Faraday Discuss. Chem. SOC., 1986,81, 117-126 A Study of Phospholipid Phosphate Groups in Model Membranes by Fourier Transform Infrared Spectroscopy Felix M. Goii* and Jose L. R. Arrondo Departamento de Bioqur'mica, Facultad de Ciencias, Universidad del Par's Vasco, Aptdo. 644, 48080 Bilbao, Spain The phosphate region (1000-1300 cm-') of the infrared spectrum of aqueous phospholipid dispersions has been studied by the infrared Fourier transform technique. The main features of this spectral region have been described for various phospholipids, including phosphatidylcholine, phosphatidyl- ethanolamine, cardiolipin and phosphatidylglycerophosphate (alkyl ether), the major phospholipid of Halobacterium purple membranes. No changes in the phosphate region are observed due to lipid polymorphism or as a consequence of changes in fatty acyl chain structure. Shifts in the phosphoryl stretching bands are interpreted in terms of changes in hydrogen bonding, while a contribution from R-0-P vibration is considered to reflect changes in phospholipid headgroup conformation.When added in equimolar amounts to phosphatidylcholine or phosphatidylethanolamine bilayers, sur- factants (Triton X-100, sodium cholate) modify the degree of hydration and/or the orientation of the headgroup with respect to the bilayer plane. Phospholipids containing two phosphate groups give rise to more complex spectral features. Infrared data suggest that the two phosphate groups of cardiolipin are conformationally non-identical when incorporated into a lipid bilayer.Controlled enzyme hydrolysis (e.g. with phospholipase D) may help in the study of these complex phospholipid headgroups. The study of model membranes consisting of pure phospholipid bilayers has been the object of much effort in the last two decades. In particular, the application of spectro- scopic techniques to such model systems has shed light on a number of physical properties of phospholipids in aqueous dispersions which, in turn, give us some insight into the structure and behaviour of cell membranes.' 1.r. spectroscopy was one of the first physical techniques to be applied to the study of phospholipids;2 however, the strong absorption of water in the i.r. region prevented further applications of this technique to aqueous membrane dispersions until the advent of computerized spectroscopy, when spectral subtraction of water was made possible.At that stage, both interfer~metric~ and monochromator4 i.r. spectrometers were applied to model and biomembrane studies. Recent reviews576 have summarized the salient features of these investigations. A compulsory initial step in the i.r. investigation of biomembranes consists of the unequivocal assignment of spectral absorption bands and absorption maxima from the main membrane components. However, this objective is still far from being accom- plished. Methylene stretching vibration bands have been extensively characterized, partly because of their sharp appearance, partly because of their sensitivity to ther- motropic phase transitions, the most widely studied of all phospholipid properties.Other spectral regions, however, have not been explored in so much detail. One such region corresponds to the antisymmetric and symmetric stretching vibrations of the PO 2 bond, between 1000 and 1300 cm-'. 117118 I. R. Studies of Phospholipids Phosphate stretching in oriented multilayers was studied by Fringeli and GU~~thard;~ some phosphate stretching frequencies in natural and model biomembranes have also been described.