Anticytomegalovirus Peptides Point to New Insights for CMV Entry Mechanisms and the Limitations of In Vitro Screenings

In the absence of an effective vaccine to prevent HCMV infections, alternative interventions must be developed. Prevention of viral entry into susceptible cells is an attractive alternative strategy. Here we report that heparan sulfate-binding peptides effectively inhibit entry into fibroblasts of in vitro-derived CMVs and partially inhibit in vivo-derived CMVs. This includes the inhibition of urine-derived HCMV (uCMV), which is highly resistant to antibody neutralization. While these antiviral peptides are highly effective at inhibiting cell-free virus, they do not inhibit MCMV cell-to-cell spread. This underscores the need to understand the mechanism of cell-to-cell spread and differences between in vivo-derived versus in vitro-derived CMV entry to effectively prevent CMV’s spread.

H uman cytomegalovirus (HCMV) is a significant pathogen within immunocompromised groups. Disease in these populations can result from primary infection or spontaneous latent virus reactivation (1,2). As 60 to 90% of adults are latently infected with HCMV, there is a substantial population at risk for complications if their immune system becomes compromised (3,4). HCMV infection/reactivation in immunocompromised persons can result in mononucleosis-like symptoms, interstitial pneumonia, gastroenteritis, retinitis, or organ transplant rejection in transplant patients (1,3). HCMV is also the leading cause of congenital disease (5,6). In utero infection may result in fetal abnormalities such as microcephaly or severe sequelae that can evolve over time in the form of progressive deafness, mental retardation, or learning disabilities (7,8). HCMV infections impose a yearly 1-to 2-billion-dollar economic burden; therefore, development of effective treatment and preventive strategies is a high priority (5,9). Because there is no effective vaccine, treatment of infected immunocompromised patients primarily consists of nucleoside analogs such as ganciclovir (GCV), foscarnet, or cidofovir which inhibit DNA replication (10)(11)(12). Unfortunately, GCV treatment can be myelosuppressive, while foscarnet and cidofovir are nephrotoxic (13). All DNA polymerase inhibitors select for resistant HCMV mutants, and cases of GCV-resistant HCMV infections are on the rise (1,14,15). This has led to the development of novel treatments such as the recently FDA-approved terminase inhibitor, letermovir (16).
Antiviral peptides (APs) are an attractive alternative treatment for inhibiting viral infections. Indeed, peptide therapeutics are being investigated for respiratory viruses and HIV (17)(18)(19). APs have different mechanisms for virus inhibition from inhibiting viral attachment, entry, replication, or egress (20). HCMV attaches to a host cell via heparan sulfate proteoglycans (HSPGs) (21). Viral glycoproteins gB and gM/gN initially interact with negatively charged sulfate moieties, which serve to "dock" the HCMV virion to the host cell (21). Docking triggers a signal cascade within the cell allowing for subsequent viral entry. HSPGs are ubiquitously expressed on most host cells, supporting the idea that HCMV can infect almost any human cell type (22).
HSPGs have a myriad of functions, including binding chemokines and cytokines and serving as scaffolds for ligand receptors, growth factors, and other cell adhesion molecules (23). Cell surface HSPGs are also major components of host-mediated endocytosis and cell membrane fusion processes. HSPG functions have been exploited for malarial and viral infections, including HCMV and herpes simplex virus 1 (24)(25)(26). Because of their major role in the early stages of HCMV replication, heparan sulfates (HSs) are an attractive target for intervention. HS-binding peptides effectively inhibit HCMV infection (27). However, these peptides were not tested against the more virulent in vivo-derived virus or in an in vivo setting (28).
We have previously reported that synthetic heparin-binding peptides bind pathological amyloid deposits in vitro and in vivo (29,30). As HCMV attaches to cells via HS, we investigated whether these peptides could inhibit virus attachment. In this study, we demonstrate that these synthetic polybasic peptides are efficient at inhibiting viral entry of tissue culture-derived HCMV and murine cytomegalovirus (MCMV). We also provide evidence of effectively inhibiting an HCMV clinical isolate obtained from infected bodily secretions. However, these peptides could not prevent cell-to-cell spread of MCMV, potentially explaining the need to further investigate additional antiviral peptides for efficiency in vivo.

