Fusion Peptide−Phospholipid Noncovalent Interactions As Observed

Viral infections are propagated by the fusing of the viral membrane with a host cell membrane. Initiation of the fusion process occurs upon perturbati...
0 downloads 0 Views 212KB Size
Anal. Chem. 2005, 77, 1556-1565

Fusion Peptide-Phospholipid Noncovalent Interactions As Observed by Nanoelectrospray FTICR-MS Yan Li,† Fre´de´ric Heitz,‡ Christian Le Grimellec,§ and Richard B. Cole*,†

Department of Chemistry, University of New Orleans, New Orleans, Louisiana 70148, and CRBM, CNRS-FRE 2593, Montpellier Cedex 34293, and CBS INSERM U554, Montpellier Cedex 34090, France

Viral infections are propagated by the fusing of the viral membrane with a host cell membrane. Initiation of the fusion process occurs upon perturbation of the membrane of the cell under attack by a subunit of the viral protein known as a fusion peptide. Fusion peptides must insert into the lipid-rich host cell membrane to initiate rupture and merging of the two entities, but much remains unknown about the details of the fusion process. We present detailed electrospray mass spectrometry studies of binding specificities of model fusion peptides P294 and P326 with cell membrane phospholipids, i.e., phosphatidylcholines (PCs, such as 1,2-dimyristoyl-sn-glycero-3phosphocholine (DMPC)) and phosphatidylglycerols (PGs, such as 1,2-dimyristoyl-sn-glycero-3-[phospho-rac-(1glycerol)] (DMPG)). The fusion peptides clearly bind more strongly to negatively charged DMPG than to zwitterionic DMPC. Detected binding between P294/P326 and PC/ PG in 100% aqueous solution was disrupted by addition of methanol, which is known to weaken hydrophobic interactions; a higher percentage of methanol was needed to destroy a stronger initial binding. Further increases in the methanol volume fraction generally resulted in a reappearance of peptide-lipid binding, with binding strength quotients of 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC)/1,2-dilauroyl-sn-glycero-3-[phospho-rac(1-glycerol)] (DLPG)-peptide complexes rising more steeply than those of DMPC/DMPG-peptide complexes. Compared to fusion peptides P294 and P326, a hydrophilic peptide, fibrinopeptide B, showed much weaker affinity for zwitterionic DMPC, but had moderate binding affinity to negatively charged DMPG in 100% aqueous solutions. However, upon progressive addition of methanol, this hydrophilic peptide showed only a minor initial decrease in binding to DMPG before the detected binding eventually increased. These results contrast with those obtained for the hydrophobic peptides, and offer corroborative evidence that hydrophobic interactions play a key role in the mass spectrometrically observed binding between fusion peptides and phospholipids. Because the * To whom correspondence should be addressed. † University of New Orleans. ‡ CNRS-FRE. § CBS INSERM.

1556 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

rate of viral infection has been found to be pH-dependent, the effect of initial solution pH on peptide-lipid binding was also studied. As the pH was lowered, P326-DMPC binding had a steep and immediate weakening, whereas the P294-DMPC binding was slightly strengthened at pH 3.7 and then gradually weakened with a further decrease in pH. Both P326 and P294 exhibited affinities toward unsaturated lipids; (18:1)PC bound slightly more strongly to P294 than (18:3)PC. These experiments offer further evidence of the ability of electrospray mass spectrometry to provide binding information concerning noncovalent interactions that were established principally by the hydrophobic effect in solution. A critical step in infection by viruses such as HIV and influenza is the fusing of the viral membrane with the host cell membrane. This membrane fusion process is facilitated by viral envelope glycoproteins called fusion proteins. Investigations into ways to block viral membrane fusion are now being pursued as means to combat the spread of disease,1 which requires a detailed knowledge of the fusion mechanism. Fusion peptides correspond to short regions within the ectodomain of these proteins, which are mainly composed of hydrophobic residues. It is believed that fusion peptides exist as a central motif in the mechanism of fusion in all viral membrane proteins.2 They serve to initiate membrane fusion by leading insertion into the host cell membrane.1,3,4 Investigation of interactions between fusion peptides and lipid bilayers is essential for improving the understanding of the membrane fusion process. In addition, when covalently linked to a polar nuclear localization sequence (NLS)5 that targets the cell nucleus, fusion peptides can act as efficient cargo carriers by facilitating transport and passage across the cellular membrane.6 The NLS is strongly hydrophilic and thus offers amphipathic character to the fusion peptide, which facilitates the water solubility of the hydrophobic domain. In this study, the hydrophobic sequences of the model fusion peptides were derived from (1) Eckert, D. M.; Kim, P. S. Annu. Rev. Biochem. 2001, 70, 777-810. (2) Cohen, F. S.; Melikyan, G. B. Nat. Struct. Biol. 2001, 8, 653-655. (3) Peisajovich, S. G.; Shai, Y. Trends Biochem. Sci. 2002, 27, 183-190. (4) Epand, R. M. Biochim. Biophys. Acta 2003, 1614, 116-121. (5) Goldfarb, D. S.; Gariepy, J.; Schoolnik, G.; Kornberg, R. D. Nature 1986, 322, 641-644. (6) Voglino, L.; McIntosh, T. J.; Simon, S. A. Biochemistry 1998, 37, 1224112252. 10.1021/ac040084k CCC: $30.25