*-’’ In a recent study,12 we investigated the 1000-1300 cm-’ region of the i.r. spectrum of aqueous dipalmitoylphosphatidylcholine (DPPC) and other phos- phate-containing molecules (glycerophosphorylcholine, phosphorylcholine, L-a- glycerophosphate etc.) by the Fourier transform technique. Buffered DPPC displays two maxima, at 1086 and 1222 cm-’, corresponding, respectively, to symmetric and asymmetric PO, stretching; these values are the same above and below the gel-to-liquid crystalline T, transition temperature of the phospholipid.The present paper summarizes the main results from a series of similar studies carried out with other naturally occurring phospholipids, i.e. phosphatidylethanolamine, cardiolipin and phosphatidylgly- cerophosphate (alkyl ether), the major phospholipid of Halobacterium purple membranes. Experiment a1 1,2-Dipalmitoyl-sn-glycero-3 -phosphocholine (DPPC) and 1,2-dimyristoyl-sn-g1ycero- 3-phosphoethanolamine (DMPE) were purchased from Fluka and their purity was checked by thin-layer chromatography and differential scanning calorimetry. Cardiolipin was from Sigma, egg-yolk phosphatidylcholine (EPC) was purified according to Single- ton et d;’’ bacterial phosphatidylethanolamine (BPE), from E.coZi, was type V from Sigma. The purity of lipids from natural origin was checked by thin-layer chromatogra- phy. Purple membrane from Halobacterium halobium was prepared as described by Muga14 and their major phospholipid (I) purified, in the form of sodium salt, according to Kates et aZ.” The buffer used throughout this work was 10 mmol dm-3 Hepes, pH 7.0. Phospholipid suspensions (30 mg cm-’) were prepared, unless otherwise stated, accord- ing to Cortijo et ~ 1 . ’ ~ When required, purple membrane lipids, in the form of sonicated suspensions, were treated with phospholipase D (type 111, from Sigma) in order to obtain a phospholipid (11) containing a single phosphate group.The resulting lipid was 0 R-O-CH2 II R-0-CH 0 I I I I 11 I I II CH2-O-P-O-CH, I I 0- 0- R-0-CH2 CH20- P-0- I CH2-0-P-0- R-0-CH 0 OHCH 0- R = phytyl chain R = phytyl chain (1) (11) purified by preparative thin-layer chromatography (A. Muga, to be published), and resuspended in Hepes buffer as above. Phospholipid-surfactant mixtures were prepared in organic solvents then freeze-dried overnight. The solvent-free mixture was later resuspended in water to a final phospholipid concentration of 30mgcmP3, at a tem- perature well above T, of the pure lipid with vortexing. The samples were introduced into a thermostatted microcell (Beckman FH-O1C FT) with CaF2 windows and a pathlength of 7 pm. A 10-MX Nicolet f.t.i.r. spectrometer was used; 512 spectra were routinely accumulated and averaged with a standard resolution better than 2 cm-’.Spectral subtraction of pure buffer was performed in order to remove the scattering background.F. M. Go% and J. L. R. Arrondo 119 wavenumber/cm-' Fig. 1. F.t.i.r. absorbance spectra of aqueous dispersions of ( a ) EPC and (b) BPE, in the region between 1000 and 1300 cm-', at 19 "C. Results and Discussion Studies with Phosphatidylcholine and Phosphatidylethanolamine Various preparations of phosphatidylcholine and phosphatidylethanolamine of natural and synthetic origin have been examined in order to determine the effect of phosphate group substituents and fatty acyl chains on the phospholipid phosphate vibrations. The 1000-1300 cm-' region of the i.r.spectrum of aqueous BPE and EPC at 19 "C is shown in fig. 1. EPC shows two main absorption bands, with maxima at 1086 and 1221 cm-', and a shoulder at 1060 cm-'; these are virtually the same values found previously for DPPC,12 that were assigned, respectively, to asymmetric and symmetric PO, stretching, and to the R-0-P-0-R' group. The spectrum of BPE looks very similar, except that the symmetric band is shifted towards lower frequencies and the R-0-P-0-R' signal now appears as a well resolved band, the three maxima being located at 1026, 1076 and 1220 cm-'. Fookson and Wallach17 examined dry films of DPPC and DPPE and found a significant shift (32 cm-') between the asymmetric PO, stretching bands of both substances, DPPE being shifted to lower frequencies, but no difference is found in our case, when fully hydrated samples are observed.The shift was attributed to intermolecular hydrogen bonding of PE. It is possible