RESULTS
Peptide characteristics. Three polybasic peptides, designated p5 (coil) , p5 (coil)D , and p5 ϩ 14 (coil) , were synthesized using a glycine-rich backbone to enhance flexibility of the peptide chain ( Table 1). The p5 (coil) peptide is the parental peptide (31,32) from which the derivative p5 (coil)D and p5 ϩ 14 (coil) peptides were designed. p5 (coil)D is the D form of p5 (coil) . Because D-form peptides are more proteolytically stable and are equally effective as L-form peptides at inhibiting HCMV entry, we focused on p5 (coil)D for the majority of this study (33). We also utilized the peptide p5 ϩ 14 (coil) , which is p5 (coil) with an additional repeat of the last 14 amino acids. The addition of 14 amino acids has been shown to increase the efficacy of peptide-induced HCMV inhibition (27). At a peptide concentration of 50 M, both p5 (coil)D and p5 ϩ 14 (coil) inhibited HCMV and MCMV infection of fibroblasts (Table 1). We chose 50 M as the initial concentration for the screening the peptides because another polybasic D-form peptide could inhibit MCMV in vivo at this dose (33). All three peptides were predicted to adopt a flexible coil secondary structure, which is different from previously published peptides and may increase their efficacy (34,35).
As p5 (coil) was the peptide from which the others were generated, we tested the binding of a biotinylated variant to a panel of HS moieties using a synthetic glycoarray (Fig. 1A). p5 (coil) bound significantly more effectively to sulfated glycans (black bars) than to unsulfated glycans (red bars) (Fig. 1A), with the exception of the 5-sugar HS008 glycan. Statistical analysis showed significantly enhanced binding of p5 (coil) to almost all sulfated glycans relative to nonsulfated species (see Table S1 in the supplemental material). In general, no significant difference was observed between levels of peptide binding to the structurally different sulfated HSs. Figure 1B lists the structure of the glycans used in the glycan array. These results highlight that p5 (coil) preferentially binds sulfated glycans.
Efficacy of inhibition of CMV infection. The blockade of HCMV and MCMV attachment to cells was studied in the presence of increasing concentrations of peptide  These data indicate that our peptides are capable of inhibiting both HCMV and MCMV; however, MCMV is inhibited to a lesser extent. Peptide inhibition of infection of mice was evaluated using the MCMV mouse model (Fig. 3). BALB/c mice were pretreated with p5 (coil)D or p5 ϩ 14 (coil) at 250 g per mouse 1 h prior to infection. Evaluation of the viral titer in spleens harvested 4 days postinfection (dpi) indicated no significant difference in viral burden (Fig. 3). In order to confirm peptide was present at the time of infection and following infection, we evaluated peptide biodistribution postadministration (see Table S2 in the supplemental material). Our biodistribution assay confirms that p5 (coil)D or p5 ϩ 14 (coil) is present within the host at the time of infection and following infection. While there is a difference in biodistributions between the two peptides, there is no difference in levels of viral dissemination to the spleen. We have previously reported the inability of p5R D , another antiviral polybasic peptide, which is significantly more stable in vivo, to substantially inhibit infection of any primary dissemination organs (e.g., spleen, liver, and lung) (33). The inability of these peptides to reduce viral load in vivo could be due in vivo dosage/timing effect, but an alternative explanation is that the peptides differ in their ability to block in vivo-derived virus versus in vitro-derived viruses. Previous studies have reported differences between MCMVs grown in culture compared with those harvested in vivo, which was related to their HSPG usage for entry (36)(37)(38)(39)(40).

Peptide inhibition of in vivo-and in vitro-derived MCMV.