© 2005 American Chemical Society Published on Web 02/17/2005

the protein gp41 of HIV17 (GALFL GFLGA AGSTM GA for P294 and GALFL AFLAA ALSLM GL for P326). The large T-antigen of SV40 (PKKKR KV) was employed as the NLS sequence.8 The hydrophobic sequence and the NLS sequence were linked through the three amino acid WSQ spacer, which can improve the flexibility and integrity of the two sequences. It has been shown by a variety of methods, such as circular dichroism,9-12 fluorescence,12-15 NMR,11,14 FT-IR,16,17 and AFM,18 that hydrophobic interactions, electrostatic interactions, and conformational changes of both the peptide and the membrane all contribute to the binding of the peptide with the lipid membrane.9-19 However, reports of the use of mass spectrometry to observe noncovalent complexes between lipids and soluble proteins,20 membrane proteins,21 or peptides22 have appeared only very recently. Compared to older mass spectrometric ionization techniques, electrospray mass spectrometry (ES-MS) has the advantage of enabling the preservation of noncovalent associations that exist in solution.23-25 However, there is often concern that the gas-phase ions representing noncovalent complexes observed by mass spectrometry may not reflect the status of the component molecules in solution.26-30 Moreover, it has been established that, as solvent molecules escape from the final charged electrospray (7) Slepushkin, V. A.; Andreev, S. M.; Sidorova, M. V.; Melikyan, G. B.; Grigor’ev, V. B.; Chumakov, V. M.; Grinfel’dt, A. E.; Manukyan, R. A.; Karamov, E. V. AIDS Res. Hum. Retroviruses 1992, 8, 9-18. (8) Chaloin, L.; De, E.; Charnet, P.; Molle, G.; Heitz, F. Biochim. Biophys. Acta 1998, 1375, 52-60. (9) Chen, T. C.; Sparrow, J. T.; Gotto, A. M., Jr.; Morrisett, J. D. Biochemistry 1979, 18, 1617-1622. (10) Gierasch, L. M.; Lacy, J. E.; Anderle, G.; LaLancette, R.; Mendelsohn, R. Biopolymers 1983, 22, 381-385. (11) Epand, R. M.; Epand, R. F.; Orlowski, R. C.; Schlueter, R. J.; Boni, L. T.; Hui, S. W. Biochemistry 1983, 22, 5074-5084. (12) Ponsin, G.; Strong, K.; Gotto, A. M., Jr.; Sparrow, J. T.; Pownall, H. J. Biochemistry 1984, 23, 5337-5342. (13) Surewicz, W. K.; Epand, R. M. Biochemistry 1984, 23, 6072-6077. (14) Jacobs, R. E.; White, S. H. Biochemistry 1986, 25 (9), 2605-12. (15) Jones, J. D.; Gierasch, L. M. Biophys. J. 1994, 67, 1546-61. (16) Demel, R. A.; Goormaghtigh, E.; De Kruijff, B. Biochim. Biophys. Acta 1990, 1027, 155-162. (17) Butler, D. H.; McNamee, M. G. Biochim. Biophys. Acta 1993, 1150, 1724. (18) Van Mau, N.; Vie, V.; Chaloin, L.; Lesniewska, E.; Heitz, F.; Le Grimellec, C. J. Membr. Biol. 1999, 167, 241-249. (19) Morris, M. C.; Chaloin, L.; Heitz, F.; Divita, G. Curr. Opin. Biotechnol. 2000, 11, 461-466. (20) Demmers, J. A. A.; van Dalen, A.; de Kruijff, B.; Heck, A. J. R.; Killian, J. A. FEBS Lett. 2003, 541, 28-32. (21) de Brouwer, A. P. M.; Versluis, C.; Westerman, J.; Roelofsen, B.; Heck, A. J. R.; Wirtz, K. W. A. Biochemistry 2002, 41, 8013-8018. (22) Li, Y.; Heitz, F.; Le Grimellec, C.; Cole, R. B. Rapid Commun. Mass Spectrom. 2004, 18, 135-137. (23) Ganem, B.; Li, Y. T.; Henion, J. D. J. Am. Chem. Soc. 1991, 113, 7818-19. (24) Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1991, 113, 8534-5. (25) Daniel, J. M.; Friess, S. D.; Rajagopalan, S.; Wendt, S.; Zenobi, R. Int. J. Mass Spectrom. 2002, 216, 1-27. (26) Light-Wahl, K. J.; Winger, B. E.; Smith, R. D. J. Am. Chem. Soc. 1993, 115, 5869-70. (27) Loo, R. R. O.; Goodlett, D. R.; Smith, R. D.; Loo, J. A. J. Am. Chem. Soc. 1993, 115, 4391-2. (28) Knight, W. B.; Swiderek, K. M.; Sakuma, T.; Calaycay, J.; Shively, J. E.; Lee, T. D.; Covey, T. R.; Shushan, B.; Green, B. G.; Chabin, R.; Shah, S.; Mumford, R.; Dickinson, T. A.; Griffin, P. R. Biochemistry 1993, 32, 20315. (29) Fitzgerald, M. C.; Chernushevich, I.; Standing, K. G.; Whitman, C. P.; Kent, S. B. H. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 6851-6856. (30) Huang, H. W.; Wang, K. T. Biochem. Biophys. Res. Commun. 1996, 227, 615-21.