that, in aqueous suspension, both PC and PE bind extensively to water through hydrogen bonds. Mendelsohn and Mantsch'* mention the preseace of a symmetric ester C-0 stretching vibration band centred at 1070k3 cm I , while the phosphate ester stretch (C-0-P) would be located at 1047* 1 cm-' for most phospholipids. In our previous paper,12 we demonstrated the contribution of the C-0-P stretching vibration to the 1060 cm-' signal in fully hydrated DPPC bilayers (1069 cm-' for anhydrous DPPC). A contribution from the C-0 stretching vibration to the same signal cannot be ruled out, h~wever.~ Mantsch et al.19 have been able to characterize both a gel-to-liquid crystalline and a liquid-to-inverted hexagonal phase transition of egg-yolk phosphatidylethanolamines by i.r.spectroscopy through changes in CH2 stretching vibrations. Fig. 2 shows the120 I. R. Studies of Phospholipids 0.59E 0 .LL 7 0 s 2 0.298 si -2 0.1.49 0.000 1300 1200 1100 1000 wavenumber/cm-' Fig. 2. F.t.i.r. absorbance spectra of aqueous dispersions of BPE at ( a ) 12, ( b ) 19 and (c) 50 "C. phosphate region of the i.r. spectrum of aqueous BPE at 10, 19 and 50°C. These temperatures correspond, respectively, to phosphatidylethanolamine in the gel phase, the liquid-crystalline phase and the inverted hexagonal phase. (Phase transitions were monitored through changes in the symmetric CH2 stretching vibration of BPE; data not shown.) No variation in the maximum wavenumber of any of the bands is observed as a result of the phase transitions.The gel-to-liquid crystalline transition of DPPC also failed to produce any change in the i.r. phosphate spectrum.12 The influence of the nature of fatty acyl chains on the PO, stretching vibration bands has also been explored. We have seen that the phosphate region of spectra from EPC and DPPC are virtually undistinguishable (fig. 1). The same is true of dimyristoylphos- phatidylcholine in aqueous dispersion. Similar results are obtained when the spectra of BPE and DMPE are compared in the temperature range between 10 and 60°C; in no case did changes in fatty acyl chains modify the phosphate vibration bands of the i.r. spectrum of aqueous phospholipid (spectra not shown).The Interaction of Surfactants with Phospholipid Headgroups Surfactants are commonly used for membrane solubilization in the process of purification and reconstitution of integral membrane proteins; they also find an application as 'activators' of lipase enzyme activities. The details of surfactant-phospholipid interac- tions are unknown, although they are important in order to rationalize many studies on membrane reconstitution and phospholipase enzyme kinetics. We have examined equimolar mixtures of DPPC or DMPE with a surfactant, either Triton X-100 (a non-ionic detergent) or sodium cholate (a bile salt) both widely used in biochemical studies. According to a variety of physical and biochemical data,20.2' the amounts of surfactantF.M. Go% and J. L. R. Arrondo 121 wavenumber/cm-' Fig. 3. F.t.i.r. absorbance spectra of aqueous dispersions of DPPC and DPPC-surfactant mixtures. ( a ) Pure DPPC; ( b ) DPPC-Triton X-100, equimolar mixture; ( c ) DPPC-sodium cholate, equimolar mixture. used in the present study are not enough to produce bilayer solubilization, although they do produce dramatic changes of bilayer fluidity and permeability, together with high increases in phospholipase activity. According to our i.r. spectroscopic studies, surfactants modify the phosphate vibra- tion modes of phospholipids. Fig. 3 shows the phosphate region of the i.r. spectrum of DPPC, DPPC:Triton X-100 and DPPC:cholate at 50°C, i.e. in the fluid state. Each surfactant acts in its own way.Triton X-100 [fig. 3(b)] shifts the two main phosphate bands of DPPC towards higher frequencies, from 1220 to 1239 cm-' and from 1085 to 1090 cm-', respectively; also, the relative intensity of the R-0-P-0-R' shoulder at 1060 cm-' is obviously