To evaluate the possibility of differential inhibition between in vitro-derived virus and in vivo-derived virus, we performed plaque reduction assays using MCMV salivary gland-isolated virus (SGV) and MCMV passaged on cultured cells (tissue culture-derived virus [TCV]) ( Additionally, in both cases p5 ϩ 14 (coil) was significantly more efficacious than p5 (coil)D . When used at 50 M, p5 ϩ 14 (coil) inhibited murine TCV (75%) more effectively than SGV (40%) (Fig. 4C). Our data support observations made by Ravindranath and Graves (36) and indicate SGV and TCV use different entry strategies. One entry mechanism is inhibited by peptides (i.e., TCV), and the other is only partially blocked (i.e., SGV).
To further investigate the differences in SGV and TCV entry identified by the peptide inhibition studies, mouse embryonic fibroblasts (MEFs) were treated with 50 mM sodium chlorate prior to infection to remove 2-O-and 6-O-linked HS sulfations (41). We focused on these sulfation patterns based on observations from HCMV, which indicated that these O-linked sulfations were important for viral attachment (28). This treatment resulted in inhibition of infection of both SGV and TCV, with the latter being significantly more impacted (Fig. 4D). It is known that incubation of MCMV with heparin blocks cellular entry; therefore, we studied the effect of increasing heparin concentration on infection efficiency of TCV (Fig. 4E) and SGV (Fig. 4F). Pretreatment of TCV with heparin resulted in a dose-dependent decrease in infection, with 50% loss of efficiency in the presence of 40 g/ml heparin (Fig. 4E). In contrast, there was no significant decrease in the infectivity of murine SGV following pretreatment with 40 g/ml heparin (Fig. 4F). Because viruses derived from different tissues vary in their susceptibility to antibody neutralization (37), we speculated that perhaps not all in vivo-derived MCMVs would be resistant to peptide inhibition. Therefore, we infected mice subcutaneously in the footpad, and SGV was isolated on day 14 postinfection, splenic virus (SPV) at day 5, and the footpad virus (FPV) at day 3. The different time points were necessary to maximize viral load in the given organ. Once MCMV organ titers were determined, whole-organ homogenates were plated at ϳ100 PFU on peptide-treated MEF 10.1 cells (Fig. 5A). As expected there was some inhibition of in vivo-grown virus, but not to the same levels as TCV. Interestingly, there was no difference between the different organs in regard to  peptide inhibition susceptibility. Because there is no difference between viruses isolated from different organs, the SGV phenotype is not the result of SGV sample preparation.
We speculated that the differences in susceptibility to the peptides correlated with differences in entry complexes that determine cellular tropism (42). MCMVs grown on macrophages contain different entry complexes compared to MCMVs grown on fibroblasts (42). Therefore, we tested whether MCMVs derived from macrophages in vitro would mimic peptide inhibition of SGV or TCV. In vitro-derived MCMV was grown on the macrophage cell line RAW 264.7 or bone marrow-derived macrophages (BMDMs). The progeny viruses were subjected to a plaque reduction assay (Fig. 5B). Peptide treatment of cells blocked RAW 264.7-grown virus (ϳ75% inhibition) but only partially blocked BMDM virus (ϳ50% inhibition). Not only are BMDMs capable of generating "in vivo-like" virus, but when SGV virus was grown on BMDMs, the peptide was even less efficient in blocking MCMV infection (Fig. 5C). Regardless of the initial MCMV input into BMDMs, they produced "in vivo-like" virus ( Fig. 5B to D). Additionally, these results indicate that the phenotype observed with SGV and virus isolated from other tissues is not an artifact of sample preparation.
Peptide inhibition of in vivo-derived HCMV. We sought to determine whether p5 ϩ 14 (coil) could inhibit infection of an HCMV clinical isolate. This virus was used directly from a patient's urine (11). This is important because the tropism/entry complex can mutate after a single passage in vitro (43,44). Peptide p5 ϩ 14 (coil) significantly inhibited the entry of a clinical isolates of HCMV into human fibroblasts (ϳ70% inhibition [ Fig. 6A]), but was significantly more effective when tissue culture-grownTB40/E virus was used (ϳ90% inhibition). Because these results mimicked MC-MV's peptide inhibition, we tested which HS moieties were important for viral entry. MRC-5 fibroblasts were treated with 50 mM sodium chlorate to remove 2-O and 6-O sulfations (Fig. 6B). Despite the small but significant difference between HCMVs from in vitro-or in vivo-grown viruses, removing the 2-O and 6-O sulfations prevented infection of both HCMVs, albeit more inhibition of tissue culture-grown HCMVs than the clinical isolate was observed. These results corroborate the MCMV data. Both viruses highly rely on sulfated surface glycans for entry.