Chart 1. Structures of (a) DMPC (a Saturated Phosphatidylcholine), (b) (14:1)PC (an Unsaturated Phosphatidylcholine), and (c) DMPG (a Saturated Phosphatidylglycerol)

droplets via evaporation, hydrophobic interactions are weakened, whereas electrostatic interactions are strengthened.25 In biological systems, noncovalent lipid-peptide or lipidprotein interactions are characterized by both electrostatic and hydrophobic components. Our initial efforts22 specifically targeted the hydrophobic aspect of initial binding between lipids and peptides. In the current study, we broaden and deepen our investigations into the detailed binding specificities between selected phospholipids and model fusion peptides. EXPERIMENTAL SECTION Lipids. Selected phosphatidylcholines (PCs) and phosphatidylglycerols (PGs) were purchased from Avanti Polar Lipids (Alabaster, AL) and were employed without further purification. The structure of DMPC (1,2-dimyristoyl-sn-glycero-3-phosphocholine) appears in Chart 1a. DLPC (1,2-dilauroyl-sn-glycero-3phosphocholine) and DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphocholine) differ from DMPC only in the lengths of the two alkyl chains. Each of the two alkyl chains of DLPC is shorter by a -CH2CH2- group, whereas each alkyl chain of DPPC contains an additional -CH2CH2- unit. (14:1)PC (Chart 1b) denotes two fatty acyl chains each containing 14 carbon atoms with a single unsaturation in each chain in addition to the carbonyl group. The structure of DMPG (1,2-dimyristoyl-sn-glycero-3-[phospho-rac-(1glycerol)]) is shown in Chart 1c. DLPG (1,2-dilauroyl-sn-glycero3-[phospho-rac-(1-glycerol)]) is the analogue wherein each alkyl chain contains one less -CH2CH2- unit. Fusion Peptides. Fusion peptide P294 (Ac-GALFL GFLGA AGSTM GAWSQ PKKKR KV-Cya, where Ac ) CH3CO and Cya ) NHCH2CH2SH, Mr ) 2906.58) and P326 (Ac-GALFL AFLAA ALSLM GLWSQ PKKKRKV-Cya, Mr ) 3044.75) were synthesized as described previously.31 Table 1 summarizes the amino acid sequence differences between P326 and P294. The order of the side chain length is glycine (G) < alanine (A) < leucine (L). Also threonine (T) is polar, while leucine is nonpolar. So, generally, in the synthesis of P326, several residues (31) Vidal, P.; Chaloin, L.; Mery, J.; Lamb, N.; Lautredou, N.; Bennes, R.; Heitz, F. J. Pept. Sci. 1996, 2, 125-133.

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1557

Table 1. Differences in Primary Structure between P326 and P294a

P294 P326

6

9

12

14

17

G A

G A

G L

T L

A L

a The numbers in the column head indicate the residues where the differences occur.

with shorter side chains in P294 were replaced by those with longer side chains, and one polar residue in P294 was substituted by nonpolar leucine. It must be mentioned that, in addition to the peaks corresponding to intact P326 (Mr ) 3044.75), there were peaks of even higher abundance corresponding to a single impurity whose Mr ) 3012.68. From the results of an MS/MS/ MS study, and information concerning the synthesis process, we were able to unequivocally assign this impurity as an analogue P326 with a C-terminus of -NHCH2CH3 instead of -NHCH2CH2SH that we are calling “P326-S”. Calculations on five spectra showed that 1:1 [P326-S + DMPC] complexes had binding strength quotients (i.e., the ratio of total complexed to total unbound peptide) that were virtually identical to those of 1:1 [P326 + DMPC] complexes (0.09 ( 0.01 for [P326-S + DMPC] complexes, 0.08 ( 0.01 for [P326 + DMPC] complexes) obtained in the same spectrum. Calculated in the same way, the binding strengths for [P326-S + DMPG] and [P326 + DMPG] were 0.15 ( 0.01 and 0.16 ( 0.02, respectively. Because peptide P326-S and its complexes gave more intense signals than those of P326, we considered them to be more reliable; hence, they were used in subsequent calculations of the binding strengths of P326 + lipid complexes. Solution Preparation. Stock solutions (200 µM) of fusion peptides were prepared in methanol and then immediately divided into several portions, followed by removal of methanol. The dried peptides were stored at -80 °C. For each set of comparison experiments, solutions of the same fusion peptides were made from the same sample batch. The stock solutions of lipids were made to 2.0 mM in chloroform and stored at -80 °C for no longer than 10 days. For each set of comparison experiments, solutions of the same lipid were made from the same batch of stock solution. The aqueous solutions containing mixtures of peptide(s) + lipid(s) were made in three steps: first, delivering a certain volume of lipid(s) from stock solution(s) into a vial, and removing the chloroform by vacuum centrifuge, followed by exposure to dry N2 stream for about 2 min; second, dissolving the fusion peptide(s) in pure water to a certain concentration and then combining the resulting solution with the lipid(s) to the desired concentrations; third, vortexing the solution mixture at room temperature for 1 h. Mass Spectrometry and Binding Strength Evaluation. All experiments were performed in the positive mode on a Bruker (Billerica, MA) Apex II 7.0 T Fourier transform ion cyclotron resonance (FTICR) mass spectrometer. To obtain a balance between stable strong signals and maintaining “soft” ES conditions, voltages on the capillary exit and the skimmer were fixed at 40 and 12 V, respectively. The electrospray current was maintained at 10-25 nA. 1558

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

Nanospray tips were purchased from New Objective (Woburn, MA). The inner diameter (i.d.) of the intact tips was nominally 1 µm. To reduce clogging, nanospray tips were broken to widen the aperture. For each solution, at least three mass spectra were acquired under the same conditions. Each new spectrum was obtained after the ES high voltage was turned off and then on. The binding strength quotient, i.e., the ratio of the sum of complexed peptide peak abundances to the sum of unbound peptide abundances employing all detected charge states with H+ charge carriers, was used to evaluate the relative binding strengths of lipid-peptide complexes.

