Inhibition of Arenavirus Infection by a Glycoprotein-Derived Peptide with a Novel Mechanism

ABSTRACT

The family Arenaviridae includes a number of viruses of public health importance, such as the category A hemorrhagic fever viruses Lassa virus, Junin virus, Machupo virus, Guanarito virus, and Sabia virus. Current chemotherapy for arenavirus infection is limited to the nucleoside analogue ribavirin, which is characterized by considerable toxicity and treatment failure. Using Pichinde virus as a model arenavirus, we attempted to design glycoprotein-derived fusion inhibitors similar to the FDA-approved anti-HIV peptide enfuvirtide. We have identified a GP2-derived peptide, AVP-p, with antiviral activity and no acute cytotoxicity. The 50% inhibitory dose (IC₅₀) for the peptide is 7 μM, with complete inhibition of viral plaque formation at approximately 20 μM, and its antiviral activity is largely sequence dependent. AVP-p demonstrates activity against viruses with the Old and New World arenavirus viral glycoprotein complex but not against enveloped viruses of other families. Unexpectedly, fusion assays reveal that the peptide induces virus-liposome fusion at neutral pH and that the process is strictly glycoprotein mediated. As observed in cryo-electron micrographs, AVP-p treatment causes morphological changes consistent with fusion protein activation in virions, including the disappearance of prefusion glycoprotein spikes and increased particle diameters, and fluorescence microscopy shows reduced binding by peptide-treated virus. Steady-state fluorescence anisotropy measurements suggest that glycoproteins are destabilized by peptide-induced alterations in viral membrane order. We conclude that untimely deployment of fusion machinery by the peptide could render virions less able to engage in on-pathway receptor binding or endosomal fusion. AVP-p may represent a potent, highly specific, novel therapeutic strategy for arenavirus infection.

IMPORTANCE Because the only drug available to combat infection by Lassa virus, a highly pathogenic arenavirus, is toxic and prone to treatment failure, we identified a peptide, AVP-p, derived from the fusion glycoprotein of a nonpathogenic model arenavirus, which demonstrates antiviral activity and no acute cytotoxicity. AVP-p is unique among self-derived inhibitory peptides in that it shows broad, specific activity against pseudoviruses bearing Old and New World arenavirus glycoproteins but not against viruses from other families. Further, the peptide's mechanism of action is highly novel. Biochemical assays and cryo-electron microscopy indicate that AVP-p induces premature activation of viral fusion proteins through membrane perturbance. Peptide treatment, however, does not increase the infectivity of cell-bound virus. We hypothesize that prematurely activated virions are less fit for receptor binding and membrane fusion and that AVP-p may represent a viable therapeutic strategy for arenavirus infection.

INTRODUCTION

The Arenaviridae family of enveloped, negative-stranded RNA viruses encompasses a number of hemorrhagic fever (HF) viruses, five of which have been designated category A agents by the CDC and NIAID (1). Lassa virus (LASV) is the most prevalent of the HF viruses, with up to an estimated 300,000 cases occurring annually in western Africa (2). Outbreaks of arenavirus HF occur sporadically in South America, as well, and mortality rates in hospitalized cases can exceed 40% (3–5). Clinical treatment of arenavirus infection is currently limited to administration of the nucleoside analogue ribavirin, the use of which is marked by significant toxicity and suboptimal efficacy (6, 7).

Arenavirus infection is mediated by the viral glycoprotein complex (GPC), which is expressed as a single polypeptide and is cleaved into three segments by a signal peptidase and SKI-1/S1P. The mature glycoprotein spike consists of a receptor-binding subunit (GP1), a membrane-anchored fusion protein (GP2), and a unique stable signal peptide (SSP). At 58 amino acids in length, the arenavirus SSP is two to four times longer than most viral signal peptides. It features two transmembrane domains and remains associated with GP2, with a putative role in spike stability.

