Insertion of the Membrane-proximal Region of the Neuronal SNARE Coiled Coil into the Membrane*

In the neuron, solubleN-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins assemble into an α-helical coiled coil that bridges the synaptic vesicle to the plasma membrane and drives membrane fusion, a required process for neurotransmitter release at the nerve terminal. How does coiled coil formation drive membrane fusion? To investigate the structural and energetic coupling between the coiled coil and membrane, the recombinant SNARE complex in the phospholipid bilayer was studied using fluorescence quenching and site-directed spin labeling EPR. Fluorescence analysis revealed that two native Trp residues at the membrane-proximal region of the coiled coil are inserted into the membrane, tightly coupling the coiled coil to the membrane. The EPR results indicate that the coiled coil penetrates into the membrane with an oblique angle, providing a favorable geometry for the basic residues to interact with negatively charged lipids. The result supports the proposition that core complex formation directly leads to the apposition of two membranes, which could facilitate lipid mixing. Trp residues and basic residues are abundant at the membrane-proximal region of transmembrane SNARE proteins, suggesting the generality of the proposed mechanism for the SNARE complex-membrane coupling.

Neurotransmitter release at synapses requires the fusion of neurotransmitter-containing vesicles to the presynaptic plasma membrane. Membrane fusion is, however, an exceedingly difficult process to go through without the assistant of specific proteins, because of the protective nature of the biological membranes. In the neuron, soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) 1 proteins play an essential role in promoting membrane fusion (1). It is proposed that assembly of the SNARE complex initially bridges two membranes, induces lipid mixing, and leads to the hemifusion state and the fusion pore, of which the detailed mechanism is largely unknown (2)(3)(4)(5).
Progress has been made in understanding the biophysical principles of SNARE assembly. SNARE assembly starts with the interaction of vesicle-associated membrane protein 2 (VAMP2 or synaptobrevin) with target plasma membrane SNAREs Syntaxin 1A and SNAP-25. Interactions between SNARE proteins are mediated by "SNARE motifs" that are essentially coiled coil sequences and are present in all SNARE proteins (3). For the SNARE complex, one SNARE motif each from Syntaxin 1A and VAMP2 and two from SNAP-25 assemble into a 110-Å-long four-stranded coiled coil (6,7). It is worthwhile to note that target plasma membrane SNAREs Syntaxin 1A and SNAP-25 also spontaneously assemble into a similar but less stable four-stranded coiled coil (8 -10).
How does coiled coil formation lead to membrane fusion? There are two features of the SNARE coiled coil that might be important. First, the helices are all aligned parallel, suggesting the co-location of two membrane attachment points, which sets up a favorable geometry for membrane fusion (6,7,(11)(12)(13). Second, the coiled coil is highly stable (14,15). Therefore, coiled coil formation might have the capacity to overcome the repulsive force between two apposing membranes. Although this mechanistic model appears to be structurally and energetically attractive, there are caveats that require careful consideration. For example, if the SNARE core were tethered with flexible linkers to membrane domains, coiled coil formation might not be able to bring about membrane apposition no matter how strong the pulling force it generates because the energy would be dissipated.
To validate this model, a direct coupling between the coiled coil and membranes appears to be necessary. Previously, Brunger and co-workers proposed a hypothetical model for the coiled coil-to-membrane coupling (7). In this model, the coiled coil is linked to transmembrane domains (TMD) as continuous helices. This model arbitrarily assumes some bending flexibility of helices in short amino acid stretches at the membraneproximal region. Furthermore, helix-disrupting mutations or amino acid insertions in the linker region have little or only moderate effect on the SNARE fusion activity, inconsistent with this model (16,17). How then is the coiled coil energetically coupled to membranes? The answer to this fundamental question hinges on structural information of the connection of the coiled coils to the membranes.
Recent EPR investigations of intact SNAREs using site-directed spin labeling EPR have yielded new results that not only confirm the existence of coupling between the coiled coil and the membrane but also suggest a tentative mechanism of the SNARE core-membrane coupling. EPR analysis indicated that the linker region of Syntaxin 1A, enriched with basic amino * This work was supported by National Institute of Health Grant GM51290. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In this work, we report the EPR and fluorescence investigations of the membrane topology of the recombinant SNARE complex. Fluorescence quenching analysis revealed that the native Trp residues at positions 89 and 90 in VAMP2 are inserted into the acyl chain region of the bilayer. Further, the EPR results reveal that the core domain maintains the coiled coil structure up to residue 92, suggesting that the SNARE coiled coil is partially inserted into the head group region of the bilayer. The EPR data also suggest that the coiled coil penetrates into the membrane with an oblique angle. Taken together, the new results further establish the concept of the tight SNARE core-membrane coupling, providing structural basis for the force transmission from the core region to the membrane during SNARE assembly.