decreased; finally, a sharp band with a maximum at 1014 cm-' appears. The observed phosphoryl shifts are characteristic of hydrogen bonding22 and strongly suggest a decrease in hydrogen bonding to water of the phosphate group in the presence of Triton X-100. We had previously observed similar shifts in DPPC samples with various degrees of hydration. l 2 Surfactant-dependent changes in the R-0-P-0-R' shoulder are very interesting since the orientation of this part of the molecule strongly depends on the headgroup conformation.The observed shoulder is the result of coupling between the two P-0-C vibrations, depending in turn on headgroup packing, hydrogen bonding etc.17 Different transition vectors of the two P-0-C groups as the headgroup changes from a parallel to a perpendicular orientation122 I. R. Studies of Phospholipids wavenumber/cm-' Fig. 4. F.t.i.r. absorbance spectra of aqueous dispersions of DMPE and DMPE-surfactant mixtures. ( a ) Pure DMPE; ( b ) DMPE-Triton X-100, equimolar mixture; (c) DMPE-sodium cholate, equimolar mixture. with respect to the bilayer plane would also give rise to changes in the shoulder at 1060 cm-'. In summary, changes in phosphate spectral features in the presence of Triton X-100 can be interpreted in terms of a conformational change of the DPPC headgroup, involving a decrease in phosphate hydrogen bonding to water.The effect of sodium cholate on the phosphate spectrum of DPPC [fig. 3 ( c ) ] is smaller than that of Triton X-100. The intensity of the R-0-P-0-R' shoulder is decreased, but no shifts in the phosphate stretching vibration bands are observed. If the above interpretation is correct, this would mean that cholate does induce some kind of conformational change in the DPPC headgroup without altering the degree of hydration of the phosphate groups. This is in agreement with the different structure proposed for DPPC-Triton X- 100 and DPPC-cholate mixed micelle~.~~ Our results concerning DMPE-surfactant mixtures are shown in fig. 4. These experi- ments were carried out at 50 "C, i.e.with the bilayer in the fluid state, and are therefore comparable to those in fig. 3. Both surfactants act in this case in a similar way. Their main effect consists of inducing the appearance of much fine structure in the spectral bands, of the kind seen in dehydrated phospholipid samples. Why this fine structure becomes apparent in the presence of surfactants in excess water cannot be easily explained. The surfactants are also responsible for the presence of a new shoulder at the high-frequency side of the asymmetric PO, band, at 1250 cm-'. This new band inF. M. Go% and J. L. R. Arrondo 123 0. 312 0.294 e, E .fl 2 0.156 3 0.078 0.000 1 1300 1220 1140 1060 980 w aven u m be r/ cm - Fig. 5. F.t.i.r. absorbance spectrum of an aqueous dispersion of cardiolipin in the region between 1000 and 1300 cm-'. sucn a nigniy resoivea spectrum may well correspona to tne nign-rrequency sniIt ooservea in DPPC [fig.3 ( b ) ] ; also the low-frequency side of the asymmetric PO, band of DMPE-Triton X-100, but not that of DMPE-cholate mixtures, is clearly depressed as compared to that of pure DMPE (fig. 4), as was the case with DPPC and DPPC-surfactant mixtures (fig. 3). Both surfactants decrease the relative intensity of the DMPE band at 1010 cm-', attributed to R-0-P-0-R' vibrations, as was also the case with DPPE. Therefore, surfactants appear to act in a similar way in DMPE and DPPC headgroups, changing the orientation of the P-0-C bonds and, at least in some cases, reducing the extent of hydrogen bonding between phosphate and water.Phospholipids Containing Two Phosphate Groups We have extended our study of the phospholipid phosphate vibrations to the case of DhosDholiDids containing two DhosDhate crouDs. We shall first examine cardiolioin. a I 1 I " & I Y I 1 , phospholipid containing two chemically identical phosphate groups, and then proceed to phosphatidylglycerophosphate (alkyl ether), bearing two non-identical phosphates. The 1000-1300 cm-' region of the i.r. spectrum of cardiolipin is shown in fig. 5 . As in the previous cases, two main bands are seen. The one corresponding to the asymmetric PO, stretching vibrations has a maximum at 1215 cm-'; this would correspond to a high degree of hydrogen bonding17922 and, given the structure of this phospholipid, it is tempting to speculate that even in water dispersion some degree of intramolecular hydrogen bonding may exist.The lower frequency band is very wide and complex; at least three components are resolved, with maxima at 1092, 1072 and 1044 cm-', respec- tively. The maximum at 1092cm-' reveals most probably the contribution from the PO, symmetric vibration, while the other two maxima may be attributed to the four C-0- P vibrations in this phospholipid. The details of cardiolipin headgroup confor- mation in bilayers are not known, but it is possible that, if not chemically, both phosphate groups are conformationally non-identical with respect to the bilayer plane. This would explain the complexity of cardiolipin spectral features in the region around 1050 cm-'.124 I. R. Studies of Phospholipids 0 -097 0.073 6) r: -fi 0.048 B D 0.024 o * o o o 1300 1220 1140 1060 980 wavenumber/cm-' Fig.6. F.t.i.r. absorbance spectra of aqueous dispersions of phospholipids. ( a ) Phosphatidyl- glycerophosphate (alkyl ether); ( b ) phosphatidylglycerol (alkyl ether). The spectrum of phosphatidylglycerophosphate (alkyl ether), the major phospholipid in purple membranes, (I) can be seen in fig. 6 ( a ) . In this phospholipid the two phosphate groups are clearly different, one of them being monoesterified, and the other diesterified. However, the phosphate region of the i.r. spectrum is not very different from cardiolipin, with an asymmetric PO, band (maxima at 1215 cm-') and a complex band at lower frequencies, showing a maximum at 1064 and shoulders at 1038, 1085 and 1113 cm-'.Anhydrous films of the same phospholipid allow the resolution of this band into four peaks. In this case, the complexity of the band is easily understood, since there are three obviously different C-0-P groups in the molecule. Enzymic digestion of the purple membrane phospholipid provides us with a new tool for the study of i.r. phosphate vibrations. We have recently developed a method14 for cleaving an L- a-glycerophosphate moiety from phosphatidylglycerophosphate (alkyl ether), by using phospholipase D. As a result of enzyme action, an ether analogue of phosphatidic acid (11) is produced. The phosphate region of the corresponding i.r. spectrum is shown in fig. 6 ( 6 ) . Two striking features of this spectrum are: (a) the high frequency shift of the asymmetric PO, stretching band, with a maximum at 1273 cm-', and (6) the greatly simplified low-frequency band, confirming that the wide band seen in curve (a) contained the contributions from two other C-0-P groups.It is puzzling, however, that the spectrum of phospholipid (11) looks very different from that of its diester homologue, i.e. phosphatidic acid. In fact, the two maxima in the spectrum of the latter occur at 1181 and 1076crn-'.'* Among the factors that may explain such differences we should mention the smaller possibility of hydrogen bonding when the ester carbonyl groups are absent, and the sensitivity of monoesterified phosphate groups to changes in pH or counterions. Conclusions The results summarized here constitute a further step in the application of Fourier transform i.r.spectroscopy to the study of model and biomembranes. This techniqueE M. Go% and J. L. R. Arrondo 125 Table 1. Maximum wavenumbers of the main bands appearing in the 1000-1300 cm-’ region of f.t.i.r. spectra of aqueous phospholipid dispersions compound maximum wavenumber /cm-l assignment DPPC DMPE phosphatidic acid cardiolipin phosphatidylglycerophosphate (alkyl ether) 1060sh 1086 1180sh 1222 1013 1076 1177sh 