The arenavirus GP2 is considered a class I viral fusion protein due to the α-helical structure of its major domains (8, 9). Low-pH activation of the fusion protein follows receptor binding and endocytosis. Dissociation of GP1 exposes the fusion peptide region of GP2, which can insert into the endosomal membrane. Virus-cell fusion is mediated by the rearrangement of GP2 trimers into a lower-energy conformation, the six-helix bundle (6-HB), bringing together the viral and endosomal bilayers. The entry process represents a potential target for antiviral agents. One of the most notable viral entry inhibitors is the HIV drug enfuvirtide. Derived from the C-terminal heptad repeat (CHR) of HIV gp41, enfuvirtide is a peptide inhibitor of viral fusion. It associates with exposed NHR to prevent stable 6-HB formation (10). Fusion inhibitors of similar design have been reported for coronaviruses, orthomyxoviruses, paramyxoviruses, and filoviruses (11–14), all of which also bear class I fusion proteins. Because peptide drugs generally possess high specificity and low toxicity, we wished to extend this model to arenaviruses.

In this report, we describe a peptide derived from the N-terminal heptad repeat (NHR) of Pichinde virus (PICV) GP2 with in vitro activity against Old and New World arenaviruses. While ultimately inhibitory with respect to viral infection, this peptide, at neutral pH, promotes fusogenicity in biochemical assays and causes morphological changes in virions that are consistent with fusion protein deployment, suggesting a novel mechanism of action.

MATERIALS AND METHODS

Peptides.

Membrane Protein Explorer (http://blanco.biomol.uci.edu/mpex) was used to predict GP2 sequences involved in intramolecular interface formation. Peptides were synthesized by BioSynthesis (Lewisville, TX). Lyophilized peptides were resuspended in dimethyl sulfoxide (DMSO) as 5 mM stock solutions.

Cells and viruses.

Vero cells and human foreskin fibroblasts (HFF) were maintained in Dulbecco's modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 mg/ml streptomycin.

Frozen stocks of PICV (CoAn 3739), Tamiami virus (TAMV), vesicular stomatitis virus (VSV), VSV expressing LASV GPC (VSVΔLASV), and measles virus (MV) were provided by Thomas Voss. Human cytomegalovirus (CMV; TR strain) was obtained from Daniel Streblow (Oregon Health & Science University). Vaccinia virus expressing GPC from Junin virus (VACVΔJUNV) or Machupo virus (VACVΔMACV) was obtained through the National Institutes of Health Biodefense and Emerging Infections Research Resources Repository. Viruses were propagated in Vero cells or HFF (CMV) in Dulbecco's modified Eagle's medium (DMEM) with 1% fetal bovine serum (FBS) until cytopathic effects were observed (2 to 6 days). The supernatants were clarified by centrifugation at 1,500 × g for 5 min, and aliquots were stored at −80°C.

Cytotoxicity assay.

TACS 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) cell proliferation assays (R&D Systems Inc.) were carried out according to the manufacturer's instructions. Vero cells or HFF in phenol red-free minimal essential medium (MEM) with 1% FBS were seeded in a 96-well plate at a density of 5 × 10⁴ cells/well and incubated overnight at 37°C and 5% CO₂. Cells were incubated for 24 h with peptide prior to the addition of MTT assay reagents.

Infectivity assays.

Vero cells were seeded at a density of 2.75 × 10⁵ cells/well in 12-well plates and incubated for 24 h to produce confluent monolayers. Viruses (50 PFU/well) were incubated in serum-free DMEM 1 h at 37°C with peptide concentrations as indicated for each experiment. Virus-peptide inocula were added to washed monolayers and allowed to adsorb for 1 h at 37°C prior to removal. Monolayers were overlaid with 2% methylcellulose (VSV, VSVΔLASV, VACVΔJUNV, VACVΔMACV, and MV) or 1.2% Avicel (PICV and TAMV) in DMEM with 1% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin. Plates were incubated for 2 (VSV, VACVΔJUNV, VACVΔMACV), 3 (VSVΔLASV), 4 (MV), 6 (PICV), or 8 (TAMV) days at 37°C to enable formation of visible plaques. After overlay removal, monolayers were washed, fixed with 4% buffered formalin, and stained with crystal violet. Statistical analyses were carried out using GraphPad Prism software.

Assays for CMV were performed in 96-well plates seeded with HFF at a density of 1 × 10⁴ cells/well. Peptide treatment and infection were carried out as described above. Plates were immunostained 24 h postinfection with mouse anti-CMV immediate early antigen 1/2 antibody (Millipore), rabbit anti-mouse horseradish peroxidase-conjugated antibody (Dako), and AEC Substrate-Chromogen (Dako). Total numbers of infected cells were quantified using an enzyme-linked immunosorbent spot (ELISpot) plate reader (Cellular Technology Limited).