Plasmid Constructs and Site-directed Mutagenesis-Full-length VAMP2 (amino acids 1-116), the soluble SNARE motif of Syntaxin 1A (or Syntaxin H3) (amino acids 191-266), and the C-terminal SNARE motif of SNAP-25 (SNAP-25(C)) (amino acids 125-206) are inserted in the pGEX-KG vector as glutathione S-transferase fusion proteins (22). On the other hand, the N-terminal SNARE motif of SNAP-25 (SNAP-25(N)) (amino acids 1-82) is in the pQE-30 vector (Qiagen) as a Nterminal His 6 -tagged protein. To introduce a unique cysteine site for the specific nitroxide attachment, native cysteine 103 of VAMP2 was changed to alanine. All of the mutants were generated by QuikChange site-directed mutagenesis (Stratagene) and confirmed by DNA sequencing (Iowa State University DNA Sequencing Facility).
Protein Expression, Purification, and Spin Labeling-Recombinant glutathione S-transferase fusion proteins were expressed in E. coli BL21-CodonPlus RIL and purified using glutathione-agarose chromatography. Briefly, the cells were grown at 37°C in LB medium with glucose (2 g/liter), ampicillin (100 g/ml), and chloramphenicol (50 g/ml) until the A 600 reached 0.6 -0.8. After the addition of isopropyl-␤-D-thiogalactopyranoside (0.2 mM), the cells were grown further for 5 h more at 30°C for SNAP-25(C), at 22°C for full-length VAMP2 but at 16°C for the Syntaxin 1A H3 domain.
To purify the protein, the frozen cell pellet was resuspended in PBST-Met buffer (phosphate-buffered saline, pH 7.4, with Tween 20 (0.05%) (percentages are v/v unless otherwise mentioned), 10 mM Lmethionine, 2 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 M leupeptin, and 2 mM dithiothreitol. The cells were broken by sonication on the ice bath. For full-length VAMP2, Triton X-100 (0.5% (v/v) and n-lauroyl sarcosine (0.5%) were added to the solution before sonication. After nutation for 30 min, the cell lysate was centrifuged at 15,000 ϫ g for 15 min at 4°C. The supernatant was then mixed with glutathioneagarose beads in PBST-Met buffer, and the mixture was left at 4°C for 40 min. The protein bound-beads were washed with an excess volume of PBST-Met buffer for SNAP-25(C) and Syntaxin H3 or with PBST-Met-Triton buffer (PBST-Met with 0.5% Triton X-100) for full-length VAMP2.
The cysteine mutants were spin-labeled, while the protein was bound to the beads. After washing the beads with PBST-Met-Triton buffer without dithiothreitol, a 10-fold molar excess of MTSSL was added. The sample was initially reacted with MTSSL at room temperature for 1 h and left to stand at 4°C overnight. Free MTSSL was removed by washing with PBST-Met-Triton buffer, and the spin-labeled protein was cleaved off from the resin with thrombin in the cleavage buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2.5 mM CaCl 2 , 0.1% Triton X-100).
For protein purification, the frozen cell pellet was resuspended in buffer A (50 mM NaH 2 PO 4 , pH 8.0, 300 mM NaCl, 20 mM imidazole, 2 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 M leupeptin). After sonication on ice, the cell lysate was centrifuged at 15,000 ϫ g for 15 min at 4°C. The supernatant was mixed with nickel-nitrilotriacetic acid-agarose beads in buffer A. The mixture was left equilibrated at 4°C for 40 min. After equilibration the beads were washed with an excess volume of buffer A.
The protein concentration was estimated by Bio-Rad protein assay kit using bovine serum albumin as a standard. The spin labeling efficiency was estimated by comparing the double integration with the standard Tempo sample at 100 M. Spin labeling was nearly quantitative for all VAMP2 mutants.
Preparation of the Recombinant SNARE Complex-Two separate SNARE motifs of SNAP-25 lacking the long interhelical loop were used to avoid the formation of domain-swapped complex (23). While His 6tagged SNAP-25(N) was bound to nickel-nitrilotriacetic acid-agarose resin, excess amounts of purified Syntaxin 1A H3, SNAP-25(C), and full-length VAMP2 (molar ratio of 1:1:2) were added to the solution. The solution contained 0.2% Triton X-100. The mixture was left to stand at 4°C for 60 h. After extensive washing with buffer A containing 0.6% (w/v) OG, the recombinant SNARE complex was eluted with buffer A with 250 mM imidazole and 0.6% (w/v) OG. The complex was concentrated using a Centricon (5 K cut-off). During the concentration process buffer was changed to buffer B (25 mM HEPES, pH 7.7, 100 mM KCl, 10% glycerol) with 0.6% (w/v) OG. The final protein concentration was in the range of 80 -100 M.
