers to a standard state of 1/nm² (assuming an area of 0.76 nm² per POPC headgroup (31) in 100-nm lipid vesicles), we obtain a K_d of 1.3 × 10^-9 nm^-2 (at a lipid-to-protein molar ratio of 1,000) and a of 12.1 kcal/mol. Thus, the GpATM appears to form a much stronger dimer in model POPC bilayers than is observed in a natural membrane environment.

Why this discrepancy? It is possible that natural membrane compositions alter and weaken the structure of the dimer. Alternatively, and we think more likely, in a natural membrane that is filled with other proteins (32), the monomer form may readily find alternative binding partners, creating competition that effectively stabilizes the dissociated form (18). Soluble proteins also appear to be destabilized in the crowded cellular environment, contrary to expectations from excluded volume arguments (33). These comparisons illustrate how physical studies in purified systems inherently address different issues than in vivo systems and have the potential to inform each other.

The steric trap method reported here provides a previously undescribed tool for physically characterizing interactions that have been largely inaccessible in the past. We have described one implementation of the basic principle, but any approach in which binding and oligomerization are mutually exclusive would achieve the same objective. For example, an antibody that binds near the oligomer interface could be employed. Our approach using streptavidin binding described here has the advantage of flexibility and a wide dynamic range. We were able to measure apparent dissociation constants of 10^-7 to 10^-13 M, without altering the basic experimental design. This is possible because the affinity range can be tuned using different mSA mutants. Most importantly, the method can clearly access high affinity interactions in bilayers. For SNGpATM, we were able to measure a K_d of 1.3 × 10^-9 nm^-2 in POPC bilayers. To measure dissociation constants this low by dilution methods would require vesicles with at least 100 to 1,000 times the surface area of an E. coli cell. We expect that the method will provide a unique and powerful tool to study various problems in membrane protein research, including kinetic and thermodynamic folding studies of larger membrane proteins and their complexes as well as the control of specific cellular signaling events involving membrane-associated proteins and receptors.

Materials and Methods

Cloning, Expression, and Purification of SNGpATM.

The pET11a vector encoding glycorphorinA transmembrane domain (GpATM) fused to staphylococcal nuclease in the N terminus was a kind gift from Karen Fleming (Johns Hopkins University, Baltimore, MD) (10, 34). The SNGpATM gene was amplified by PCR with two primers containing NdeI and XhoI restriction sites, respectively. The PCR product was ligated after digestion to the kanamycin-resistant pET30a vector to fuse the His₆ tag to the C terminus of SNGpATM. The intrinsic XmaI site in the pET30a vector was eliminated by the site-directed mutagenesis to generate pET30aΔXmaI. A DNA cassette encoding the biotin acceptor peptide (20) (BAP: GGLNDIFEAQKIEWHEDGSP) was then inserted into the single XmaI restriction site, which was located right before the 5′ end of GpATM. A unique Cys residue was introduced at Gly72 by site-directed mutagenesis. I76A, V80A, G83I, and T87A SNGpATM mutants were generated by site-directed mutagenesis. All site-directed mutagenesis was carried out using QuickChange (Stratagene). For the detailed procedures of expression and purification of SNGpATM, see SI Appendix.

In Vitro Biotinylation and Fluorescent Labeling of SNGpATM.

Forty milliliters of 100 μM SNGpATM containing a BAP insertion were biotinylated in 50 mM TrisHCl, 200 mM NaCl, 200 mM imidazole buffer (pH 8.5) containing 5 mM ATP (Sigma), 1 mM D biotin (Sigma), 5 mM magnesium acetate, and 2 μM BirA (35, 36) overnight at room temperature. Low molecular weight solutes including excess biotin were removed by an EconoPac 10DG desalting column (Bio-rad), which was equilibrated with 1% DM, 50 mM Tris HCl, and 200 mM NaCl buffer solution (pH 8.5).

For fluorescent labeling of the biotinylated SNGpATM, 50 μM of SNGpATM-G72C in 2% DM, 50 mM TrisHCl, 200 mM NaCl (pH 8.5) was reduced with 15 times molar excess of Tris(2-carboxyethyl)-phosphine hydrochloride (Pierce) for 90 min at room temperature. A 20 times molar excess of thiol-reactive N-(1-pyrenemethyl)iodoacetamide (Invitrogen) dissolved in DMSO (∼20 mg/mL) was added to the SNGpATM solution in the dark. The reaction was slowly stirred at room temperature for ∼16 h in the dark. To remove unreacted pyrene probes, the SNGpATM was immobilized on NiNTA resin and extensively washed with 1% DM, 5 mM imidazole, 20 mM sodium phosphate, and 200 mM NaCl buffer (pH 7.5) followed by an elution with the same buffer containing 200 mM imidazole. The excess imidazole was removed using a Bio-rad 10DG desalting column. Pyrene-labeling efficiency was generally more than 90% and the biotinylation efficiency was close to 100%. For the full description of the procedures, see SI Appendix.

