Interaction between dengue virus fusion peptide and lipid bilayers depends on peptide clustering

Abstract

Dengue fever is one of the most widespread tropical diseases in the world. The disease is caused by a virus member of the Flaviviridae family, a group of enveloped positive sense single-stranded RNA viruses. Dengue virus infection is mediated by virus glycoprotein E, which binds to the cell surface. After uptake by endocytosis, this protein induces the fusion between viral envelope and endosomal membrane at the acidic environment of the endosomal compartment. In this work, we evaluated by steady-state and time-resolved fluorescence spectroscopy the interaction between the peptide believed to be the dengue virus fusion peptide and large unilamellar vesicles, studying the extent of partition, fusion capacity and depth of insertion in membranes. The roles of the bilayer composition (neutral and anionic phospholipids), ionic strength and pH of the medium were also studied. Our results indicate that dengue virus fusion peptide has a high affinity to vesicles composed of anionic lipids and that the interaction is mainly electrostatic. Both partition coefficient and fusion index are enhanced by negatively charged phospholipids. The location determined by differential fluorescence quenching using lipophilic probes demonstrated that the peptide is in an intermediate depth in the hemilayers, in-between the bilayer core and its surface. Ultimately, these data provide novel insights on the interaction between dengue virus fusion peptide and its target membranes, namely, the role of oligomerization and specific types of membranes.

• Dengue virus fusion peptide
• oligomerization
• partition
• membrane fusion
• lipid membrane

Introduction

Membrane fusion is the central molecular event during the entry of enveloped viruses into cells. The critical agents of this process are viral surface proteins, primed to facilitate bilayer fusion and triggered to do so by the conditions of viral interaction with the target cell. Dengue virus, an enveloped virus, belongs to the Flaviviridae family, together with other pathogenic viruses such as Yellow Fever, Saint Louis, West Nile and Tick-Borne Encephalitis (TBE) [1]. The viral genomic material is composed of a positive sense single-stranded RNA molecule, which encodes a polyprotein that is processed co- and post-translationally by proteases into at least ten discrete products. Three of them are associated with the virions: the E (envelope), M (membrane), and C (capsid) proteins [1]. Dengue virus enters into a host cell when the E glycoprotein binds to a receptor [2] and undergoes conformational rearrangement due to the reduced pH of the endosomal medium.

In the mature virions, E protein forms dimers that lie on the viral membrane [3]. The determination of E protein structure at the postfusion conformation revealed that the dimers are converted to trimers after the fusion, with the fusion peptide located at the tip of the trimer [4]. It has been proposed that the conversion from the dimers to the trimers is a two step process [5]. The first step is a reversible dissociation of the ectodomains, which is important to make the tip of domain-II (putative fusion peptide) accessible for the interactions with the target membrane and the second one is the irreversible trimerization. The self-associated proteins bear three ‘fusion loops’ at the tip of the trimers to insert them into the host-cell membrane. After that, viral nucleocapsid can be released into the host cell cytoplasm.

The purpose of the present work is to study dengue virus fusion peptide in aqueous solution and its interaction with different membrane model systems. Steady-state and time-resolved fluorescence spectroscopy were used to obtain structural information and to evaluate the fundamental principles that govern dengue virus fusion peptide (DEN Fpep) incorporation in the membrane model systems and its location in the phospholipid bilayer. Ultimately, these data provide novel insights on the mechanism of action of DEN Fpep, namely, the role of oligomerization and specific types of membranes.

Materials and methods

Buffer 10 mM MES, 20 mM Tris buffer, pH 5.5, was used throughout this study, unless otherwise stated.

Chemicals

POPC (1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine) and POPG (1-palmitoyl-2-oleyl-sn-glycero-3-[phospho-rac-(1-glycerol)]) were purchased from Avanti Polar-Lipids (Alabaster, AL). Cholesterol was from Sigma (St. Louis, MO) and 5NS (5-doxyl-stearic acid) and 16NS (16-doxyl-stearic acid) were from Aldrich Chem. Co. (Milwaukee, WT). All other reagents were of analytical grade.