Membrane Reconstitution of Recombinant SNARE Complex-Large unilamella vesicles with a 100-nm diameter (100 mM total lipids) were prepared in buffer B without OG using an extruder. Vesicles of POPC containing 15 mol % of DOPS were first mixed with two volumes of the SNARE complex. OG was added to the mixture to a final concentration of 0.6%. After dilution with an equal volume of buffer B, the samples were dialyzed against the buffer B containing Bio-beads SM2 adsorbent at 4°C for 40 h. During the dialysis, the buffer was changed three times. The samples were centrifuged at 100,000 ϫ g for 5 min to remove both the protein precipitates and the fraction of large vesicles.
EPR Data Collection and Accessibility Measurements-EPR spectra were obtained using a Bruker ESP 300 spectrometer (Bruker, Germany) equipped with a low noise microwave amplifier (Miteq, Hauppauge, NY) and a loop-gap resonator (Medical Advances, Milwaukee, WI). The modulation amplitude was set at no greater than one-fourth of the line width. The spectra were collected at room temperature in first derivative mode. The gas exchange to the protein sample was achieved with the TPX EPR tube for the loop-gap resonator. For individual mutants, the power saturation curves were obtained from the peak-to-peak amplitude of the central line (M ϭ 0) of the first derivative EPR spectrum as a function of incident microwave power in the range 0.1-40 mW. Three power saturation curves were obtained after equilibration: (i) with N 2 , (ii) with air (O 2 ), and (iii) with N 2 in the presence of 200 mM NiEDDA. From saturation curves, the microwave power P1 ⁄2 , where the first derivative amplitude is reduced to one-half of its unsaturated value, was calculated. The quantity ⌬P1 ⁄2 is the difference in P1 ⁄2 values in the presence and absence of a paramagnetic reagent. The ⌬P1 ⁄2 value is proportional to the diffusion co-efficient times the collision frequency of the nitroxide to the freely diffusing reagents such as oxygen and NiEDDA. Thus, ⌬P1 ⁄2 is considered to be equivalent to the accessibility W. The immersion depth is calculated based on the reference curves determined from a set of lipid molecules spin-labeled at different acyl chain positions.
Fluorescence Quenching Experiment-For fluorescence measurements, membrane samples were prepared from the recombinant SNARE complex with the wild-type sequences. Total lipid concentration was ϳ2.5 mM, whereas the concentration of the SNARE complex was 5 M. For acrylamide quenching, appropriate amounts of the acrylamide stock solution (2 M) were added to the membrane sample to make the final concentration in the range of 0 -160 mM, whereas lipids and protein concentrations remain constant among samples. The fluorescence measurements were carried out with PerkinElmer fluorescence spectrophotometer. The samples were excited at 285 nm, and the emission spectra were collected in the range of 300 -400 nm. The total fluorescence intensity F was obtained by integrating the intensity in this spectral range. The degree of quenching was analyzed according to the following Stern-Volmer equation.  (24). After fluorescence measurements, the membrane samples were treated with Triton X-100 (final concentration, 1%) to subsequently measure the acrylamide quenching in a detergent-solubilized state.
To measure the membrane immersion depth of Trp residues, membrane samples containing two types of lipid quenchers were prepared. Lipid quencher, 6,7-Br 2 -PC or 11,12-Br 2 -PC, was added in replacement of part of POPC while maintaining the DOPC mole fraction at 15%. The reconstitution of SNARE complex was carried out by the same procedure described above. The degree of quenching was determined as a function of the mole fraction of added brominated PC.
The averaged immersion depth of two Trp residues was calculated according to the parallax analysis (25,26). The distance of the Trp residue from the bilayer center Z CF is given by the following.
where L C1 represents the distance from the bilayer center to the shallow quencher (11 Å for 6,7-Br 2 -PC), C is the mole fraction of the quencher divided by the area of the lipid molecule (70 Å 2 ), F 1 and F 2 are the relative fluorescence intensities of the shallow (6,7-Br 2 -PC) and deep quenchers (11,12-Br 2 -PC), respectively, and L is the difference in the depth of the two quenchers (0.9 Å/CH 2 or CBr 2 group). The thickness of the hydrophobic region was approximated to be ϳ29 Å (27). For the calculation of immersion depth, the data collected for the 0.4 molar fraction quencher were used.

