This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Zheng, S. Articles by Blasie, J. K. Search for Related Content PubMed PubMed Citation Articles by Zheng, S. Articles by Blasie, J. K.

Comparative Structural Studies of Vpu Peptides in Phospholipid Monolayers by X-Ray Scattering

Department of Chemistry, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6323; and ^* Department of Chemistry and Biochemistry, University of California, San Diego, California 92093-0307

Correspondence: Address reprint requests to J. Kent Blasie, University of Pennsylvania, 231 S. 34th St., Philadelphia, PA 19104-6323. Tel.: 215-898-6208; Fax: 215-898-6242; E-mail: jkblasie{at}sas.upenn.edu.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS DATA ANALYSIS RESULTS DISCUSSION CONCLUSIONS APPENDIX A APPENDIX B ACKNOWLEDGEMENTS REFERENCES

 
Vpu is an 81-residue HIV-1 accessory protein, its transmembrane and cytoplasmic domains each responsible for one of its two functions. Langmuir monolayers of phospholipid incorporating a membrane protein with a unidirectional vectorial orientation, on a semiinfinite aqueous subphase, provide one "membranelike" environment for the protein. The cytoplasmic domain's interaction with the surface of the phospholipid monolayer in determining the tertiary structure of the peptide within the monolayer was investigated, employing a comparative structural study of Vpu with its submolecular fragments Tm and TmCy truncated to different extents in the cytoplasmic domain, via synchrotron x-ray scattering utilizing a new method of analysis. Localizations of the transmembrane and cytoplasmic domains within the monolayer profile structure were similar for all three proteins, the hydrophobic transmembrane helix within the hydrocarbon chain region tilted with respect to the monolayer plane and the helices of the cytoplasmic domains lying on the surface of the headgroups parallel to the monolayer plane. The thickness of the hydrocarbon chain region, determined by the tilt of the hydrocarbon chains and transmembrane domain with respect to the monolayer plane, was slightly different for Tm, TmCy, and Vpu systematically with protein/lipid mole ratio. Localization of the helices in the cytoplasmic domains of the three proteins relative to the headgroups depends on their extents and amphipathicities. Thus, the interaction of the cytoplasmic domain of Vpu on the surface may affect the tilt of the transmembrane helix within the hydrocarbon chain region in determining its tertiary structure in the membrane.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS DATA ANALYSIS RESULTS DISCUSSION CONCLUSIONS APPENDIX A APPENDIX B ACKNOWLEDGEMENTS REFERENCES

 
The human immunodeficiency virus type 1 (HIV-1) genome encodes six accessory proteins (Emerman and Malim, 1998). Vpu, one of those accessory proteins, is an 81-amino acid phosphoprotein with an N-terminal extremely hydrophobic domain containing 27 residues, and a C-terminal hydrophilic domain containing 51 residues with a high number of charged polar amino acids. The hydrophobic domain is mostly a single 23-residue -helix (Schubert et al., 1996), whereas a minimum of 20 residues and a maximum of 32 residues in the cytoplasmic domain occur within a well-ordered -helical form comprising two to three separate amphipathic helices, depending on the experimental conditions (Wray et al., 1995; Federau et al., 1996; Willbold et al., 1997). Vpu_1–81 is phosphorylated by endogenous casein kinase-2 at residues Ser52 and Ser56 within its hydrophilic domain in HIV-1 infected cells (Schubert and Strebel, 1994).

Vpu has two different activities, namely the enhancement of the release of virus from the infected cell surface (Schubert et al., 1996) and the degradation of the CD4 molecule in the endoplasmic reticulum (ER) (Willey et al., 1992; Schubert and Strebel, 1994). Vpu's hydrophobic domain exhibits nonspecific cation channel activity (Ewart et al., 1996), presumably requiring an oligomeric form (Maldarelli et al., 1993). Since Vpu is synthesized from bicistronic mRNA and cotranslationally inserted into membrane of the ER, it is most desirable to determine Vpu's structure within a membranelike environment. Langmuir monolayers of phospholipid at the water/air interface incorporating Vpu protein represent one such model system. This system has the two distinct advantages: the water subphase is effectively infinite in extent beneath the phospholipid headgroups, and the protein is unidirectionally incorporated into the monolayer, thereby providing a good approximation to the surface of an intracellular membrane, especially for the protein's cytoplasmic domain. The only disadvantage is that the transmembrane domain of the protein is solvated by the hydrocarbon chains of a single phospholipid monolayer, instead of by those of a bilayer. However, unidirectional incorporation of the protein into phospholipid bilayer model systems, including unilamellar vesicles and multilayers, is generally difficult to achieve and the water spaces within even fully hydrated multilayers is limited (Marassi et al., 1999). Such unidirectional incorporation may be essential to structural studies of the oligomerization of the transmembrane domain necessary for its cation channel activity. Another advantage of this system is the ability to investigate systematically the structural nature of the interaction of Vpu with the other membrane proteins, or submolecular domains thereof, essential to its function in HIV infection (Schubert et al., 1996; Willey et al., 1992; Schubert and Strebel, 1994).

Previously, we have used this model system for investigating the contribution of a recombinant Vpu (Vpu_2–81) to the profile structure of the host phospholipid monolayer as a function of Vpu/lipid mole ratio. This study demonstrated that the transmembrane -helix is localized in the hydrocarbon chain layer of the host phospholipid monolayer and amphipathic -helices of cytoplasmic domain lie on the surface of the phospholipid headgroups in the water subphase at higher surface pressures (Zheng et al., 2001).