Peptides synthesis

The putative dengue virus fusion peptide (amino acid sequence between 98 and 112) and the peptide fragment of dengue E protein corresponding to the amino acid sequence between 88 and 123, KRFVCKHSMVDRGWGNGCGLFGKGGIVTCAMFTCKK (named DEN Fpep), which contains the putative fusion loop (underlined amino acids), were synthesized by solid phase with the substitution of the cysteine residues (92, 116 and 121) for serine residues. Fmoc methodology was used and all protected amino acids were purchased from Calbiochem–Novabiochem (San Diego, USA) or from Neosystem (Strasbourg, France). The syntheses were carried out in an automated bench-top simultaneous multiple solid-phase peptide synthesizer (PSSM 8 system from Shimadzu). The final deprotected peptides were purified by semipreparative HPLC using an Econosil C-18 column (10 µm, 22.5×250 mm) and a two-solvent system: (A) trifluoroacetic acid/H₂O (1:1000, v/v) and (B) trifluoroacetic acid/acetonitrile/H₂O (1:900:100, v/v/v). The column was eluted at a flow rate of 5 ml/min with a 10 or 30 to 50 or 60% gradient of solvent B over 30 or 45 min. Analytical HPLC was performed using a binary HPLC system from Shimadzu with a SPD-10AV Shimadzu UV/VIS detector, coupled to an Ultrasphere C-18 column (5 µm, 4.6×150 mm), which was eluted with solvent systems A1 (H₃PO₄/H₂O, 1:1000, v/v) and B1 (acetonitrile/H₂O/H₃PO₄, 900:100:1, v/v/v) at a flow rate of 1.7 ml/min and a 10–80% gradient of B1 over 15 min. The HPLC column eluted materials were monitored by their absorbance at 220 nm. The molecular mass and purity of synthesized peptides were checked by MALDI-TOF mass spectrometry (TofSpec-E, Micromass) and/or peptide sequencing using a protein sequencer PPSQ-23 (Shimadzu Tokyo, Japan).

Preparation of lipid vesicles

Large unilamellar vesicles (LUVs), with typical 100 nm diameter [6] were prepared by the extrusion method described elsewhere [7] and used as models of biological membranes.

Steady-state fluorescence studies

All fluorescence measurements were performed with a Fluorolog-3 Spectrofluorimeter from Jobin-Yvon/Horiba, and acquired with DataMax v2.20 software programme. Samples were excited at 280 nm (unless stated otherwise) and emission spectra were collected from 300 to 450 nm and were blank corrected. All spectra were also corrected with the instrumental correction function. Excitation and emission slits with 4-nm bandpass were used for all measurements.

Time-resolved fluorescence studies

Fluorescence lifetimes acquisitions were monitored using the time-correlated single photon counting, TCSPC, technique with a 280-nm LED laser source (IBH, UK). Lifetimes were calculated from time-resolved fluorescence intensity decays using 10 Kcounts in the peak channel. Fluorescence intensity decay curves were deconvoluted with the instrument software package DAS6 (IBH, UK) and analyzed as a sum of three exponential terms. The mean average lifetime, <τ >, is:1 where a_i is the pre-exponential factor and τ_i is the fluorescence lifetime of each i component [8]. The goodness of the fit was judged from the global chi-square value and weighted residuals distribution.

Extent of partition in LUV

The extent and kinetics of partition assays of DEN Fpep (18 µM) were carried out with LUVs of POPC, POPC:POPG (4:1) and POPC:cholesterol (18, 25 and 33% molar of cholesterol). Titrations of DEN Fpep with lipidic suspensions (up to 4.5 mM) were used to evaluate the extent of partition. Samples were incubated for 10 min after each addition of lipid suspension. The partition coefficient, K_p, is calculated from the experimental data fitting with equation 2 as described elsewhere [9], unless critical concentration-dependent phenomena is observed at low global lipid:peptide ratios in the samples [10].2 where I_W and I_L are the fluorescence intensities in aqueous solution and in lipid solution, respectively, γ_L is the molar volume of lipid, and [L] is the lipidic concentration [9]. (γ_L used was 7.63×10⁻¹ dm³mol⁻¹ for vesicles containing POPC [11].

Determination of extent of partition when concentration-dependent phenomena occur

Titration curves of 9, 18 and 36 µM DEN Fpep were carried out with LUVs of POPC:POPG (4:1). Critical points were taken as the maximum value in each curve and were used to determine the K_p from the experimental data fitting with equation 3 as described by [10]. This equation describes the dependence of peptide concentration at which a critical point occurs as a linear function of the phospholipid concentration in the system at that point.3 where σ is the constant local P:L proportion in a saturated membrane, γ_L is the molar volume of lipid, [L] is the lipid concentration, and [P] is the peptide concentration at the saturation point when the lipid concentration is [L]. γ_L used was 7.63×10⁻¹ dm³.mol⁻¹ for vesicles containing POPC [11].