Virus purification.

PICV was purified by methods adapted from Neuman et al. (15). Total protein was precipitated from clarified supernatant by the addition of 10% polyethylene glycol (PEG) 8000 (wt/vol) and 1% NaCl (wt/vol) with gentle stirring overnight at 4°C. Following centrifugation at 10,000 × g for 30 min at 4°C, pellets were solubilized in phosphate-buffered saline (PBS), and the virus was further purified through a 10%-to-20%-to-30% sucrose gradient at 100,000 × g for 90 min at 4°C. The virus was resuspended in PBS at an approximate protein concentration of 2 mg/ml and stored as 100-μg aliquots at −80°C.

CD spectroscopy.

Circular dichroism (CD) measurements were recorded on a Jasco J-810 spectropolarimeter. Spectra were recorded in a 1-mm-path-length cell with a peptide concentration of 10 μM in 10 mM sodium phosphate buffer (pH 7.4), averaged over three scans, and corrected with a buffer blank. Trifluoroethanol (TFE) was added as indicated in order to stabilize the secondary structure in the peptide samples.

Liposome preparation.

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC; Avanti Polar Lipids) in chloroform was evaporated under a nitrogen stream and lyophilized overnight to remove all traces of solvent. The lipid film was hydrated with PBS to a POPC concentration of 10 mM. Large unilamellar vesicles were formed by extrusion under conditions of nitrogen pressure through two stacked 0.4-mm-pore-size Nuclepore polycarbonate membranes (GE Healthcare) followed by extrusion through two stacked 0.1-mm-pore-size membranes (with the latter step repeated a total of 10 times) in a Lipex Biomembranes extruder.

Virus-liposome fusion assay.

Octadecyl rhodamine B chloride (R18) in ethyl alcohol was added to 100-μl aliquots of purified PICV (1 mg/ml) in PBS at a concentration of 100 nM. The solution was gently shaken in the dark at room temperature (RT) for 1 h prior to use to ensure complete uptake of the label. This concentration of R18 enabled self-quenching of the dye in viral membranes with no measurable free dye in solution. Higher concentrations of R18 necessitated removal of excess dye by additional ultracentrifugation or gel filtration, resulting in decreased fusion activity or excessive dilution of the sample, respectively. For the assays, 10 μg labeled viral protein was incubated for 1 h with peptide in PBS. Where indicated, labeled PICV was replaced with 4 μM POPC liposomes labeled with 1 mol% R18. Unlabeled liposomes were added to a concentration of 100 μM POPC. The solution was acidified by the addition of predetermined volumes of 1 or 0.1 N HCl with a 5-min incubation period between measurements. Total dilution of the PICV-peptide solution did not exceed 5%. The final volume in the cuvette was 250 μl. Complete dequenching of dye was achieved with the addition of 1% (vol/vol) Triton X-100. Assays were performed using a SLM-Aminco fluorescence spectrophotometer thermostatted at 37°C. Fluorescence emission spectra were obtained over 565 to 635 nm with fixed excitation at 555 nm. Spectra are presented as the averages of the results of three scans. Percentage dequenching at 582 nm was calculated by the following equation: 100(F_t − F₀)/(F_Triton − F₀).

Cryo-EM.

Purified PICV in PBS was incubated with 0.1 N HCl or peptide for 1 h at 37°C and then subjected to UV irradiation for 30 min prior to cryo-electron microscopy (cryo-EM). The final concentration of viral protein was 1 mg/ml. Approximately 3 to 5 μl of solution was adsorbed onto lacy carbon film 200-mesh copper grids (Electron Microscopy Sciences), blotted, and plunged into liquid ethane using a Vitrobot Mark IV (FEI). Samples were imaged with a Tecnai G2 F30 Twin transmission electron microscope (FEI) at 200 keV under low-dose conditions.

Fluorescence microscopy.

