A Peptide Inhibitor of Cytomegalovirus Infection from Human Hemofiltrate

Abstract

Naturally occurring substances with antimicrobial activity can serve as a starting point for the rational design of new drugs to treat infectious diseases. Here, we screened a library of peptides derived from human hemofiltrate for inhibitory effects on human cytomegalovirus (CMV) infection. We isolated a previously unknown derivative of the neutrophil-activating peptide 2, which we termed CYVIP, for CMV-inhibiting peptide. The peptide blocked infection with human and mouse CMV as well as with herpes simplex virus type 1 in different cell types. We found that CYVIP interferes with virus attachment to the cell surface, and structure-activity relationship studies revealed that positively charged lysine and arginine residues of CYVIP are essential for its inhibitory activity. The N-terminal 29 amino acids of the peptide were sufficient for inhibition, and substitution with an acidic residue further improved its activity. The target structure of CYVIP on the cell surface seems to be the sulfate residues of heparan sulfate proteoglycans, which are known to serve as herpesvirus attachment receptors. Our data suggest that O-sulfation of heparan sulfate is required for binding of CYVIP, and furthermore, that the initial interaction of CMV particles with cells takes place preferentially via 6-O-linked sulfate groups. These findings about CYVIP's mode of action lay the basis for further development of antivirals interfering with attachment of CMV to cells, a crucial step of the infection cycle.

INTRODUCTION

Human cytomegalovirus (HCMV) is a member of the Betaherpesvirus family, with high seroprevalence rates among the human population, ranging from ∼50% in industrialized countries up to 100% in the developing world (1). Although the course of infection is usually asymptomatic in healthy individuals or is accompanied by only mild flu-like symptoms, CMV infection in immunocompromised persons often leads to serious problems, such as retinitis in AIDS patients or organ loss in transplant recipients, as well as sometimes life-threatening complications, like gastrointestinal disease, hepatitis, or pneumonia (reviewed in references 2 to 5). CMV is also the most frequent viral cause of malformations in newborns, leading to deafness or mental retardation (6). Since a protective CMV vaccine is still not available, the current therapy for CMV disease encompasses mainly nucleoside analogs, such as ganciclovir or valganciclovir, foscarnet, and cidofovir. However, these drugs are of limited efficacy, and upon long-term application they can cause severe side effects, such as neutropenia, thrombocytopenia, and renal dysfunction. Furthermore, CMV strains resistant to antiviral treatment can arise, which can no longer be controlled by the currently available therapeutic options. Consequently, there is a strong need for the development of additional antiviral medications.

The discovery of naturally occurring antimicrobial agents (e.g., the defensins) led to the realization that the body itself may produce an arsenal of substances active against various pathogens (reviewed in references 7 and 8). Peptides obtained from human tissues and body fluids may therefore constitute a valuable reservoir in the search for new starting points for drug design. To identify compounds that can be used as bases for novel antiviral therapies, we screened a peptide library derived from human hemofiltrate (HF) for inhibitory activities against CMV. This library contained highly concentrated and purified peptides isolated from 10,000 liters of HF obtained from patients undergoing dialysis due to renal dysfunction. The library exhibits a complexity of approximately 1 million different peptides smaller than 30 kDa (9–12), and the relative concentrations of the peptides to each other parallel those found in human plasma (10, 13). Systematic screening of peptide libraries for modulators of viral infection has already identified several peptides that modulate infection with the human immunodeficiency virus type 1 (HIV-1) (14–16).

Here, we report on the isolation and characterization of a 71-amino-acid (aa) peptide from human HF that blocked CMV infection. The peptide, termed CYVIP, turned out to be a derivative of the CXC chemokine neutrophil-activating peptide 2 (NAP-2). Investigation into the mechanism of action suggested that the inhibitory effect is conveyed by binding of the peptide to O-sulfated residues in heparan sulfate (HS) on the cell surface, thereby interfering with the attachment of CMV particles that use the same target structures. Structure-activity studies of CYVIP allowed us to narrow the inhibitory activity to a smaller N-terminal peptide comprising several cationic residues. Our data imply that CYVIP mimics the binding of CMV particles to HS.

MATERIALS AND METHODS

Isolation of CYVIP from human hemofiltrate.

Human blood ultrafiltrate (HF) was obtained from patients with chronic renal insufficiency. Hemofilters with a cutoff of 30 kDa were used, and the filtrate was immediately acidified with HCl to pH 3.0 and cooled to inhibit proteolysis. Peptides from 10,000 liters were extracted and processed as described previously (11). In brief, the extracts were pooled for the first separation step by using a 10-liter cation exchange column. Stepwise batch elution was performed using eight buffers with different pHs (increasing from pH 3.6 to pH 9 [pools 1 to 8]). The resulting pools (each equivalent to 15 to 25 liters of HF) were further separated using reverse-phase (RP) chromatography. Each pool was applied to a Source RP-C column (15 μm; 10 by 12.5 cm; Pharmacia), and separation was performed at a flow rate of 200 ml/min with an 8-l gradient from 100% A to 60% B (A was water, 10 mM HCl; B was 80% acetonitrile, 10 mM HCl). Fractions of 200 ml were collected, and we monitored the absorbance at 280 nm. Aliquots of these peptide bank fractions corresponding to a 5-liter equivalent of HF were lyophilized and used for the screening experiments described below.