Location in lipidic membranes

Quenching assays were followed by fluorescence intensity and lifetime with excitation at 280 nm and emission at 340 nm unless stated otherwise. Fluorescence quenching by acrylamide was carried out using wavelength λ_exc=290 nm to minimize the relative quencher/fluorophore light absorption ratio. Nevertheless, the quenching data were corrected for the simultaneous light absorption of fluorophore and quencher [12]. Quenching assays data were analyzed by the Stern–Volmer equation (equation 4),4 where I and I₀ are the fluorescence intensity or lifetime of the sample in the presence and absence of quencher, respectively, K_SV is the Stern–Volmer constant, and [Q] the concentration of quencher [8], [13].

Acrylamide is unable to efficiently quench the fluorescence of Trp residues deeply buried in the bilayer; titration of peptide in the presence of LUVs with this quencher gives initial insight on peptide in-depth location [14]. Fluorescence emission quenching (λ_exc=290 nm, λ_em=340 nm) with acrylamide was carried out with LUVs composed of POPC:POPG (4:1), both below (0.77 mM lipid in buffer) and above (3.5 mM lipid in buffer) the critical global lipid:peptide ratio in the sample.

To further evaluate the membrane in-depth location of the DEN Fpep Trp residue, differential quenching methodologies were used. 5NS and 16NS are quenchers of Trp fluorescence, which have different locations in the lipidic bilayer. 5NS is located near the interface whereas 16NS buries more deeply into the bilayer core [15]. Titration of peptide samples (18 µM), in the presence of LUVs composed of POPC:POPG (4:1), both below (0.77 mM lipid in buffer) and above (3.5 mM lipid in buffer) the critical global lipid:peptide ratio in the sample, was carried out with small aliquots of ethanolic solution of 5NS and 16NS (70 mM stock); final ethanol concentration was kept below 2%). The assays were followed by fluorescence emission intensity (λ_exc=280 nm, λ_em=340 nm). Data were corrected for simultaneous absorption of fluorophore and quencher [12]. The effective quencher concentration in the lipidic bilayer matrix, [Q]_L, was used in the Stern–Volmer plots (equation 5);5 where [Q]_T is the quencher concentration in total sample volume and K_p,q is the quencher partition coefficient [16]. For gel phase vesicles, K_p,Q equals 12570 and 3340 for 5 NS and 16 NS, respectively. In crystal-liquid-like phase, the values used for 5 NS and 16 NS were 89000 and 9730, respectively [17].

Membrane fusion induced by Dengue virus fusion peptide

Fusion was tracked using the Forster's resonance energy transfer (FRET)-based methodology [18–20]. Briefly, vesicles doped with both 1% N-NBD-PE (donor) and 1% N-Rh-PE (acceptor) and unlabeled vesicles were mixed; after that, 18 µM of the peptides was added. If fusion between unlabeled vesicles and donor/acceptor-labeled vesicles occurs, the average distance between donors and acceptors increases, that is, FRET efficiency decreases. POPC, POPC:POPI (4:1), POPC:POPS (4:1) and POPC:POPG (4:1) were used, both below (0.77 mM lipid in buffer) and above (3.5 mM lipid in buffer) the critical global lipid:peptide ratio in the sample. Fluorescence intensity was followed with λ_exc=470 nm (NBD absorption) and λ_em=590 nm (Rh emission). Control experiments (peptide absence for spontaneous background fusion and 0.2% triton X-100 for total fusion) were carried out in all cases.

Results

Amino acid sequence alignment of E proteins indicates 62–77% homology among the four dengue virus serotypes and 40–50% homology among the different flaviviruses. The segment between residues 98 and 112 of E glycoprotein forms a loop in domain II of dengue E glycoprotein [5] and has been considered to be the fusion peptide of the flaviviruses because:(i) it presents a very high homology among all the members of the Flaviviridae family (it is identical in all of them, except for a single residue in the TBE virus), and (ii) site-directed mutagenesis in that region prevents the virus fusion [21]. However, when we tested the fusion activity of a synthetic peptide corresponding to this segment, a very low peptide-induced fusion was obtained (Figure 1A). This occurred probably because this putative fusion peptide has only 13 residues and, therefore, it is too flexible to maintain the active structure. Thus, to stabilize the fusion peptide loop we decided to flank each side of the sequence with amino acids forming part of the β-strand structure, according to the crystallographic data [5]. The larger peptide was much more efficient in promoting fusion (Figure 1A), and so it was used throughout this study referred as DEN Fpep (Figure 1B).