Although the recombinant Vpu's submolecular fragments Tm (Vpu_2–37) and TmCy (Vpu_2–51) have been truncated into shorter sequences, namely 36 residues for Tm and 50 residues for TmCy, respectively, all three proteins possess identical transmembrane domains. In phospholipid micellar and multilayer environments, the cytoplasmic domain of Vpu (Vpu_2–81) contains an amphipathic-helix-loop-amphipathic-helix secondary structure, whereas Tm (Vpu_2–37) contains only a very short portion of the first amphipathic helix and TmCy (Vpu_2–51) contains the entire first amphipathic helix of Vpu's cytoplasmic domain. Thus, a comparative study of Vpu with its submolecular fragments Tm and TmCy is an effective way to investigate the possible effects of the interactions of the cytoplasmic domain with the surface of the host phospholipid monolayer and the transmembrane domain in determining the tertiary structure of the peptide within the monolayer.

Here, we present an analysis of x-ray reflectivity data from mixed Langmuir monolayers of Vpu and its two submolecular fragments Tm and TmCy with a long-chain, diacylphosphatidylcholine DLgPC, performed independently for each protein over the same range of six different lipid/protein mole ratios, all at a constant, relatively high surface pressure of 45 mN/m. The gradients of the electron density profiles, and their analytic integration to the absolute electron density profiles, have been derived employing the model-independent Box Refinement method from the Fresnel normalized x-ray reflectivity data for each of the mixed protein/DLgPC monolayers. This approach was necessary to obtain objective experimental measures of the thicknesses and the essential features of both the hydrocarbon chain region and the polar headgroup region within the absolute electron density profile structures for each of the 18 mixed monolayers. Construction of simple slab models for the contributions of the protein's transmembrane domain, and the lipid hydrocarbon chains to the hydrocarbon chain region of the mixed monolayer electron density profiles as a function of protein/DLgPC mole ratio for the three proteins compared with their experimental counterparts, established the structural localization of the transmembrane domain within the hydrocarbon chains in mixed protein/DLgPC monolayers, including the estimated tilt of the hydrophobic helix relative to the monolayer surface normal. Structural parameters obtained from objectively fitting general Gaussian function models representing the lipid polar headgroups and the cytoplasmic domains of the three proteins within the headgroup region of the experimentally determined absolute electron density profiles as functions of the protein/DLgPC mole ratio established that the -helical portions of the cytoplasmic domains of Vpu and its submolecular fragments Tm and TmCy lie parallel to, but at significantly different distances from, the polar headgroups on the surface of the host phospholipid monolayer. Moreover, the extents of the helices within the cytoplasmic domains of the three proteins were found to be consistent with their otherwise known secondary structures. Overall, the structural similarities for the localization of the cytoplasmic and transmembrane domains of these three proteins within the phospholipid monolayer is consistent with both structural studies of these proteins in other model membrane systems via NMR spectroscopy and functional channel activity studies (Marassi et al., 1999; Ma et al., 2001). However, the quantitative modeling of the monolayer electron density profiles further demonstrates that the localization of the amphipathic helices of the protein's cytoplasmic domain with respect to the polar headgroups within the profile structure of the host phospholipid monolayer is dependent on their lengths and amphipathicities. These different localizations, arising from the differing interactions, are presumably responsible for the slightly different tilt behavior of the transmembrane helix with respect to the monolayer surface normal, systematic as a function of lipid/protein mole ratio. Finally, we present models describing the localization of Tm, TmCy, and Vpu within the phospholipid monolayer environment at the water/helium interface that are fully consistent with the electron density profiles derived from the x-ray reflectivity data from 18 different mixed monolayers, assuming no interrelationships among the data sets.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS DATA ANALYSIS RESULTS DISCUSSION CONCLUSIONS APPENDIX A APPENDIX B ACKNOWLEDGEMENTS REFERENCES

 
Purification of Vpu (Vpu_2–81) and its submolecular fragments Tm (Vpu_2–37) and TmCy (Vpu_2–51)
The detailed procedure for the cloning, expression, and purification of Vpu has been published in other papers (Marassi et al., 1999; Ma et al., 2001). Briefly, Escherichia coli strain BL21 (DE3) cells were transformed with the vectors carrying the Vpu gene and grown in minimal media. Nickel affinity chromatography (His•Bind Resin, Novagen) enabled the purification of the His-tagged fusion protein from other proteins in cell lysate. Cyanogen bromide was used to cleave Vpu from the fusion partner (Gross and Witkop, 1961). To facilitate this cleavage, the two Met residues in the Vpu sequence were mutated to Leu. Reverse-phase HPLC was subsequently used to purify Vpu. The purity and integrity of Vpu was confirmed by mass spectrometry. The biological activity of the double mutant protein was similar to that of authentic Vpu (Ma et al., 2001). The sequence of cleaved recombinant Vpu polypeptide is QPIQIAIVAL VVAIIIAIVV WSIVIIEYRK ILRQRKIDRL IDRLIERAED SGNESEGEIS ALVELGVELG HHAPWDVDDL and the sequences of Tm and TmCy are QPIQIAIVAL VVAIIIAIVV WSIVIIEYRK ILRQRK and QPIQIAIVAL VVAIIIAIVV WSIVIIEYRK ILRQRKIDRL IDRLIERAED, respectively.