∑I Binding strength quotient )

complexi+

i

∑I

peptidei+

i

where Icomplexi+ is the mass spectral signal intensity of a peak representing a peptide-lipid complex of any stoichiometry having a charge state ‘i’, and Ipeptidei+ is the signal intensity of a peak of unbound peptide of charge state ‘i’. For the calculation of the binding strength quotient of [Fib-B + PG] (Fib-B ) fibrinopeptide B) complexes only, clearly detectable peaks with Na+ charge carriers were also present; hence, they were included in the calculation. For all presented plots, each point represents the average of at least three measurements, and error bars show the standard deviations of the three measurements. RESULTS AND DISCUSSION Influence of the Nature of the Phospholipid. It has been shown that the type of headgroup, acyl chain length, and degree of unsaturation each affect the mass spectrometric response of lipids.32 The electrospray ionization efficiencies of phospholipids decreased with increasing chain length.32 For phosphatidylcholine, at low concentration (0.2 µM), as the chain length increased from 12 to 24, the ES ionization efficiency decreased linearly, whereas, at high concentration (10 µM), the ionization efficiency vs chain length curve had an exponential decay.33 Because of the weak nature of the noncovalent binding between fusion peptides and lipids, a high lipid concentration (typically 20 µM) is required to mass spectrometrically detect binding. Although the chain length of naturally occurring lipids is usually equal to or larger than 12 carbon atoms, limited by the nonnegligible difference between their ionization efficiencies, DLPC/DMPC and DLPG/DMPG were chosen as model lipids to study the effect of chain length on peptide-lipid binding. As for the peptides used in this study, circular dichroism spectroscopy shows that P294 has a random coil conformation in water, but has a β-sheet conformation in lipid, while P326 is in an R-helix conformation both in water and in lipid. (32) Koivusalo, M.; Haimi, P.; Heikinheimo, L.; Kostiainen, R.; Somerharju, P. J. Lipid Res. 2001, 42, 663-672. (33) Kofeler, H. C.; Rechberger, G. N.; Fauler, G.; Windischhofer, W.; Leis, H.J. Proceedings of the 51st ASMS Conference on Mass Spectrometry and Allied Topics, Montreal, Canada, June 9, 2003; American Society for Mass Spectrometry: Santa Fe, NM, 2003.

As the volume of a micelle increases, the surface/volume ratio always decreases.34 The repulsive force coming primarily from similarly charged headgroups increases as the area per headgroup decreases. Thus, for a stable micelle, there is an optimal area per headgroup. Model phospholipids employed in this study have two hydrocarbon tails; thus, the area per headgroup is about twice as large as that for phospholipids with one tail. The employed phospholipids tend to form large disklike bilayer micelles in aqueous solution.34 It has been reported that the critical micelle concentrations (cmc’s) of didecanoylphosphatidylcholine and dipalmitoylphosphatidylcholine (DPPC) are 5 × 10-6 M35 and 4.7 × 10-10 M,36 respectively. It is estimated that the cmc decreases by about 10-fold when the hydrocarbon chain length increases by a -CH2CH2- unit.34 Using this approximation, the cmc values for DLPC and DMPC are about 10-7 and 10-8 M, respectively. As mentioned before, a typical lipid concentration employed in this experiment is 20 µM. At this concentration, in pure aqueous solutions, DLPC and DMPC exist as bilayers. Kleinschmidt and Tamm37 showed that cmc values for DLPG and DMPG are 2.3 × 10-6 and 2.1 × 10-7, respectively, at 25 °C in 10 mM sodium borate, pH 10. King et al.38 showed that the cmc value of PG is about 1.7 times higher than that of PC with the same tail chain length. On the basis of these data, at the 20 µM concentrations employed in this study, DLPG and DMPG are present in the form of bilayers in aqueous solution. Different Phase States of Lipids. At temperatures lower than the transition temperature (Tm), the lipid molecules yield more orderly arrays and form a gel-like solid. Above the Tm, lipids are in a “liquid crystal” state, and the hydrophobic core of the bilayer can be treated as a hydrocarbon liquid because lipid molecules are highly mobile.34 Studies have shown that the nature of the lipid phase can significantly affect the peptide/protein-to-lipid binding.39-42 In many cases, peptides were found to associate more strongly with lipids in the liquid crystal state.39,41,42 However, in some studies, peptides displayed a stronger interaction with membranes in the gel solid phase.40 We have shown that binding between P294 and DMPC steadily increased upon storage at 0 °C for up to 4 h. That does not necessarily mean that P294 binds more strongly to gel-solid-state DMPC. It may just be that these molecules in aqueous solution required some time to orient themselves into low-energy conformations. The fact that DMPC required significantly longer time intervals to form complexes than DMPG, combined with the overall weaker binding of DMPC vs DMPG, is the first indication that hydrophobic interactions play a key role in forming [P294 + DMPC] solution-phase complexes. In this study, spectra were (34) Tanford, C. The Hydrophobic Effect: Formation of Micelles an Biological Membranes; John Wiley & Sons: New York, 1980. (35) Reynolds, J. A.; Tanford, C.; Stone, W. L. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 3796-3799. (36) Smith, R.; Tanford, C. J. Mol. Biol. 1972, 67, 75-83. (37) Kleinschmidt, J. H.; Tamm, L. K. Biophys. J. 2002, 83, 994-1003. (38) King, M. D.; Marsh, D. Biochemistry 1987, 26, 1224-1231. (39) Lewis, R. N. A. H.; Prenner, E. J.; Kondejewski, L. H.; Flach, C. R.; Mendelsohn, R.; Hodges, R. S.; McElhaney, R. N. Biochemistry 1999, 38, 15193-15203. (40) Turchiello, R. F.; Juliano, L.; Ito, A. S.; Lamy-Freund, M. T. Biopolymers 2000, 54 (3), 211-221. (41) Pedersen, T. B.; Sabra, M. C.; Frokjaer, S.; Mouritsen, O. G.; Jorgensen, K. Chem. Phys. Lipids 2001, 113, 83-95. (42) Tomczak, M. M.; Hincha, D. K.; Crowe, J. H.; Harding, M. M.; Haymet, A. D. J. FEBS Lett. 2003, 551, 13-19.