Viruses and cells.

The HCMV strain TB40/E-EGFP, which expresses enhanced green fluorescent protein (EGFP) was generated by mutagenesis (see below) of the bacterial artificial chromosome (BAC)-cloned genome of the endotheliotropic HCMV isolate TB40/E (17) (a kind gift of C. Sinzger [University of Tübingen, Tübingen, Germany] and G. Hahn [Clinical Center of Ingolstadt, Ingolstadt, Germany]). The HCMV laboratory strain AD169-EGFP and the mouse CMV (MCMV) strain MCMV-EGFP were described previously (18, 19). For construction of a recombinant herpes simplex virus 1 (HSV-1), a clinical isolate (provided by S. Dewhurst, Rochester, NY) was BAC cloned in Escherichia coli, with simultaneous introduction of the EGFP open reading frame (ORF) controlled by the MCMV immediate-early promoter (E. M. Borst, unpublished data). Telomerase-immortalized microvascular endothelial cells (TIME) (20) were cultivated in EGM-2-MV Bullet kit medium (CC-3202; Cambrex Bio Science, Walkersville, MD). ARPE-19 retina pigmented epithelial cells (ATCC CRL-2302) were propagated in Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 (1:1; D6421; Sigma) with 5% fetal calf serum (FCS), 100 U/ml of penicillin, 100 μg/ml of streptomycin sulfate, 0.348% sodium bicarbonate, and 2 mM glutamine. The telomerase-immortalized human fibroblast cell line hTERT-BJ-1 (Clontech) was grown in a medium consisting of four parts of DMEM and one part of medium 199 supplemented with 5% FCS, 100 U/ml of penicillin, and 100 μg/ml of streptomycin sulfate. Murine embryonic fibroblasts were prepared from pregnant mice as described previously (21) and kept in DMEM with 10% FCS. Vero cells (African green monkey kidney epithelial cells) and HeLa cells were cultivated in the same medium, and CHO (Chinese hamster ovary) and the mutant CHO cell line pgsA-745 (22) were cultured in minimal essential medium supplemented with 5% FCS, 100 U/ml of penicillin, and 100 μg/ml of streptomycin sulfate.

BAC mutagenesis and generation of the TB40/E-EGFP mutant.

The TB40/E-EGFP mutant was constructed in E. coli by homologous recombination between the BAC-cloned virus genome (17) and a linear fragment comprising the EGFP ORF and the MCMV major immediate-early promoter plus a kanamycin resistance (knR) cassette flanked by Flp recognition target (FRT) sites (23). Primers to amplify the linear fragment were GFP3.for (5′-GGATCCAGACATGATAAGATACATTG-3′) and GFP3.rev (5′-GCGATCTATCGATGCATGCCATGGTACCCGGGAGCTCGAATTCGAAGCTTCTAGGACGACGACGACAAGTAA-3′), resulting in a PCR product that provided sequences homologous to the desired integration site in the TB40/E BAC, namely, the poly(A) signal sequences and the promoter region of the guanine phosphoribosyltransferase gene present within the BAC (17). Mutagenesis utilizing the phage λ recombination proteins Red-α, -β, and -γ and subsequent excision of knR with Flp recombinase were performed as described previously (23). Reconstitution of mutant virus from the recombinant BAC was done by adenovirus-mediated DNA delivery as reported previously (24).

Screening of the peptide library.

Peptide fractions were dissolved in 500 μl of sterile water by incubation at 4°C overnight. TIME cells were seeded in 96-well plates and incubated until they reached confluence. The peptide fractions were tested in triplicates by adding 30 μl of the respective peptide mixture to cells cultivated with 50 μl medium, followed by incubation at 37°C for 1 h. The TB40/E-EGFP virus was then added to the cultures at a multiplicity of infection (MOI) of 0.5, and cells were further kept at 37°C for an additional hour. Afterwards, the inocula were replaced with 100 μl of cell culture medium per well. Three days later, the cells were washed with phosphate-buffered saline (PBS) and fixed with 3% paraformaldehyde–PBS for 15 min at room temperature. EGFP expression in the cells was measured by fluorometry by using the Synergy 2 multimode plate reader (BioTek, Bad Friedrichshall, Germany). The background signal of noninfected cells was subtracted from each data point, and the EGFP signal obtained in the absence of peptides was set as 100%. Wells displaying a reduced EGFP signal in the fluorometric assay were additionally checked by light microscopy to assess cell viability and by fluorescence microscopy to examine EGFP expression at the single-cell level. Peptide fractions that led to a reduced number of infected cells in the absence of cytotoxicity were tested again for inhibition of HCMV infection by adding serial dilutions of the peptides to the cells. Peptide fractions reproducibly inhibiting infection in a dose-dependent manner were considered hits. Cytotoxicity of these fractions was tested by the 3-(4,5-dimethyl-2-thiozolyl)-2,5-diphenyl-2H-tetrazolium bromide cell viability assay (25). Fractions that led to decreased cell viability were excluded from further analysis.

Purification of peptides from hit fractions, peptide sequencing, and chemical synthesis of CYVIP.