Do polybasic peptides prevent MCMV cell-to-cell spread? Cell-to-cell spread in vivo is important for viral dissemination within the host (11,45,46). We evaluated the effectiveness of p5 (coil)D and p5 ϩ 14 (coil) on cell-to-cell spread. effect on cell-to-cell spread (Fig. 7A). Because most of the PEC population consists of immune cells (data not shown), we tested our peptide's ability to inhibit cell-to-cell spread from infected to uninfected MEF 10.1 cells in a plaque reduction assay. As seen with the PECs, peptide treatment did not inhibit cell-to-cell spread regardless of cell type (Fig. 7B). To evaluate whether or not HS mediates cell-to-cell spread, we treated murine fibroblasts with heparinase, sodium chlorate, and heparin. Cell-associated virus spread was marginally affected in the absence of HS, pointing to a different entry mechanism from cell-free MCMV, as heparin treatment was ineffective at inhibiting infection (Fig. 7C). These data highlight the complexity of CMV entry whether the virus enters from cell-to-cell spread or cell-free virus.

DISCUSSION
Decades of investment and innovation have failed to generate a protective HCMV vaccine (10,11). Without a vaccine, treatment of HCMV relies on drugs that target two different stages of HCMV replication. First, nucleoside analogues (e.g., GCV, foscarnet, and cidofovir) have been the standard treatment for HCMV disease, but these compounds are myelosuppressive and nephrotoxic. The recently approved letermovir (trade name Prevymis), which inhibits the terminase complex, is not myelosuppressive or nephrotoxic (16). All of these agents potentially select for resistant viruses (15,47). Therefore, a multifaceted approach may be required to inhibit additional stages during HCMV replication, and this could include APs that block viral entry. Other anti-CMV APs inhibit viral entry of in vitro-derived CMV, but none have been shown to inhibit a clinical isolate of HCMV (27,28,33,48). We demonstrated that our p5 (coil) peptides are capable of inhibiting infection of both HCMV and MCMV and that these peptides can partially inhibit infection of in vivo-derived CMVs. This is the first report of an effective inhibition of HCMV isolated directly from urine. Clinical isolates have previously been shown to be resistant to anti-HCMV antibody neutralization (49). Although inhibition of in vivo- derived virus was not as efficient as the blockade of tissue culture-derived virus, greater than 50% reduction in infectivity was achieved, which is comparable to that observed following sodium chlorate treatment of cells. These results support previous observations and point to the important role of 2-O-or 6-O-sulfated cell surface glycans in viral entry via HS moieties (28).
Our data indicate that the p5 (coil) peptides, which presumably inhibit the interactions of viral attachment proteins with HS, were unable to inhibit cell-to-cell spread. Also, heparin treatment of infected cells did not inhibit cell-to-cell spread. These data provide insights into why polyclonal antibodies generated in response to the gB vaccine are of limited efficacy (50) (i.e., the mechanism of HCMV cell-to-cell spread may be independent of HS interactions involving the viral gB protein). The studies of peptide-mediated inhibition of viral attachment and infection have elucidated several important aspects of CMV infection pathways. First, these peptides, which bind preferentially to sulfated HS, effectively inhibited TCV entry into cultured cells. There was one HS exception, HS008, to which the peptide did not bind. This glycan consisted of GlcA-GlcNS-ClcA-GlcNS-GlcA, while HS007 is one sugar moiety shorter but binds to the peptide very well. The full set of features that are important for viral entry, (i.e., length, number of repeating sugar units, etc.) and whether it depends on in vivo-or cell culture-grown virus remain to be determined. Interestingly, another AP, p5 ϩ 14, also did not bind to HS008 (data not shown) (27). Could the lack of binding to HS008 be the reason that the peptides cannot block 100% of the infection? Without knowing the composition of the HS on the cell surface, we can only speculate the lack of complete inhibition of infectivity (i.e., 5% remaining infectivity for HCMV and 30% for MCMV). Entry differences could be due to differences in entry complexes, glycosylations of the virion proteins, or other factors during replication in vivo. Because we have demonstrated that virus grown on BMDMs recapitulates the "in vivo" virus phenotype, this will allow an in-depth investigation of these possibilities. Our data demonstrate the significant biochemical differences between in vivo-grown virus and virus cultured in vitro. Previously, differences between in vivo-derived and tissue culture-grown MCMVs were defined as "virulent" for in vivo-grown virus or "attenuated" when grown in vitro. These differences were not due to selection of mutated viruses, but rather differential usage of sugar moieties for entry (36,39,40), which could explain why one may be more virulent than the other. Interestingly, it appears that all in vivo-derived viruses, regardless of the tissue from which they were isolated, utilize discrete surface HS compared to tissue culture-derived virus. These differences in HS usage may explain the lack of peptide-mediated inhibition of in vivo-grown MCMV. Furthermore, it is noteworthy that cell attachment by in vivo-grown virus did not depend on 2-O or 6-O glycans, as indicated by the differential inhibition following sodium chlorate treatment of the cells (Fig. 6).
