Membrane Interaction of the Glycosyltransferase WaaG

552

Biophysical Journal

Volume 109

August 2015

552–563

Article Membrane Interaction of the Glycosyltransferase WaaG Jobst Liebau,1 Pontus Pettersson,1 Scarlett Szpryngiel,1 and Lena Ma¨ler1,* 1

Department of Biochemistry and Biophysics, Stockholm University, Stockholm, Sweden

ABSTRACT The glycosyltransferase WaaG is involved in the synthesis of lipopolysaccharides that constitute the outer leaflet of the outer membrane in Gram-negative bacteria such as Escherichia coli. WaaG has been identified as a potential antibiotic target, and inhibitor scaffolds have previously been investigated. WaaG is located at the cytosolic side of the inner membrane, where the enzyme catalyzes the transfer of the first outer-core glucose to the inner core of nascent lipopolysaccharides. Here, we characterized the binding of WaaG to membrane models designed to mimic the inner membrane of E. coli. Based on the crystal structure, we identified an exposed and largely a-helical 30-residue sequence, with a net positive charge and several aromatic amino acids, as a putative membrane-interacting region of WaaG (MIR-WaaG). We studied the peptide corresponding to this sequence, along with its bilayer interactions, using circular dichroism, fluorescence quenching, fluorescence anisotropy, and NMR. In the presence of dodecylphosphocholine, MIR-WaaG was observed to adopt a three-dimensional structure remarkably similar to the segment in the crystal structure. We found that the membrane interaction of WaaG is conferred at least in part by MIR-WaaG and that electrostatic interactions play a key role in binding. Moreover, we propose a mechanism of anchoring WaaG to the inner membrane of E. coli, where the central part of MIR-WaaG inserts into one leaflet of the bilayer. In this model, electrostatic interactions as well as surface-exposed Tyr residues bind WaaG to the membrane.

INTRODUCTION Glycosyltransferases (GTs) form a large and diverse class of enzymes that catalyze a multitude of transfer reactions by the formation of glycosidic bonds. Most GTs adopt either a GT-A or a GT-B fold (1,2). However, crystal structures of representatives of a third fold, GT-C, have recently been solved (3,4), and some GTs are predicted to adopt none of these folds (5). GTs are further grouped into retaining or inverting enzymes depending on whether the stereochemistry of the donor molecule is retained during the transfer reaction or not. No correlation exists between the fold and the mechanism since there are inverting and retaining GTs of both major folds (6). The GT WaaG adds the first outer-core D-glucose from UDP-glucose to the inner-core L-glycero-D-manno heptose II of lipopolysaccharides, which are the main constituents of the outer leaflet of the outer membrane of Gram-negative bacteria and form a protective barrier against environmental stress and antimicrobial compounds (7). The transfer reaction catalyzed by WaaG retains the a-stereochemistry of the donor in the product. WaaG is thus classified as a retaining a-1,3-GT belonging to the CAZy (Carbohydrate Active Enzymes database, http://www.cazy.org/) (6,8) family GT4 of GTs, which may be the ancestral family of retaining GTs (6,9). The protein sequence of WaaG is highly conserved in

Submitted February 27, 2015, and accepted for publication June 19, 2015. *Correspondence: [email protected] This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/).

Escherichia coli and Salmonella, with sequence identities above 85%. Moreover, deletion of WaaG leads to destabilization of the outer membrane and to nonmotile E. coli strains with an increased sensitivity to novobiocin (10,11). WaaG has therefore been identified as a potential antibiotic target, and inhibitor scaffolds have been studied (12). As evidenced by the crystal structure (Fig. 1), WaaG adopts a GT-B fold characterized by two domains, both of which have a Rossman fold. The active site is located within the cleft formed between the N- and C-terminal domains. WaaG localizes at the cytosolic side of the inner membrane in Gram-negative bacteria such as E. coli (9). The inner membrane of E. coli consists mainly of three lipids. The most abundant lipid is phosphoethanolamine (PE, 70
80%), followed by phosphoglycerol (PG, 20
25%) and cardiolipin (CL, 5
10%) (13,14). None of these lipids are absolutely essential for the growth of E. coli, as previous studies showed that mutants deficient in each one of these lipids could be constructed and grown under laboratory conditions (15,16). For adaptation to environmental stress, the acyl chain lengths and the degree of unsaturation can vary quite significantly. Under normal growth conditions, 16:0 acyl chains are the most abundant and the ratio of unsaturated to saturated fatty acids is ~1:3 (14,17,18). The membrane interaction of monotopic GTs of the GT-B fold has been of special interest because such proteins are able to catalyze the transfer of sugars from hydrophilic donor molecules to membrane-associated acceptors at the membrane surface. Electrostatic and hydrophobic

Editor: Kalina Hristova. Ó 2015 The Authors 0006-3495/15/08/0552/12

http://dx.doi.org/10.1016/j.bpj.2015.06.036

Membrane Interaction of WaaG

553 TABLE 1 Chemicals used in this study ordered by their abbreviation Abbreviation 5-doxyl PC 10-doxyl PC DHPC-d22 DMPC DMPG DPC-d38 or DPC Gd(DTPA-BMA) POPC POPE

FIGURE 1 Crystal structure of WaaG (PDB accession code 2IW1), with MIR-WaaG indicated. The figure was prepared using PyMOL. To see this figure in color, go online.

interactions by surface-exposed cationic and hydrophobic residues, as well as by Trp residues in the N-terminal domain, have been identified as key contributors to membrane anchoring (19). Here, we characterized the binding of WaaG to the inner membrane of E. coli using a set of biophysical methods. For this purpose, we identified a 30-residue membrane-interacting region located in the N-terminal domain of WaaG (MIRWaaG) as a putative anchor of WaaG to the membrane (Fig. 1). The identification criteria were that the region 1) had to be surface exposed, 2) had a net positive charge, and 3) had to contain aromatic amino acids. To study membrane interactions, we prepared bicelles and large unilamellar vesicles (LUVs) with varying relative concentrations of anionic lipids, but kept the fatty acid composition constant (14:0-14:0 in bicelles and 16:0-18:1 in LUVs). The bicelles were prepared with relatively short, saturated fatty acid chains because it is difficult to prepare and control the properties of this mimetic with other lipids (20,21). Vesicles can be prepared in a controlled way with a wider range of lipids, and therefore 16:0-18:1 chains were chosen. MATERIALS AND METHODS Materials MIR-WaaG peptide (YAEKVAQEKGFLYRLTSRYRHYAAFERATF) corresponding to residues 103–132 in the crystal structure (PDB accession code 2IW1) was synthesized by PolyPeptide Laboratories (Strasbourg, France) with a purity of 93%. The lipids, detergents, and paramagnetic agents used in this study are listed in Table 1 together with their abbreviations used in the text. All chemicals were used without further purification.