RESULTS
Fluorescence Quenching Experiments-For fluorescence measurement the recombinant complex was assembled from full-length VAMP2, soluble Syntaxin 1A, and two separate SNARE motifs from SNAP-25 (Fig. 1). The TMD of Syntaxin 1A was not included this time to avoid the coexistence of two TMDs in one membrane and to best mimic the "hypothetical" trans-SNARE complex in which two TMDs are separately anchored to two apposing membranes. In the recombinant SNARE complex, there are total two native Trp residues. Both of them reside at the membrane-proximal region of VAMP2 (Trp 89 and Trp 90 ) and belong to the coiled coil in the core structure (Fig. 1). This is ideal for the investigation of the possible coiled coil-membrane coupling using fluorescence.
First, to examine whether these residues are exposed to the solvent or not, we monitored Trp fluorescence in the presence of an added quencher acrylamide that is hydrophilic and partitions heavily into the solution phase. For the detergent-solubilized SNARE complex, the fluorescence intensity (F) decreases sharply as the acrylamide concentration increases, suggesting that Trp residues are solvent-exposed (Fig. 2a, open   circles). In contrast, for the membrane-reconstituted complex, F decreases little in the presence of added acrylamide (Fig. 2a,  closed circles). This result strongly implies that Trp residues in the SNARE complex are sequestered from the water phase, suggesting the possibility of insertion into the membrane.
Next, the insertion of Trp residues into the membrane is probed utilizing brominated lipids in which bromines are attached to the acyl chain of the lipid. In the presence of a brominated lipid, Trp fluorescence is effectively quenched only when Trp is in contact with the acyl chain region of the membrane. Otherwise, we would expect a negligible effect. As the mole fraction of the brominated lipid increases, a significant decrease in F was observed (Fig. 2b), suggesting that Trp residues are inserted into the membrane. The immersion depth of Trp residues was calculated based on quenching efficiencies by the shallow lipid quencher 6,7-Br 2 -PC (the lipid with bromines at the sixth and seventh carbon positions) and the deep lipid quencher 11,12-Br 2 -PC. Trp residues are ϳ8.8 Å below the phosphate groups of lipids. Because there are two Trp residues, the immersion depth determined here must be an average depth of two residues. In conclusion, Trp 89 and Trp 90 that belong to the coiled coil are inserted in the membrane in the SNARE complex.
Site-directed Spin Labeling EPR-To investigate the membrane topology of the SNARE complex further using site-directed spin labeling EPR, residues of full-length VAMP2 near the membrane-water interface were replaced with cysteines, to which a nitroxide spin label was attached. We prepared 13 spin-labeled mutants (K83C-M95C) to explore the interfacial region with EPR (Fig. 1). For EPR measurements, the recombinant complex was assembled from spin-labeled VAMP2, soluble Syntaxin 1A, and two separate SNARE motifs from SNAP-25. All spin-labeled recombinant complexes were capable of forming the SDS-resistant complex, which is one characteristic feature of the core complex, as confirmed with SDS-PAGE (data not shown).
After reconstitution of the SNARE complex into POPC vesicles containing 15 mol % of DOPS, the EPR spectra were collected for spin-labeled mutants at room temperature. The EPR spectrum is sensitive to the tumbling rate of the nitroxide. EPR spectra shown in Fig. 3 are all relatively broad, indicating slow motion. Slow motional spectra represent motionally restricted nitroxides. There are three structural factors that might have contributed to the immobilization of the nitroxide (28 -30): (i) The motional restriction of the peptide backbone, because of the ␣-helical secondary structure, could have re- duced the tumbling rate of the nitroxide. (ii) Tertiary interactions with other parts of the protein would slow down the motion of the nitroxide significantly. On the basis of the crystal structure (7), we expect that the four-stranded coiled coil structure extends up to residue 92, which gives rise to many potential tertiary contacts between helices. (iii) From the fluorescence measurement it is clearly shown that Trp 89 and Trp 90 are inserted into the acyl chain region of the bilayer. Therefore, we expect that a significant part of the region is immersed in the membrane, which exposes nitroxides into the viscous membrane environment and the high density head group region. It is likely that the combination of all three factors contributed to the EPR spectral broadening.
Additionally, to examine the possibility of intermolecular interactions between SNARE complexes, we measured low temperature (130 K) EPR spectra in which the spectral broad-ening caused by the spin-spin interaction is readily identified (31). Comparison of the low temperature EPR spectra with the standard confirmed that SNARE complexes are separated from each other beyond the detectable distance range (less than 25 Å) (data not shown), eliminating the possibility of self-aggregation of the SNARE complex.