Table 2. Model Lipids and Their Transition Temperatures

Tm (°C)

DLPC

DMPC

(18:1)PC

(18:3)PC

DLPG

DMPG

-1

23

-20

-60

-3

23

obtained after the peptide(s) + lipid(s) mixture solutions were stored at 0 °C for 4-6 h; Table 2 lists the model compounds employed along with their Tm values. DLPC and DLPG are in the liquid crystal phase during mixing or upon storage at 0 °C, whereas DMPC and DMPG are in the gel solid state during storage at 0 °C. This could introduce a potential problem in comparing the binding strengths of complexes formed by peptides with DLPC/DLPG as opposed to those formed with DMPC/ DMPG. However, studies have shown that adding in peptide could change the Tm of the lipid through cooperative interaction,42-44 commonly increasing the Tm when the peptide penetrates deep into the hydrophobic core.43 So it is possible that, with the addition of P294/P326, DLPC and DLPG are also in the gel solid state upon storage at 0 °C. PC vs PG in Fusion Peptide Binding. In comparing the binding strength quotients of corresponding complexes of P294 + PC mixtures with those of separate P294 + PG mixtures, it was noted that the PGs generally showed stronger binding affinities for P294 than the PCs; the dominant complexes detected under a variety of conditions were 1:1 P294-lipid complexes. In a more direct comparison, from the spectrum obtained from a P294-DMPC-DMPG ) 20:20:20 µM aqueous solution (Figure 1) where competition for peptide binding between DMPC and DMPG can be directly probed, the binding strength quotient of [P294 + DMPG] complexes was clearly higher than that of [P294 + DMPC] complexes. The zwitterionic PC headgroups are overall neutral, whereas PGs are negatively charged. Thus, the increased electrostatic interaction between PG and positively charged peptide must play a significant role in stabilizing peptide-PG complexes. Like P294, P326 showed affinity for both PC (Figure 2a) and PG (Figure 2b), although, as was the case for P294, PC binds less tightly with P326 than PG as manifested by binding strength quotients of 0.09 ( 0.01 and 0.15 ( 0.01, respectively. It is again clear that electrostatic interactions contributed more heavily to stabilizing the binding between fusion peptides and PG as compared to PC. Fusion Peptides vs Hydrophilic Peptides. Fib-B (EGVNDNEEGFFSAR, Mr ) 1569.67) is a peptide with obvious hydrophilic character. To further examine the contribution of the hydrophobic interaction to lipid-peptide binding, experiments were performed using equimolar concentrations of P294 and Fib-B competitively binding to DMPC (Figure 3a) and DMPG (Figure 3b). In the spectrum obtained from a P294-Fib-B-DMPC ) 20:20:40 µM aqueous solution (Figure 3a), the main bound complexes detected were [P294 + DMPC + 3H]3+ (m/z ) 1195.7), [P294 + Fib-B + 4H]4+ (m/z ) 1120.1), and [P294 + Fib-B + 3H]3+ (m/z ) 1493.1). (43) Brandenburg, K.; Harris, F.; Dennison, S.; Seydel, U.; Phoenix, D. Eur. J. Biochem. 2002, 269, 5414-5422. (44) Morein, S.; Killian, J. A.; Sperotto, M. M. Biophys. J. 2002, 82, 1405-1417.

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1559

Figure 1. Electrospray mass spectrum of 20:20:20 µM P294-DMPC-DMPG in aqueous solution. “C1-1” and “C1-2” represent [P294 + DMPG - Na + H] and [P294 + DMPC] complexes, respectively.

Figure 2. Electrospray mass spectra of (a) P326-DMPC (20:20 µM) aqueous solution and (b) P326-DMPG (20:20 µM) aqueous solution. “C2-1” and “C2-2” represent [P326 + DMPC] and [P326 + DMPG - Na + H] complexes, respectively. For C2-1 and C2-2 the binding strength quotients calculated on five spectra were 0.09 ( 0.01 and 0.15 ( 0.01, respectively. Each pair of peaks labeled in the figure represents a “P326” and “P326-S” pair (see the Experimental Section). P326-S and P326 showed similar binding strengths to both PC and PG.

The peak corresponding to [Fib-B + DMPC]2+ (m/z ) 1124.6) was very weak and not stable, and in some spectra, it was not even visible above the noise level. The descending order of absolute peak intensity is [P294 + DMPC] > [P294 + Fib-B] . [Fib-B + DMPC]. This result implies that, under the employed conditions, DMPC binds much more strongly to P294 (binding 1560

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

quotient 0.11 ( 0.02) than to Fib-B (binding quotient 0.04 ( 0.04). In fact, Fib-B actually binds much more strongly to P294 than to DMPC. The fact that an increased hydrophilicity of the peptide has a negative impact on the binding affinity to DMPC provides further evidence that hydrophobic interactions play a key role in establishing binding between the fusion peptide and DMPC.

Figure 3. Electrospray mass spectra of (a) P294-Fib-B-DMPC (20:20:40 µM) aqueous solution and (b) P294-Fib-B-DMPG (20:20:40 µM) aqueous solution. In (a), “C3-1” and “C3-2” represent [P294 + DMPC] and [Fib-B + DMPC] complexes, respectively, with binding strength quotients of 0.11 ( 0.02 and 0.04 ( 0.04, respectively. “C3-3” represents [P294 + Fib-B] complexes. In (b), “C3-4” represents the [Fib-B + DMPG] complex (binding strength quotient 0.14 ( 0.02).