Choice of phospholipid
The phospholipid used in these studies was 1,2-Dilignoceroyl-sn-Glycero-3-Phosphocholine (abbreviated herein as DLgPC, chromatographically pure, from Avanti Polar Lipids) which has C24-saturated hydrocarbon chains, the longest chain length commercially available. Since the length of Vpu's hydrophobic -helix is 34.5 Å containing 23 residues (Willbold et al., 1997), the use of DLgPC provides a nonpolar core for the host phospholipid monolayer whose maximal thickness (29.4 Å for untilted fully extended all trans chains) roughly matches the length of Vpu's hydrophobic -helix, although the N-terminus of the transmembrane helix would be only in a moist helium environment in the Langmuir monolayer instead of that provided by phospholipid headgroups and water in a bilayer. In addition, the glycero-phosphorylcholine headgroup is generally the predominant species in biological membranes (Yu et al., 1983). The fact that the diacyl hydrocarbon chains are saturated for DLgPC is mitigated for the higher protein/lipid mole ratios employed in this study, which preclude the formation of the gel phase, as first described in our earlier work (Zheng et al., 2001) and more extensively here. Each of the three proteins and the DLgPC lipid were dissolved in a 3:1 chloroform:methanol solution to obtain the desired lipid/protein mole ratio. Monolayers were prepared by spreading the chloroform/methanol solutions of either pure phospholipid or the mixtures of each of the three proteins with DLgPC onto a Millipore-filtered water subphase. The monolayers were kept at a constant temperature of 20°C during the x-ray reflectivity measurements.

Langmuir trough
A custom-built Langmuir trough, fabricated from a copper block and coated with Teflon, contained the water subphase and the spread monolayer. The temperature of the water subphase was controlled by cooled water circulation in the copper block and was measured with a resistance thermometer probe. Surface pressure was measured by a Wilhelmy plate and controlled by a movable barrier with feedback. Inasmuch as high quality x-ray reflectivity data can only be obtained from Langmuir monolayers when the aqueous subphase surface is relatively smooth, the Langmuir trough sat on a vibration isolation stage in the liquid surface spectrometer described below, a delay time of several seconds was employed between any motion of the spectrometer and data collection, and a flat, smooth silicon block was also submerged slightly below the water surface to dampen long wavelength excitations in the local height of the water surface. This resulted in reflectivity data collected (as described below) from a clean water surface which typically exhibited a minimal surface roughness of 3 Å; this was regularly ascertained for each experimental setup utilized over the course of this work. During the x-ray reflectivity measurements, moist helium gas was circulated inside the trough to replace the air, thereby reducing the x-ray background scattering.

Liquid-surface spectrometer
The x-ray reflectivity and grazing incidence diffraction (GID) experiments were performed mostly on beamline X-22B at the National Synchrotron Light Source at Brookhaven National Laboratory (Upton, New York), although some of the work was performed on entirely analogous instrumentation at the Complex Materials Consortium, Sector 09 at the Advanced Photon Source, Argonne National Laboratory (Argonne, Illinois). Details of the liquid surface spectrometer have been reported elsewhere (Als-Nielsen and Pershan, 1983; Braslau et al., 1988; Ocko et al., 1997). Here we give only a brief description. The synchrotron x-ray source was a bending magnet in the electron storage ring operating at an energy of 2.8 GeV and currents of 150–250 mA. Monochromatic x-rays were obtained via a horizontally reflecting Si (111) crystal monochromator to provide a wavelength = 1.546 Å. X-rays were reflected downward onto the horizontal liquid surface via a Ge (111) crystal to provide an angle of incidence, . Incident beam slits were set to collect the full horizontal beamwidth and vertically to limit the beam footprint on the liquid surface. A scintillation detector recorded the scattering from a thin Kapton film in the incident beam to provide an incident beam flux monitor. The specularly reflected beam from the liquid surface was measured at an angle ß with respect to the liquid surface with another scintillation detector for = ß in the vertical scattering plane at 2_xy = 0°. Scattered beam slits were set to accept the full specularly reflected beam. Off-specular background was measured at = ß with 2_xy = ±0.3°. The difference (specular minus off-specular background) provided the reflectivity R(q_z) for photon momentum transfer q_z perpendicular to the liquid surface with q_z = (4/)sin. We collected grazing incidence x-ray diffraction using a one-dimensional position-sensitive detector (Brookhaven National Laboratory) and evacuated Soller slits (JJ X-Ray, Denmark) that provided large vertical acceptance (q_z 0.8 Å^-1) and fine horizontal resolution (2 = 1.4 mrad, q_xy = 0.006 Å^-1). Both the total counts integrated over the length of the detector (i.e., q_z-integrated) and the counts as a function of channel number (17 channels/degree; i.e., q_z-resolved) were recorded. The direct beam full width at half maximum (FWHM) measured 0.14° horizontally and <0.12° (two channels) vertically. For GID, the incident angle was = 0.12° (i.e., 0.8 _c where _c is the critical angle for the water subphase), and the angle of the detector arm with respect to the surface was typically ß = 0° (accessing the q_z-range 0.0 Å^-1 q_z 0.8 Å^-1), although in some cases we scanned to larger q_z (0.4 Å^-1 q_z 1.2 Å^-1) by setting ß = 6°. To ensure that the monolayers were unaffected by radiation damage, we scanned regions of 2_xy four times with 5 seconds per 2_xy-value and averaged the four scans together after verifying that they were all similar to within counting statistics.