It is interesting to note that EPR spectra for positions 93-95 are less broad than others. In fact, EPR spectra for these three positions closely resemble those observed for the nitroxide attached to the membrane-inserted linker region of Syntaxin 1A (18,19).
EPR Accessibility Measurements-The EPR line shape is a useful parameter for the tumbling rate of the nitroxide that, in many cases, provides a qualitative assessment of the local environment surrounding the nitroxide (29). However, the line shape alone is often not sufficient to yield information pinpointing the secondary and the tertiary structures. Furthermore, the partial insertion of the protein into the bilayer makes the matters complicated for the SNARE complex. Here, we utilized the EPR saturation method to assess the local structure and the membrane topology of the linker region. For the nitroxide, the EPR saturation method measures the accessibility to a water-soluble paramagnetic reagent such as NiEDDA (W NiEDDA ) to estimate the solvent exposure of the spin-labeled site, or the accessibility to a nonpolar paramagnetic reagent such as molecular oxygen (W O 2 ) to probe, for example, the insertion into the membrane (32)(33)(34).
In Fig. 4 W NiEDDA and W O 2 for the SNARE complex are plotted against the residue number, respectively. We observe an overall decrease of W NiEDDA , whereas we detect an overall increase of W O 2 . Interestingly, however, there are quite significant increases and decreases for W NiEDDA along the sequence, which might imply a secondary structure such as ␣-helix. Further, the W O 2 values show variations that appear to be in opposite directions to those of W NiEDDA , although the trend is much less clear. Such an out-of-phase oscillatory behavior of W NiEDDA and W O 2 has been previously found for ␣-helical peptides residing at membrane-water interface, which includes the fusion peptide of influenza hemagglutinin and a synthetic amphiphilic peptide (32,35,36).
Quantitatively, the ratio of W NiEDDA to W O 2 has been shown to be a useful parameter to characterize the secondary structure. For a ␣-helix we expect a periodical behavior of this parameter along the sequence with a periodicity of 3.5. In Fig. 5, the ⌽ value, which is defined as the logarithm of the ratio of W O 2 to W NiEDDA , is plotted as a function of residue number. We observe a significant variation of the ⌽ value along the sequence. In particular, it appears that there is a periodic oscillation of the ⌽ values in the region of residues 83-92. To better represent this oscillatory behavior, we fit the data with a sine function of the 3.5 residue repeat, which represents the ␣-helical geometry. In this fit, we also take into account the overall decreasing trend of the ⌽ values along the sequence as a linear term added to the sine function. The EPR data are shown overlapped with the fit in Fig.  5. Despite the qualitative nature of the ⌽ value, the EPR results clearly fall into the pattern of the ␣-helical geometry, consistent with the structure inferred from the crystal structure in which the last ordered residue of VAMP2 is 92. The fit also indicates that the helix is tilted with respect to the membrane surface, although the exact magnitude of the tilt could not be estimated from the fit.
For lipid-exposed nitroxides, the ratio of W NiEDDA to W O 2 has been shown to hold a quantitative relationship to the membrane immersion depth. However, EPR spectra for positions 84, 87, 90, and 92 contain highly immobilized spectral components with large outer hyperfine splitting. Such high immobilization might be due to some tertiary interactions of the nitroxide in the complex. This certainly limits the application of the membrane depth analysis in the region of residues 83-92. To our advantage, it is already shown that Trp 89 and Trp 90 are inserted into the membrane 8.8 Å deep from the lipid phosphate group.
As mentioned before EPR spectra for positions 93-95 exhibit intermediate motional rates, characteristic of lipid-exposed nitroxides. It is also highly likely that these three residues are disordered when judged from the disagreement with the continued ␣-helical geometry. For these three positions, we compared the ⌽ values with those obtained from the spin-labeled lipids in the membrane of the same lipid composition containing the similar concentration of the unlabeled SNARE complex. The immersion depth analysis revealed that these three positions are all inserted into the membrane (Fig. 5, panel to the right of the dashed line).
Combining the EPR data and the fluorescence data, we conclude that the C-terminal part of the coiled coil is inserted into the membrane at an oblique angle. Further, we propose that FIG. 6. Model for the trans-SNARE complex. The highlight of this model is the insertion of the C-terminal ends of coiled coil helices into two apposing membranes. The insertion of interfacial Trp residues (yellow side chains) and the electrostatic interactions between the basic residues (green spheres) and the membrane surface charge contribute to the stability. Coiled coil formation tightly pins two membranes to allow the juxtaposition. SNARE proteins are color-coded: red, Syntaxin 1A; blue, VAMP2; and gray, SNAP-25.