In the spectrum obtained from a P294-Fib-B-DMPG ) 20:20:40 µM solution (Figure 3b), the dominant peak was [Fib-B + 2H]2+ (m/z ) 785.8). [Fib-B + DMPG] complexes showed a moderate tendency toward binding ([Fib-B + DMPG + 3H Na]2+, m/z ) 1119.1; [Fib-B + DMPG + 2H]2+, m/z ) 1130.1; [Fib-B + DMPG + H + Na]2+, m/z ) 1141.0), whereas neither unbound P294 nor [P294 + DMPG] complexes showed detectable peaks. Because PG is negatively charged in the solution, these results can be explained by considering that electrostatic interactions contributed most heavily to the binding which brought hydrophilic Fib-B and DMPG together. Signals for P294 and its complexes were suppressed in the P294-Fib-B-DMPG (20:20:40 µM) aqueous solution, whereas P294 showed strong signals in both P294-Fib-B (20:20 µM) and P294-DMPG (20:20 µM) aqueous mixtures. Effects of Methanol Addition. It has been shown that when the retention behavior of analytes with hydrophobic character on a nonpolar column during reversed-phase liquid chromatography is studied, an increase in the methanol volume weakens the hydrophobic interactions between the C18 stationary phase and the analytes.45 Methanol addition can thus be considered as a means to disfavor hydrophobic interactions in aqueous systems. Five different peptide-lipid combinations with fixed peptide-lipid (45) Hearn, M. T. W.; Zhao, G. Anal. Chem. 1999, 71, 4874-4885.

concentration ratios were tested in a series of experiments where the methanol content was varied in otherwise fixed-composition mixtures. Figure 4 shows the influence of the addition of methanol on the binding quotient of each combination. In a P294-DLPC-DMPC ) 20:20:20 µM solution (Figure 4a), increasing the methanol volume fraction to 0.20 totally disrupted the binding between P294 and DLPC/DMPC. A further increase in the methanol volume fraction to 0.30 also resulted in no detectable binding. The fact that the binding between P294 and PC was destroyed by addition of 20% methanol offers further evidence that, in 100% aqueous solution, [P294 + PC] complexes were already formed (prior to intervention of any gas-phase phenomena), with the hydrophobic interactions acting as the primary driving force promoting the interaction. If gas-phase processes (i.e., increased electrostatic attraction at the moment when the final solvent molecules depart) rather than hydrophobic interaction (in the initial solution) were responsible for enabling detection of these complexes, the improved desolvation conditions (owing to methanol addition) would not be expected to disrupt binding. For a P326-DLPC-DMPC ) 20:20:20 µM solution (Figure 4b), the binding strength quotient of [P326 + PC] complexes in 100% aqueous solution was stronger than that of [P294 + PC] complexes (Figure 4a). For the former, a higher methanol volume Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1561

Figure 4. Change of binding strength of [peptide + lipid] complexes for five peptide-lipid combinations as a function of the methanol volume fraction: (a) P294-DLPC-DMPC ) 20:20:20 µM solution; (b) P326-DLPC-DMPC ) 20:20:20 µM solution; (c) P294-DLPGDMPG ) 20:20:20 µM solution; (d) P326-DLPG-DMPG ) 20:20: 20 µM solution; (e) Fib-B-DMPG ) 20:20 µM solution. Lines linking the points are meant only to serve as a visual guide.

fraction (∼0.37) was needed to diminish the binding to zero (for DLPC) or near zero (for DMPC). Further increases in the methanol volume fraction resulted in the reappearance of binding for [P326 + DLPC] and, to a lesser degree, for [P326 + DMPC]. 1562

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

We deduce that a higher percentage of methanol is needed to disrupt stronger hydrophobic interactions. Smith et al.23 showed that, as the methanol percentage increased from 0 to 50% and then to 90%, the cmc value of DPPC increased sharply from 4.6 × 10-10 to about 1 × 10-8 and then to about 1 × 10-4 M, respectively. It is reasonable to expect that the cmc values of DLPC and DMPC also have a ∼20-fold increase as the solvent is changed from pure water to a 50% methanol/water mixture. As the concentration of free lipid monomer largely increased in solution at 50% methanol content, electrostatic interactions between the peptide and lipid became more prevalent, and thus raised the amount of complex detected. An increased propensity for electrostatic interactions could also occur following faster solvent evaporation at a higher methanol volume fraction. When using PG instead of PC, i.e., in a P294-DLPG-DMPG ) 20:20:20 µM solution (Figure 4c), again a stronger binding was shown between P294 and PG relative to P294 and PC (Figure 4a). The higher binding quotient for PG is attributed to a stronger electrostatic interaction. Higher methanol volume fractions (∼0.31 for DLPG, ∼0.33 for DMPG) were needed to lower the binding quotient of [P294 + PG] complexes to zero relative to DLPC and DMPC complexes (∼0.20 methanol volume fraction required for each). At even higher methanol percentages, both [P294 + DLPG] and [P294 + DMPG] complexes reappeared, with the binding strength quotient of the former rising much more steeply than that of the latter (Figure 4c). PG is negatively charged in the solution. This reappearance of the complexes may be explained, as above, by considering that higher concentrations of lipid monomer occur in solvents of higher methanol content (e.g., 50%), and desolvation is improved such that electrostatic interactions become strengthened between the species present in the final droplets. For a P326-DLPG-DMPG ) 20:20:20 µM solution (Figure 4d), in 100% aqueous solution, again PG is bound more strongly to P326 than PC (Figure 4b). In addition, the signal of the complexes diminished to zero (for DMPG) or near zero (for DLPG) only with a higher methanol volume fraction (∼0.41). Upon further increase of the methanol portion, the complexes again reappeared, with the binding strength quotient of detected [P326 + DLPG] complexes rising slightly faster than that of [P326 + DMPG] complexes. A sharp contrast could be seen when the response of the detected binding between the hydrophilic peptide Fib-B and DMPG upon addition of methanol was examined (Figure 4e). As the methanol volume fraction increased, after a small initial decrease, the binding strength quotient eventually increased. This contrasting result from a hydrophilic peptide that showed only a minor decrease in binding to lipids in changing from 0 to 10% methanol offers additional evidence that the hydrophobic effect plays a key role in the mass spectrometrically observed binding between the fusion peptides (P294 and P326) and DMPG. Binding at Low pH. It has been reported that a fusion peptide found in the fusion protein haemagglutinin of the influenza virus only interacted with the target membrane at low pH,2 whereas the fusion peptide of HIV viruses is active at both low and neutral pH.46 In the experiments that follow, the “acidified solutions” were (46) Pritsker, M.; Rucker, J.; Hoffman, T. L.; Doms, R. W.; Shai, Y. Biochemistry 1999, 38, 11359-11371.