1221 1076 1181 1044 1072 1092 1136 1215 1038 1064 1085 1113 1215 R-0-P-0-R’ symmetric PO, stretch asymmetric PO, stretch symmetric PO, stretch R-0-P-0-R’ asymmetric PO, stretch symmetric PO, stretch asymmetric PO, stretch R-0-P-0-R(?) R-0-P-0-R‘ symmetric PO, stretch asymmetric PO; stretch R-0- P-0 - R’( ?) R-0-P-0-R symmetric PO, stretch asymmetric PO; stretch provides spectra of excellent quality from dilute samples in a short time and buffer, temperature and other sample conditions can be regulated easily.In our case, the technique has been applied to the description of the phosphate region of phospholipid i.r. spectra. Phospholipids have been studied in the form of liposomes. For some of them, tentative assignments of bands have been made (table 1). Our results show that phosphate stretching vibrations are not influenced by the length or unsaturation of fatty acyl chains, nor by lipid polymorphism (in excess water). However, the nature of the phospholipid headgroup, or the presence of surfactants, does influence the phosphate vibration bands. The contribution from the R-0-P-0-R’ appears to be particularly sensitive to conformational changes, C-0- P vibrational bands suggest that the two phosphate groups of cardiolipin are conformationally non-identical when the phos- pholipid is integrated in a bilayer.More complex lipids, such as the major phospholipid of Hulobucterium purple membrane, may also be studied by this technique; enzymic hydrolysis under controlled conditions provides an additional tool for the analysis of the resulting complex spectra. We thank CAICYT (grant no. 0992-84) and Diputaci6n Foral de Vizcaya (O.F. no. 2813/85) for support. References 1 Biological Membranes, ed. D. Chapman (Academic Press, New York, 5 vols, 1968-1985). 2 D. Chapman, R. M. Williams and B. D. Ladbrooke, Chem. Phys. Lipids, 1967, 1, 445.3 D. G. Cameron, H. L. Casal and H. H. Mantsch, J. Biochem. Biophys. Methods, 1979, 1, 21. 4 D. Chapman, J. C. G6mez-Fernhdeq F. M. Goiii and M. Barnard, J. Biochem. Biophys. Methods, 1980, 2, 315.126 I. R. Studies of Phospholipids 5 R. L. h e y and D. Chapman, in Biomembrane Structure and Function, ed. D. Chapman (Macmillan, 6 H. L. Casal and H. H. Mantsch, Biochim. Biophys. Acta, 1984, 779, 381. 7 V. P. Fringeli and H. H. Gunthard, in Membrane Spectroscopy, ed. E. S. Grell (Springer, Berlin, 1981), 8 D. G. Cameron and H. H. Mantsch, Biochem. Biophys. Res. Commun., 1978, 83, 886. 9 H. H. Mantsch, D. G. Cameron, T. A. Tremblay and M. Kates, Biochirn. Biophys. Acta, 1982, 689, 63. 10 H. H. Mantsch, S. C. Hsi and D. G. Cameron, Biochim. Biophys. Acta, 1983, 728, 325. 11 K. J. Rothschild, W. J. De Grip and R. Sanches, Biochim. Biophys. Acta, 1980, 596, 338. 12 J. L. R. Arrondo, F. M. Goiii and J. M. Macarulla, Biochim. Biophys. Acta, 1984, 794, 165. 13 W. S. Singleton, M. S. Gray, M. L. Brown and J. C. White, J. Am. Oil Chem. Soc., 1965, 92, 52. 14 A. Muga, M.Sc. Thesis (University of the Basque Country, Bilbao, 1984). 15 M. Kates, S. C. Kushwaha and G. D. Sprott, in Methods in Enzymology, ed. S. P. Colowick and N. 0. 16 M. Cortijo, A. Alonso, J. C. G6mez-Fernkndez and D. Chapman, J. MoZ. BioL, 1982, 157, 598. 17 J. E. Fookson and D. F. H. Wallach, Arch. Biochem. Biophys., 1978, 189, 195. 18 R. Mendelsohn and H. H. Mantsch, in Protein-Lipid Interactions, ed. A. Watts (Elsevier, Amsterdam, 19 H. H. Mantsch, A. Martin and D. G. Cameron, Biochemistry, 1981, 20, 3138. 20 M. A. Urbaneja, A. Alonso, J. L. R. Arrondo and F. M. Goiii, Surfactants in Solution (Plenum Press, 21 D. Lichtenberg, Y. Zilberman, P. Greenzaid and S. Zamir, Biochemistry, 1979, 18, 3517. 22 L. J. Bellamy, Advances in Infrared Group Frequencies (Methuen, London, 1968). 23 A. Helenius and K. Simons, Biochim. Biophys. Acta, 1975, 415, 29. London, 1983), pp. 199-256. pp. 270-332. Kaplan (Academic Press, New York, 1982), pp. 98-111. in press). New York, 1986). Received 23rd December, 1985