Preparation of LUVs Six different types of LUVs were prepared for circular dichroism (CD) and fluorescence studies according to a previously described protocol (22,23). These consisted of zwitterionic LUVs (100% POPC); anionic LUVs in

POPG Tempo-PC TOCL

Chemical 1-palmitoyl-2-stearoyl-(5-doxyl)-sn-glycero-3phosphocholine 1-palmitoyl-2-stearoyl-(10-doxyl)-sn-glycero-3phosphocholine tail-deuterated 1,2-dihexanoyl-sn-glycero-3phosphocholine 1,2-dimyristoyl-sn-glycero-3-phosphocholine 1,2-dimyristoyl-sn-glycero-3-phospho-(10 -racglycerol) deuterated and undeuterated dodecylphosphocholine Gd3þ diethylenetriamine pentaacetic acidbismethylamide 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine 1-palmitoyl-2-oleoyl-sn-glycero-3phosphoethanolamine 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(10 -racglycerol) 1,2-dipalmitoyl-sn-glycero-3-phospho(tempo) choline tetraoleoyl cardiolipin

All chemicals were purchased from Avanti Polar Lipids except for Gd(DTPABMA), which was acquired from AK Scientific (Union City, CA). which 10%, 20%, 30%, or 40% of the POPC was substituted for the anionic lipid POPG; and LUVs mimicking the inner-membrane composition of E. coli with 60% POPE, 30% POPG, and 10% TOCL. The required amounts of lipids were dissolved in chloroform and mixed to achieve the desired lipid molar ratios. Chloroform was then evaporated under a stream of nitrogen gas. The lipid mixtures were resuspended in 50 mM of phosphate buffer (pH 7.2) by thorough vortexing for 15 min. Next, the samples were subjected to five cycles of freezing in liquid nitrogen and thawing in a warm water bath. Finally, vesicles were extruded through polycarbonate membranes (100 nm pore diameter; Avanti Polar Lipids, Alabaster, AL) to obtain a homogeneous size distribution (24). The total lipid concentration was 1 mM. The size distribution of LUVs in all samples was checked by dynamic laser light scattering and found to be homogeneous with a typical hydrodynamic radius of ~60 nm.

Preparation of bicelles Bicelles were prepared as described previously (23,25). Briefly, the required amounts of DMPC and DMPG were mixed in 50 mM of phosphate buffer (pH 7.2) in 100% D2O and subsequently vortexed. DHPC-d22 was then added to obtain 150 mM (total lipid þ DHPC concentration) of q ¼ 0.5 bicelles with either 100% DMPC or 70% DMPC þ 30% DMPG. Here, q indicates the molar ratio of lipids (DMPC or DMPC þ DMPG) to DHPC-d22. Finally, the samples were subjected to three vortex-centrifuge cycles.

CD CD spectra of MIR-WaaG were measured on a Chirascan CD spectrometer (Applied Photophysics, Leatherhead, UK) at 25 C in the range of 190– 260 nm. A quartz cuvette of 1 mm path length was used. Thirty spectra of MIR-WaaG in each LUV environment were recorded and averaged. The background spectrum (average of 30 spectra with only LUVs in buffer) was subtracted from the peptide spectrum. The concentrations were 1 mM of lipids and 30 mM of peptide, giving a peptide/lipid ratio of 1:33. CD spectra of MIR-WaaG in DPC micelles were acquired at a 50 mM DPC and 0.4 mM MIR-WaaG concentration in a quartz cuvette with a 5 mm path length. Biophysical Journal 109(3) 552–563

554

Liebau et al.

Fluorescence studies Fluorescence measurements were carried out on a Fluorolog fluorescence spectrometer (Horiba Scientific, Edison, NJ) at 25 C using a quartz cuvette with a 1 cm path length. All recorded spectra were corrected for background signal. The concentrations were 1 mM of lipids and 20 mM of peptide, giving a peptide/lipid ratio of 1:50. Tyr fluorescence quenching was observed by titration of the water-soluble quencher acrylamide and the lipophilic quencher 5-doxyl PC to a final concentration of 0.2 M and 0.1 mM, respectively. 5-Doxyl PC was titrated from a 2.1 mM ethanolic stock solution. The ethanol concentration in the sample never exceeded 5% (v/v). The excitation wave length was 280 nm and the maximum emission was detected at ~307 nm. The Stern-Volmer quenching constant, KSV, reports on the quencher accessibility of the fluorophore (the Tyr residues), as is evident in the equation given by I0 =IQ ¼ KSV ½Q þ 1 (26), where IQ and I0 are the fluorescence intensities at quencher concentration [Q] and in the absence of quencher, respectively. This relation was used to investigate the association of the peptide with LUVs with varying PG content (0–40%) and with a native-like E. coli inner-membrane composition. To obtain a quantitative estimate of the immersion depth of MIR-WaaG in E. coli membrane-mimicking LUVs, we employed the parallax method. Tempo-PC, 5-doxyl PC, and 10-doxyl PC were titrated from a 2.1 mM ethanolic stock solution to 1 mM LUVs and 30 mM MIR-WaaG to obtain LUVs containing 16 mol % spin label. As described in detail previously (27,28), the immersion depth z of the fluorophores can be obtained from

  lnðF1 =F2 Þ z ¼ Lc1 þ L221
2L21 ; pC

(1)

where Lc1 is the distance of the shallow quencher from the center of the bilayer, L21 is the distance between the shallow quencher and the deep quencher, C is the quencher concentration in the membrane plane, and F1 and F2 are the fluorescence intensities in the presence of the shallow and deep quenchers, respectively. The parameters were chosen as described previously (27,28), i.e., the Lc1 of Tempo-PC, 5-doxyl PC, and 10-doxyl PC ˚ , respectively. C is defined as the molar ratio of was 19.5, 12.2, and 7.7 A spin-labeled lipids to total lipids per lipid area (which is assumed to be ˚ 2). 70 A For steady-state fluorescence anisotropy measurements, Glan-Thompson polarizers were employed and the measured signals were corrected for background scattering. Each component of the anisotropy was measured 10 times. To determine the dissociation constant KD of MIR-WaaG in E. coli membrane-mimicking vesicles, 500 nM of peptide was dissolved in 50 mM of phosphate buffer (pH 7.2) and LUVs were titrated from a 1 mM stock solution up to a concentration of 0.5 mM. The fluorescence intensity was corrected for background emission and fitted to a binding function that assumes a single binding site:

F
F0 ½L ; ¼ Bmax KD þ ½L F0

(2)

where F0 is the maximum fluorescence intensity in the absence of LUVs, F is the maximum fluorescence intensity at lipid concentration [L], and Bmax is the maximum binding (29).

Translational diffusion We measured the translational diffusion coefficients of 0.6 mM MIR-WaaG in buffer, as well as in solutions containing 50 mM of DPC micelles (peptide/detergent ratio of 1:83) and zwitterionic (DMPC) and anionic (70% DMPC þ 30% DMPG) bicelles with 50 mM of lipids and 100 mM of DHPC-d22 (at a peptide/lipid ratio of 1:83), at 25 C by using a modified Stejskal-Tanner spin-echo pulse sequence on a 600 MHz Bruker Avance Biophysical Journal 109(3) 552–563

spectrometer. The diffusion time was kept constant while the pulsed field gradient (PFG) was increased in 32 linear steps (30–32). A 1% H2O in D2O and 0.1 mg/mL GdCl3 standard sample was used to calibrate the gradient strengths. The translational diffusion coefficient was obtained from a fit of the signal attenuation to a modified Stejskal-Tanner equation that accounts for field inhomogeneities (33). The diffusion coefficients of MIR-WaaG were determined by taking the average of four dominant 1H aromatic peaks, and lipids were monitored on three 1H acyl chain peaks (H3, H4-H13, and H14). The diffusion coefficients were corrected for viscosity differences by multiplying with the diffusion coefficient of trace HDO in the sample and dividing with the diffusion coefficient of H2O.

