Phosphorothioate-Modified Oligodeoxynucleotides Inhibit Human Cytomegalovirus Replication by Blocking Virus Entry

Abstract

Studies in animal models have provided evidence that Toll-like receptor 9 (TLR9) agonists, such as synthetic oligodeoxynucleotides (ODNs) that contain immunostimulatory deoxycytidyl-deoxyguanosine (CpG) motifs (CpG ODNs), protect against a wide range of viral pathogens. This antiviral activity has been suggested to be indirect and secondary to CpG-induced cytokines and inflammatory responses triggered through TLR9 activation. However, few studies have addressed the potential of CpG ODNs as direct antiviral agents. Here, we report on the ability of some CpG ODNs to directly suppress, almost completely, human cytomegalovirus (HCMV) replication in both primary fibroblasts and endothelial cells. Murine CMV replication was inhibited as well, whereas no inhibition was observed for herpes simplex virus type 1, adenovirus, or vesicular stomatitis virus. The antiviral activity of these ODNs was significantly reduced when they were added after virus adsorption, indicating that their action may be primarily targeted to the very early phases of the HCMV cycle. In fact, the B-class prototype CpG ODN 2006 effectively prevented the nuclear localization of pp65 and input viral DNA, which suggests that it inhibits HCMV entry. Moreover, a CpG 2006 control, ODN 2137 without CpG motifs, also showed a potent inhibitory activity on the HCMV entry phase, indicating that the anticytomegaloviral activity is independent of the CpG motif. In contrast, a phosphodiester version of CpG 2006 showed reduced antiviral activity, indicating that the inhibitory activity is dependent on the phosphorothioate backbone of the ODN. These results suggest that this yet-unrecognized activity of CpG ODNs may be of interest in the development of novel anticytomegaloviral molecules.

In vertebrates, Toll-like receptors (TLRs) detect highly conserved pathogen-associated molecular patterns and initiate the appropriate host immune responses. Among the TLRs, TLR9 has evolved to recognize unmethylated deoxycytidyl-deoxyguanosine (CpG) dinucleotides that are relatively common in bacterial and viral genomic DNA, but not in vertebrate genomes (9-11). Recognition of such a pathogen-associated molecular pattern during an infection triggers an immunomodulatory cascade that provides an alarm signal to the innate immune system and stimulates the host to eliminate the pathogen (34). Short synthetic oligodeoxynucleotides (ODN) containing one or more unmethylated CpG dinucleotides (CpG ODNs) can mimic bacterial and viral DNA to stimulate TLR9 in vitro and in vivo and activate innate and adaptive immune responses. CpG ODNs are single-stranded sequences 20 to 30 nucleotides long, containing two to three CpG motifs, usually with phosphorothioate (PTO) modifications in their backbone to enhance stability against nucleases, cellular uptake, and immunomodulatory activity (11). Three classes of CpG ODNs, known as A, B, and C classes, with distinct structural and biological features, have been described. The A class CpG ODNs are potent inducers of alpha interferon (IFN-α) secretion from plasmatocytoid dendritic cells but weakly stimulate B cells. They contain palindromic CpG phosphodiester sequences with phosphorothioate G-rich ends that result in the formation of a high-order secondary structure. The B class CpG ODNs (also known as K type), which contain multiple CpG motifs on a full PTO backbone, trigger the differentiation of plasmacytoid dendritic cells and strongly induce B-cell proliferation and differentiation. The C class CpG ODNs have a completely PTO backbone and show immune properties intermediate between the A and B classes (9-11).

Studies in animal models have confirmed that the immunomodulatory activities of CpG ODNs confer protection against infectious diseases, allergy, and cancer. The safety of CpG ODNs in humans has been assessed in clinical trials (11), and multiple phase II and III clinical trials are currently testing the potential therapeutic use of B class CpG 2006 (also known as CpG 7909 or PF-3512676) as an adjuvant to virus, toxin, and bacterial vaccines as well as cancer vaccines, alone and in combination with conventional therapies (11, 14).

Due to their immunomodulatory properties, CpG ODNs have been considered for prophylactic or therapeutic treatment against intracellular pathogens. Studies in mice have revealed that they can provide a potent innate protection against herpes simplex virus type 2 (HSV-2) acute infection, particularly when administered before infectious challenge (22). Moreover, B class CpG ODNs have been shown to induce rapid suppression of hepatitis B virus (HBV) replication in an experimental transgenic mouse model, which suggests a potential use in the treatment of chronic viral infections (8). The mechanisms of protection elicited by these ODNs depend on the pathogen and the site of infection. In the HSV-2 experimental model, protection was associated with modifications of the vaginal epithelium and recruitment of inflammatory cells to the submucosa (1), and a role for IFN-β (5), but not IFN-γ (1), was identified. In the HBV murine transgenic model, a role for CpG ODN-induced type I IFN (IFN-α and/or -β) secretion has been suggested by the lack of effects on HBV replication in mice genetically deficient in the type I IFN receptor (8). Thus, in these in vivo models, the antiviral effects of CpG ODNs seem to be indirect and secondary to CpG ODNs’ internalization, binding to TLR, and activation of TLR-induced cytokine production. These steps are generally thought to be indispensable for all the immunostimulatory activities of CpG ODNs. However, the restricted intracellular expression of TLR9 to B cells and plasmacytoid dendritic cells limits the full response to CpG to these cell types (34).