Figure 1.  Fusion activity (A) and amino acid sequence (B) of two peptides fragments of dengue E glycoprotein corresponding to the putative fusion peptide alone (amino acids between 98 and 112, underlined in B) or with part of two flanking β-strands, named DEN Fpep (amino acids between 88 and 123). DEN Fpep arrangement as it appears in the structure of E glycoprotein (C) solved by Modis et al. [5]. Hydrophobic (blue), non-charged polar (green) and charged polar (red) residues are represented. This figure is reproduced in colour in Molecular Membrane Biology online.

Photophysical characterization of DEN Fpep interaction with membrane model systems

The interaction of DEN Fpep with LUV was followed by the changes in several spectroscopic parameters of Trp residues, namely, fluorescence intensity, fluorescence spectral shifts, and fluorescence lifetime. A significant blue-shift is observed in the emission spectra of the DEN Fpep in the presence of phospholipids vesicles (Figure 2A). The effect is more pronounced for vesicles containing negatively charged phospholipids. This spectral shift is known to be due to the incorporation of Trp residues in a more hydrophobic environment. This result is supported by the fluorescence quenching experiments using the hydrophilic quencher acrylamide (Figure 2B). These experiments were carried out with 18 µM peptide in aqueous solution and in the presence of 0.77 or 3.45 mM POPC:POPG (4:1) LUVs. Results from the linear Stern–Volmer plots showed lower K_SV values in the presence of lipids than in aqueous solution, meaning that the fluorescence emission of DEN Fpep Trp in the presence of LUVs is not so efficiently quenched by acrylamide. Thus, the Trp residue is at least partially inserted in the lipid bilayer.

Figure 2.  Interaction of DEN Fpep with LUVs. (A) Normalized fluorescence emission spectra of DEN Fpep in buffer pH 5.5 (black solid line) or in the presence of 3.45 mM LUVs of POPC:POPG (4:1) (dotted line) or POPC (long dashes). (B) Quenching of fluorescence emission of fusion peptide by acrylamide in aqueous solution (closed circle), in the presence of 0.77 (open circle) or 3.45 mM (closed triangle) POPC:POPG 4:1 (LUVs). (C) LUV fusion induced by DEN virus fusion peptide. Fusion reaction was tracked using the Forster resonance energy transfer-based methodology. We used 0.77mM POPC:POPG (4:1) (closed circle), POPC:POPI (4:1) (closed up triangle), POPC:POPS (4:1) (closed square) and POPC 100% (closed down triangle) unlabeled vesicles or labeled with N-NBD-PE and N-Rh-PE. The final peptide concentration was 18 mM in 20 mM MES, 30 mM Tris buffer, pH 5.5.

POPC:POPG was chosen as the anionic phospholipid-containing liposomes because peptide-induced fusion activity was evident only when these anionic vesicles were used (Figure 2C), although the peptide was able to interact with other liposomes systems tested (as followed by the blue shift in Trp fluorescence spectra).

Upon the partitioning of DEN Fpep to vesicles (Figure 3A), there was a decrease in fluorescence intensity for the neutral POPC LUV, with or without cholesterol. For the anionic system, the fluorescence intensity of DEN Fpep initially increases until reaching a maximum at about 1 mM lipid concentration. Then, a decrease at higher lipid concentrations occurs (Figure 3A; closed squares). On the other hand, average fluorescence excited state lifetime, <τ>, shows a regular increase (Figure 3A; open squares). Lifetimes gradually increase upon POPG-containing vesicles addition, even after overcoming 1 mM lipid concentration. <τ> for other LUV compositions (POPC, POPC with cholesterol) were determined with no significant changes compared to DEN Fpep in aqueous buffer (data not show).