Structure determination The NMR solution structure of MIR-WaaG was determined in 50 mM of DPC-d38 and 50 mM of phosphate buffer (pH 5.7), with a peptide concentration of 0.6 mM at 25 C and containing 10% D2O for frequency locking. Experiments were conducted on a 700 MHz Bruker Avance spectrometer equipped with a cryoprobe. Homonuclear 2D experiments were acquired with 64 scans and 512 increments, and 13C-HSQC spectra were obtained with 128 scans and 700 increments. TOCSY spectra were acquired with mixing times of 30 and 95 ms. Structural constraints were obtained from a NOESY spectrum with a 50 ms mixing time (distances) and from natural-abundance 13C-HSQC (chemical shifts). All spectra were processed in TopSpin 3.0 and assigned with Sparky. Protein backbone torsion angles were predicted by TALOSþ (34,35) based on Ha, Ca, and Cb chemical shifts. Only predictions classified as good for residues located within secondary structure elements, predicted using Ha and Ca secondary chemical shifts as described by Wishart and Sykes (36), were used as torsion-angle constraints. We calculated 100 structures by simulated annealing using CYANA 2.1 with automated assignment of NOESY peaks when the peaks could not be unambiguously assigned manually. The final structure ensemble consisted of the 20 energetically most-favored structures ranked by their constraint function values. The structures were validated using the Protein Structure Validation Suite (PSVS, http://psvs-1_5-dev.nesg.org/) (37).

Paramagnetic NMR studies The reduction of 1H-13Ca crosspeak intensities in natural-abundance 13CHSQC spectra acquired on a 700 MHz Bruker Avance spectrometer equipped with a cryoprobe (600 scans, 200 increments in the Ca region of the spectrum) was studied in the presence of Gd(DTPA-BMA). The required amounts of phosphate buffer (50 mM, pH 7.2) and Gd(DTPA-BMA) (10 mM) were mixed and lyophilized. A solution of 0.6 mM of MIRWaaG and 50 mM of DPC-d38 dissolved in 100% D2O was prepared, added to the lyophilized samples, and thoroughly vortexed. This ensured that the peptide and micelle concentrations were the same in both the sample containing Gd(DTPA-BMA) and the reference sample. Reduced crosspeak intensities were referenced to the crosspeak intensities of a spin-label-free sample acquired under the exact same conditions. To estimate how deep the residues of MIR-WaaG inserted into the micelle, we also measured the reduction of 1H-13C crosspeak intensities in natural-abundance 13C-HSQC spectra of nondeuterated DPC in the presence of Gd(DTPA-BMA). These experiments were acquired with eight scans and 500 increments on a 600 MHz Bruker Avance spectrometer equipped with a triple-resonance probe.

CLEANEX studies CLEAN chemical exchange (CLEANEX) experiments report on proton exchange on a millisecond timescale and on NOE crosspeaks to water (38,39). For these experiments, 0.6 mM of MIR-WaaG and 50 mM of DPC-d38 were

Membrane Interaction of WaaG dissolved in 50 mM of phosphate buffer (pH 5.7) with 90% H2O and 10% D2O. A 1D CLEANEX experiment was recorded at 25 C with a 200 ms mixing time using a 600 MHz Bruker Avance spectrometer equipped with a triple-resonance probe.

RESULTS Anionic lipids induce ordered structure formation in MIR-WaaG The secondary structure of MIR-WaaG was determined in buffer and in the presence of zwitterionic (100% POPC), anionic (10–40% POPG), and E. coli inner-membrane-like LUVs using far-UV CD spectroscopy of the peptide bond. The dominant secondary structure can be inferred from the characteristic features of such spectra (40). The spectra are shown in Fig. 2. For MIR-WaaG in buffer and in the presence of zwitterionic and slightly anionic (up to 10% POPG) LUVs, we observed an overall spectrum characteristic of random coil structures with minima between 199 and 200 nm. As the anionic charge increased in the LUVs, the minimum was red-shifted and a maximum below 200 nm, as well as a second minimum above 220 nm, appeared. These features are characteristic of a-helical structures. At high anionic charge (40% POPG) and in E. coli inner-membrane-like LUVs, MIR-WaaG adopts a mostly a-helical conformation. We conclude that the presence of anionically charged membranes leads to the induction of ordered secondary structure in MIR-WaaG. To investigate whether the negative charge was responsible for inducing the structure, we conducted a similar titration study in which we substituted up to 40% of POPC for POPE (data not shown). The presence of POPE did not have any effect on the secondary structure of MIR-WaaG, which remained unstructured in LUVs that were enriched in POPE.

FIGURE 2 CD spectra of MIR-WaaG in buffer and zwitterionic, anionic, and E. coli inner-membrane-like vesicles. The ratios in the legend refer to the molar ratio of POPC/POPG in the vesicles. E. coli refers to a lipid mixture (60% POPE, 30% POPG, 10% TOCL) that mimics the composition of the inner membrane of E. coli. To see this figure in color, go online.

555

Tyr residues are shielded from the aqueous environment in the presence of anionically charged membranes The quenching of tyrosine fluorescence by the hydrophilic quencher acrylamide and the lipophilic quencher 5-doxyl PC was measured with LUVs of increasing anionic lipid content and in E. coli inner-membrane-like LUVs. With increasing anionic charge, the quenching of acrylamide became less efficient (Fig. 3 A), whereas the quenching efficiency of 5-doxyl PC increased (Fig. 3 B). Thus, in highly anionic LUVs, Tyr residues are shielded from the aqueous environment but are accessible to the lipid quencher, i.e., MIR-WaaG binds to anionically charged LUVs. Note that the absorbance of acrylamide at the excitation wavelength is not negligible. Part of the intensity decrease with

FIGURE 3 (A and B) Fluorescence quenching of MIR-WaaG in buffer and zwitterionic, anionic, and E. coli inner-membrane-like vesicles by (A) acrylamide and (B) 5-doxyl PC. The ratios in the legend refer to the molar ratio of POPC/POPG in the vesicles. E. coli refers to a lipid mixture (60% POPE, 30% POPG, 10% TOCL) that mimics the composition of the inner membrane of E. coli. To see this figure in color, go online. Biophysical Journal 109(3) 552–563

556

increasing acrylamide concentration was thus due to acrylamide absorption, but this contribution was the same under all sample conditions. Thus, the slope differences of the Stern-Volmer plots in Fig. 3 A cannot be attributed to the absorbance of acrylamide. It is, however, not useful to compute or interpret quenching constants, since these depend on this contribution. Moreover, there are four Tyr residues located throughout the sequence (Y1, Y13, Y19, and Y22). Nevertheless, the results clearly demonstrate an overall trend that the association is charge dependent, supporting the CD data. The anisotropy of MIR-WaaG increases in the presence of anionically charged membranes To further study the lipid binding of MIR-WaaG, we measured the steady-state fluorescence anisotropy of the peptide in the presence of zwitterionic (100% POPC), anionic (10–40% POPG), and E. coli inner-membrane-like LUVs. The results are shown in Fig. 4. The anisotropy in buffer was found to be 0.06 5 0.01. From the Perrin equation, a rough estimate of the expected anisotropy is hri ¼ 0:05
0:07 for a monomeric MIR-WaaG assuming a rotational correlation time of 1.1 ns (41) and a typical lifetime for tyrosine of 3.5 ns (42). Within the range of uncertainty, zwitterionic and slightly anionic LUVs (10% POPG) gave rise to the same anisotropy for the peptide as in buffer, and these values were below the intrinsic anisotropy of immobilized tyrosines, which is in the range of 0.19–0.32 depending on the spatial proximity of the Tyr residues (43). Thus, it can be concluded that MIR-WaaG does not interact with zwitterionic or slightly anionic LUVs. A significant increase in anisotropy was observed as the anionic charge of the LUVs increased above 10%. As an increase in anisot-