However, CpG ODNs can potentially affect cell physiology, as well as replication of intracellular pathogens, in cells that lack TLR9 expression, since these compounds are spontaneously internalized in cultured cells without the need for uptake enhancers or transfection (11). Indeed, few studies have addressed the capability of CpG ODNs to directly interfere with the replication of viral pathogens in vitro, and their potential as direct antiviral agents remains to be evaluated (23, 24).

In this study, we addressed the effects of CpG ODNs on the replication of the important human opportunistic viral pathogen human cytomegalovirus (HCMV). We report that the in vitro replication of HCMV was potently suppressed by several CpG ODNs by a mechanism that was TLR9 independent and targeted the entry phase of the viral replicative cycle.

MATERIALS AND METHODS

Cells, culture conditions, and viruses.

Low-passage human embryonic lung fibroblasts (HELFs) were grown as monolayers in Eagle's minimal essential medium (Gibco-BRL) supplemented with 10% fetal bovine serum (FBS; Gibco-BRL), 1 mM sodium pyruvate, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin sulfate. Human umbilical vein endothelial cells (HUVECs) obtained by trypsin treatment of umbilical cord veins were cultured in endothelial growth medium (EGM-2; Cambrex Bio Science, Walkersville, MD) supplemented with 2% FBS, human recombinant vascular endothelial growth factor, basic fibroblast growth factor, human epidermal growth factor, insulin growth factor 1, hydrocortisone, ascorbic acid, heparin, gentamicin, and amphotericin B (1 μg/ml each). Experiments were performed with cells at passages two to six. Vero and NIH 3T3 cells were grown as monolayers in Dulbecco's modified Eagle's medium (Gibco-BRL) supplemented with 10% donor bovine serum (Gibco-BRL), 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin sulfate.

HCMV strain AD169 was purchased from ATCC (VR538). It was propagated in HELFs infected at a multiplicity of infection (MOI) of 0.01, incubated in minimal essential medium supplemented with 1% heat-inactivated FBS, and cultured until a marked cytopathic effect was seen. Virus stocks were then prepared by sonicating the cells, followed by centrifugal clarification, and titrated by standard plaque assay on HELFs. HCMV VR1814 (kindly provided by G. Gerna) and HCMV AL1 are derivatives of clinical isolates recovered from a cervical swab from a pregnant woman (20) and the bronchoalveolar lavage fluid of a lung transplant recipient, respectively. These strains were propagated in HUVECs and titrated by the indirect immunoperoxidase staining procedure on HELFs using a monoclonal antibody (MAb) reactive to the HCMV IE1 and IE2 proteins (clone E13; Argene Biosoft) (20). HCMV TB40 UL32-EGFP (kindly provided by C. Singzer) is a recombinant HCMV TB40 strain in which enhanced green fluorescent protein (EGFP) is fused to the C-terminal end of the UL32 gene (21). It was propagated on HELFs, pelleted through a 20% sorbitol cushion, and titrated as described above for HCMV AD169. Murine cytomegalovirus (MCMV) strain Smith (ATCC VR194) was propagated in NIH 3T3 cells and titrated by standard plaque assay. A clinical isolate of HSV-1 was propagated and titrated by standard plaque assay on Vero cells. Vesicular stomatitis virus (VSV) serotype Indiana and a clinical isolate of adenovirus were propagated and titrated by standard plaque assay on HELF cells.

ODNs and antiviral substances.

Synthetic ODNs (Metabion International, Germany) were of high-performance liquid chromatography-purified quality and were dissolved in Tris-EDTA buffer (pH 8.0) at a concentration of 1 mM. The following sequences were used (bold letters indicate CpG motifs): ODN 2216 (5′-GGgggacgatcgtcgGGGGG-3′), ODN 2006 (5′-TCGTCGTTTTGTCGTTTTGTCGTT-3′), ODN 2006 PD (5′-tcgtcgttttgtcgttttgtcgtt-3′) ODN 2137 (5′-TGCTGCTTTTGTGCTTTTGTGCTT-3′), ODN 10103 (5′-TCGTCGTTTTTCGGTCGTTTT-3′), ODN 10104 (5′-TCGTCGTTTCGTCGTTTTGTCGTT-3′), ODN 10105 (5′-TCGTCGTTTTGTCGTTTTTTTCGA-3′), ODN 2007 (5′-TCGTCGTTGTCGTTTTGTCGTT-3′), ODN 1826 (5′-TCCATGACGTTCCTGACGTT-3′), ODN 2138 (5′-TCCATGAGCTTCCTGAGCTT-3′), and ODN 2395 (5′- TCGTCGTTTTCGGCGCGCGCCG-3′). Capital letters in ODN sequences indicate 3′ phosphorothioate internucleotide linkages, and lowercase letters indicate 3′ phosphodiester internucleotide linkages (see Table 1, below).