Figure 3.  Partition of DEN Fpep into LUVs (A) and extent of partition into POPC:POPG LUVs (B and C). (A) Fluorescence emission intensity of DEN Fpep normalized to [L] = 0 (I/Iw) upon titration with LUVs of POPC (open triangles), POPC:POPG (4:1) (filled squares), or POPC:Cholesterol 18% (filled triangle), 25% (open circle) and 33% (filled circle) – Equation 2 was used to fit the data. Fluorescence lifetimes upon titration with LUVs of POPC:POPG (4:1) are shown in open squares. (B) Fluorescence emission intensity of DEN Fpep at concentration of 9 (closed circle), 18 (open circle) and 36 mM (closed triangle) DEN Fpep normalized to [L] = 0 (I/Iw) upon titration with LUVs of POPC:POPG (4:1) in 20 µM MES, 30 µM Tris buffer, pH 5.5. (C) Linear relationship between [peptide] and [lipid] at the critical point. Total peptide and phospholipid concentrations at critical points for the POPG:POPC system, together with the corresponding fitting by Equation 3 (solid lines). Saturation points were obtained from the partition curves at different peptide concentrations (Figure 3B).

The fluorescence intensity recorded in the titration of aqueous suspensions of the DEN Fpep (18 µM) with lipid vesicles of different lipid mixtures (Figure 3A) was used to calculate the partition coefficient (Table I) by fitting the data with equation 2. In case of anionic lipid system, application use is not possible due to the concentration-dependent critical phenomena occurring at about 1 mM lipid; equation 3 was used instead. This equation results from a model in which two regimes are considered [10]: (1) a ‘saturation’ regime, at high global peptide:lipid ratios in the sample, where the local concentration of the peptide in the membrane is constant and not dependent on the concentration of the peptide in the aqueous phase (the constant parameter that describes the system is the local membrane peptide:lipid ratio, σ), and (2) an ‘excess lipid’ regime, at high global lipid: peptide ratios in the sample, where the system is described by the constant Nernst-like partition coefficient, K_p=[peptide]_L/[peptide]_W [9]; [peptide]_L and [peptide]_W are the peptide concentrations in the local lipidic and ‘bulk’ aqueous environment, respectively. The critical points in the curves represent the exact condition where both σ and K_p are valid to describe the system, i.e., the exact condition where the local concentration of the peptide in the membrane and in aqueous phase are predicted by σ and K_p simultaneously. Critical points are the border lines between both regimes.

Table I.  Partition constants for different lipid compositions.

Vesicle composition	Kp (× 10^−3)	σ	Equation used
POPC	0.4	–	2
POPC:Cholesterol (2:1)	0.5	–	2
POPC:POPG (4:1)	8.7	0.011	3

The critical points (Figure 3C) were obtained from the partition curves with different peptide concentrations (Figure 3B) and were fitted using equation 3. The σ value of 0.011 means a local concentration in the membrane of ∼90 lipids per peptide. This value is quite large and very different from those values obtained with antimicrobial peptides and cannot be assigned to spatial saturation of the membrane surface or charge equivalence (18 negatively charged lipids per peptide inserted in the membrane). The results are indicative of oligomerization driven by the high local concentration of the peptide in the membrane but not strictly motivated by saturation phenomena.

In order to investigate the role of electrostatic interactions on the partition of DEN Fpep to vesicles, the effect of high ionic strength on its interaction with POPC:POPG vesicles was evaluated. Titration of aqueous suspensions of the DEN Fpep (18 µM) in the presence of 200 mM NaCl with lipidic vesicles was carried out (Figure 4A). At this salt concentration, the increase of fluorescence intensity at lower peptide/lipid ratios was abolished. In addition, we studied the partition curves at low (5.5) and slightly alkaline (8.0) pH. The contour of the curve is the same at both pH, but the maximum fluorescence intensity is higher at pH 5.5 (Figure 4B).

Figure 4.  Effect of high ionic strength and pH on DEN Fpep partition into lipidic vesicles. Fluorescence emission intensity of DEN Fpep normalized to [L] = 0 (I/I_w) upon titration with LUVs of POPC:POPG (4:1): (A) in the presence of low (filled circles; no added NaCl) and high (200 mM NaCl, open circles) ionic strength; (B) at pH 5.5 (filled square) and pH 8.0 (open square). The final peptide concentration was 18 µM in 20 mM MES, 30 mM Tris buffer. 200 mM NaCl (A) leads to an intermediate situation, between the one obtained with POPC and POPC:POPG 4:1 without NaCl.