Vero cells were seeded onto collagen-coated 35-mm-diameter glass-bottom dishes (MatTek) 24 h in advance. Cells were incubated with F-12 media for 1 h prior to use to reduce autofluorescence, and 2 mg/ml Hoechst 33342 (Invitrogen) was added during the last 5 min to stain nuclei. Dishes were placed on ice for several minutes before spinoculation. Purified PICV was labeled as previously described with 25 nM R18 to minimize self-quenching. Labeled virus (1 μg), either left untreated or preincubated with 50 μM peptide for 1 h at 37°C, was spinoculated onto cells at 2,100 × g and 4°C for 20 min. Unbound virus was removed by four washes with cold PBS, and 2.5 ml cold imaging buffer (140 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 20 mM HEPES, 2 mg/ml Hoechst 33342, and 2% FBS) was added to the dish. Cells were imaged immediately using a Zeiss Axioplan upright microscope with a 40× water immersion objective and 0.5-μm-thick Z-slices. Deconvolution and intensity thresholding were performed using Volocity software (PerkinElmer). Puncta smaller than 0.1 μm² in diameter and larger than 1 μm² were excluded from particle counts as nonviral background and aggregates, respectively.

Cross-linking.

Purified PICV (50 μg) was treated with 5 μM 6-carboxytetramethylrhodamine (TAMRA)-conjugated peptide for 30 min at 37°C. The cross-linking agent bis(sulfosuccinimidyl) suberate (BS³; Thermo Scientific Pierce) was added at a concentration of 0.5, 1, or 5 mM and incubated at RT for 30 min. The cross-linking reaction was quenched by addition of 50 mM Tris (pH 6.8) for 15 min at RT. Total protein was precipitated with 10% trichloroacetic acid overnight at 4°C. Protein was pelleted by centrifugation at 20,000 × g for 15 min and washed with 1 volume of cold ethyl alcohol. SDS-PAGE was performed on a 4% to 10% Nuvex bis-tris gel (Bio-Rad). Gels were imaged under UV light prior to fixation and colloidal Coomassie staining.

Steady-state fluorescence anisotropy assay.

1,6-Diphenylhexatriene (DPH; Invitrogen) or 1-(4-trimethylammoniumphenyl)-6-phenyl-1,3,5-hexatriene p-toluenesulfonate (TMA-DPH;Invitrogen) in dimethylformamide was incorporated into POPC liposomes at a 1:100 molar ratio or into purified PICV at a ratio of 3.3 pmol per 1 μg viral protein. The probes were incorporated into membranes by 90 min of incubation at RT in the dark. The final probe concentrations were 1 μM in 100 μM POPC and 400 nM in 30 μg PICV. Labeled liposomes or virions were incubated for 10 min with peptide prior to measurements. Assays were performed at 37°C, and the final cuvette volume was 250 μl. Fluorescence emission spectra were obtained over 400 to 600 nm with fixed excitation at 395 nm. L-configuration polarization was used. Anisotropy (r) data were calculated by the instrument's software using the following equation: r = (I_Vv − GI_Vh)/(I_Vv + 2GI_Vh), where I_Vv and I_Vh are the intensities of the emitted polarized light with the emission polarizer parallel and perpendicular, respectively, to the excitation polarizer and G (G = I_Hv/I_Hh) is the correction factor.

RESULTS

Identification of a peptide inhibitor specific for arenaviruses.

An initial screen of peptide candidates with significant interfacial hydrophobicity identified several with antiviral activity. Among the most potent was a 19-mer designated AVP-p, comprising amino acids 348 to 366 (LNLFKKTINGLISDSLVIR) of the PICV GPC precursor. Antiviral activity was assessed through plaque reduction assays, and AVP-p was determined to inhibit PICV with a 50% inhibitory dose (IC₅₀) of 7.05 ± 2.19 μM (Fig. 1A). Complete inhibition of plaque formation occurred in the presence of approximately 20 μM peptide. Although scrambled peptides (scrambled 1, RTILLFIGVKDLLKNSNSI; scrambled 2, KLTILNKDGILRSVILSFN) demonstrated some inhibitory activity, it was considerably reduced versus that of the template peptide, which suggests that the antiviral mechanism of AVP-p is sequence dependent to some extent. AVP-p was similarly effective at inhibiting viruses and pseudoviruses expressing GPC from New World (TAMV, JUNV, MACV) and Old World (LASV) arenaviruses (Fig. 1B). The range for IC₅₀s was approximately 2 to 10 μM. A panel comprised of VSV, CMV, and MV was used to assess the breadth of peptide activity against other enveloped viruses; however, infectivity of these viruses was not significantly inhibited, indicating that the activity is specific against viruses bearing arenavirus glycoproteins (Fig. 1B). No significant loss of viability was observed in Vero cells or HFF, which are particularly sensitive to cytotoxic agents, following incubation with peptide (Fig. 1C).