TABLE 1.

Antiviral activities of CpG ODNs against HCMV AD169 and cytotoxicity in HELFs

ODN	ODN class	Sequencea	CC50 (μM)b	IC50 (μM)b	IC90 (μM)b
2006	B	TCGTCGTTTTGTCGTTTTGTCGTT	>10	0.020 ± 0.01	0.065 ± 0.01
2137	B	TGCTGCTTTTGTGCTTTTGTGCTT	>10	0.030 ± 0.01	0.220 ± 0.02
2007	B	TCGTCGTTGTCGTTTTGTCGTT	>10	0.200 ± 0.04	4 ± 0.5
10103	B	TCGTCGTTTTTCGGTCGTTTT	>10	0.025 ± 0.01	0.065 ± 0.01
10104	B	TCGTCGTTTCGTCGTTTTGTCGTT	>10	0.050 ± 0.02	0.800 ± 0.05
10105	B	TCGTCGTTTTGTCGTTTTTTTCGA	>10	0.020 ± 0.01	0.100 ± 0.01
1826	B	TCCATGACGTTCCTGACGTT	>10	0.040 ± 0.01	0.900 ± 0.1
2138	B	TCCATGAGCTTCCTGAGCTT	>10	0.200 ± 0.04	>5
2006 PD	B	tcgtcgttttgtcgttttgtcgtt	>10	3.8 ± 0.4	>10
2216	A	GGgggacgatcgtcgGGGGG	>10	0.140 ± 0.04	3.6 ± 0.3
2395	C	TCGTCGTTTTCGGCGCGCGCCG	>10	0.100 ± 0.02	3.2 ± 0.3

^aCapital letters in ODN sequences indicate 3′-phosphorothioate internucleotide linkage; lowercase letters indicate 3′-phosphodiester internucleotide linkage; underlining indicates self-complementary palindromes; the CpG motifs thought to contribute the most immune stimulation are in bold letters.

^bCC₅₀, the ODN concentration that reduces cell viability to 50% in the MTT assay; IC₅₀ and IC₉₀ values show inhibition of HCMV yield at 6 days p.i. ODNs were incubated for 1 h before, during, and after HCMV infection. The IC₅₀ and IC₉₀ values are means ± standard deviations of three independent experiments.

ODNs were either incubated with infected cells after virus adsorption (posttreatment) or added to the cells 1 h before infection, and then they were maintained throughout the antiviral assay (pre- and posttreatment).

Ganciclovir (GCV; Cymevene; Roche) was used as control reference drug with known antiviral activity (17). IFN-α (PegIntron; Schering-Plough) was used at 1,000 U/ml; polyinosine-poly(C) [poly(I:C); InvivoGen] was used at 10 μg/ml.

Antiviral assay.

Untreated cells or cells incubated with different concentrations of ODNs for 1 h before infection were infected with HCMV, MCMV, adenovirus, or VSV at an MOI of 1 or with HSV-1 at an MOI of 0.1. Following virus adsorption (2 h at 37°C), cultures were maintained in medium containing the corresponding ODN and then incubated until control cultures displayed extensive cytopathology (6 days postinfection [p.i.] for HCMV and MCMV, 4 days p.i. for adenovirus, and 48 h p.i. for HSV-1 and VSV). Thereafter, the cells and supernatants from the antiviral assay were harvested and disrupted by sonication. The extent of virus replication was then assessed by titrating the infectivity of supernatants of cell suspensions by standard plaque assay on HELFs for HCMV, adenovirus, and VSV, on Vero cells for HSV-1, or on NIH 3T3 cells for MCMV. Plaques were microscopically counted, and the mean plaque count for each drug concentration was expressed as a percentage of the mean plaque count of the control virus or as the viral titer. The number of plaques was plotted as a function of drug concentration; concentrations producing 50 and 90% reductions in plaque formation (IC₅₀ and IC₉₀) were determined. To determine cell viability, HELFs were exposed to increasing concentrations of ODNs. After 6 days of incubation, the number of viable cells was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) method, as previously described (18).

Immunoblotting.

Whole-cell protein extracts were prepared as previously described (3). Proteins were separated by 8% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred to Immobilon-P membranes (Millipore). Filters were blocked overnight in 5% nonfat dry milk in 10 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 0.1% Tween 20 and immunostained with the mouse anti-HCMV IE1 and IE2 MAb (clone E13; Argene Biosoft), mouse anti-HCMV UL44 MAb (clone 1202; Goodwin Institute, Plantation, FL), mouse anti-HCMV UL99 MAb (clone 1207; Goodwin Institute), or mouse antiactin MAb (Chemicon International). Immunocomplexes were detected with sheep anti-mouse immunoglobulin Ab conjugated to horseradish peroxidase (Amersham) and visualized by enhanced chemiluminescence (Super Signal; Pierce).