DEN Fpep location in the lipid bilayer

The previous results obtained from fluorescence quenching experiments using the hydrophilic quencher acrylamide (Figure 2B) showed that the Trp residue is at least partially located inside the lipid bilayer. To further evaluate the in-depth location of the DEN Fpep Trp residue when it is interacting with the POPC:POPG vesicles, fluorescence quenching by the lipophilic probes 5NS and 16NS was used. These two derivatized fatty acids differ in the position of the quencher moiety (doxyl) in the hydrocarbon chain. They are used to evaluate the depth of the fluorophore in the membrane, by comparing the quenching results obtained with each of them. The closer the Trp residues are to the quencher group, the more efficient quenching is. Thus, 5NS probes the bilayer interface whereas 16NS probes its core. Initial experiments using steady-state fluorescence intensities showed a pronounced static quenching between the fluorophore and the quencher. This prompted us to use the parallax method [21] to find the depth of the Trp residues in the membranes. The quenching experiments were performed with fixed concentrations of peptide and lipid (0.77 or 3.45 mM), and increasing concentration of quencher. The Stern-Volmer plots with the effective local concentration of 5NS and 16NS in the lipid bilayers are presented in Figure 5A and 5B.

Figure 5.  In-depth location of DEN Fpep in LUVs using Stern-Volmer Plots. Quenching by the derivatized lipophilic molecules 5NS (closed circles) and 16NS (open circles) in the presence of 0.77 mM (A) or 3.45 mM (B) POPC:POPG (4:1) vesicles. (C) Schematic representation of DEN Fpep location in membranes, using Parallax method [22]. P represents the location of fusion peptide Trp in the external monolayer. The final peptide concentration was 18 µM in 20 mM MES, 30 mM Tris buffer. This figure is reproduced in colour in Molecular Membrane Biology online.

DEN Fpep induced vesicle fusion

Vesicle fusion implies that (1) the inner content of two or more vesicles is mixed and (2) lipids from previously separated bilayers coexist in the same bilayer after fusion [23]. Fusion may result from a variety of stimuli. The FRET-based methodology described elsewhere was used to study vesicle fusion [18–20]. The results obtained in the saturated and non-saturated conditions are presented in Figure 6. Fusion is more efficient using a high [peptide]/[lipid] ratio (membrane ‘saturation’ range).

Figure 6.  LUV fusion induced by DEN virus fusion peptide. Fusion reaction was tracked using the Forster resonance energy transfer-based methodology. We used POPC:POPG (4:1) unlabeled vesicles or labeled with N-NBD-PE and N-Rh-PE, in the ‘saturated’ (high pep/lip) or non-saturated (low pep/lip) condition. The process was initiated by addition of the peptide and the FRET efficiency was accompanied for 10 min. The fusion index of 100% was calculated by adding 0.2% Triton final concentration. The final peptide concentration was 18 µM peptide in 10 mM MES, 20 mM TRIS buffer, pH 5.5.

Discussion

Interaction with membrane model systems

The different fluorescence spectroscopy methodologies clearly show that DEN Fpep is incorporated into the membrane model systems studied, independently of the phospholipids composition. However, there is more extensive peptide incorporation in LUV containing negatively charged phospholipids. Besides, there is a more pronounced blue-shift of the emission spectra of DEN Fpep in the presence of vesicles containing POPG (negatively charged phospholipid headgroup) (Figure 2A). Moreover, the partition coefficients are approximately 18 fold larger when negatively charged vesicles are present (Table I). In this situation, partition is influenced by ionic strength and pH (Figure 4), due to the positive net formal charge of the peptide at neutral pH (eight positively and only one negatively charged residues) (Figure 1B). The use of POPG increases the electrostatic component of the peptide partition constant and might stabilize its structure upon membrane incorporation [24]. Although the negatively-charged phospholipids are usually segregated in the inner side of the cellular membranes, cell surface contains many other negatively-charged molecules, as, for example, the heparan sulfate (HS). HS, the most ubiquitous member of the glycosaminoglycans (GAGs), is used by many viruses to bind to target cells [25–27], including the dengue virus. Studies have shown that dengue glycoprotein bound to highly sulfated GAGs on the surface of Vero cells and that infection of these cells could be prevented by heparin and by high-sulfate HS [28]. This negatively-charged GAG might act directly as a receptor or help to concentrate these viruses on the cell surface, facilitating the interaction with the specific high-affinity receptors. Thus, the presence of negative phospholipids might be mimicking these molecules, being important for enhancing the interaction between the peptide and membranes.