FIG 1.

Specific antiviral activity of AVP-p. (A) Dose-response curves for AVP-p and variants against PICV. AVP-p and two scrambled versions were tested for antiviral activity by the use of a plaque reduction assay. (B) Activity of AVP-p against other viruses and pseudoviruses. The peptide was tested against TAMV, VACV expressing JUNV or MACV GPC, VSV expressing LASV GPC, VSV, CMV, and MV. (C) AVP-p does not induce cytotoxicity following 24 h of incubation with cells as measured by an MTT assay.

CD spectroscopy.

AVP-p is derived from a GP2 region predicted to possess α-helical structure (Fig. 2A). CD spectroscopy revealed that the hydrophobic peptide formed a random coil in solution, as demonstrated by a single minimum at 200 nm (Fig. 2B). The less-inhibitory scrambled peptides likewise adopted random-coil conformations in buffer. Alpha-helicity was induced in all three peptides upon addition of 25% TFE (Fig. 2C). At 10% TFE, however, AVP-p remained a random coil whereas the scrambled peptides showed some helical content, indicating that AVP-p has a lower propensity for helicity than the scrambled variants.

FIG 2.

Structure of AVP-p. (A) The location of the AVP-p sequence (pink) in the N-terminal helix is depicted in a PyMol rendering of the PICV postfusion GP2 ectodomain. (B) CD spectra for AVP-p and the scrambled peptides in phosphate buffer (pH 7.4) showing the random-coil structure. (C) CD spectra for AVP-p and the scrambled peptides in phosphate buffer with TFE.

Peptide-induced fusion of PICV and liposomes.

In order to determine whether peptide-treated virions remain fusion competent, purified PICV was labeled with R18, the use of which has been well established in bilayer-fusion assays (16). The results seen with POPC liposomes approximated those seen with biological membranes in our assay. Labeled virus (not treated with peptide) exhibited approximately 52% dequenching following acidification to pH 5.0, with the highest rates of fusion occurring between pH 6.5 and 5.5 (Fig. 3A). The kinetics of PICV dequenching are in good agreement with the results of Di Simone et al. (17) for lymphocytic choriomeningitis virus (LCMV) fusion. Lengthening the incubation time with each incremental pH decrease did not affect the extent of dequenching (not shown). When AVP-p was incubated with R18-labeled virus, we observed extensive dequenching in the sample at pH 7.4, prior to the addition of liposomes and acid (Fig. 3B). Hypothesizing that the dequenching resulted from fusion among unevenly labeled virions, we tested AVP-p-induced fusion of R18-labeled PICV with liposomes at neutral pH. Concentration-dependent dequenching of labeled PICV occurred at a linear rate (r² values of 0.9534 for 5 μM and 0.9405 for 50 μM) over the hour following the addition of the peptide (Fig. 3C). Labeled virus incubated with liposomes and 5 or 50 μM peptide at pH 7.4 exhibited approximately 24% or 54% dequenching, respectively, after 1 h. The R18 label was stable once incorporated into viral membranes, with only 5% dequenching over the duration of the experiment in the absence of peptide. AVP-p did not induce significant dequenching of PICV preincubated at pH 5.0 prior to labeling or of R18-labeled liposomes, indicating that the peptide itself does not directly cause fusion of lipid bilayers and that dequenching by AVP-p is strictly glycoprotein mediated.

FIG 3.

Fusion of PICV and liposomes in the presence of AVP-p. (A) R18-labeled virus (10 μg) was added to POPC liposomes (100 μM) at pH 7.4, and the pH was lowered incrementally to trigger fusion. Complete dequenching was achieved by the addition of Triton X-100 at the end of the experiment. a.u., arbitrary units. (B) Dequenching of labeled virus was observed following peptide treatment (50 μM). Addition of liposomes immediately prior to acidification yielded only slight further dequenching. (C) AVP-p was added to labeled PICV (10 μg) or liposomes (4 μM) at t = 0 in the presence of 100 μM unlabeled liposomes. Inactivated virus was preincubated at pH 5.0 prior to labeling and peptide treatment. Asterisks indicate dequenching values that are significantly greater than those determined for the untreated control.