Quantitative viral nucleic acid analysis.

To determine the number of viral DNA genomes per nanogram of cellular reference DNA (18S rRNA gene), viral DNA levels were measured by quantitative real-time PCR, using the previously described probe and primers amplifying a segment of the IE1 gene (31). Briefly, 15 ng DNA from the samples was added to the Real Master Mix Probe Rox with 5 mM Mg²⁺ (Applied Biosystems) with oligonucleotide primers and TaqMan dual-labeled IE1 (5′ 6-carboxyfluorescein and 3′ 6-carboxytetramethylrhodamine quencher) probe; Applied Biosystems). After activation of the AmpliTaq Gold for 10 min at 95°C, 50 cycles of 15 s at 95°C and 1 min at 62°C were carried out in an Mx 3000 P apparatus (Stratagene). HCMV DNA copy numbers were normalized by dividing by the amount of human 18S rRNA gene (Assay-on-Demand, 18S, assay no. HS99999901_s1; Applied Biosystems) amplified per reaction mixture. Standard curves were constructed using values from the serially diluted genomic DNA mixed with an IE1-encoding plasmid (6).

Real-time quantitative reverse transcription-PCR (RT-PCR) analysis was performed on an Mx 3000 P apparatus (Stratagene). After HCMV infection and cell treatment, total cellular RNA was isolated using the Eurozol reagent (Euroclone Ltd., United Kingdom), and RNA samples (1 μg) were then retrotranscribed at 42°C for 60 min in PCR buffer (1.5 mM MgCl₂) containing 5 μM random primers, 0.5 mM deoxynucleoside triphosphates, and 100 U Moloney murine leukemia virus reverse transcriptase (Ambion), in a final volume of 20 μl. Reverse-transcribed cDNAs, or water as a control, were then amplified in duplicate using the Brilliant Sybr green QPCR master mix (Stratagene). Primer sequences were as follows: IE1 (sense, 5′-CAA GTG ACC GAG GAT TGC AA-3′; antisense, 5′-CAC CAT GTC CAC TCG AAC CTT-3′); IE2 (sense, 5′-TGA CCG AGG ATT GCA ACG A-3′; antisense, 5′-CGG CAT GAT TGA CAG CCT G-3′); human TLR9 (sense, 5′- ACTTCACCTTGGATCTGTCACG-3′; antisense, 5′-GCTTATTGCGGGACAGGTCTA-3′); β-actin (sense, 5′-CAA AAG CCT TCA TAC ATC TC-3′; antisense, 5′-TCA TGT TTG AGA CCT TCA A-3′); human MxA (sense, 5′-TTCAGCACCTGATGGCCTATC-3-prime; antisense, 5′-TGGATGATCAAAGGGATGTGG-3′); human 2′,5′-oligoadenylate synthetase (2′,5′-OAS) (sense, 5′-CTACCTGCTTCACGGAGCTC-3′; antisense, 5′-CTCCTTACACAGTTGGTACCAG-3′); human IFN-α (sense, 5′-GTGAGGAAATACTTCCAAAGAATCAC-3′; antisense, 5′-TCTCATGATTTCTGCTCTGACAA-3′).

Following an initial denaturing step at 95°C for 2 min to activate 0.75 U Platinum Taq DNA polymerase (Invitrogen), the cDNAs were amplified for 40 cycles (95°C for 1 min, 60°C for 1 min, and 72°C for 1 min). For quantitative analysis, semilogarithmic plots were constructed of delta fluorescence versus cycle number, and a threshold was set for the changes in fluorescence at a point in the linear PCR amplification phase (C_T). The C_T values for each gene were normalized to those for β-actin with the ΔC_T equation. The level of target RNA, normalized to the endogenous reference and relative to that of the mock-infected and untreated cells, was calculated by the comparative C_T method with the 2^−ΔΔC^T equation.

HCMV EGFP binding and pp65 translocation assay.

HELFs were grown to semiconfluence on glass coverslips in 24-well plates. Prechilled cell monolayers were then treated with ODNs or heparin (30 μg/ml) for 1 h at 4°C and then infected with precooled HCMV TB40 UL32-EGFP (21) at an MOI of 20 for 2 h at 4°C. The cells were gently washed twice with ice-cold phosphate-buffered saline and fixed in 4% formaldehyde for 10 min on ice. Cells infected with EGFP-expressing virus were examined by fluorescence microscopy in combination with conventional phase contrast or following nuclear counterstaining with propidium iodide (Sigma). Images were recorded with an Olympus Fluoview-IX70 inverted confocal laser scanning microscope.