	DATA ANALYSIS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS DATA ANALYSIS RESULTS DISCUSSION CONCLUSIONS APPENDIX A APPENDIX B ACKNOWLEDGEMENTS REFERENCES

 
The normalized reflectivity R(q_z)/R_F(q_z) from a liquid surface arises from, in the first Born approximation, the modulus square of the Fourier transform of the gradient d(z)/dz of the electron density profile (z) across the air-water interface averaged over the in-plane coherence length of the incident x-rays (Als-Nielsen and Pershan, 1983; Helm et al., 1991), namely

(1)

where R(q_z) is the experimental reflectivity (normalized only by the incident beam flux), R_F(q_z) is Fresnel reflectivity from a single infinitely sharp (ideal) interface, the electron density of the semiinfinite bulk subphase is , and q_c is q_z at the critical angle for the subphase _c. This expression, Eq. 1, becomes progressively less valid as q_z approaches q_c, which is mitigated to a large extent by the use of q'_z, via the distorted-wave Born approximation (Zhou, 1995), where (q'_z)² = [(q_z)² - (q_c)²] (also Lösche et al., 1993).

Given the potential complexity of the mixed monolayers studied here, the complexity increasing for the proteins from Tm to TmCy to Vpu, the so-called slab model refinement method traditionally employed for the analysis of reflectivity data from more simple systems might be viewed as less than totally objective. This is because initial models for the electron density profile of the monolayer using this method must be constructed based on one's physical-chemical knowledge of the system of interest, the model then refined against the normalized reflectivity data via Eq. 1, and the method necessarily refines to the profile structure for the monolayer most similar to the initial model, as fully described in our earlier publication (Zheng et al., 2001). Thus, we have utilized the model-independent Box Refinement method, also described in our earlier publications (Strzalka et al., 2000, Zheng et al., 2001), to derive, with no a priori assumptions, the gradient of the monolayer electron density profiles from the experimental normalized reflectivity data via Eq. 1, independently, for each of the three proteins at each of the six different lipid/protein mole ratios investigated, namely pure DLgPC and DLgPC/protein of 50:1, 40:1, 20:1, 15:1, and 10:1, all at the same constant surface pressure of 45 mN/m and a temperature of 20°C. These gradients of the monolayer electron density profiles d(z)/dz were then integrated, both numerically and (with greater precision) analytically, to provide the electron density profiles for each of the monolayers. This model-independent approach with no a priori assumptions thereby provided for the subsequent objective extraction of the relevant structural parameters of these mixed monolayer systems of increasing structural complexity for the three proteins Tm to TmCy to Vpu.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS DATA ANALYSIS RESULTS DISCUSSION CONCLUSIONS APPENDIX A APPENDIX B ACKNOWLEDGEMENTS REFERENCES

 
Isotherm data
Monolayers of pure DLgPC and its mixtures with each of the proteins Tm (Vpu_2–37), TmCy (Vpu_2–51), and Vpu (Vpu_2–81) were spread from 3:1 chloroform-methanol solutions on a pure water subphase at 20°C. Surface pressure-area (-A) isotherms for pure DLgPC and the mixtures of Tm/DLgPC, TmCy/DLgPC, and Vpu/DLgPC are shown in Fig. 1, A, B, and C, respectively. The area of the mixed monolayers is described in terms of the area per average molecule, the average molecule here defined to include one DLgPC lipid molecule of area A_L and a fractional protein molecule of area A_P, the fractional contribution determined by N_P/N_L denoting the mole ratio of spread protein and DLgPC molecules:

(2)

The monolayer can be thus considered to consist of a known large number of the average [DLgPC + (N_P/N_L)protein] molecules. For the range of mole ratios employed in this study, this description of the monolayer area is a very close approximation to the more usual definition of the average molecule in terms of the mole fractions of each component, i.e., N_P/N_T and N_L/N_T for N_T = N_P + N_L.

(3)

Surface pressure isotherms of Tm/DLgPC mixtures (Fig. 1 A) are shifted systematically to larger mean areas per molecule with increasing Tm/DLgPC mole ratio with characteristic shape otherwise similar to that of the pure DLgPC. This systematic increase implies that the Tm molecules compete with the phospholipids for area at the water/air interface over the full range of Tm/DLgPC mole ratios investigated. This is not unexpected due to the highly hydrophobic nature of the Tm protein dominated by the transmembrane helix of Vpu. Compared with the Tm/DLgPC mixtures, surface-pressure area isotherms for both TmCy/DLgPC mixtures (Fig. 1 B) and Vpu/DLgPC mixtures (Fig. 1 C) exhibit an additional plateaulike feature for surface pressures in the range of 18 mN/m 25 mN/m which extends to progressively larger areas per average molecule with increasing protein content. One might expect that this additional plateaulike feature is due to the helices of the cytoplasmic domain competing with the DLgPC molecule for area at the water/air interface due to their amphipathic nature, instead of simply dissolving into water subphase below the phospholipid headgroups at these relatively lower surface pressures of <25 mN/m. Quantitative comparison of the areas per average molecule for the TmCy/DLgPC and Vpu/DLgPC mixtures at, for example, the lower surface pressures of 10 mN/m and 25 mN/m (see Table 1) with those for the Vpu/DLgPC mixture indicates that the magnitude of the area increase for Vpu/DLgPC is larger than that for TmCy/DLgPC. This further supports the interpretation that the cytoplasmic domains of proteins give rise to the plateaulike feature in the isotherms because the cytoplasmic domain of Vpu containing 51 residues is substantially larger than that of TmCy containing 21 residues. At higher surface pressures above 25 mN/m, the area per average molecule for TmCy/DLgPC mixtures and Vpu/DLgPC mixtures decreases sharply with the same characteristic shape as that of Tm/DLgPC mixtures. This suggests that a majority of cytoplasmic domains of TmCy/DLgPC and Vpu/DLgPC mixtures have been squeezed out of the water/air interface. However, quantitative comparison of the areas per average molecule at, for example, a higher surface pressure of 45 mN/m as listed in Table 1 shows that Vpu, with its relatively longer cytoplasmic domain, still occupies greater area per average molecule than does TmCy with its shorter cytoplasmic domain, whereas the area per average molecule for the latter TmCy is comparable to that of Tm. This result suggests that the longer cytoplasmic domain of Vpu is substantially more effective in competing with phospholipid for area at the water/air interface than the shorter cytoplasmic domain of TmCy.