Evidence of fusion protein activation by cryo-electron microscopy.

Viral morphology was assessed by cryo-EM following treatment. Untreated control samples showed pleiomorphic, spherical virions with distinct glycoprotein spikes (Fig. 4A). Incubation at pH 5.0 yielded particles lacking visible glycoprotein spikes (Fig. 4B). Loss of the prefusion spike structure is indicative of GP1 dissociation (18). Reorganization of the internal protein matrix in some acidified PICV particles was also observed. Virions treated with AVP-p at neutral pH displayed considerable similarity to those exposed to low pH (Fig. 4C). Disappearance of prefusion glycoprotein spikes was observed in samples incubated with peptide, but the phenomenon was less complete than with low pH (Fig. 4D). Low-pH-treated virions showed a trend toward diameters larger than those seen with untreated PICV, which is consistent with fusion among virions (Fig. 4E). Peptide-treated PICV demonstrated an even more substantial shift toward larger particles, suggesting more extensive fusion with peptide treatment than with acidification.

FIG 4.

Cryo-electron microscopy of PICV. (A) Untreated virus shows prominent prefusion glycoprotein spikes. (B) Disappearance of spike structure due to dissociation of GP1 following exposure to pH 5.0. (C) Partial loss of prefusion spikes after incubation with 50 μM AVP-p. Bars, 50 μm. (D) Percentages of spike-bearing particles present are decreased in acidified and peptide-treated samples versus the untreated control. (E) Acidified (n = 113) and peptide-treated (n = 113) samples exhibit a trend toward larger particle sizes versus the control (n = 110).

Effect of peptide treatment on viral binding.

R18-labeled purified PICV was incubated with peptide and bound to cells by spinoculation. Because most membrane labels are incompatible with fixation, the cells were imaged live and in ice-cold buffer in order to prevent endocytosis and fusion. The total numbers of particles bound to 100 cells in randomly selected fields in three separate experiments were counted for both untreated and peptide-treated samples. Incubation with AVP-p reduced but did not eliminate overall binding of individual virions (Fig. 5).

FIG 5.

Binding of peptide-treated PICV to cells. (A and B) Untreated (A) or 50 μM AVP-p-treated (B) R18-labeled PICV was spinoculated onto Vero cells at 4°C. Cells were imaged live in ice-cold buffer to prevent internalization and fusion of virus. Bars, 10 μm. (C) Total particle binding for 100 cells in each treatment group is presented as a percentage of the untreated control level. Large puncta (>1 μm²) were excluded from particle counts as aggregates.

Interaction of peptide and GPC ectodomains.

To determine whether AVP-p interacts directly with glycoprotein spikes, purified virus was incubated with a TAMRA-conjugated peptide prior to cross-linking by BS³. TAMRA is UV visible at low-picogram quantities, making the assay highly sensitive to peptide-glycoprotein binding. Because BS³ does not cross membranes, it is a useful agent for detecting spike-bound peptide. No interaction, however, was observed between peptide and glycoprotein species (Fig. 6). Although the yield of BS³-cross-linked peptide was relatively low, binding of 0.1% to 0.01% of the available TAMRA-peptide to glycoproteins would have produced highly visible bands. Additionally, cross-linking by glutaraldehyde failed to reveal peptide-glycoprotein binding (results not shown), indicating that the specificity of BS³ for primary amine groups was not a limiting factor in this experiment.

FIG 6.

Cross-linking by BS³ following TAMRA-peptide treatment of PICV. An SDS-PAGE gel is shown following colloidal Coomassie staining (left) and under UV light (right). TAMRA-conjugated AVP-p was incubated with PICV prior to addition of the cross-linking agent BS³. Per lane, 50 μg viral protein and 5 μM TAMRA peptide were used.

Membrane ordering following peptide treatment.