Figure 5. Binding between PCs and (a) P294 (P294-DLPCDMPC ) 20:20:20 µM) and (b) P326 (P326-DLPC-DMPC ) 20:20:20 µM) as a function of the initial solution pH. Curves linking the points have been added only to aid in visualization of trends.

prepared by adding acetic acid to pure aqueous solution. The pH’s of 0, 0.01, 0.05, 0.1, and 1% acetic acid solutions were measured to be 5.5, 3.7, 3.4, 3.2, and 2.8, respectively. P294 showed a slightly higher affinity for PC at pH 3.7 relative to the solution devoid of acetic acid (pH 5.5, Figure 5a). With a further lowering of the solution pH, the detected binding decreased gradually. If one considers that addition of acetic acid also increases the ionic

strength of the solution, the ionic strengths of 0.01, 0.05, and 0.1% acetic acid aqueous solutions were 1.8 × 10-4, 4.0 × 10-4, and 5.6 × 10-4 M, respectively. The binding strength quotient of the complex reached a maximum value when the ionic strength of the solution was around 200 µM. On the other hand, peptidelipid binding was weaker in acidified solution than in a buffered solution of equivalent ionic strength.22 As compared to P294, the binding between P326 and PC showed a much sharper immediate decrease once 0.01% acetic acid was added (Figure 5b). Further lowering of the pH to 2.8 (1% acetic acid aqueous solution) caused only a slight further decrease in the binding strength quotient of [P326 + PC] complexes. Binding to Unsaturated PC. More than half of the plant and animal lipids are unsaturated, and they are often polyunsaturated.34 Double bonds in these lipids usually occur in the cis-configuration. Because unsaturated lipids are abundant in nature, it is important to investigate whether the model fusion peptides have affinities for unsaturated lipids, and compare the results with those of the saturated varieties. PC, with a hydrophobic component important to peptide binding, was chosen to test the effects of lipid chain unsaturations. PCs with tail chain lengths up to 18 carbon atoms were employed: 1,2-dimyristoleoyl-sn-glycero-3-phosphocholine, 1,2-dioleoyl-sn-glycero-3-phosphocholine, and 1,2-dilinolenoyl-snglycero-3-phosphocholine, simplified as (14:1)PC, (18:1)PC, and (18:3)PC, respectively. In this set of experiments, (14:1)PC (which has a 9-cis double bond on each tail chain, Chart 1b) was employed for a direct comparison with its saturated homologue, DMPC. P294 (Figure 6a) and P326 (Figure 6b) were found to each bind to unsaturated PC with binding affinities approximately the same as those of

Figure 6. Electrospray mass spectra of (a) P294-(14:1)PC-DMPC (20:20:20 µM) aqueous solution and (b) P326-(14:1)PC-DMPC (20:20:20 µM) aqueous solution. Both P294 and P326 showed similar levels of affinity for unsaturated (14:1)PC as compared to saturated DMPC. In (a), “C6-1” represents the [P294 + (14:1)PC] complex, and “C6-2” represents the [P294 + DMPC] complex. In (b), “C6-3” represents the [P326 + (14:1)PC] complex, and “C6-4” represents the [P326 + DMPC] complex. The isotopic patterns corresponding to [peptide + DMPC] and [peptide + (14:1)PC] complexes exhibited considerable overlap.

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1563

Figure 7. Electrospray mass spectra of P294-(18:1)PC- (18:3)PC (20:40:40 µM) aqueous solution. “(18:1)” represents (18:1)PC, “(18:3)” represents (18:3)PC, “C(18:1)” represents the [P294 + (18:1)PC] complex, and “C(18:3)” represents the [P294 + (18:3)PC] complex.