Active fractions were further purified on an analytical cation exchange column [40 to 90 μm; 20 by 125 mm; Fractogel TSK SP(M); Merck, Darmstadt, Germany] using a gradient of 100% A to 100% B in 30 min (buffer A was 20 mM sodium phosphate [pH 7.2], and B was buffer A plus 1 M sodium chloride). Aliquots of these fractions were screened for inhibition of HCMV infection, and the resulting active fractions were applied to further analytical RP chromatographic steps on C₁₈ columns as follows: (i) a 5-μm, 30-nm, 1- by 25-cm column (Vydac, Hesperia, CA) and (ii) a 5-μm, 30-nm, 0.46- by 25-cm column (YMC, Schernbeck, Germany). Chromatography was performed with RP solvents using trifluoroacetic acid (TFA, 0.1%) as a counterion in the gradient elution. Active fractions were pooled and analyzed for inhibitory activity. The high-performance liquid chromatography (HPLC) equipment used was a Kontron (Neufahrn, Germany) system 400 with HPLC pumps 420 and HPLC detector 430. Mass analysis of HPLC fractions and purified peptides was performed with a Voyager (DE) Pro matrix-assisted laser desorption ionization–time of flight mass spectrometry apparatus (PerSeptive Biosystems, Freiburg, Germany), using α-cyanohydroxycinamic acid as a matrix and standard instrument conditions as described previously (26). Sequence analysis of the purified active peptide was carried out by conventional Edman sequencing. Solid-phase synthesis of peptide chains was carried out using 9-fluoroenylmethoxy carbonyl chemistry according to the methodology of Merrifield (27) on an ABI 433 synthesizer (Applied Biosystems) at room temperature. The crude peptide chain was purified by RP-HPLC, and the obtained product was analyzed by mass spectrometry and analytical HPLC.

pp65 uptake and attachment assays.

BJ-1 fibroblasts were incubated with CYVIP (80 μM) or without CYVIP for 1 h at 37°C. Then, the supernatant was removed and the cells were inoculated with HCMV for 15 min (MOI, 1). After 6 h of further incubation in fresh medium, cells were fixed with 3% paraformaldehyd–PBS for 15 min at room temperature and analyzed by immunofluorescence microscopy with a mouse monoclonal antibody against the HCMV tegument protein pp65 (dilution, 1:500; catalog number NB110-57244; Novus Biologicals, Littleton, CO) and an Alexa Fluor 488-labeled secondary anti-mouse antibody (dilution, 1:500; catalog number A11029; Molecular Probes). Cell nuclei were stained with TO-PRO-3. Pictures of randomly chosen areas of the slides were taken with a Zeiss Observer.Z1 microscope and further processed using Adobe Photoshop CS4. Total cell numbers were determined by counting TO-PRO-3-stained nuclei, and cells that had taken up the virus were assessed by counting pp65-positive nuclei. To examine HCMV attachment to cell surfaces, cells were incubated with CYVIP at different concentrations or without CYVIP for 1 h at 37°C. Then, the cells were cooled on wet ice, and virus (MOI, 1) was allowed to bind to the cells for 1 h at 4°C. Cells were washed five times with PBS at 4°C, lysates were prepared, and samples corresponding to ∼50,000 cells were loaded per lane on a 10% SDS-polyacrylamide gel. Immunoblotting was performed using the pp65 antibody (1:4,000 dilution) and a horseradish peroxidase-labeled anti-mouse antibody (catalog number NIF 825; Amersham Pharmacia). Signals were visualized using the ECL Advance detection kit (catalog number RPN2135; Amersham Pharmacia), and pictures taken with a Fuji LAS-3000 image reader were further processed using ImageJ 1.41 and Adobe Photoshop CS4.

Interference of sulfated polysaccharides and de-O-sulfated heparins with CYVIP and HCMV particles.

Heparin (catalog number H4784) and dextransulfate (catalog number D6924) were purchased from Sigma, and fucoidan was extracted from Fucus vesiculosus. 2-O-desulfated, 6-O-desulfated, and fully de-O-sulfated heparin was obtained from Neoparin Inc., Alameda, CA (catalog numbers GT6012, GT6013, and GT6011, respectively). CYVIP (80 μM) or HCMV particles (8,000 PFU/sample) were preincubated with the polysaccharides at different concentrations for 1 h at 37°C. The samples were then added to the cells and further incubated for 1 h at 37°C. The supernatants were aspirated, and either fresh medium was added (when the interaction with HCMV particles was analyzed) or cells were inoculated with HCMV (MOI, 0.5) for 15 min. EGFP expression of the cells was analyzed 3 days later by fluorometry. All experiments were done in triplicates.

RESULTS

Identification of a neutrophil-activating peptide 2 derivative as an inhibitor of HCMV infection.