FIGURE 4 Fluorescence anisotropy of MIR-WaaG in buffer and zwitterionic, anionic, and E. coli inner-membrane-like vesicles. The ratios in the different categories refer to the molar ratio of POPC/POPG in the vesicles. E. coli refers to a lipid mixture (60% POPE, 30% POPG, 10% TOCL) that mimics the composition of the inner membrane of E. coli. To see this figure in color, go online. Biophysical Journal 109(3) 552–563

Liebau et al.

ropy originates from a decrease in rotational correlation time of the fluorophore, we interpret this observation as indicating the binding of MIR-WaaG to anionic LUVs. At 40% POPG and in E. coli inner-membrane-like vesicles, the anisotropy was 0.11 5 0.01 and 0.10 5 0.01, respectively. These values are similar to the anisotropy that was determined for the 20-residue ShB peptide upon interaction with anionic LUVs (hri ¼ 0:12050:003) by Poveda and coworkers (44). Dissociation constant and immersion depth of MIR-WaaG in E. coli inner-membrane-like LUVs Fig. 5 shows the increase in fluorescence intensity upon binding of MIR-WaaG to E. coli membrane-mimicking LUVs. By employing a simple hyperbolic model assuming one binding site (Eq. 2), we obtained a dissociation constant of 35 5 6 mM. The distances of the fluorophores from the center of the bilayer at 16 mol % quencher concentration calculated with Eq. 1 are shown in Table 2. Quenching was most efficient for Tempo-PC (77% residual signal), followed by 5-doxyl PC (81%) and 10-doxyl PC (83%). The distance estimate is best for quencher pairs in the vicinity of the fluorophore and is worse if the quenchers are farther away (28), in particular if they are beyond the quenching radius Rc. ˚ . Therefore, we From our data, we find that Rc ¼ 7 A conclude that the distance estimate of the 5-doxyl/10-doxyl pair is unreliable and that the Tempo-PC/5-doxyl pair provides the best estimate. This means that the Tyr side chains ˚ from the bilayer center near are on average located 16 A the glycerol backbone of the lipids, i.e., at the surface of the bilayer.

FIGURE 5 Binding of MIR-WaaG to E. coli inner-membrane-mimicking LUVs monitored by tyrosine fluorescence. The fluorescence intensity F is normalized by the fluorescence intensity F0 of MIR-WaaG in buffer only. The curve represents the fit of the data to Eq. 2, from which KD ¼ 35 5 6 mM was obtained. To see this figure in color, go online.

Membrane Interaction of WaaG TABLE 2 bilayer

557

Distances of the Tyr residues from the center of the

Spin-Label Pair

˚) Distance (A

Tempo-PC/5-doxyl PC 5-doxyl PC/10-doxyl PC

16.3 5 0.2 10.4 5 0.4

MIR-WaaG interacts with DPC micelles and zwitterionic and anionic bicelles The translational diffusion coefficient of MIR-WaaG in buffer, with DPC micelles as well as with zwitterionic and anionic bicelles, was measured by PFG NMR. Table 3 shows the viscosity-corrected translational diffusion coefficients for MIR-WaaG and DPC micelles and bicelles. The diffusion coefficient for MIR-WaaG in buffer corresponds to what is expected for a 30-residue monomeric peptide that is free in solution (45). The diffusion coefficients for DPC micelles and both bicelle types are similar to those previously reported for the same or similar systems (25,46,47). In the presence of DPC micelles, as well as anionic and zwitterionic bicelles, the diffusion coefficients of MIRWaaG were drastically reduced to almost the same diffusion coefficients as observed for the membrane mimetic. By using a two-state model (free and bound peptide), one can estimate the fraction of bicelle-bound peptide (48). Using this relationship, we found that 98% of MIR-WaaG bound to both bicelle systems, which is within the error limits equal to fully bound. The same calculation for DPC binding is complicated by the fact that a fraction of DPC exists as monomers in solution. Since the diffusion of MIR-WaaG is slower than DPC diffusion, we can still conclude that also in this case the peptide is fully bound. We thus conclude that MIR-WaaG binds micelles and bicelles irrespective of the surface charge. Structure of MIR-WaaG To investigate the 3D structure of MIR-WaaG, we used standard NMR methods on a sample containing the peptide in 50 mM DPC-d38. CD spectra showed that the peptide in DPC adopted a structure very similar to that observed in LUVs mimicking the inner membrane of E. coli (Fig. S1 TABLE 3 Translational diffusion coefficients for MIR-WaaG and the membrane mimetics

a

Insertion depth of MIR-WaaG residues into DPC micelles The intensity reductions of 1Ha-13Ca crosspeaks in the presence of Gd(DTPA-BMA) relative to a spin-label-free reference sample report on the insertion depths of individual amino acids into the peptide. The intensity reductions of 1 H-13C DPC crosspeaks were used as a molecular ruler to measure the insertion depths of individual residues of MIR-WaaG (Fig. 7). Residual 1H-13C DPC crosspeak intensities in the presence of Gd(DTPA-BMA) are shown in Fig. 7 B. The peak intensity of Cg (see Fig. 7 C for the nomenclature of DPC carbons) was most strongly reduced, whereas residual intensities increased toward the center of the micelle. Gd(DTPA-BMA) is designed to exhibit weak electrostatic or hydrophobic interactions (51–53), and no specific interactions with micelles were observed here. TABLE 4

Structure Calculation for MIR-WaaG

Number of upper-limit constraints Number of torsion-angle constraints 1 H assignment completeness Maximum violation

293 16 98% ˚ 0.27 A

˚) Backbone Atom RMSD (A

Dt  10
11 m2 s
1

MIR-WaaG Membrane mimeticb

in the Supporting Material). The structure of MIR-WaaG was computed by simulated annealing using CYANA 2.1 with 293 upper-distance constraints and 16 torsion-angle constraints predicted from Ha, Ca, and Cb chemical shifts by TALOSþ. Table 4 summarizes data from the structure calculation, and Fig. 6 shows the structure ensemble of the 20 mostfavored conformers superimposed on the crystal structure of the corresponding segment of WaaG (9). MIR-WaaG consists of three a-helical segments arranged in a b-shape. The first nine N-terminal residues are rather ill defined, most likely due to the small number of constraints in this region. Still, an a-helical propensity for residues K4
E9 was determined by using the DSSP program (49,50). Via a two-residue linker, this helix connects to a short 310-helix at L12–R14. The peptide is kinked at T16, connecting to a third helix that stretches from S17 to F25. Remarkably, both the overall fold and the individual a-helices of the isolated MIR-WaaG sequence correspond well to the crystal structure (helix I: A2–E8; helix II: F11–L15; helix III: S17–T29).