saturated PCs of the same tail chain length. Because there was only a 4 mass unit difference between DMPC and (14:1)PC, the isotopic patterns corresponding to [peptide + DMPC] and [peptide + (14:1)PC] complexes exhibited considerable overlap. This made it inconvenient to readily compare the relative binding strengths. Degree of Unsaturation. The hydrophobic interaction between peptides and lipids should be strengthened as the lipid tail chain length increases. So far, in neat aqueous solution, P294 showed only a slightly higher affinity toward DMPC vs DLPC (see Figure 4a, 0.00 methanol fraction), whereas P326 showed higher affinity for DLPC than DMPC (see Figure 4b, 0.00 methanol fraction). As mentioned earlier, in comparing the binding strengths of fusion peptide complexes formed with DLPC as opposed to those formed with DMPC, a potential problem is that DLPC is in the liquid crystal phase, whereas DMPC is in the gel solid state during mixing or upon storage at 0 °C. In theory, one could avoid this phase difference problem by comparing the binding strengths of [peptide + DMPC] complexes with those of [peptide + DPPC] complexes, because both lipids are in the gel solid state. In practice, however, ES desorption of DPPC and its complexes, globally, was found to be very poor. A plot of ES ionization efficiency of phospholipids vs alkyl chain length of the phospholipids has been reported to exhibit an exponential decay.33 Thus, as the alkyl chain length is increased, a stronger initial hydrophobic interaction appears to be counterbalanced and even overwhelmed by a decrease in the ionization efficiency. Moreover, the cmc value of DMPC is ∼10-8 M, while that of DPPC is ∼10-10 M.36 Thus, at a fixed lipid concentration above the cmc, the availability of free lipid monomer will be reduced for the longer chain phospholipids. Alternatively, the magnitude of hydrophobic interactions can also be weakened by the introduction of double bonds into the lipids’ tail chains. In terms of reduced hydrophobicity, it is estimated that the effect of introducing two double bonds is equivalent to the removal of a -CH2- group.34 Under certain conditions, including high concentration, unsaturation can increase the ES ionization efficiency of lipids.32 Moreover, all unsaturated model lipids have very low Tm values (-20 and -60 °C for (18:1)PC and (18:3)PC, respectively). Upon mixing and storage at 0 °C, there is no lipid phase difference between (18:1)PC and (18:3)PC. Thus, these lipids offer the opportunity to test the effect 1564

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

of lipid hydrophobicity on lipid-peptide binding, with fewer potential complications than may arise when alkyl chain lengths are altered. The first double bond of the polyunsaturated lipid commonly occurs at carbon 9, with additional double bonds potentially occurring at every third carbon atom.34,47 The model lipids chosen in this set of experiments are 1,2-dioleoyl-sn-glycero-3-phosphocholine and 1,2-dilinolenoyl-sn-glycero-3-phosphocholine, simplified as (18:1)PC and (18:3)PC, respectively. In a P294-(18:1)PC-(18: 3)PC ) 20:20:20 µM aqueous solution, the [P294 + (18:1)PC] and [P294 + (18:3)PC] complexes exhibited isotope patterns that were totally resolved from one another. Results indicate that (18:1)PC binds slightly more strongly to P294 than (18:3)PC, yielding binding strength quotients of 0.12 ( 0.01 and 0.09 ( 0.01, respectively. To address the possibility that the binding strength difference may come from an imprecise concentration ratio of (18:1)PC and (18:3)PC, a separate batch of P294-(18:1)PC(18:3)PC ) 20:40:40 µM aqueous solution was made. The binding strength quotients (Figure 7) for [P294 + (18:1)PC] and [P294 + (18:3)PC] complexes were 0.16 ( 0.01 and 0.13 ( 0.01, respectively; thus observed differences do not appear to be artifacts. CONCLUSION Detailed binding specificities between selected phospholipids and model fusion peptides have been examined. DMPG, which carries a negative charge, exhibits a stronger binding to P294 and P326 than DMPC, which is zwitterionic in nature. The increased electrostatic interaction clearly played a significant role in stabilizing peptide-PG complexes. Methanol addition (known to weaken hydrophobic interactions) disrupted binding between P294/P326 and PC/PG that had been observed from 100% aqueous solutions; a stronger initial interaction required a higher percentage of methanol to destroy binding. These results indicate that detected P294/P326-lipid complexes were already formed in 100% aqueous solution, with the hydrophobic effect being a primary driving force promoting the interaction. The fact that the increased hydrophilicity of fibrinopeptide B resulted in a much weaker binding affinity to zwitterionic DMPC as compared to that of the fusion peptides P294 and P326 offered further evidence that hydrophobic interactions in solution contributed heavily to the formation of [P294/ (47) Voet, D.; Voet, J. G.; Pratt, C. Fundamentals of Biochemistry, John Wiley & Sons, Inc.: New York; 1998.

P326 + PC] complexes. Fibrinopeptide B had moderate binding affinity for DMPG in 100% aqueous solution. Upon addition of methanol, however, it showed an initial slight decrease in binding to lipids that was followed by an increase in detected binding upon further methanol addition. This behavior contrasts with that observed for the fusion peptides and PG, and thus offers additional evidence that hydrophobic interactions play a key role in allowing the mass spectrometric observation of the latter complexes. Furthermore, with a lowering of pH, detected P326-DMPC complexes exhibited an immediate and steep drop in binding, whereas the detected P294-DMPC binding was slightly augmented at pH 3.7 and then decreased gradually at even higher acidity. A comparison of lipid-peptide binding as a function of the degree of unsaturation offered the opportunity to test the effect of the hydrophobicity of the lipid on binding. While both P326 and P294 exhibited affinities for unsaturated lipid, (18:1)PC bound slightly more strongly to P294 than the less hydrophobic (18:3)-

PC. Again, evidence for the importance of an initial solution hydrophobic effect in establishing binding between model fusion peptides and cell membrane phospholipids is offered. Study findings corroborate the notion that hydrophobic interactions between fusion proteins and cell membrane phospholipids can serve to initiate membrane perturbation in the early stages of viral fusion. At the same time, the ability of ES-MS to provide information regarding the strength of noncovalent interactions that originated from a hydrophobic effect is established. ACKNOWLEDGMENT This research was supported by the U.S. Department of Energy through Grant No. DE-FG02-02ER63378 (Y.L. and R.B.C.), and by EU Grant QLK2-2001-01451 (F.H. and C.L.G.). Received for review April 27, 2004. Accepted November 10, 2004. AC040084K

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1565