Electrostaic interaction with the inner side of the cellular membrane surface cannot be discarded. Crystallographic data obtained with the dengue virus fusion protein [5] show that clustered fusion loops form a nonpolar apex with a hydrophobic core, suggested to penetrate the hydrophobic region of the lipid bilayers. The fluid and dynamic nature of the cell membrane favours interaction of the fusion peptide with the inner layer of (negatively) charged lipids. Once the lipid bilayer structure is perturbed due to the challenge imposed by the charged peptide at its surface, it is possible that the peptide comes in direct contact with the inner layer of lipids. This interaction has been the subject of intense study and is quite well demonstrated for other classes of membrane-active peptides, such as antimicrobial and cell-penetrating peptides [29], [30]. Fusion peptides are short hydrophobic but frequently cationic sequences, which are characteristics found in most membrane-active peptides. Upon contact of this kind of peptides with one surface of the lipid bilayer, a local charge and concentration gradient is created. These gradients may be driving forces for lipid flip-flop and/or peptide. Gibbons et al. [31] showed for the Semliki Forest virus that the glycine-rich main chain interacts tightly with the lipid heads, projecting aromatic side chains into the aliphatic region of the lipid bilayer. Therefore, perturbation of the bilayer may reach its core and extend to the other leaflet through electrostatic interactions.

Combined analysis of steady state and time resolved fluorescence partition curves indicate that partition into vesicles without negatively charged phospholipids induces a static intramolecular (conformational) quenching of DEN Fpep, whereas POPG-containing vesicles cause an increase on fluorescence intensity first (lipid up to 1 mM) and a pronounced static intramolecular quenching afterwards. This data suggests that there is probably a great peptide conformational flexibility that may be facilitated by the high content of glycines. Other fusion peptides, such as avian sarcoma leukosis virus [32] and influenza virus [33] fusion peptides, reveal similar conformational flexibility that is critical for membrane fusion [34]. For the Semliki Forest Virus, the fusion loops show considerable plasticity also being a gly-rich flexible sequence [30]. Moreover, fusion loop clustering modulates interaction with lipid bilayers, which is in agreement with our findings (see end of this section).

It is noteworthy the saturation-like phenomenon (Figure 3) that usually occurs when the interaction with lipids is very strong, like those between antimicrobial peptide and membranes [10], for instance. In a titration of DEN Fpep with vesicles, at low lipid concentrations, the incorporation of peptide into the vesicles containing POPG seems to be regulated by the constant lipid:peptide ratio, since a linear relationship between fluorescence intensity and lipid concentration was observed at low lipid concentrations. The overall titration curve is not hyperbolic-like, in contrast to expectation from simple partition without membrane saturation [9]. After the maximal fluorescence intensity critical point, which was dependent on the peptide concentration, a non-linear relationship resulted. This new regime is assigned to far-from-saturating conditions due to the increased lipid concentration. In this regime, partition is regulated by K_p. Membrane saturation may occur at low lipid concentrations and high K_p values. Combination of both conditions leads to membrane saturation because the bound peptide concentration hypothetically dictated by K_p is higher than what the membrane can accommodate. In the case of DEN Fpep, however, a membrane saturation due to complete crowding of the membrane or peptide:lipid charge equivalence in the membrane surface is not compatible with a local 90 lipid:peptide ratio in the membrane containing 20% anionic lipid. Oligomerization precedes complete saturation, triggering a conformational change that improves membranetropism (see below).

Localization in the membrane

Fluorescence quenching data show that the DEN Fpep is equally quenched (similar K_SV values) by the doxyl group in 5NS and in 16NS (Figure 5A and 5B), independently of the peptide:phospholipid ratio. Thus, it can be reasoned that the fluorophore is located in an intermediate depth of the hemilayer, in between the positions 5 and 16 of the acyl chains. The quenching data were used for the measurement of the depth of insertion of the Trp side chain by Parallax method [22], indicating that the Trp is located approximately 12.5Å from membrane surface (Figure 5C).

Membrane fusion induced by DEN Fpep

During dengue virus infection, the pH induced membrane fusion is a crucial step for the viral RNA evasion from the endosome into the cytoplasm. This process is catalyzed by E glycoprotein, more specifically its fusion peptide. In this study, fusion induced by DEN Fpep was more efficient using anionic phospholipids (POPG) and a high macroscopic peptide:lipid ratio (Figure 6). Anionic lipids may not have a direct role in fusion. POPG interference with fusion may result from enhanced partition into the membrane as showed in Table I. The fusion index is greater in the saturated-like membrane concentration range, indicating that self-associated fusion peptides are more membranotropic. This may help to explain why fusion occurs with oligomerized E protein. At the local microscopic level, at the point of insertion of the fusion peptides during fusion, a high peptide:lipid ratio is created, similar to the one that occurs in nearly saturated membranes. It is worth mentioning that, at high local peptide concentration in the membrane, the depth of the peptide in the lipid bilayers does not change significantly. Peptide-peptide interaction (such as forced by crowding effects in vesicles or protein oligomerization during virus-cell fusion) probably triggers conformational changes responsible for distinct photophysical parameters (e.g., fluorescence quantum yield).