For the pp65 translocation assay, HELFs were grown to semiconfluence on glass coverslips in 24-well plates. Prechilled cell monolayers were then treated with ODNs or heparin (30 μg/ml) for 1 h at 4°C and then infected with precooled AD169 at an MOI of 5 for 2 h at 4°C, ensuring viral attachment but not entry, as previously described (4, 19). The cells were then transferred to 37°C for 2 h to allow viral entry. Thereafter, they were fixed in 4% paraformaldehyde (10 min, room temperature) and permeabilized with 0.2% Triton X-100 in phosphate-buffered saline (20 min, 4°C). Indirect immunofluorescence analysis was performed by incubating fixed cells with the mouse anti-pp65 MAb (Cinapool; Argene Biosoft) for 2 h at 37°C, followed by secondary antibody incubation (anti-mouse immunoglobulin G-fluorescein isothiocyanate; Sigma) for 1 h at room temperature. Nuclear counterstaining was performed with propidium iodide. Images were then taken using an Olympus Fluoview-IX70 inverted confocal laser scanning microscope.

Virucidal assay.

To assess the effect of ODNs on viral infectivity, the procedure described by Shogan et al. (26) was used. Briefly, CpG ODNs (0.1 μM) were added to aliquots of HCMV AD169 (10⁴ PFU), and the virus-ODN samples were then incubated at either 4 or 37°C for various lengths of time. After incubation, the samples were diluted with medium to reduce the concentration of ODN to that not active in an antiviral assay. HCMV was then titrated on HELF cells.

Data analysis.

All data were generated from duplicate wells in at least three independent experiments. The effects of CpG ODNs at different concentrations were expressed as PFU/ml on a log₁₀ scale, or the mean plaque count for each drug concentration was expressed as a percentage of the mean plaque count of the control virus. Concentrations producing 50 and 90% reductions in plaque formation (IC₅₀ and IC₉₀) were calculated by linear regression using the computer program GraphPad Prism version 4.0.

RESULTS

Effects of CpG ODNs on HCMV productive infection in cultured cells.

As shown in Fig. 1, pretreatment of HELFs with CpG ODNs 1 h before infection produced a significant concentration-dependent inhibitory effect on HCMV AD169 replication at 6 days p.i. The IC₅₀ and IC₉₀ concentrations of CpG ODNs for HCMV replication are shown in Table 1. All ODNs showed an IC₅₀ from 1 to 2 orders of magnitude lower than that of the reference drug GCV (its IC₅₀ when tested on HCMV AD169 replication in HELFs was 2.95 μM). However, the calculated IC₅₀ values of the ODNs varied from the lowest measured for the B class prototype CpG 2006 (0.02 μM) to that of the less active CpG 2007 (0.20 μM). The A class (CpG 2216) and C class (CpG 2395) prototypes were less effective than most of the B class ODNs (Fig. 1 and Table 1). The inhibitory effect of CpG 2006 was neither virus strain specific nor cell type specific, because it was also observed in HELFs infected with the clinical isolates VR1814 (IC₅₀, 0.038 μM) and AL1 (IC₅₀, 0.044 μM) and in HUVECs infected with the endotheliotropic VR1814 strain (IC₅₀, 0.025 μM).

FIG. 1.

Effects of CpG ODNs on productive infection by HCMV, MCMV, and HSV-1. (A) CpG ODNs inhibit HCMV replication. HELFs were infected with HCMV AD169 (MOI of 1) and, where indicated, the cells were pretreated and treated with increasing concentrations of the different ODNs 1 h prior to and during infection, until an extensive viral cytopathic effect was observed in the untreated controls. The extent of AD169 replication was then assessed by titrating the infectivity of supernatants of HELF suspensions by standard plaque assay. Plaques were microscopically counted, and the mean plaque counts for each ODN concentration were expressed as PFU/ml on a log₁₀ scale. The number of plaques was plotted as a function of drug concentration, and the IC₅₀ and IC₉₀ concentrations were determined. The data shown represent means ± standard deviations (error bars) of three independent experiments. ODN 2137 is the non-CpG control ODN for CpG 2006. CpG 2006 PD is the pure phosphodiester version of CpG 2006.

None of the ODNs analyzed significantly affected the viability of HELFs in the relevant range of concentrations, since >90% of cells were viable after 6 days treatment with ODNs up to a concentration of 10 μM (Table 1), demonstrating that the antiviral activity was not due to cytotoxicity for the target cells themselves.

Moreover, a control ODN for CpG 2006, ODN 2137, in which the CpG motifs have been changed to GpC, inhibited HCMV replication to a similar degree as CpG 2006, which suggests that its antiviral activity against HCMV is independent of the CpG motif (Fig. 1 and Table 1). A similar effect, albeit less evident, was also observed with ODN 2138, a non-CpG sequence used as a control for CpG 1826, which still showed significant antiviral activity (Table 1). Consistent with these results, when the TLR9 expression was evaluated in HELFs and HUVECs by measuring its mRNA, no detectable signals were observed in cDNA samples prepared from total RNA purified from these two cell types, which confirms that TLR9 was not expressed in our fibroblast or endothelial cell models (Fig. 2A). Moreover, the expression of IFN-α and of some representative IFN-stimulatable genes, such as MxA and 2′,5′-OAS, was investigated by real-time RT-PCR in CpG 2006-treated HELF cells. As positive controls, HELF cells were treated with the IFN-inducer poly(I:C) or with exogenous IFN-α. As shown in Fig. 2B, treatment with poly(I:C), as expected, significantly induced the expression of IFN-α, MxA, and 2′,5′-OAS mRNA. By contrast, CpG 2006 did not stimulate the expression of these genes at any of the time points analyzed. Thus, the anti-HCMV activity of CpG 2006 in fibroblasts and endothelial cells is independent of the binding and activation of TLR9 and/or induction of an IFN response.