Buffer

Zwitterionic Bicellesa

Anionic Bicellesa

DPC Micelles

First cycle, residues 4
27 Final cycle, residues 1
30 Final cycle, ordered residues

17 5 1 n.a.

6.5 5 0.2 6.3 5 0.1

6.2 5 0.2 6.0 5 0.3

8.5 5 0.2 9.9 5 0.1

Ramachandran Regions (%)

Diffusion coefficients for the bicelles were measured for DMPC/DMPG signals. These lipids fully integrate into bicelles. b The translational diffusion of the membrane mimetic indicates DPC micelles, zwitterionic, and anionic bicelles.

Most favored regions Additionally allowed regions Generously allowed regions Disallowed regions

2.6 1.3 5 0.3 0.8 5 0.3

71.1 28.3 0.6 0

Biophysical Journal 109(3) 552–563

558

Liebau et al.

FIGURE 6 NMR structure of MIR-WaaG. Ribbon representation of the 20 lowest-energy conformers of MIR-WaaG in DPC-d38 micelles (orange) aligned with the crystal structure of MIR-WaaG (blue).

Based on the DPC ruler, the insertion depth of individual MIR-WaaG residues was analyzed. Fig. 7 A shows the retained peak intensities of nonoverlapping and unambiguously assigned 1Ha-13Ca crosspeaks (except for T16 and T29, where 1Ha-13Cb are shown). Residual intensities for all observable residues in the region A2–A6 in the peptide were reduced by 50% or more compared with those obtained without Gd(DTPA-BMA), which means that they locate outside the micelle. In contrast, very little reduction in the intensity of crosspeaks belonging to Y13–T16 was observed, which indicates that this part resides inside the micelle. This trend is broken by S17 and R18, which again appear to be surface exposed, while the intermediate reduction of crosspeaks for Y19–A23 may indicate a location at the micelle interface. Like the N-terminus, the C-terminal residues A24–F30 had strongly reduced crosspeak intensities, suggesting a location outside the micelle. A 1D CLEANEX experiment with a 200 ms mixing time is shown in Fig. 8. The most prominent exchange peaks that appeared were due to guanidinium proton exchange in all arginine side chains. Due to overlap in the 1D spectrum, only a few unambiguous assignments of backbone amide protons could be made. Among these, E3, H21, and T29 showed exchange. Moreover, NOE peaks to water were observed for the aromatic ring protons. DISCUSSION The binding of monotopic GTs of the GT-B fold to membranes has previously attracted attention due to the peculiar ability of such enzymes to recognize both a water-soluble Biophysical Journal 109(3) 552–563

FIGURE 7 Localization studies of MIR-WaaG by paramagnetic relaxation. (A) Residual intensities of 1Ha-13Ca (1Ha-13Cb for T16 and T29) crosspeaks for residues in MIR-WaaG upon addition of Gd(DTPA-BMA). (B) Residual intensities of the 1H-13C crosspeaks of DPC upon addition of Gd(DTPA-BMA). The intensities in (B) serve as a ruler to determine the insertion depths of MIR-WaaG residues. (C) Structure of DPC and nomenclature of carbon atoms. To see this figure in color, go online.

donor and a lipophilic acceptor molecule. Despite the lack of characteristic motifs, it has been suggested that monotopic GTs adopting a GT-B fold possess characteristic structural features and properties that confer membrane interaction. In particular, clusters of basic residues, as well as surface-exposed tryptophans and hydrophobic amino acids in the N-terminal domain, could play a crucial role in permanently anchoring such GTs to membranes (19). The GT WaaG is involved in the core synthesis of lipopolysaccharides and is located at the cytosolic leaflet of the inner membrane of Gram-negative bacteria. Here, we analyzed the membrane interaction of a putative membrane-anchoring region of WaaG, termed MIR-WaaG, using a set of biophysical methods. MIR-WaaG interacts electrostatically with anionically charged membranes We prepared model systems that mimicked the characteristic anionic charge of the E. coli inner membrane conveyed by PG and CL, by enriching vesicles with POPG. In the presence of vesicles enriched in POPG, MIR-WaaG adopts a largely a-helical structure, similar to what is also observed

Membrane Interaction of WaaG

FIGURE 8 1D CLEANEX spectrum of MIR-WaaG in DPC micelles obtained with a mixing time of 200 ms. Peaks that could be assigned unambiguously are labeled.

in vesicles composed of an E. coli inner-membrane lipid mixture. Peptide-membrane interactions are also evident from an increase in fluorescence anisotropy as well as from an increase in quenching efficiency in the presence of a lipophilic quencher or, conversely, a decrease in quenching efficiency in the presence of a hydrophilic quencher. To quantify this interaction, we determined the dissociation constant and found that the affinity of MIRWaaG to E. coli inner-membrane LUVs is high (KD ¼ 35 mM) (44). This implies that under the conditions used in this study, all of MIR-WaaG is bound. Moreover, the KD is similar to what has been reported for peripheral membrane proteins (29) or membrane-associated peptides (44). Our data clearly suggest that binding is driven by anionically charged lipids, but we cannot directly determine whether specific (cationic) amino acid residues are involved in this interaction, or whether there is a preference for the peptide to interact with PG or CL. Thus, electrostatic interactions play a crucial role in binding MIR-WaaG to membranes, and mere hydrophobic interactions are not sufficient to accommodate MIR-WaaG in membranes or induce a native-like structure. Still, the localization of Tyr residues at the interface of vesicles and micelles, as observed by immersion-depth measurements in vesicles and from the structure of MIR-WaaG in DPC (see below), indicates that these residues are involved in binding even though we did not obtain any direct evidence for this here. The intrinsic anisotropy of free tyrosine is ~0.30 (43,54), but it can be drastically reduced by resonance energy transfer if Tyr residues are in close spatial proximity (41,43). Gryczynski and co-workers (43) reported a value of 0.19 for neighboring Tyr residues. Here, we find that the anisotropy of MIR-WaaG increases from 0.06 in buffer to 0.11

559

with anionic vesicles. This increase is significant, but the maximum anisotropy is still below the putative intrinsic anisotropy. Since the typical fluorescence lifetimes of tyrosines are ~3.5 ns (42), this means that the Tyr residues in MIR-WaaG undergo rotational motions on a nanosecond timescale even when the peptide is bound to LUVs. These motions are orders of magnitude faster than the value expected for vesicles of 60 nm radius, which is on the order of microseconds (55). At the same time, KD measurements indicate that all of the peptide associates with vesicles with a high POPG content. This means that the peptide undergoes considerable local motions even when it is bound to anionically charged membranes. From the NMR diffusion experiments, it can be concluded that MIR-WaaG associates fully with DPC micelles, as well as with zwitterionic and anionic bicelles. Within the limitations of the experimental setup, no free peptide is detectable. At concentrations above the critical micelle concentration, free DPC coexists with micelles and exchange between the two states is rapid (56). Therefore, only one set of NMR peaks is observed. Consequently, the overall diffusion coefficient of DPC is somewhat larger than the diffusion coefficient of MIR-WaaG bound to DPC micelles. The observation that MIR-WaaG binds to anionic bicelles is in agreement with the CD and fluorescence data for the peptide in the presence of the corresponding LUVs. Yet, MIR-WaaG was also observed to bind to zwitterionic bicelles, but no interaction with zwitterionic vesicles was observed. We speculate that this result may arise from the high edge curvature of the bicelles, as it was previously shown that some peptides interact with highly curved micelles (e.g., a membrane-binding helix of the GT AlMGS (57) and Ab (1–40) (58)). In fact, we may observe curvature-induced binding here, as DPC micelles give rise to the same secondary structure as E. coli membranemimicking LUVs (Fig. S1). Since we cannot distinguish between the contributions to binding due to curvature effects and those due to electrostatic effects, the results from our NMR diffusion experiments remain inconclusive. They underline, however, a general problem of working with bicelles, i.e., that interactions with the rim might cause artifacts in the results. Anchoring of WaaG to membranes The NMR solution structure of MIR-WaaG in the presence of DPC micelles is remarkably similar to that of the sequence in the crystal structure, indicating that this membrane mimetic confers a realistic structure induction in the peptide. In contrast, a quantitative analysis of the CD spectra in Fig. S1 revealed some b-sheet content. If unstructured peptide were present, the CD spectra would have a higher contribution from that state than what is observed in the NMR-derived structure (which favors a Biophysical Journal 109(3) 552–563