View larger version (24K): [in this window] [in a new window]   FIGURE 1  Surface pressure area isotherms of pure DLgPC and mixed protein/DLgPC monolayers for the various mole ratios indicated on a pure water subphase at T = 20°C. (A) The isotherms for the Tm/DLgPC mixtures at various mole ratios. (B) The isotherms for the TmCy/DLgPC mixtures at various mole ratios. (C) The isotherms for the Vpu/DLgPC mixtures at various mole ratios. The abscissa is defined here in terms of the area-per-average-molecule according to Eq. 2, in which the average molecule includes one DLgPC molecule and a fractional protein molecule, the fraction defined by NP/NL denoting the mole ratio of spread protein and DLgPC molecules.

In addition, interactions between the protein and lipid components in the monolayer within these three different regions of its profile structure may render the areas per lipid A_L and protein A_P within each of these regions dependent on lipid/protein mole ratio, even when the pressure area isotherms for the pure lipid and protein components are known experimentally, as is the case here. As a result of these considerations, the x-ray reflectivity from the same Langmuir monolayer systems was systematically investigated.

Grazing incidence diffraction data
Fig. 2 A shows two-dimensional grazing incidence x-ray diffraction (GID) data as a function of (q_xy, q_z) from a DLgPC monolayer at 45 mN/m and 20°C. It is characterized by an intense maximum at q_xy = 1.429 Å^-1, q_z = 0 Å^-1 and a less intense maximum at q_xy = 1.383 Å^-1, q_z = 0.8 Å^-1, as defined in both q_z-integrated and q_z-resolved scans obtained with ß = 0° and ß = 6°, respectively, in Fig. 2, g–j. These features are indicative of a distorted hexagonal in-plane packing of all trans hydrocarbon chains tilted in the nearest neighbor direction 35°–36° with respect to the normal to the plane of the monolayer, as characteristic of a gel phase for a saturated diacylphospholipid. The former diffraction maximum centered at q_z = 0 Å^-1 appears as a sharp peak along q_xy somewhat broader than the q_xy-resolution, whereas the latter appears as a broader maximum along q_xy, in the q_z-integrated scans for ß = 0°. Both are characterized by Lorentzian line shapes characteristic of disordered chain-to-chain in-plane correlations decaying exponentially with distance, as opposed to a Gaussian line shape characteristic of well-ordered domains of finite in-plane size (Helm et al., 1991). Allowing for the experimental q_xy-resolution (via deconvolution of the incident beam line shape), the correlation lengths are 275 Å in the next nearest neighbor direction and only 33 Å in the nearest neighbor direction.

View larger version (49K): [in this window] [in a new window]   FIGURE 2  Grazing incidence x-ray diffraction data (GID) from monolayers of pure DLgPC and its binary mixtures with Vpu at = 45 mN/m and T = 20°C. (a) GID for pure DLgPC, shown here in two dimensions from qz-resolved scans in a gray-scale representation as a function of qxy, qz over the ranges indicated. (b–f) GID for DLgPC/Vpu mixtures at mole ratios 50:1, 40:1, 20:1, 15:1, and 10:1, shown here in two dimensions from qz-resolved scans in a gray-scale representation as a function of qxy, qz over the ranges indicated. (g) GID shown here as a qz-integrated (0.0 Å-1 qz 0.8 Å-1) scan at ß = 0° as a function of qxy over the range indicated. (h) qz-resolved GID data over the range 0.0 Å-1 qz 0.8 Å-1 centered on the maximum at qxy = 1.429 Å-1. (i) GID shown here as a qz-integrated (0.4 Å-1 qz 1.2 Å-1) scan at ß = 6° as a function of qxy over the range 0.4 Å-1 qz 1.2 Å-1. (j) qz-resolved GID data over the range 0.4 Å-1 qz 1.2 Å-1 centered on the maximum at qxy = 1.383 Å-1.

View larger version (31K): [in this window] [in a new window]   FIGURE 3  Grazing incidence x-ray diffraction (GID) from the monolayers of pure DLgPC and mixtures of Vpu/DLgPC for various mole ratios, all at a surface pressure of = 45 mN/m and T = 20°C, shown here as qz-integrated (0.0 Å-1 qz 0.8 Å-1) scans at ß = 0° as a function of qxy over the range indicated. Solid lines are the best fits of Lorentzian line shapes to the GID data characteristic of exponentially decaying in-plane interchain correlations.