FIG. 2.

TLR9 is not expressed in HELFs or HUVECs, and the IFN response is not activated on stimulation with CpG 2006 ODN. (A) Lack of TLR9 mRNA expression in HELFs and HUVECs. Total RNA was isolated from HELFs (lane 3) and HUVECs (lane 4) and reverse transcribed. Real-time RT-PCR was then performed with the appropriate TLR9 and β-actin primers. PCR products were then fractionated by electrophoresis in a 2.0% agarose gel and stained with ethidium bromide. A negative water control (lane 1) and a positive control for TLR9 mRNA expression (a total RNA sample from human plasmacytoid dendritic cells [pDC]; lane 2) are shown. TLR9 mRNA levels were normalized according to expression of the actin gene. The TLR9 mRNA expression in HELFs and HUVECs is expressed on a log₁₀ scale relative to the level measured in pDC, which was set at 1. A representative experiment is shown. (B) Lack of IFN response in CpG 2006-treated HELF cells. HELF cells were treated with CpG 2006 (1 μM) or stimulated with IFN-α (1,000 U/ml) or poly(I:C) (10 μg/ml) as positive controls. Total RNA was isolated at the indicated times and reverse transcribed. Real-time RT-PCR was then performed with the appropriate IFN-α, MxA, 2′5′-OAS, and β-actin primers. For each time point, IFN-α, MxA, and 2′,5′-OAS mRNA levels were normalized according to expression of the actin gene. The data shown are the means ± standard deviations (error bars) from triplicate analyses.

Furthermore, the pure phosphodiester ODN CpG 2006 PD showed a low inhibitory effect on HCMV replication (IC₅₀, 3.8 μM) (Fig. 1 and Table 1), which suggests that the antiviral activity of CpG 2006 may be related to the chemical structure of the PTO-DNA backbone, rather than the CpG-containing sequence alone.

We next investigated whether the CpG ODNs interfere with the replication of other herpesviruses, namely, MCMV and HSV-1. The results obtained with MCMV were similar to those obtained with HCMV. When preincubated with NIH 3T3 cells, the B class CpG prototype 2006 ODN and its control 2137 showed significant inhibition of MCMV replication (Fig. 3), and the calculated IC₅₀s were 0.05 and 0.02 μM, respectively. In contrast, the replication of HSV-1 in HELF cells pretreated with either 0.1 or 1 μM of the CpG ODNs (concentrations that potently block MCMV and HCMV replication) was inhibited to a very small, and in most cases insignificant, extent (Fig. 3). Moreover, CpG ODNs did not significantly affect in HELF cells the replication of other viruses, such as a clinical isolate of adenovirus or a VSV laboratory strain (Fig. 3).

FIG. 3.

Effects of CpG ODNs on the replication of other viruses. NIH 3T3 cells were infected with MCMV Smith (MOI of 1), and HELF cells were infected with HSV-1 (MOI of 0.1), adenovirus (MOI of 1), or VSV (MOI of 1). Where indicated, the cells were pretreated and treated with either 0.1 or 1 μM ODN as described in the text. The extent of MCMV, HSV-1, adenovirus, and VSV replication was then assessed by titrating the infectivities of cell suspension supernatants by standard plaque assay. Values shown represent means ± standard deviations (error bars) of three independent experiments.

Taken together, these results demonstrate that CpG ODNs potently inhibit the productive replication in vitro of HCMV and MCMV, but not HSV-1, VSV, or adenovirus, and that this antiviral activity is independent of the CpG motifs, the presence of a functional TLR9 receptor, or IFN induction.

CpG ODNs inhibit the first phases of the HCMV replicative cycle.

To gain more insight into the mechanism of CpG antiviral activity, we next examined the effects of CpG 2006 on overall HCMV gene expression. For this purpose, total cell extracts were prepared at various times postinfection from HCMV AD169-infected HELFs that were treated with CpG 2006 or, as a control, GCV. The extracts were then analyzed for their content of immediate early (IE), early (E), and late (L) proteins by immunoblotting with specific antibodies. Expression of IEA (IE1 and IE2), UL44, and UL99 was assessed as a control for IE, E, and L proteins. As shown in Fig. 4A, GCV, as expected from its mode of action, had a more pronounced inhibitory effect on the expression of early and late proteins. In contrast, CpG 2006 inhibited the expression of all the examined HCMV proteins at any of the time points analyzed, which demonstrates that its antiviral activity stems from an early inhibitory effect on the expression of adequate amounts of pivotal IE proteins. Consistent with this result, when real-time RT-PCR was used to measure IE1 and IE2 mRNA levels in HCMV-infected HELFs that were exposed to CpG 2006 or CpG 10103 (also shown to inhibit HCMV replication [Table 1]), complete prevention of both IE1 and IE2 mRNA expression was measured (Fig. 4B). This result suggests that these ODNs act at a very early stage in the HCMV replicative cycle, before the expression of IE genes.