560

structured state). This is, however, in contradiction to an analysis of the secondary chemical shifts, which also captures the average over all states in the sample (Fig. S2), showing that the peptide is either a-helical (residues Y13–L15 and S17–Y22) or unstructured. Moreover, the lipid concentrations used in the CD measurements are high enough to ensure that all of the peptide is bound. Therefore, it is reasonable to speculate that the intensities in the CD spectra are erroneous, which precludes a quantitative analysis but still permits the qualitative conclusions presented above. Hence, our data show that in the presence of DPC as well as an E. coli lipid mixture, MIR-WaaG adopts to a large extent a helical structure. It can be seen in Fig. 7 A that parts of MIR-WaaG, specifically Y13–T16, are immersed in DPC micelles, whereas Y19–A23 locate in the interface region. Although the backbone of Y13 is located inside the micelle, the side chain extends toward its interface, in agreement with our findings from fluorescence studies. A24–F30 and A2–A6 locate outside the micelle. The amide protons of E3 and T29 exchange with water, confirming that the N- and C-termini are exposed to an aqueous environment. CLEANEX peaks indicate that all arginine side chains stretch out toward the solvent. Moreover, the aromatic protons of the Tyr residues show NOE peaks to water. On the other hand, Tyr residues are on average exposed to the lipophilic environment of anionic vesicles, as shown by fluorescence quenching experiments. These observations lead to the conclusion that the Tyr residues are located at the interface between wateraccessible and lipophilic environments. This conclusion is corroborated by immersion-depth measurements obtained using the parallax method. They show that the Tyr side ˚ from the chains are located at the bilayer interface, ~16 A bilayer center, when bound to LUVs. This means that the localization of the peptide in DPC micelles is similar to that in anionic LUVs, confirming that DPC micelles are a valid membrane mimetic for studying the immersion depth of MIR-WaaG in detail. Based on the results of this study, we propose that MIRWaaG anchors WaaG to membranes in a charge-dependent way. Although interactions involving hydrophobic residues and interface Tyr residues may contribute to the association, their contribution is not sufficient for binding. The localization of MIR-WaaG in DPC micelles consistent with the qualitative constraints obtained in this study is shown in Fig. 9 A. In this model, which is based on the combined results, residues in the two termini, A1–A6 and A24–F30, are outside of the micelle, G10–T16 are buried inside, and the remaining parts of the sequence are at or close to the interface. Based on this model, a mechanism for anchoring WaaG to a bilayer is proposed in Fig. 9 B. In this model, MIR-WaaG, which is located in the N-terminal domain of WaaG, anchors WaaG to the cytosolic leaflet of the inner membrane by the aforementioned mechanisms. Biophysical Journal 109(3) 552–563

Liebau et al.

FIGURE 9 (A) Model of the localization of MIR-WaaG in DPC micelles (70 DPC molecules). The lowest-energy conformer of MIR-WaaG is depicted in cartoon representation. Arg residues are in blue and Tyr residues are in green. Note that the results in this study do not provide any information about the micelle shape, size, or aggregation state. Nevertheless, the localization of MIR-WaaG in this micelle is consistent with our experimental results. (B) Model of the mechanism for anchoring WaaG to membranes. WaaG is depicted in blue and MIR-WaaG is in orange. The bilayer consists of 75% POPE, 20% POPG, and 5% TOCL. The micelle and bilayer were generated using CHARMM-GUI (59–63) and the figures were prepared in PyMOL. Atom coloring scheme: C, gray; P, red; O, orange; and N, blue.

Our results are in agreement with previous studies of monotopic GT-B GTs. It was previously shown that AlMGS, a monotopic GT-B enzyme that is involved in the synthesis of a glycolipid in A. laidlawii, recruits anionic lipids (21) and contains basic and hydrophobic residues in an amphipathic a-helix that associates with anionic membranes (57). Moreover, AlMGS tightly associates with (64) and penetrates into (65) one leaflet of anionic membranes. Hydrophobic interactions between Trp residues and membranes, as well as electrostatic interactions of basic residues and anionic lipid headgroups, play a decisive role in binding AtDGD2 (another monotopic GT-B enzyme, found in thylakoid membranes of plants) to membranes. It has been proposed that its

Membrane Interaction of WaaG

561

N-terminal domain anchors tightly to the membrane, whereas the C-terminal domain associates more loosely, e.g., it may perform a bending motion of the entire domain to accommodate the acceptor molecule (66,67). Despite this agreement, a subtle difference in the binding mechanism of WaaG compared with that of AtDGD2 is that MIR-WaaG does not contain any Trp residues. Moreover, an in silico search for membrane-interacting domains in WaaG conducted with MPEx (68) did not identify MIR-WaaG as a putative membrane-interacting region. We suspect that this is due to the lack of Trp residues in MIR-WaaG. The particular importance of Trp residues in binding to membranes is well known (69,70), and Trp residues are included in the model that Albesa-Jove´ et al. (19) and Ge et al. (66) suggested for the binding of AtDGD2 to membranes. Here, we find that Trp residues are not necessary for the membrane interaction of a monotopic GT. Electrostatic interactions conferred by basic amino acids, complemented by interactions of Tyr residues at the interface and hydrophobic residues inside the membrane, suffice for insertion of MIRWaaG into membranes.

REFERENCES

CONCLUSIONS

10. Parker, C. T., A. W. Kloser, ., B. W. Gibson. 1992. Role of the rfaG and rfaP genes in determining the lipopolysaccharide core structure and cell surface properties of Escherichia coli K-12. J. Bacteriol. 174:2525–2538.

It has been suggested that the membrane anchor of monotopic GT-B GTs is located in their N-terminal domain, and that binding is achieved by electrostatic interactions of positively charged clusters with anionic lipids and membrane interactions of Trp residues. Here, we have shown that in general this hypothesis holds true in the case of WaaG, but that Trp residues can be replaced by Tyr residues. We identified an exposed, positively charged, Tyr-enriched region of WaaG, the NMR structure of which is similar to the respective region in the crystal structure, and found that membrane anchoring of WaaG is conferred at least in part by the amino acids in this region. SUPPORTING MATERIAL Two figures are available at http://www.biophysj.org/biophysj/supplemental/ S0006-3495(15)00619-0.