Perhaps most notably, we have shown that the HS-binding peptides do not inhibit cell-to-cell spread of CMV, which represents the main mechanism of cellular transmission in vivo (45,(51)(52)(53). Our data indicate that cell-to-cell spread does not require HS (i.e., heparinase and sodium chlorate treatment of uninfected cells prior to being cocultured with infected fibroblasts only achieved approximately 15% inhibition). These data once again indicate that our current understanding of CMV entry remains incomplete. Further understanding of the mechanism of cell-to-cell spread may provide a path to improve treatment strategies for HCMV.
The HCMV TB40/E and MCMV K181 (55) were used in this study. HCMV TB40/E expressing luciferase under the control of the UL18 promoter was a gift from Christine O'Connor and Eain Murphy (University of Buffalo and Forge Life Science, LLC). TB40/E viruses were propagated on HUVECs. MCMV was produced in vitro using MEF 10.1 cells. All viruses were stored at Ϫ80°C until use. Viral titer was assessed by plaque assay (described below) on MEF 10.1 cells (MCMV) or MRC-5 cells (HCMV). The HCMV clinical isolates were collected at Johns Hopkins University without any identifiers that can link them to a specific patient. To generate BMDM virus, differentiated BMDMs were infected with K181 and virus was harvested after a 100% cytopathic effect (CPE) was observed. Salivary gland-derived virus was harvested from mouse salivary glands at 14 dpi with K181 via the intraperitoneal (i.p.) route. Spleen-derived virus was obtained by harvesting the organ 5 dpi after i.p. infection, while footpad-derived virus was obtained by harvesting footpads 3 dpi following footpad inoculation with K181. All organs were homogenized, and titers were determined via plaque assay.
BMDM differentiation. Bone marrow from naive mice was harvested and incubated in RPMI 1640 for 4 h. The medium was then changed and supplemented with macrophage colony-stimulating factor (M-CSF; PeproTech, Rocky Hill, NJ) at 1 ng/l for 7 days. The medium was changed every 3 days. Differentiated cells were infected with K181 at a multiplicity of infection (MOI) of ϳ0.1. Virus was harvested 1 day after cells showed 100% CPE.
Treatment of cells and viruses. Cells were washed with PBS prior to addition of treatment. Heparinase I (NEB, Ipswich, MA) was used at 12 U/l in medium. Cells were pretreated for 1 h at 37°C before addition of virus or cells. MEF 10.1 or MRC-5 cells were treated for 1 h with 50 mM sodium chlorate to remove 2-O-and 6-O-linked HS as previously described (41). Heparin at various concentrations was preincubated with virus or infected fibroblasts for 30 min at 4°C. Following pretreatment, viral infectivity was assayed by plaque assay or cell-to-cell transfer assay as described below.
Peptides. Peptides were purchased and purified as previously described (27). Briefly, peptides were purified by high-performance liquid chromatography (HPLC), and purity was confirmed by mass spectrometry (MS). Purified peptides were lyophilized and resuspended in PBS prior to use.