Site-Directed Mutagenesis and Purification of Monovalent Streptavidin.

The ampicillin-resistant pET21a vectors encoding wild-type streptavidin with a C-terminal His₆ tag (active, K_d,biotin = 4 × 10^-14 M) and N23A/S27D/S45A triple mutant (inactive, K_d,biotin = 1.2 × 10^-3 M) streptavidin without His₆ tag were kindly provided by Alice Ting [Massachusetts Institute of Technology (MIT), Boston, MA] (21). To modify the intrinsic biotin binding affinity (K_d,bioitin) of mSA, site-direct mutagenesis (QuickChange, Stratagene) was performed on the active streptavidin gene to generate eight variants [S45A (37), Y43F/S45A, E44Q/S45A, W79M, W79A (38), W79Q, N23A/S45A, and S27R]. For the detailed procedures of expression, purification, and refolding of monovalent streptavidin, see SI Appendix.

Preparation of Proteoliposomes.

Total amount of 15.2 to 60.8 mg of POPC (Avanti polar lipids) dissolved in chloroform (25 mg/mL) were added to glass tubes and dried under a stream of nitrogen gas. The residual solvent was further removed in a vacuum dessiccator for 2 h. The dried lipid films were hydrated and solubilized in 2 mL of 2–6% β-octylglucoside (Anatrace, Anagrade), 20 mM 3-(N-morpholino)propanesulfuric acid (MOPS), 200 mM NaCl (pH 7.5). The SNGpATM stock solution in DM was added to the solubilized POPC at a final concentration of 10 μM and incubated at room temperature for 1 h. SNGpATM was reconstituted into POPC vesicles by dialyzing against 250 sample volumes of 20 mM MOPS, 200 mM NaCl buffer solution (pH 7.5), with three equivalent buffer exchanges over the course of 48 h at 5 °C (Spectra/Por 25 kDa cutoff dialysis membrane, 7.5-mm diameter). The resulting proteoliposomes were passed through Nucleo track etch membrane (Whatman) with 200-nm pore sizes using a miniextruder (Avanti polar lipids) 15 times. The typical average diameter of proteoliposomes was 110 ± 60 nm as measured by dynamic light scattering (DynaPro light-scattering systems, Protein Solutions). The liposomal solutions were stored at 4 °C.

Binding Measurements in Detergent Micelles and POPC Bilayers.

For binding measurements in detergent micelles, 2 μM of biotinylated pyrene-labeled wild-type or mutant SNGpATMs were titrated with mSA in buffer solutions containing 10 mM to 70 mM DM (Anatrace, Anagrade), 20 mM sodium phosphate, and 200 mM NaCl (pH 7.5). A total volume of 70 μL of each sample was transferred to a 96-well UV-compatible microplate (BD-Falcon), sealed with a polyolefin tape, and incubated overnight at room temperature. Binding of mSA coupled with the dissociation of the SNGpATM dimer was monitored by pyrene monomer fluorescence at 390 nm with an excitation wavelength of 330 nm using a SpectraMax M5 plate reader (Molecular Devices). All the fluorescence intensities were corrected by a micellar solution without SNGpATM. The background-subtracted data were fitted to Eq. S1 to extract the apparent dissociation constants (K_d,GpA) of the SNGpATM dimer.

For binding measurements in POPC proteoliposomal solution, 30 μL of proteliposomes was mixed with various compositions of buffer/BSA (Sigma)/mSA mixtures. BSA was added to match the total protein concentrations of the sample solutions, which can affect osmolarity. The total molar concentration of BSA and mSA was adjusted to 100 μM in all sample solutions. A total volume of 100 μL of each sample was transferred to a microplate, sealed with a polyolefin tape, and incubated overnight at room temperature. The sample solutions in the wells were carefully stirred to resuspend settled proteliposomes before measurements. The same instrumental parameters as in the micellar solutions were used. The mSA-induced changes in the pyrene-fluorescence intensity were measured five times and averaged. At the end of the measurements, excess free biotin was added to a final concentration of 2 mM and incubated more than 6 h at room temperature to dissociate bound mSA from the biotinylated SNGpATM. The fluorescence data from the biotin-blocked samples, which were averaged from five time measurements, served as a background. The background-subtracted data were fitted to Eq. S4 to extract K_d,GpA of the SNGpATM dimer.