Since HCMV is a blood-borne virus, we reasoned that HF may constitute a promising source for the identification of peptides that interfere with HCMV infection. TIME cells were incubated with a total of 384 different fractions of the HF peptide library and subsequently infected with the endotheliotropic HCMV strain TB40/E-EGFP expressing EGFP. Peptide fractions that led to a reduced EGFP signal in infected cells at day 3 postinfection (p.i.) and that did not display cytotoxicity were defined as hit fractions. Peptide fractions with the strongest inhibitory activities were found in pH pool 6 of the peptide library, which was eluted at pH 7.4 (Fig. 1A). Further consecutive purification of fraction 16 followed by peptide sequencing allowed us to pin down the inhibitory activity to a 71-aa peptide that was almost identical to neutrophil-activating peptide 2 (NAP-2), except that an additional tyrosine residue was present at the N terminus (Fig. 1B). We termed this peptide CYVIP, for cytomegalovirus-inhibiting peptide. NAP-2 is a CXC chemokine that belongs to the family of β-thromboglobulins, which are platelet α-granule-derived polypeptides (28, 29). To confirm that CYVIP was the peptide responsible for inhibition of HCMV infection, we synthesized the peptide with the additional N-terminal tyrosine. The synthetic CYVIP inhibited the infection of TIME cells with TB40/E-EGFP at least down to a concentration of 10 μM (Fig. 1C).

Fig 1.

Identification of hit fractions in the hemofiltrate-derived peptide library and properties of the inhibiting peptide. (A) Screening of fractions of the hemofiltrate pH pool 6 for antiviral activity against HCMV strain TB40/E-EGFP on the TIME cell line. Infection rates from triplicate cultures (means ± standard deviations) were determined by fluorometry using GFP expression of infected cells at day 3 p.i. The most efficient inhibition of infection was found with fractions 16 and 17. (B) Sequence of the inhibiting peptide purified from fraction 16 of pH pool 6. (C) Concentration-dependent effect of CYVIP. Cultures of TIME cells (triplicates) were incubated with the indicated concentrations of synthetic CYVIP followed by infection as described for panel A.

CYVIP inhibits infection of different cell types with several herpesviruses.

Next, we examined whether inhibition by CYVIP was cell type specific or dependent on a specific HCMV strain and whether CYVIP also blocked infection with other herpesviruses. As can be seen in Fig. 2, CYVIP interfered with the infection by the TB40/E strain of not only TIME cells but also ARPE epithelial cells and BJ-1 fibroblasts. It also blocked infection of fibroblasts by the HCMV laboratory strain AD169 and infection of murine fibroblasts with mouse CMV. Moreover, infection of Vero and HeLa cells with HSV-1 was affected by CYVIP. These experiments also revealed that the effective dose needed for inhibition varied between cell types, e.g., less CYVIP was needed to inhibit HCMV on endothelial cells than on epithelial cells or fibroblasts, and likewise, the effect of CYVIP on HSV-1 infection of HeLa cells was stronger than on infection of Vero cells (Fig. 2).

Fig 2.

CYVIP inhibits infection of several herpesviruses for different cell types. Cell cultures were preincubated with CYVIP in triplicate at the concentrations indicated, followed by infection with the indicated viruses. EGFP expression was measured on day 3 (HCMV) or day 2 p.i. (MCMV and HSV-1). Error bars indicate standard deviations. TIME, human microvascular endothelial cells; ARPE, human retina pigmented epithelial cells; BJ-1, human foreskin fibroblasts; MEF, murine embryonic fibroblasts; Vero, African green monkey kidney epithelial cells; HeLa, human cervix carcinoma epithelial cells; TB40/E, endotheliotropic HCMV strain; AD169, laboratory HCMV strain.

CYVIP acts on the cells, not on virus particles.

The cell type context of CYVIP's activity suggested that its mechanism of action operates on the cells rather than on virus particles. To discriminate between these possibilities, we preincubated HCMV particles with CYVIP, followed by dilution of the sample below the active concentration of CYVIP, before adding it to the cells. For this and subsequent experiments, we used the HCMV laboratory strain AD169 and human fibroblasts as target cells, since this is the most convenient HCMV infection system, although higher CYVIP concentrations were needed for fibroblasts than endothelial cells. Upon preincubation of HCMV particles with CYVIP, the infection rate of cells was comparable to that observed in the absence of the peptide (Fig. 3A), implying that CYVIP does not act directly on virions. As a positive control for inhibition, virus was preincubated with neutralizing antibodies, which completely abolished infection. Next, we preincubated fibroblasts with different amounts of CYVIP and removed the unbound peptide before infecting the cells. Inhibition of infection in a dose-dependent manner was observed, with 96% inhibition at the highest concentration of CYVIP tested (Fig. 3B), indicating that the mechanism of action of CYVIP is by targeting the cells.

Fig 3.

Mode of action of CYVIP. (A) HCMV particles were preincubated with medium (without CYVIP) or with CYVIP (80 μM) or with a neutralizing antiserum against HCMV. The samples were then diluted 16-fold below the active concentration of CYVIP and incubated with human fibroblasts for 3 h at 37°C, followed by replacement of the medium. EGFP expression of the cells (means from triplicates, ± standard deviations) was measured 3 days later. n.i., noninfected cells. (B) Fibroblasts were incubated with graded dilutions of CYVIP, before the medium was replaced and cells were infected with AD169-EGFP (MOI, 0.5). (C) CYVIP (80 μM) was added to fibroblasts and incubated with the cells at 37°C for the time periods indicated. At the given time points, the peptide was removed, and cells were infected and further processed as described for panel A. (D) Fibroblasts were incubated with CYVIP (80 μM) for 1 h at 37°C. Then, the peptide was removed, and cells were infected after the indicated times. EGFP expression in the cells was measured 3 days later by fluorometry.