Conclusion

DEN Fpep is the putative fusion peptide in the middle of two beta sheets (Figure 1). In this work, we used fluorescence spectroscopy-based methodologies to study the interaction of the peptide with model membranes. Analysis of our results shows: (i) the peptide interacts with model membranes, independently of its lipid composition, as supported by changes in fluorescence intensity and blue shifts occurring upon addition of vesicles, but the presence of anionic lipids leads to a more extensive partition; (ii) partition curves with vesicles containing negatively charged phospholipids show that the fluorescence intensity reaches a maximum at ∼1 mM of lipids, followed by a progressive decrease, which is not accompanied by lifetimes. Lifetimes reach a plateau. This suggests that the interaction depends on the global peptide:lipid ratio in the sample. Moreover, that result indirectly suggests that the peptide undergoes structural changes, probably small, dependent of that ratio, leading to a static quenching of Trp fluorescence; (iii) quenching studies indicate that the DEN Fpep Trp residue is buried in the middle of the outer hemilayer, independently of the peptide:lipid ratio (5NS and 16NS have similar quenching efficiency); (iv) DEN Fpep is able to induce membrane fusion, but the process is more efficient using vesicles containing anionic lipids and at high macroscopic peptide:lipid ratio.

Based on the data described above, one can hypothesize in general terms what might be the mechanism of interaction between DEN Fpep and membranes at the molecular level. As depicted in Figure 7, the interaction depends on the global peptide:lipid ratio in the sample. At high ratios, it is possible to detect oligomerization. Until a certain critical ratio, oligomerized peptides will be buried in an intermediate depth in the outer hemilayer in a conformation constrained by packing. This stage of interaction only occurs with vesicles containing anionic phospholipids, since quasi-saturation results from the high affinity between the peptide and the lipid. Upon addition of excess lipid, at low peptide:lipid ratios, the peptide is conformationally unrestricted because crowding disappears and undergoes structural changes, probably small, without changing its in-depth location. The oligomerized, quasi-saturated situation seems to be a suitable model for studying the interaction of viral particles and cell membranes. The micro-environment composed of the cell membrane phospholipids and the viral surface proteins has a high local peptide:lipid ratio. We believe this may be the reason behind protein E oligomerization for viral fusion, the fusion peptides forming the tip of a trimer. Our results indicate that membrane fusion is more efficient in this situation. Moreover, a preferential interaction of the fusion peptides of the E protein with anionic lipids may drive the depth of insertion further to the interior of the membrane, closer to the negatively charged inner (cytoplasmatic) lipid hemilayer, which further perturbs membranes and favors fusion.

Figure 7.  Schematic representation of the proposed mechanism of interaction between DEN Fpep and lipidic vesicles. DEN Fpep clearly interacts more extensively with anionic lipid containing vesicles. The mechanism of interaction depends on the macroscopic peptide:lipid ratio in the sample. At high ratio, oligomerization occurs due to quasi-complete crowding and is buried in an intermediate position of the outer hemilayer, in a compact (closely packed) conformation. Upon addition of lipid, at low peptide:lipid ratio, the peptide suffers a conformational change, probably small, leading to static quenching of Trp fluorescence, without changing its in-depth location. This figure is reproduced in colour in Molecular Membrane Biology online.

Altogether, these data provide novel insights on the mechanism of dengue virus fusion peptide, namely the frequently overlooked role of viral proteins oligomerization.

Acknowledgements

We thank Sónia Troeira Henriques from Lisbon University (Portugal) for valuable help.  This work was supported by grants from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ), Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Gabinete de Relações Internacionais da Ciência e do Ensino Superior (GRICES), Special Programme for Research and Training in Tropical Diseases (TDR) of World health Organization (WHO), Third World Academy of Sciences (TWAS) and Fundação para a Ciência e Tecnologia (Portugal), including a grant to M. Melo (SFRH/BD/24778/2005). F. Stauffer was recipient of a fellowship from CAPES (BEX 1279/05).