1. Liu, J., and A. Mushegian. 2003. Three monophyletic superfamilies account for the majority of the known glycosyltransferases. Protein Sci. 12:1418–1431. 2. Breton, C., L. Snajdrova´, ., A. Imberty. 2006. Structures and mechanisms of glycosyltransferases. Glycobiology. 16:29R–37R. 3. Lizak, C., S. Gerber, ., K. P. Locher. 2011. X-ray structure of a bacterial oligosaccharyltransferase. Nature. 474:350–355. 4. Chung, B. C., J. Zhao, ., S. Y. Lee. 2013. Crystal structure of MraY, an essential membrane enzyme for bacterial cell wall synthesis. Science. 341:1012–1016. 5. Lairson, L. L., B. Henrissat, ., S. G. Withers. 2008. Glycosyltransferases: structures, functions, and mechanisms. Annu. Rev. Biochem. 77:521–555. 6. Coutinho, P. M., E. Deleury, ., B. Henrissat. 2003. An evolving hierarchical family classification for glycosyltransferases. J. Mol. Biol. 328:307–317. 7. Silipo, A., and A. Molinaro. 2010. The diversity of the core oligosaccharide in lipopolysaccharides. Subcell. Biochem. 53:69–99. 8. Campbell, J. A., G. J. Davies, ., B. Henrissat. 1997. A classification of nucleotide-diphospho-sugar glycosyltransferases based on amino acid sequence similarities. Biochem. J. 326:929–939. 9. Martinez-Fleites, C., M. Proctor, ., G. J. Davies. 2006. Insights into the synthesis of lipopolysaccharide and antibiotics through the structures of two retaining glycosyltransferases from family GT4. Chem. Biol. 13:1143–1152.

11. Yethon, J. A., E. Vinogradov, ., C. Whitfield. 2000. Mutation of the lipopolysaccharide core glycosyltransferase encoded by waaG destabilizes the outer membrane of Escherichia coli by interfering with core phosphorylation. J. Bacteriol. 182:5620–5623. 12. Landstro¨m, J., K. Persson, ., G. Widmalm. 2012. Small molecules containing hetero-bicyclic ring systems compete with UDP-Glc for binding to WaaG glycosyltransferase. Glycoconj. J. 29:491–502. 13. Dowhan, W., and M. Bogdanov. 2002. Functional roles of lipids in membranes. In Biochemistry of Lipids, Lipoproteins and Membranes, 4th ed.. D. E. Vance and J. E. Vance, editors. Elsevier Science B.V., Amsterdam. 14. Luckey, M. 2008. Membrane Structural Biology. Cambridge University Press, New York. 15. Cronan, J. E. 2003. Bacterial membrane lipids: where do we stand? Annu. Rev. Microbiol. 57:203–224. 16. Parsons, J. B., and C. O. Rock. 2013. Bacterial lipids: metabolism and membrane homeostasis. Prog. Lipid Res. 52:249–276. 17. Marr, A. G., and J. L. Ingraham. 1962. Effect of temperature on the composition of fatty acids in Escherichia coli. J. Bacteriol. 84:1260– 1267.

L.M. and J.L. designed the research and wrote the manuscript. J.L., P.P., and S.S. performed the research. J.L. and P.P. analyzed the data.

18. Wikstro¨m, M., A. A. Kelly, ., A. Wieslander. 2009. Lipid-engineered Escherichia coli membranes reveal critical lipid headgroup size for protein function. J. Biol. Chem. 284:954–965. 19. Albesa-Jove´, D., D. Giganti, ., M. E. Guerin. 2014. Structure-function relationships of membrane-associated GT-B glycosyltransferases. Glycobiology. 24:108–124.

ACKNOWLEDGMENTS

20. Ma¨ler, L. 2012. Solution NMR studies of peptide-lipid interactions in model membranes. Mol. Membr. Biol. 29:155–176.

AUTHOR CONTRIBUTIONS

We thank Minghui Zhang for assistance with fluorescence experiments, ˚ ke Wieslander for fruitful discussions in the early stages of this and A work. This work was supported by a grant from the Swedish Research Council (contract 621-2011-5964). The Knut and Alice Wallenberg Foundation provided support for the NMR spectrometers.

21. Ario¨z, C., W. Ye, ., L. Ma¨ler. 2013. Anionic lipid binding to the foreign protein MGS provides a tight coupling between phospholipid synthesis and protein overexpression in Escherichia coli. Biochemistry. 52:5533–5544. 22. Biversta˚hl, H., J. Lind, ., L. Ma¨ler. 2009. Biophysical studies of the membrane location of the voltage-gated sensors in the HsapBK and KvAP K(þ) channels. Biochim. Biophys. Acta. 1788:1976–1986. Biophysical Journal 109(3) 552–563

562 23. Ma¨ler, L., and A. Gra¨slund. 2009. Artificial membrane models for the study of macromolecular delivery. Methods Mol. Biol. 480:129–139. 24. Hope, M. J., M. B. Bally, ., P. R. Cullis. 1985. Production of large unilamellar vesicles by a rapid extrusion procedure: characterization of size distribution, trapped volume and ability to maintain a membrane potential. Biochim. Biophys. Acta. 812:55–65. 25. Ye, W., J. Liebau, and L. Ma¨ler. 2013. New membrane mimetics with galactolipids: lipid properties in fast-tumbling bicelles. J. Phys. Chem. B. 117:1044–1050. 26. Stern, O., and M. Volmer. 1919. On the quenching-time of fluorescence. Phys. Z. 20:183–188. 27. Chattopadhyay, A., and E. London. 1987. Parallax method for direct measurement of membrane penetration depth utilizing fluorescence quenching by spin-labeled phospholipids. Biochemistry. 26:39–45. 28. Abrams, F. S., and E. London. 1993. Extension of the parallax analysis of membrane penetration depth to the polar region of model membranes: use of fluorescence quenching by a spin-label attached to the phospholipid polar headgroup. Biochemistry. 32:10826–10831. 29. Chaudhuri, A., and A. Chattopadhyay. 2014. Lipid binding specificity of bovine a-lactalbumin: a multidimensional approach. Biochim. Biophys. Acta. 1838:2078–2086. 30. Stejskal, E., and J. Tanner. 1965. Spin diffusion measurements: spin echoes in the presence of a time-dependent field gradient. J. Chem. Phys. 42:288–292. 31. Von Meerwall, E., and M. Kamat. 1989. Effect of residual field gradients on pulsed-gradient NMR diffusion measurements. J. Magn. Reson. 83:309–323. 32. Callaghan, P. T., M. E. Komlosh, and M. Nyden. 1998. High magnetic field gradient PGSE NMR in the presence of a large polarizing field. J. Magn. Reson. 133:177–182.