To learn about the kinetics of the inhibitory activity of CYVIP, cells were incubated with the peptide, followed by infection at different time points thereafter. Figure 3C demonstrates that preincubation with CYVIP for only 15 min was already sufficient to inhibit infection to ∼80%. In order to test how long the inhibitory effect of CYVIP persisted, the peptide was incubated with the fibroblasts for 1 h before the medium was exchanged, and cells were infected at different time points thereafter (Fig. 3D). Even after 7 h, infection was still inhibited to about 40%. This shows that the inhibitory effect of CYVIP is rapidly accomplished and then persists for a remarkable time period.

CYVIP interferes with uptake of HCMV into cells.

The drastically reduced number of infected cells observed in the presence of the peptide could arise from inhibition of virus binding to the cells, or from blocking of the fusion of viral and cellular membranes, impaired intracellular transport of capsids to the nucleus, or impeded viral gene expression. To study whether CYVIP interferes with virus entry, cells were incubated with or without the peptide prior to infection. Six hours later, translocation of the HCMV structural protein pp65 to the nucleus was investigated by immunofluorescence microscopy. pp65 is an abundant HCMV tegument protein that upon entry of virus particles into cells rapidly migrates to the nucleus. pp65-positive nuclei can therefore be used as a surrogate marker for virus uptake. As can be seen in Fig. 4A, the number of cell nuclei that stained positive for pp65 was markedly diminished when the cells were pretreated with CYVIP, indicating that CYVIP blocks virus entry into cells.

Fig 4.

CYVIP interferes with HCMV entry into cells. (A) Fibroblasts were incubated with or without CYVIP (80 μM) and then infected with HCMV. Six hours later, the cells were stained with an antibody specific for the viral tegument protein pp65 (left). The diagram shows the percentage of pp65-positive nuclei determined by analyzing 10 images each of cells incubated without CYVIP (n = 726) or with the peptide (n = 765) (the error bars indicate the standard deviations of the percentage of pp65-positive nuclei observed between individual images). To-Pro-3, DNA stain. (B) Fibroblasts were incubated with decreasing concentrations of CYVIP (lanes 4 to 6) or without CYVIP (lanes 1 and 2) for 1 h at 37°C and then cooled on wet ice. Virus was added and allowed to adsorb to the cells for 1 h at 4°C. In the heparin lane, HCMV particles were preincubated with heparin (5 μg/ml) before being added to the cells. The amount of the viral tegument protein pp65 associated with cells was assessed by immunoblotting. The signals resulting from cross-reactivity of the antibody with several cellular proteins migrating below the viral pp65 protein served as loading controls. n.i., lysate of noninfected cells.

Two scenarios can account for the impaired virus uptake: either (i) infection is inhibited at the level of virus binding, or (ii) the subsequent fusion of the viral envelope with the plasma membrane is affected. We therefore investigated binding of HCMV to the cells in the presence of CYVIP by using an attachment assay based on the detection of the pp65 protein associated with the cells. Fibroblasts were treated with different amounts of CYVIP, followed by incubation with virus at 4°C. At this temperature, virus particles can bind to the cells but are not taken up. As a control, virus particles were preincubated with heparin before being added to the cells, a treatment that is known to inhibit virus attachment (30). Immunoblotting revealed that in the presence of CYVIP concentrations that inhibit HCMV infection (see Fig. 2), the levels of cell-associated pp65 were substantially reduced compared to cells incubated without the peptide or with CYVIP at a noneffective concentration (Fig. 4B). We concluded that CYVIP exerts its inhibitory effect already during the attachment of virus particles to the cells. When CYVIP was added after the cells were infected with HCMV, plaque numbers or sizes did not differ in comparison to cultures that were kept without the peptide, and quantification of the EGFP signals revealed no differences (data not shown). Thus, subsequent to virus entry, CYVIP does not exert additional inhibitory effects on the viral infection cycle.

Structure-activity relationship of CYVIP.

It is known that binding of HCMV particles to cells occurs first via interaction with negatively charged HS on the cell surface, followed by subsequent interaction of the virus with a specific receptor(s) (1, 31). CYVIP exhibits a high degree of sequence similarity to platelet factor 4, which is known to bind to heparin and HS (32), and the arginine and lysine residues of platelet factor 4 involved in heparin binding are located at almost identical positions as those in CYVIP (Fig. 5A). We reasoned therefore that CYVIP binds to HS and occupies the viral attachment sites. To find out which residues are required for inhibition of infection, CYVIP was subjected to structure-activity relationship analysis. We first synthesized the C-terminal (frg 1) and N-terminal (frg 3) parts of CYVIP, and we found the N-terminal 34 aa (frg 3) to be sufficient for inhibition (Fig. 5B, compare frg 1 and frg 3). The ELR motif present within the N-terminal part of CYVIP has been implicated in binding of chemokines to the CXCR-1 and CXCR-2 receptors and their triggering (33–35). frg 2, which lacks the first 5 aa containing the ELR motif, was not active, and a fragment comprising only the 10 N-terminal amino acids of CYVIP (frg 4) did not inhibit infection either (data not shown). This suggested that the residues YAELR contribute to the inhibitory activity but are not sufficient. frg 11 lacked the N-terminal tyrosine, which differentiates CYVIP from NAP-2. The diminished potency of frg11 compared to frg 3 indicated that the tyrosine contributes to the activity of CYVIP but is not absolutely required. Replacement of the arginine residue with alanine within the ELR motif abolished the inhibitory function of CYVIP, whereas replacement of the glutamic acid turned out to be beneficial to inhibition (Fig. 5B, frg 12 and frg 13). This result argues against an involvement of chemokine receptor activation and rather suggests that the inhibitory activity is mediated by the positively charged residues. A peptide 5 aa shorter at the C terminus than frg 3 still inhibited infection (Fig. 5B, frg 14). Replacement of each of the lysine residues with alanine (frg 15 to 17) abrogated the activity of the resulting peptides, which was in line with the suggested role of these residues in HS binding. frg 18, which combines the favorable alanine substitution of frg3 with the small size of frg 14, displayed inhibitory properties even superior to the original CYVIP peptide.