In order to investigate the interaction of AVP-p with membranes, the fluorescent lipid probes DPH and TMA-DPH were used to monitor lipid ordering. Another PICV GP2-derived antiviral peptide was used as a positive control. GP2 peptide^289–307 (PGGYALEQWAIIWAGIKAF) corresponds to the fusion peptide region and binds to the surface of membranes (unpublished data). A GP1-derived peptide (residues 194 to 212; HLIASLAQIIGDPKIAWVG) that does not interact with membranes was included to demonstrate that the presence of peptide in the cuvette has no inherent effect on anisotropy values. TMA-DPH anisotropy increased in a concentration-dependent manner upon the addition of AVP-p and its scrambled versions, as well as GP2 peptide^289–307, indicating greater molecular order at the interface and slowing of probe rotational diffusion (Fig. 7A and B). The peptides have similar effects on model and viral membranes, although the latter are intrinsically more rigid due to their protein and cholesterol content, as shown by the higher initial anisotropy of PICV compared to liposomes. DPH was used to investigate the extent of peptide-induced perturbance in the membrane, since it accumulates in the bilayer core, whereas TMA-DPH is anchored just below the bilayer surface. While GP2 peptide^289–307 effects on acyl chain order are largely incidental, the greater increase in DPH anisotropy observed in the presence of AVP-p indicates that it inserted more deeply into membranes than GP2 peptide^289–307 (Fig. 7C and D). In contrast, neither scrambled version of AVP-p extensively altered DPH anisotropy.

FIG 7.

Lipid probe steady-state anisotropy assays. The lipophilic probes TMA-DPH and DPH were incorporated into POPC large unilamellar vesicles (LUV) and purified PICV membranes. Bilayer rigidity, as measured by probe anisotropy, was assessed following incubation with increasing concentrations of peptide. GP2 peptide^289–307, derived from the distal fusion peptide region of GP2, served as a positive control, while a non-membrane-interacting GP1-derived peptide represents a negative control.

Rate of peptide activity.

To assess the rate at which AVP-p acts on PICV, virus was preincubated with peptide for the indicated length of time prior to infection. Activity of AVP-p shows considerable time dependence, with the full effects of 5 μM peptide, a concentration near the IC₅₀ for PICV, not observed until 1 h postaddition (Fig. 8A). Because the effects of AVP-p treatment in earlier experiments were consistent with fusion protein activation, we considered whether the peptide could have a similar effect in the absence of separate pretreatment of virus before the introduction to cells. In order to verify that AVP-p does not enhance viral infectivity, monolayers were exposed to approximately 50 PFU PICV and peptide together without preincubation for 1 h. Inhibitory activity was certainly lower when virus was not pretreated with peptide prior to incubation with cells, but no increase in plaque number was observed (Fig. 8B). Virus was also incubated for 1 h with cells at 4°C, which enables binding without internalization, prior to addition of peptide for 1 h at 37°C. Again, no significant antiviral activity or increase in viral infectivity versus the control was observed, suggesting that endocytosis and fusion occur in less time than is required for the peptide to impact the process significantly.

FIG 8.

Time dependence of AVP-p activity. (A) AVP-p (5 μM) was incubated with PICV for the indicated amount of time before addition to cells. (B) PICV was added to monolayers together with AVP-p (no preincubation period) or allowed to adsorb at 4°C prior to addition of the peptide.

DISCUSSION

In this study, we have investigated the interaction of PICV with a peptide derived from its fusion protein. Assays reveal specific in vitro antiviral activity in the low micromolar range. The antiviral efficacy of the peptide extends to viruses expressing both New and Old World arenavirus GPC.

AVP-p was identified using Membrane Protein Explorer, which employs an algorithm based on the Wimley-White interfacial hydropathy scale (WWIHS) (19). We reasoned that GPC regions likely to be involved in interface formation, whether protein-protein or protein-lipid, would demonstrate high WWIHS scores. This approach has been used previously to design inhibitory peptides for ranges of enveloped viruses, and these peptides appear to possess diverse mechanisms of action, including stabilization of intermediate fusion protein conformers and viral membrane disruption leading to genome expulsion (11, 20, 21). Our findings implicate yet another possible mechanism: membrane destabilization-induced triggering of metastable fusion proteins prior to receptor binding and endocytosis.