View larger version (17K): [in this window] [in a new window]   FIGURE 4  Halfwidths (A) and amplitudes, expressed as signal/background ratios (B), extracted from the best fits of the Lorentzian line shapes to the GID data shown in Fig. 3 for various Vpu/DLgPC mole ratios. Note that the range of values for these two parameters shown on the respective ordinates in these plots does not include zero. Entirely similar results were also obtained for Tm/DLgPC mixed monolayers over the same range of mole ratios (not shown).

View larger version (33K): [in this window] [in a new window]   FIGURE 5  Fresnel normalized reflectivity R(qz)/RF(qz) data for pure DLgPC (open circles) and comparisons of such data for the three proteins Tm, TmCy, and Vpu within mixed protein/DLgPC monolayers as a function of photon momentum transfer qz at each of the various mole ratios investigated, all at a surface pressure of = 45 mN/m and T = 20°C. Data sets for different mole ratios have been offset for clarity and the counting statistics errors are smaller than the circles describing the data for pure DLgPC.

Functions derived directly from the normalized reflectivity data
The gradients of the monolayer electron density profiles d(z)/dz derived from the experimental normalized reflectivity data with no a priori assumptions via the model-independent Box Refinement method (Stroud and Agard, 1979; Makowski, 1981), exactly as described in detail in our earlier publication (Zheng et al., 2001) and therefore not repeated here, are shown on the right side (circles) of Fig. 6 for the case of Vpu at DLgPC/Vpu mole ratios of , 50:1, 40:1, 20:1, 15:1, and 10:1. These gradient profiles are fully consistent with the normalized reflectivity data from which they were derived, as shown in the left side of Fig. 6. The corresponding figures for the proteins Tm and TmCy are shown in Appendix A. (Although not shown here, it may be noted that the gradient profiles reported here are either unique, or in effect unique, solutions to the phase problem as obtained via Box Refinement. This is because all of phase space within the bounds of ± over the accessed range of momentum transfer was systematically explored. Only one other solution was occasionally found for these data sets—and only for the larger lipid/protein mole ratios—which was less consistent with normalized reflectivity data by a factor of two or more, using standard least-squares residuals criteria, and therefore rejected.

View larger version (30K): [in this window] [in a new window]   FIGURE 6  Gradients of the monolayer electron density profiles d(z)/dz derived directly from the experimental normalized reflectivity data via the model-independent Box Refinement method, exactly as described in detail in our earlier publication (Zheng et al., 2001), are shown on the right side (circles) for the case of Vpu at DLgPC/Vpu mole ratios of , 50:1, 40:1, 20:1, 15:1, and 10:1. The hydrocarbon/helium interface is defined here as the z = 0 Å origin, which is of no other consequence in these studies. These gradient profiles are fully consistent with the normalized reflectivity data from which they were derived, as shown in the left side, where the experimental R(qz)/RF(qz) data are shown as circles and the |F(q'z)|2 calculated via Eq. 1 as solid lines. The best nonlinear least-squares fits of the sum of four Gaussian functions to the gradients of the monolayer electron density profiles d(z)/dz from Box Refinement, the minimum number of Gaussians required in the sum based on matching criteria (a) and (b) described in the text and Fig. 7, are shown as the solid lines on the right side. The corresponding figures for the proteins Tm and TmCy are shown in Appendix A.

View larger version (31K): [in this window] [in a new window]   FIGURE 7  (Left side, criteria a) Experimental R(qz)/RF(qz) data are shown as circles for the case of Vpu at DLgPC/Vpu mole ratios of , 50:1, 40:1, 20:1, 15:1, and 10:1.and the |F(q'z)|2, calculated via Eq. 1 for the best nonlinear least-squares fits of the sum of four Gaussian functions to the gradients of the monolayer electron density profiles d(z)/dz from Box Refinement, are shown as the solid lines. (Right side, criteria b) The top half shows the absolute electron density (z) profiles calculated by analytic integration of the best nonlinear least-squares fits of the sum of four Gaussian functions to the gradients of the monolayer electron density profiles d(z)/dz from Box Refinement for the case of Vpu at DLgPC/Vpu mole ratios of , 50:1, 40:1, 20:1, 15:1, and 10:1. The bottom half shows the absolute electron density (z) profiles calculated by numerical integration of the gradients of the monolayer electron density profiles d(z)/dz from Box Refinement for this case. The corresponding figures for the proteins Tm and TmCy are shown in Appendix B.