Fig 5.

Structure-activity relationship of CYVIP. (A) Alignment of the amino acid sequence of CYVIP and platelet factor 4 (PF4) is shown, with the arginine and lysine residues indicated in bold. (B) Peptide fragments 1 to 18 (frg 1 to frg 18) were synthesized and tested for their abilities to inhibit HCMV infection. The ELR motif is underlined, and alanine substitutions are indicated in bold and underlined. The cationic lysine residues in fragments 14 to 18 are shown in bold. The peptides were incubated in decreasing concentrations with human fibroblasts for 1 h at 37°C. Then, the unbound peptide was removed, and the cells were infected with the AD169-EGFP strain (MOI, 0.5). Three days later, EGFP expression was measured by fluorometry, and the mean signals of triplicate cultures (± the standard deviations) are depicted in the diagram. Activities of the peptides were also scored according to the following criteria: −, no inhibition; +, inhibition at 80 μM only; ++, inhibition down to 40 μM; +++, inhibition down to 20 μM.

The activity of CYVIP is reversed by pretreatment with heparin and other sulfated polysaccharides.

Since it is known that HCMV attachment occurs via HS and our results indicated that the positively charged residues of CYVIP are crucial for inhibition, we hypothesized that CYVIP blocks HCMV attachment by occupying negatively charged HS. If true, pretreatment of CYVIP with negatively charged compounds displaying structural similarity to HS, such as heparin, should counteract its inhibitory activity on HCMV infection. To distinguish potential effects of heparin on CYVIP from direct effects of heparin on cells or virus particles, control experiments were performed with heparin alone. When CYVIP frg 13 pretreated with heparin was applied to cells prior to inoculation with HCMV, infection occurred and depended on the amount of heparin used (Fig. 6A, group 1), i.e., heparin neutralized the inhibitory activity of CYVIP. Treatment of cells with heparin alone had little effect on infection with HCMV (Fig. 6A, group 2); at most, it partially inhibited infection at the highest heparin concentration tested. Preincubation of HCMV particles with heparin blocked infection in a dose-dependent manner (Fig. 6A, group 3). The latter finding is in line with the results in previous reports (30), and it is explained by the binding of heparin to HCMV particles, preventing their binding to HS and subsequent uptake into cells. The small effect seen after pretreatment of cells with heparin (Fig. 6A, group 2) can be explained by the same mechanism, if we assume that residual heparin remaining on the cells following replacement of the medium neutralized HCMV particles. Taken together, we conclude from these experiments that heparin interacts with CYVIP in a manner similar to that with HCMV particles, thereby interfering with its inhibitory activity.

Fig 6.

Interaction of CYVIP and HCMV particles with sulfated polysaccharides. (A) Heparin in decreasing amounts was preincubated with CYVIP fragment 13 (group 1) or with medium only (group 2) for 1 h at 37°C before being added to the cells. After 1 h, the cells were infected with HCMV. For group 3, HCMV particles were preincubated with heparin at the indicated concentrations before being incubated with fibroblasts. The percentage of infected cells from triplicate cultures represents the EGFP signal (± the standard deviation) of cells at 3 days p.i. in comparison to cells infected in the absence of the peptide. (B and C) Dextran sulfate (B) and fucoidan (C) were incubated with CYVIP (group 1), cells (group 2), or HCMV particles (group 3), and effects were assessed as described for heparin for panel A.

After preincubation of CYVIP with the anionic polysaccharides dextran sulfate or fucoidan and after pretreatment of cells and virus particles, similar results to those observed for heparin were obtained, except that the potency of the compounds differed slightly (cf. Fig. 6B and C versus A). The only qualitative difference was that pretreatment of cells with dextran sulfate led to stronger inhibition of infection than heparin and fucoidan. This suggests that dextran sulfate binds more strongly to the cell surface than the other substances tested. Since the polysaccharides used differ in their saccharide backbone, the data suggest that the interference with CYVIP relies primarily on the sulfation common to the compounds, i.e., their negative charge. Altogether, these results are in accordance with the hypothesis that CYVIP binds to HS.

Interference of desulfated heparins with the activity of CYVIP and with HCMV infection.