FIG. 4.

CpG 2006 prevents HCMV gene expression. (A) Effects of CpG 2006 on HCMV IE, E, and L proteins. HELFs were infected with HCMV AD169 (MOI of 1) or mock infected. Where indicated, the cells were pretreated and treated with 1 μM CpG 2006 1 h prior to and during infection or 100 μM GCV during infection. At the indicated times postinfection, total cell extracts were prepared, fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (50 μg protein/lane), and analyzed by immunoblotting with anti-IEA (IE1 and IE2), anti-UL44, or anti-UL99 MAb. Actin immunodetection with a MAb served as an internal control. (B) IE gene expression is blocked in cells treated with CpG ODNs. HELFs were infected with HCMV AD169 (MOI of 1) or mock infected. Where indicated, the cells were pretreated and treated with 1 μM CpG 2006 or CpG 10103 1 h prior to and during infection. Total RNA was isolated at the indicated times postinfection and reverse transcribed. Real-time RT-PCR was then performed with appropriate IE1, IE2, and β-actin primers. For each time point, IE1 and IE2 mRNA levels were normalized according to the expression of the actin gene. The data shown represent means ± standard deviations (error bars) of three independent experiments.

To further support this hypothesis, incubation with different CpG ODNs at 0.1 or 1 μM after 2 h of virus adsorption (posttreatment) did not significantly affect HCMV replication in HELFs, in contrast to the strong antiviral effects observed with the same CpG ODNs under the conditions of preincubation (pre- and posttreatment) (Fig. 5). This confirms that their antiviral activity targets the very early phases of the viral replicative cycle, such as adsorption and/or entry.

FIG. 5.

Incubation of CpG ODNs after virus adsorption reduces their anticytomegaloviral activities. HELFs were infected with HCMV AD169 (MOI of 1), and CpG ODNs (0.1 or 1 μM) were added after virus adsorption (posttreatment) or 1 h before as well as during virus adsorption and remained in the culture medium throughout the experiment (pre- and posttreatment). At 6 days p.i., the extent of AD169 replication was assessed by titrating the infectivity of HELF suspension supernatants in a standard plaque assay. The values shown represent means ± standard deviations (error bars) of three independent experiments.

CpG ODNs block HCMV entry.

To test whether the CpG ODNs act at the viral adsorption stage, we used a recombinant HCMV strain that expresses EGFP fused to the capsid-associated tegument protein UL32, thus generating green fluorescent virion particles (21). The HCMV UL32-EGFP reporter virus was incubated with HELFs that were treated with CpG ODNs, or heparin as a positive control for inhibition of viral adsorption, for 2 h at 4°C (a condition that is known to allow virus adsorption only). As shown in Fig. 6, a fine punctate cell surface pattern of green fluorescence was observed in HCMV UL32-EGFP-infected HELFs, thus indicating the successful attachment of fluorescent virion particles to the cells. As expected, heparin treatment prevented virus adsorption. In contrast, CpG 2006, its control ODN 2137 with the non-CpG sequence, and the less active CpG 2007 did not significantly reduce the binding of HCMV UL32-EGFP virions to HELFs. Thus, these results indicate that HCMV attachment to the target cells is not influenced by the presence of CpG ODNs.

FIG. 6.

CpG ODNs do not interfere with HCMV adsorption to HELFs. HELFs were infected with the reporter strain HCMV TB40 UL32-EGFP (MOI of 20) at 4°C in the presence of CpG 2006, CpG 2007, ODN 2137 (0.1 μM), or heparin (30 μg/ml) for 2 h to allow for virus adsorption only. Cells were then fixed, counterstained with propidium iodide, and examined by fluorescence microscopy to detect green fluorescent virus particles. The experiment was repeated twice, and representative images are presented.