Plaque reduction assay. Lyophilized peptides were resuspended in PBS and stored at 4°C until use. Fibroblasts were seeded into 24-well dishes. After cells reached ϳ80% confluence, medium was removed and cells were washed with PBS. Peptide in PBS plus 10% FBS or PBS plus 10% FBS alone as the control was added, and the mixture was incubated for 30 min at 37°C. Virus was added to treated and untreated control wells (ϳ100 PFU/well) and incubated for an additional hour. Following virus incubation, medium was removed and the overlay was added. Overlays consisted of 0.75% carboxymethyl cellulose (CMC; Sigma-Aldrich, St. Louis, MO) for MCMV-infected wells and 0.5% SeaKem agarose (Lonza, Rockland, ME) in complete medium for HCMV-infected wells. For MCMV assays, plates were incubated for 5 days, at which point they were stained with Coomassie stain and plaques counted using a dissecting microscope. Reduction in viral infectivity was expressed as a percentage of infectivity of PBS-treated wells. Data were analyzed using Prism 7 (GraphPad Software, La Jolla, CA).
Cell-to-cell transfer of virus. To examine fibroblast-mediated cell-to-cell spread, MEF 10.1 cells were infected with K181 at an MOI of 3. Following 16 h of incubation, cells were trypsinized, washed, and counted. Approximately 300 potentially infected cells were added to an uninfected 100% confluent monolayer of MEF 10.1 cells. The uninfected monolayer was pretreated with peptide or PBS. Infected cells were given an hour to transfer infection, medium was removed, and a CMC overlay was added. In order to evaluate immune cell-mediated cell-to-cell spread, mice were injected i.p. with 3% thioglycollate (56). After 4 days, animals were infected i.p. with MCMV K181 at 1 ϫ 10 6 PFU, and peritoneal exudate cells (PECs) were collected 12 h postinfection. PECs were added to an uninfected monolayer either treated with p5 ϩ 14 (coil) or untreated. PECs were incubated for an hour, medium was removed, and a CMC overlay was added. Analysis was performed as described for the plaque reduction assay.
HCMV infectivity assay. HCMV TB40/E infectivity was assessed using a luciferase reporter assay as previously described (33). Briefly, MRC-5 fibroblasts were seeded into a 24-well dish. After reaching ϳ80% confluence, cells were washed once with PBS and peptide with PBS plus 10% FBS or the PBS-plus-10% FBS control was incubated at 37°C for 30 min. Virus was added at ϳ1,000 relative light units (RLU) and incubated at 37°C for 1 h. Cells were washed, medium was replaced, and then the mixture was incubated at 37°C for 3 days. On day 3, cells were washed with PBS and lysed using passive lysis buffer (Promega, Madison, WI), and the cell lysates were pelleted. Luciferase reagent (Gaussian luciferase) was combined 1:1 with cell lysate in a clear-bottom 96-well plate. Luminescence was measured using a Synergy 2 plate reader (BioTek, Winooski, VT) and recorded as RLU. The RLU from untreated uninfected samples was subtracted as background, and results were normalized to untreated infected wells to 100% infection. Data were analyzed using Prism 7 (GraphPad Software, La Jolla, CA). The data were expressed as the percentage of luminescence of the untreated infected well.
Statistical analysis and IC 50 determination. Each experiment represents two or more independent experiments with at least three replicates per experimental group, unless otherwise indicated. Individual data points are shown for all graphs. Error bars represent the standard deviation (SD) from each data set. Statistical significance was determined by one-tailed Student's t test or one-way analysis of variance (ANOVA) with Tukey's multivariance analysis when appropriate. IC 50 values were calculated using a linear regression sigmoidal dose-dependent test. All statistical analysis was performed using Prism 7 (GraphPad Software, La Jolla, CA). Statistical significance was assigned to P values of Ͻ0.05. Significant values are labeled as follows: ns, not significant; *, P Յ 0.05; **, P Յ 0.01; ***, P Յ 0.001; and ****, P Յ 0.0001.