HS molecules become sulfated at different positions during biosynthesis (36). It seemed likely that these negatively charged moieties mediate the interaction with the positively charged CYVIP or HCMV particles. The same assay as before was applied to test the ability of heparin derivatives that lacked either the 2-O-sulfate residues, the 6-O-sulfate moieties, or all of the O-linked sulfate modifications, in order to (i) neutralize the inhibitory activity of CYVIP and (ii) neutralize the infectivity of HCMV particles. As is shown in Fig. 7A, pretreatment of CYVIP frg 18 with 2-O-desulfated or 6-O-desulfated heparin affected its inhibitory activity, i.e., a substantial fraction of the cells became infected with HCMV. Notably, pretreatment of frg 18 with fully desulfated heparin had no effect, as frg 18 still blocked HCMV infection to more than 90% (Fig. 7A). None of the de-O-sulfated heparins exerted a marked effect on virus infection when incubated with the cells alone (Fig. 7B). These data suggest that the 2-O- and the 6-O-linked sulfate residues mediate interaction with CYVIP, whereas fully de-O-sulfated heparin does not interfere with CYVIP.

Fig 7.

Interference of de-O-sulfated heparins with CYVIP and with HCMV particles. (A) CYVIP frg 18 was preincubated with 2-O-desulfoheparin, 6-O-desulfoheparin, or with fully de-O-sulfated heparin before being added to the fibroblasts for 1 h. Then, triplicate cultures of cells were infected with HCMV and tested for EGFP expression on day 3 p.i. Error bars indicate standard deviations. (B) The de-O-sulfated heparins were preincubated with medium before being applied to the fibroblasts. Further analysis was as described for panel A. (C) HCMV particles were preincubated with the indicated heparins for 1 h at 37°C before being incubated with the cells.

Next we analyzed the ability of the partially or fully de-O-sulfated heparins to neutralize the infectivity of HCMV particles. 2-O-desulfated heparin blocked infection in a dose-dependent manner, with 90% inhibition at the highest concentration tested (Fig. 7C). In contrast, preincubation of virus particles with 6-O-desulfated heparin had only a small effect on the infection rate (Fig. 7C). Fully de-O-sulfated heparin could not neutralize HCMV particles (Fig. 7C). Thus, the 6-O-sulfate moieties seem to be more important than 2-O-sulfate moieties. In conclusion, interaction of both HCMV and CYVIP with heparin requires O-sulfation. However, both 2-O- and 6-O-linked sulfate residues mediate interaction with CYVIP, whereas there seems to be a preference in binding of HCMV particles to 6-O-linked sulfate molecules.

Virus binding studies with cells deficient in HS biosynthesis.

To further test our hypothesis about the mechanism of action of CYVIP, we made use of the mutant CHO cell line pgsA-745, which is deficient for the synthesis of HS (and chondroitin sulfate) due to the absence of xylosyl transferase (22). We hypothesized that in comparison to wild-type CHO cells, the attachment of HCMV particles would be reduced in pgsA-745 cells due to the lack of HS and, moreover, CYVIP would not be able to influence virus binding. The HCMV attachment assay described above, utilizing the viral pp65 protein, was applied. CHO and pgsA-745 cells were incubated with different amounts of CYVIP frg 13, followed by virus binding to the cells at 4°C. Immunoblotting revealed that frg 13 blocked HCMV attachment to CHO cells in a dose-dependent manner (Fig. 8, upper part). Virus binding to pgsA-745 cells seemed to be weaker than to CHO cells and, most importantly, was not modulated by CYVIP (Fig. 8, lower part). These data demonstrate that on pgsA-745 cells the target structure for CYVIP is absent.

Fig 8.

Interaction of CYVIP and HCMV with cells deficient in heparan sulfate biosynthesis. Attachment of HCMV to wild-type CHO cells (upper part) or to pgsA-745 (mutant CHO cells deficient in HS biosynthesis [lower part]) in the presence or absence of CYVIP frg 13. The experimental procedure was as described in the legend for Fig. 4B. The cross-reactivity of the pp65 antibody with a cellular protein (lower band) served as an internal loading control. mock, medium instead of virus was added to the cells.

DISCUSSION

In an attempt to define peptide inhibitors of HCMV infectivity, we screened a peptide library from human hemofiltrate and found a peptide that affects entry of HCMV into cells and has not been implicated in antiviral activity before. Its mode of action is by binding to cells; it does not interact with virus particles and does not directly neutralize their infectivity. Blocking of the binding sites for HCMV particles of HS, a known herpesviral attachment receptor, is most likely the mechanism by which CYVIP inhibits infection. Structure-activity relationship studies revealed that the N-terminal 29 aa of CYVIP are sufficient for the activity of the peptide, and among these the cationic residues are crucial. Inhibition of the infection by CYVIP could be reversed by different kinds of O-sulfated polysaccharides, whereas HCMV particles seemed to use preferentially the 6-O-linked sulfate moieties in HS for initial binding to cells.

By screening a peptide library derived of human HF, we identified a previously unknown variant of NAP-2 that inhibits HCMV infection. About 10 years ago it became clear that some chemokines (and precursors or processed products of them) possess direct antimicrobial properties (37–39). One example is the polypeptides related to NAP-2 that are active against bacteria and fungi (40, 41). To the best of our knowledge, CYVIP is the first of this polypeptide family that displays a distinct antiviral activity. CYVIP differs from NAP-2 by an N-terminal tyrosine and, interestingly, this residue was found to be important for its inhibitory effect. In contrast, chemokine properties and triggering of the CXCR-1 and CXCR-2 receptors did not play a role, as mutagenesis of the ELR motif of CYVIP to ALR did not abolish, and instead improved, antiviral activity.