d

d

d

We then fit a sum of Gaussian functions to the so-determined gradient profiles d(z)/dz using an objective nonlinear least-squares fitting procedure employing a chi-squared criteria for the goodness of fit. (We have utilized both the method of steepest descents and the Levenberg-Marquardt algorithm to effect the nonlinear least squares fitting, as provided for example in Mathematica. As is generally the case, given the large number of parameters required to describe even a small number of interfaces at three per interface, we found it necessary to perform a so-called grid search for some of the parameters and floated the remainder to best minimize the goodness of fit when fitting the sum of only four Gaussians to the gradient profiles. Whereas four Gaussians were just barely sufficient to provide acceptable fits, as best seen via criteria (a) where the counting statistics errors in the normalized reflectivity data are clearly evident, five Gaussians can provide essentially perfect fits over the entire range of momentum transfer accessed. However, since the nonlinear least squares fitting algorithms employed cannot converge treating the resulting fifteen parameters independently, three to four Gaussians were fit to different overlapping regions within each gradient profile. The resulting best fits were then combined, which required small adjustments in the parameters of the Gaussians within the overlapped portions, to produce the best five-Gaussian fits over the full range of the gradient profile. These small adjustments may render the combining process somewhat less objective and these results are therefore not shown here.) A minimum number of such Gaussian functions was sought to represent the gradient profile d(z)/dz sufficient to make a), the modulus square of its Fourier transform match the experimental normalized reflectivity data via Eq. 1; and b), the analytic integration of the sum of Gaussian functions fit to the gradient profile d(z)/dz match the absolute electron density profile (z) obtained by numerical integration of the gradient profile d(z)/dz. For all 18 different mixed monolayers studied, it was thereby determined that four interfaces described by 12 parameters (three each) were the minimum necessary, provided that two layers (or slabs) described the hydrocarbon chain region and one layer (or slab) described the polar headgroup region of the monolayer electron density profile, with reference to the profile structure of the host DLgPC monolayer. Employing two layers in the headgroup region and only one layer in the hydrocarbon chain provided substantially poorer matches via both criteria a) and b), above.

The fitting of a sum of four Gaussian functions to the gradient profiles d(z)/dz for the Vpu case at six DLgPC/Vpu mole ratios is shown in the right side (solid lines) of Fig. 6. The matching criteria a) and b) for the Vpu case at the six DLgPC/Vpu mole ratios are shown in Fig. 7. The corresponding figures for the proteins Tm and TmCy are shown in Appendix B. Table 2 contains the best fit parameters for all three proteins at the six different lipid/protein ratios. Finally, a comparison of the absolute electron density profiles, as obtained by analytic integration as described above, for the three proteins of increasing structural complexity Tm, TmCy, and Vpu, each at the six lipid/protein mole ratios studied, are shown in Fig. 8.

View larger version (18K): [in this window] [in a new window]   FIGURE 8  Comparison of the absolute electron density profiles (z) calculated by analytic integration of the best nonlinear least-squares fits of the sum of four Gaussian functions to the gradients of the monolayer electron density profiles d(z)/dz from Box Refinement for (A) mixed Tm/DLgPC monolayers, (B) mixed TmCy/DLgPC monolayers, and (C) mixed Vpu/DLgPC monolayers, all as a function of the same six increasing protein/DLgPC mole ratios at a surface pressure of 45 mN/m and T = 20°C. The profiles are rigorously aligned via their z = 0 Å origin, chosen here as the hydrocarbon/helium interface (as identified in Fig. 6). The range of average electron densities calculated for the hydrocarbon chain region of these absolute electron density profiles for the range of mole ratios investigated, based on the modeling of that region, is indicated for the Vpu case by the two horizontal lines (at 0.328 e/Å3 and 0.352 e/Å3) and is identical for the TmCy and Tm cases.

d

d

d

View larger version (26K): [in this window] [in a new window]   FIGURE 9  Two measures of the thicknesses of the hydrocarbon core region of the monolayer electron density profiles, as determined objectively from the gradients of the monolayer electron density profiles d(z)/dz from Box Refinement, for Tm, TmCy, and Vpu as a function of increasing mole fraction of protein in the mixed monolayers. (Top) Thicknesses as determined by fitting a sum of four Gaussian functions to the gradients of the monolayer electron density profiles d(z)/dz, namely L2 + L3 from Table 2. The estimated errors in these thicknesses (see legend for Table 2) are only ±0.1 Å. (Bottom) Thicknesses as determined via the separation of the positions of the minima representing the hydrocarbon/helium and headgroup/hydrocarbon interfaces in the gradients of the monolayer electron density profiles d(z)/dz. These positions (see Fig. 6) were determined by the zeros of the derivative of the gradient profiles. The estimated errors in these thicknesses are again only ±0.1 Å.

Since the electron density and thickness obtained from the electron density profiles in the hydrocarbon chain region contain contributions from both the lipid hydrocarbon chains and the transmembrane helix and because the profiles are derived from x-ray reflectivity data sensitive to the monolayer composition averaged over an area whose lateral extent is defined by the coherence length 10⁴ Å of the synchrotron x-rays, the average electron density of that particular region within the monolayer profile structure at a particular protein/DLgPC mole ratio is given by

(4)

Similarly, the average thickness of that particular region within the monolayer profile structure at a particular protein/DLgPC mole ratio is given by

(5)