Liebau et al. 44. Poveda, J. A., M. Prieto, ., C. R. Mateo. 2003. Intrinsic tyrosine fluorescence as a tool to study the interaction of the shaker B ‘‘ball’’ peptide with anionic membranes. Biochemistry. 42:7124–7132. 45. Danielsson, J., J. Jarvet, ., A. Gra¨slund. 2002. Translational diffusion measured by PFG-NMR on full length and fragments of the Alzheimer Ab (1–40) peptide. Determination of hydrodynamic radii of random coil peptides of varying length. Magn. Reson. Chem. 40:S89–S97. 46. Lauterwein, J., C. Bo¨sch, ., K. Wu¨thrich. 1979. Physicochemical studies of the protein-lipid interactions in melittin-containing micelles. Biochim. Biophys. Acta. 556:244–264. 47. Unnersta˚le, S., J. Lind, ., L. Ma¨ler. 2009. Solution structure of the HsapBK Kþ channel voltage-sensor paddle sequence. Biochemistry. 48:5813–5821. 48. Lindman, B., M. C. Puyal, ., P. Stilbs. 1984. Micelle formation of anionic and cationic surfactants from Fourier transform proton and lithium-7 nuclear magnetic resonance and tracer self-diffusion studies. J. Phys. Chem. 88:5048–5057. 49. Kabsch, W., and C. Sander. 1983. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers. 22:2577–2637. 50. Joosten, R. P., T. A. te Beek, ., G. Vriend. 2011. A series of PDB related databases for everyday needs. Nucleic Acids Res. 39:D411– D419. 51. Bernini, A., V. Venditti, ., N. Niccolai. 2009. Probing protein surface accessibility with solvent and paramagnetic molecules. Prog. Nucl. Magn. Reson. Spectrosc. 54:278–289. 52. Zangger, K., M. Respondek, ., T. Madl. 2009. Positioning of micellebound peptides by paramagnetic relaxation enhancements. J. Phys. Chem. B. 113:4400–4406. 53. Otting, G. 2010. Protein NMR using paramagnetic ions. Annu. Rev. Biophys. 39:387–405.

33. Damberg, P., J. Jarvet, and A. Gra¨slund. 2001. Accurate measurement of translational diffusion coefficients: a practical method to account for nonlinear gradients. J. Magn. Reson. 148:343–348.

54. Kierdaszuk, B. 2013. Fluorescence anisotropy of tyrosinate anion using one-, two- and three-photon excitation: tyrosinate anion fluorescence. J. Fluoresc. 23:339–347.

34. Cornilescu, G., F. Delaglio, and A. Bax. 1999. Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J. Biomol. NMR. 13:289–302.

55. Li, L., H. Szmacinski, and J. R. Lakowicz. 1997. Synthesis and luminescence spectral characterization of long-lifetime lipid metal-ligand probes. Anal. Biochem. 244:80–85.

35. Shen, Y., F. Delaglio, ., A. Bax. 2009. TALOSþ: a hybrid method for predicting protein backbone torsion angles from NMR chemical shifts. J. Biomol. NMR. 44:213–223.

56. Patist, A., J. R. Kanicky, ., D. O. Shah. 2002. Importance of micellar kinetics in relation to technological processes. J. Colloid Interface Sci. 245:1–15.

36. Wishart, D. S., and B. D. Sykes. 1994. Chemical shifts as a tool for structure determination. Methods Enzymol. 239:363–392.

57. Lind, J., T. Ra¨mo¨, ., A. Wieslander. 2007. High cationic charge and bilayer interface-binding helices in a regulatory lipid glycosyltransferase. Biochemistry. 46:5664–5677.

37. Bhattacharya, A., R. Tejero, and G. T. Montelione. 2007. Evaluating protein structures determined by structural genomics consortia. Proteins. 66:778–795. 38. Hwang, T., S. Mori, ., P. C. van Zijl. 1997. Application of phasemodulated CLEAN chemical EXchange spectroscopy (CLEANEXPM) to detect water-protein proton exchange and intermolecular NOEs. J. Am. Chem. Soc. 119:6203–6204. 39. Lee, Y., E. Lee, ., J. Lee. 2009. Measurement of hydrogen exchange times of the RNA imino protons using by phase-modulated CLEAN chemical exchange spectroscopy. Bull. Korean Chem. Soc. 30:2197– 2198. 40. Kelly, S. M., T. J. Jess, and N. C. Price. 2005. How to study proteins by circular dichroism. Biochim. Biophys. Acta. 1751:119–139. 41. Lakowicz, J. R. 2006. Principles of Fluorescence Spectroscopy. Springer, New York. 42. Ross, J. A., W. R. Laws, ., H. R. Wyssbrod. 2002. Tyrosine fluorescence and phosphorescence from proteins and polypeptides. In Topics in Fluorescence Spectroscopy, Vol. 3 J. R. Lakowicz, editor. Springer, New York, pp. 1–64. 43. Gryczynski, I., R. F. Steiner, and J. R. Lakowicz. 1991. Intensity and anisotropy decays of the tyrosine calmodulin proteolytic fragments, as studied by GHz frequency-domain fluorescence. Biophys. Chem. 39:69–78. Biophysical Journal 109(3) 552–563

58. Jarvet, J., J. Danielsson, ., A. Gra¨slund. 2007. Positioning of the Alzheimer Abeta(1-40) peptide in SDS micelles using NMR and paramagnetic probes. J. Biomol. NMR. 39:63–72. 59. Jo, S., T. Kim, and W. Im. 2007. Automated builder and database of protein/membrane complexes for molecular dynamics simulations. PLoS One. 2:e880. 60. Jo, S., T. Kim, ., W. Im. 2008. CHARMM-GUI: a web-based graphical user interface for CHARMM. J. Comput. Chem. 29:1859–1865. 61. Jo, S., J. B. Lim, ., W. Im. 2009. CHARMM-GUI Membrane Builder for mixed bilayers and its application to yeast membranes. Biophys. J. 97:50–58. 62. Brooks, B. R., C. L. Brooks, 3rd, ., M. Karplus. 2009. CHARMM: the biomolecular simulation program. J. Comput. Chem. 30:1545–1614. 63. Wu, E. L., X. Cheng, ., W. Im. 2014. CHARMM-GUI Membrane Builder toward realistic biological membrane simulations. J. Comput. Chem. 35:1997–2004. 64. Li, L., P. Storm, ., A. Wieslander. 2003. Irreversible binding and activity control of the 1,2-diacylglycerol 3-glucosyltransferase from Acholeplasma laidlawii at an anionic lipid bilayer surface. Biochemistry. 42:9677–9686. 65. Eriksson, H. M., P. Wessman, ., A. Wieslander. 2009. Massive formation of intracellular membrane vesicles in Escherichia coli by a

Membrane Interaction of WaaG monotopic membrane-bound lipid glycosyltransferase. J. Biol. Chem. 284:33904–33914. 66. Ge, C., A. Georgiev, ., A. A. Kelly. 2011. Tryptophan residues promote membrane association for a plant lipid glycosyltransferase involved in phosphate stress. J. Biol. Chem. 286:6669–6684. 67. Szpryngiel, S., C. Ge, ., L. Ma¨ler. 2011. Lipid interacting regions in phosphate stress glycosyltransferase atDGD2 from Arabidopsis thaliana. Biochemistry. 50:4451–4466.

563 68. Snider, C., S. Jayasinghe, ., S. H. White. 2009. MPEx: a tool for exploring membrane proteins. Protein Sci. 18:2624–2628. 69. White, S. H., and W. C. Wimley. 1999. Membrane protein folding and stability: physical principles. Annu. Rev. Biophys. Biomol. Struct. 28:319–365. 70. Killian, J. A., and G. von Heijne. 2000. How proteins adapt to a membrane-water interface. Trends Biochem. Sci. 25:429–434.

Biophysical Journal 109(3) 552–563