Our data show that CYVP acts on the target cells and not on virus particles and that it interferes early with attachment of the virus to the cell surface. It is known that HCMV infection starts with tethering of the virus particles to negatively charged HS proteoglycans on the cell surface (30), and this helps the virus to engage its specific receptor(s) (31). Obviously, viral attachment is a crucial step, as it was shown that removal of HS from the cell surface by treatment with HS-degrading enzymes led to a drastic reduction in HCMV binding and infectivity (30, 42, 43). This conclusion is supported by our finding that CYVIP concentrations, that blocked virus attachment were also effective in inhibiting infection. Interference with virus attachment is therefore a promising target for antiviral intervention (44–47).

HS proteoglycans are expressed in various cell types and tissues, with tremendous heterogeneity with respect to different isoforms of the core proteins and their expression levels, variable lengths of the glycosaminoglycan side chains, and the degrees of sulfation, acetylation, and epimerization (36, 48). These differences could explain the different effective doses of CYVIP observed for the various cell types. Moreover, CYVIP may also bind to other negatively charged molecules on the cell surface, such as chondroitin sulfate, whose amounts may vary between different cell lines, too. Since HCMV attaches preferentially to HS (30), binding of CYVIP to chondroitin sulfate would not influence HCMV infection directly; however, CYVIP could be sequestered, leading to less efficient blocking of HCMV attachment to HS. Whether inhibition of HCMV infection by CYVIP is a natural defense mechanism in vivo remains an open question. Precursors of CYVIP and NAP-2 are synthesized in large quantity in megakaryocytes and platelets (28, 49). Perhaps CYVIP may reach local concentrations that are effective against HCMV under inflammatory conditions. This question has to be the subject of further investigation.

Besides the fact that HCMV attaches to HS (30), we have provided additional data suggesting that CYVIP blocks the infection via this target structure on the cell surface. CYVIP could not interfere with virus attachment to cells that lacked HS (and other glycosaminglycans), and CYVIP's activity could be reversed by compounds such as heparin that share characteristics with HS. This conclusion is further supported by the following findings of others: (i) sequestration of chemokines by glycosaminoglycans present on cells or within the extracellular matrix seems to be common (50), and (ii) CYVIP displays a high degree of homology to platelet factor 4 (Fig. 5A), which has heparin- and HS-binding properties (32) and is known to inhibit HSV-1 infection (51). To the best of our knowledge, the activity of PF4 against CMV has not been analyzed. Particularly notable is the presence of arginine and lysine residues in CYVIP at almost identical positions as in platelet factor 4, which mediate heparin binding of this factor (32). Upon exchange of these cationic residues for alanine, the inhibitory activity of CYVIP was lost. However, the presence of cationic residues alone was not sufficient, as the C-terminal part of CYVIP, which contains such amino acids as well, did not confer any inhibitory effect. This suggests that additional structural features of the N-terminal domain are required. Interestingly, the antimicrobial activity of another polypeptide related to NAP-2, namely, TC-1, also mapped to the N-terminal part (52). Although the modes of action of TC-1 and CYVIP are probably different, it is remarkable that the activities are mediated by the same domain.

Our results suggest that the sulfate moieties of HS are required for the inhibition of HCMV infection by CYVIP. This conclusion is supported by the observation that coincubation of CYVIP with heparin and other sulfated polysaccharides overcame its antiviral activity. Whereas these polysaccharides differ in their sugar backbones, sulfation is a common characteristic of all of them. We propose that CYVIP mimics HCMV particles in binding to HS. As HS carries sulfate groups at various positions of its sugar backbone (36), we asked which kind of sulfation is involved in binding of HCMV particles and in the activity of CYVIP. Previous studies performed with HSV-1 revealed that its attachment to HS involves both 2,3-O- and 6-O-sulfated residues (53), and HS sulfated at position 3 of glucosamine can even act as an entry receptor for HSV-1 (54). For HCMV, this has not been investigated before. Our data provide the first hint for a role of 6-O-sulfation in HCMV attachment to HS. Further studies are required to substantiate this finding, for instance, an analysis of HCMV binding to cell lines in which the expression of different sulfotransferases has been knocked down. CYVIP displayed less specificity with respect to sulfation than HCMV particles, as both 2-O- and 6-O-linked sulfate groups seemed to contribute to the interaction. It has been shown that cationic amino acids in herpesvirus envelope glycoproteins mediate the interaction with HS (55). Design of peptides that mimic the function of domains of herpesviral glycoproteins in cell entry (56) is therefore an alternative to the approach taken here. Isolation of naturally occurring peptides has its own appeal as, for instance, new principles of interference may be revealed.

We think that CYVIP can serve as a promising starting point for the development of more potent compounds with higher efficacy against CMV, especially when a more detailed understanding of the structural requirements for interactions of CMV glycoprotein domains with HS becomes available. Sulfated polysaccharides have been discussed as antimicrobial drugs (46, 57–59), and the potential of compounds that interfere with virus attachment to HS proteoglycans as novel antiviral agents has been investigated (44, 45, 47, 56, 60, 61). Most importantly, such compounds would be effective against viruses that have acquired resistance to currently available antiviral medications.