The initial aim of this study was to identify a heptad repeat-derived peptide capable of binding to prefusion glycoprotein spikes in order to arrest fusion. Inhibitory peptides of this type have been reported for members of every other viral family possessing class I fusion proteins; however, the structure of GP2 may make arenaviruses an exception to the rule (9). The AVP-p portion of the N-helix is completely α-helical in the postfusion configuration of GP2 as part of the common trimeric coiled-coil core and does not interact with the C-terminal helix due to the disparate lengths of the arenavirus heptad repeats. Discrepancy between the native pre- and postfusion structures of the AVP-p GP2 region and the peptide's random-coil solution structure may account for our inability to detect any high-affinity peptide-spike interaction by chemical cross-linking. It should be noted that peptides derived from NHR typically show somewhat less inhibitory activity than CHR peptides, a difference which is ascribed to the greater tendency of peptides from NHR to aggregate, since they form the 6-HB core. Interestingly, a peptide corresponding to the PICV CHR exhibited no significant antiviral activity at 100 μM, suggesting that its binding site may not be exposed prior to GP2 activation (unpublished data).

Instead of interacting with prefusion spikes, AVP-p shows a strong membranotropic propensity that appears to be related to its specific activity against arenaviruses. The scrambled versions of AVP-p not only form α-helices more readily in hydrophobic environments but also insert less deeply into lipid bilayers, which suggests that peptide sequence is primarily important to antiviral activity as it affects amphipathicity and secondary structure. In order to dissect the interaction of AVP-p with membranes more fully, another GP2-derived peptide was tested in parallel with labeled virus and liposomes. The peptides have similar effects on the lateral order of phosphate head groups. However, GP2 peptide^289–307, which does not promote fusogenicity (unpublished data), induces a substantially lower degree of acyl chain packing, while the scrambled versions of AVP-p have little effect on the membrane core at any concentration. This observation suggests that remodeling of the viral membrane, particularly the hydrophobic core, directly relates to the mechanism of action of AVP-p.

Our results implicate a premature fusogenic rearrangement of viral glycoproteins as the inhibitory mechanism of AVP-p. The effects of peptide treatment on viral ultrastructure are comparable to but less encompassing than those of exposure to low pH. The patchy appearance of peptide-treated virions suggests that the peptide acts locally, and the low rate of action may signify dependence on accumulation or diffusion of peptide within the membrane. Although inactivation by peptide proceeds much more slowly than acidification, the results are equally irreversible due to GPC metastability and noncovalent subunit associations. As GP1 must dissociate prior to or concomitantly with the conformational rearrangement of GP2, the triggering of its glycoprotein spikes leaves a virion unable to bind its cellular receptor. Incubation of virus with peptide at a concentration (50 μM) that completely blocks infectivity was found to reduce but not eliminate binding of virus to cells, although the concentration of virus used in electron and fluorescence microscopy is necessarily higher than in plaque assays, resulting in very different peptide/virion ratios. Further, endosomal fusion would be impeded for virions that had already undergone some degree of fusion protein deployment, since pore formation and expansion require engagement of a minimum number of spikes at the locus (22–25).

Unlike most fusion protein spikes, which have three transmembrane domains, the mature arenavirus spike possesses nine transmembrane domains as a result of SSP retention. This association of bilayer-spanning helices potentially renders arenaviruses uniquely sensitive to membrane-perturbing agents, as the SSP-GP2 interface appears to be highly important in maintaining native metastability. Point mutations in SSP and GP2 transmembrane domains or membrane-proximal ectodomains can result variously in viral hypo- and hyperfusogenicity (26–28). Work by York and Nunberg implicates the membrane-proximal region of GP2 and the SSP ectodomain loop as a principal site of arenavirus fusion initiation (28, 29). AVP-p-induced changes in lipid organization may affect the interaction between GPC subunits, thereby destabilizing the glycoprotein spikes sufficiently to trigger fusogenic rearrangement in the absence of low pH. We hypothesize that any mutations that might reduce arenavirus susceptibility to destabilization via the membrane would also adversely affect infectivity by raising the threshold for fusion protein activation.

There may exist concern regarding the possibility of higher rates of infectivity resulting from the use of agents that activate fusion proteins, as they could potentially facilitate infection by virions that are cell associated but not receptor bound. However, we did not observe increased plaque formation with AVP-p, even when virus was permitted to prebind to cells prior to treatment, indicating that the relatively low rate of peptide activity precludes the possibility of an additional benefit to the virus. Although premature activation of viral fusion proteins is a little-explored antiviral tactic, promising in vitro results have been reported for paramyxoviruses using agonists for the attachment protein (30, 31). Since it is specific, irreversible, and unlikely to evoke resistant mutants, premature activation may be a viable therapeutic strategy for arenavirus infection.