where N_L/N_P is the lipid/protein mole ratio, _HC is the electron density and t_HC is the profile thickness of the hydrocarbon chains of DLgPC, _TH is the electron density and t_TH is the profile thickness of Vpu's transmembrane helix, A_L is the area per DLgPC molecule and A_TH is the area per transmembrane domain. ("Profile thickness" here refers to the thickness of the hydrocarbon chains or transmembrane helix, respectively, as projected onto the profile axis normal to the plane of the monolayer.) The minimum area/molecule of the pure DLgPC at 45 mN/m is 45 Ų, based on the pressure/area isotherm and the GID data, and the average electron density of all trans hydrocarbon chains at 22 Ų/chain is 0.320 e/ų. The minimum cross-sectional area of an -helix 10 Å in diameter is 80 Ų and the average electron density of the hydrophobic transmembrane helix common to the three proteins is 0.490 e/ų. (Pressure area isotherms for the pure proteins have been determined and x-ray reflectivity data has been collected and similarly analyzed for Tm as a function of surface pressure. At higher surface pressures of 45 mN/m, the area/molecule approaches 200 Ų whereas the thickness of the monolayer profile approaches 35 Å consistent with the 23-residue length of its hydrophobic helix at 1.5 Å/residue. Nevertheless, it is unreasonable to expect the average area/molecule calculated from the isotherm to approach the value of 100 Ų as calculated for the minimum cross-sectional area of the helix (e.g., for helices 10 Å in diameter hexagonally close-packed in the plane perpendicular to their long axis, the average area per helix becomes 90 Ų, because of both the nonzero temperature, 293 K, and that the helices would have to exhibit in-plane hexagonal close-packing with long-range order over the entire area of the spread monolayer.) Assuming the most simple situation of no interactions between these two components within the hydrocarbon chain region, Eq. 4 was used to estimate the average electron density of the hydrocarbon core region with increasing mole fraction of the transmembrane helix. The range of these values (0.328 e/ų 0.352 e/ų) for mole ratios 1:50 up to 1:10, respectively, is shown in Fig. 8 and the average electron densities of the hydrocarbon chain regions of these profiles for the three proteins with increasing mole fraction of protein are thereby seen to be physically reasonable. In addition, Eq. 5, together with the 23-residue length of the hydrophobic transmembrane helix, was used to estimate the average thickness of the hydrocarbon core region, assuming a tilt of the helices and hydrocarbon chains relative to the normal to the monolayer plane sufficient to match the experimentally determined values L2 + L3 from Table 2. The resulting estimated tilt angles shown in Table 3 and their dependence on the particular protein and the lipid/protein mole ratio thus necessarily follow the same trends as exhibited by the thickness of the hydrocarbon core region (Fig. 8), the thicknesses and tilt angles being inversely related.

Therefore, a model for the electron density in excess of that of pure water within the polar headgroup region of the absolute electron density profiles for the mixed monolayers was constructed based on two Gaussian functions representing the electron density profiles of the lipid polar headgroups and the respective cytoplasmic domains of the proteins. This anticipates some amount of water (see below) within both the headgroup layer and the cytoplasmic domain layer, given by

(6)

where _h-w is excess electron density of the lipid headgroup layer including water, _c-w is the excess electron density of cytoplasmic domain layer including water, µ_h-w is the center of the distribution of electron density of the headgroups, µ_c-w is center of the distribution of electron density of the cytoplasmic domain's helices, _h-w is the halfwidth of distribution of the headgroups, and the _c-w is the halfwidth of distribution of the cytoplasmic domain's helices. Note that these Gaussian functions, and thus the model, are completely general in that no restrictions were placed on their amplitudes, widths, or positions.

A nonlinear least-squares procedure was then used to fit Eq. 6 to the polar headgroup region within the absolute electron density profiles derived from the normalized reflectivity from the various monolayers. The goodness of fit was tested using the standard R-factor, or integral of the residuals defined as

The R-factors attained were in the range of 0.02 0.06 in all cases demonstrating that two Gaussian functions in the model were sufficient. The results are shown in Fig. 10 for Vpu/DLgPC mixed monolayers, for example, and all results are listed in Table 3. From Table 3, it can readily be seen that the widths of the Gaussian representing the phospholipid headgroups (identified as that closest to the hydrocarbon chain region and decreasing in amplitude with increasing mole fraction of protein in the mixed monolayer) are generally in the range of 4–5 Å, consistent with that for a diacylphosphatidylcholine at 10 Å resolution. More importantly and also readily seen from Table 3 is that the widths of the Gaussian representing the cytoplasmic domain (identified as that farthest from the hydrocarbon chain region and increasing in amplitude with increasing mole fraction of protein in the mixed monolayer) are relatively constant with _c-w 4 Å, except for a significant increase for only the Vpu case at the higher two protein/lipid mole ratios. This width is fully consistent only with the profile of an -helix lying parallel to the monolayer plane (and certainly not that of a helix lying perpendicular to this plane, especially given the anticipated lengths of the helices in the cytoplasmic domains of TmCy and Vpu), both as calculated from the known structure of -helices and as experimentally determined via analogous x-ray reflectivity studies for Langmuir monolayers of both synthetic amphipathic helices (Strzalka et al., 2000) and the amphipathic helices of Vpu's cytoplasmic domain (Zheng et al., 2001) at relatively low surface pressures. Furthermore, the separation between the centers of the distributions of the cytoplasmic domain's helices and the lipid polar headgroups t_CH-HG depends more strongly on the particular Vpu protein, but otherwise not so strongly on the lipid/protein mole ratio, with the exception of only the Vpu case at the higher two protein/lipid mole ratios.

View larger version (27K): [in this window] [in a new window]   FIGURE 10  Best nonlinear least-squares fits of two Gaussian functions, representing the protein's cytoplasmic domain and the phospholipid headgroups, to the excess electron density within the asymmetric polar headgroup region of the absolute electron density profiles (z) obtained via analytic integration as described in the text, for the mixed Vpu/DLgPC monolayers as a function of increasing Vpu/DLgPC mole ratio at a surface pressure of 45 mN/m and T = 20°C.

(7)

where the A_C is the area occupied by the cytoplasmic domain, the A_w is the area occupied by water molecules