Next, we performed a virion content delivery assay (4, 19) to investigate whether the step subsequent to HCMV adsorption is affected by CpG ODNs. The highly abundant pp65 protein of the HCMV tegument rapidly localizes to the nucleus of infected cells after membrane fusion and viral entry. Therefore, pp65 nuclear localization can be used to assess viral penetration. Indirect immunofluorescence staining for pp65 was performed on fixed and permeabilized HELF cells that were treated with CpG 2006, CpG 2006 PD, CpG 2007, or the control ODN 2137, then infected with HCMV AD169 at 4°C for 2 h, and finally shifted to 37°C for 2 h. These experimental conditions allowed synchronized virus penetration after attachment at low temperature (4, 19). As in the binding assay, treatment of HELFs with soluble heparin inhibited the nuclear localization of pp65 (Fig. 7). Similarly, CpG 2006 and its control 2137 prevented pp65 accumulation in the nucleus, whereas the less active CpG 2007 showed a lower inhibitory effect on pp65 translocation (Fig. 7). The pure phosphodiester ODN CpG 2006 PD did not block nuclear accumulation of pp65, which suggests that the inhibitory activity of CpG 2006 on viral entry depends from the PTO-DNA backbone. To support the hypothesis that CpG 2006 inhibits HCMV entry, real-time PCR was used to quantify at 4 and 48 h p.i. the amount of viral DNA per nanogram of cellular reference DNA (18S rRNA) in HELFs treated with CpG 2006 and then infected with HCMV. As shown in Fig. 8, the number of copies of viral DNA at 4 h p.i. after viral entry, but well before the onset of genome replication, in the untreated controls was 10³, whereas in the presence of CpG 2006, the number of viral genomes at 4 h p.i. decreased to 7 × 10¹. The extent of this reduction was confirmed at 48 h p.i., when the numbers of copies of viral DNA measured in the untreated and in the CpG 2006-treated cultures were 2 × 10⁴ and 1.4 × 10³, respectively.

FIG. 7.

ODN 2006 and 2137 inhibit HCMV entry. HCMV AD169 (MOI of 5) was incubated with HELFs at 4°C in the presence of CpG 2006, CpG 2006 PD, CpG 2007, ODN 2137 (0.1 μM), or heparin (30 μg/ml) for 2 h. The cultures were shifted to 37°C for 2 h to allow viral entry. Cells were then fixed, and nuclear accumulation of viral tegument pp65 protein was detected by indirect immunofluorescence. Cell nuclei were counterstained with propidium iodide (PI). The experiment was repeated three times, and representative images are presented.

FIG. 8.

ODN 2006 prevents accumulation of HCMV DNA at immediate-early times of infection. HELFs were treated with CpG 2006 (0.1 μM) and infected with HCMV AD169 (MOI of 5). Total genomic DNA was purified at 4 and 48 h p.i. and subjected to real-time PCR. The data shown represent means ± standard deviations (error bars) of three independent experiments.

To further investigate the mechanism of action of CpG 2006, we explored the possibility that it was interacting with HCMV particles and inactivating infectivity. HCMV AD169 was incubated with 0.1 μM of CpG 2006 or CpG 2007 at 4 or 37°C for various lengths of time. After incubation, the samples were diluted to reduce the ODN concentration well below that which inhibits HCMV replication, and the virus was titrated on HELF cells (26). As shown in Fig. 9, the preincubation of virions with ODNs at either 4 or 37°C did not significantly affect HCMV AD169 infectivity. Thus, this result indicates that CpG 2006 ODN inhibits infection only if present at the time of virus entry into the cells.

FIG. 9.

Preincubation of ODN 2006 with virus does not affect HCMV infectivity. HCMV AD169 aliquots (10⁴ PFU) were incubated at either 4 or 37°C for various lengths of time with no ODN (closed squares), 0.1 μM CpG 2006 (closed triangles), or 0.1 μM CpG 2007 (open squares). After incubation, the samples were diluted to reduce the ODN concentration below that which inhibits HCMV replication, and the virus was titrated on HELF cells. The data shown represent means ± standard deviations (error bars) of three independent experiments.

Cumulatively, these findings indicate that interference with viral entry into the cells is involved in the mode of action of CpG 2006 toward HCMV replication and that this activity is independent of the CpG motifs but rather related to the PTO modification of the ODN.

DISCUSSION

This study is believed to be the first report of a novel activity of some CpG ODNs. The prototype B class CpG 2006, as well as other related ODN sequences, showed a potent dose-dependent inhibitory effect on HCMV replication, by a mechanism(s) in which suppression of the entry phase of the viral cycle seemed to make the major contribution to the overall antiviral activity. The antiviral activity likely stems from the PTO modification of the CpG ODNs backbone that causes direct interference with the activation and/or functioning of the viral fusion machinery. Several lines of evidence support this hypothesis: (i) the inhibitory activity of CpG ODNs was shown to be independent of TLR9 triggering; (ii) a control ODN lacking CpG motifs was also active in blocking HCMV entry; (iii) a control ODN with a phosphodiester backbone, even with the CpG motif (e.g., 2006 PD), showed significantly reduced inhibitory activity; (iv) a BLAST search revealed no significant alignment between the HCMV AD169 and VR1814 genomes and 2006, 2137, and 10103 ODN sequences, thus excluding antisense effects as the molecular mechanism that interferes with HCMV replication; and (v) no IFN response was measured in CpG ODN-treated cells, which excludes any contribution of these cytokines to the anti-HCMV activity. Moreover, the antiviral activity of the CpG ODNs seemed to specifically target cytomegaloviruses, since both HCMV and MCMV replication was inhibited while negligible inhibition was measured for HSV-1, adenovirus, or VSV.

In the past few years, experimental evidence has revealed the efficacy of CpG ODNs for stimulating protective innate and adaptive responses in animal models of human viral infections, thus paving the way to clinical trials in which their activity was evaluated as monotherapy or vaccine adjuvants (11, 14). In contrast, with the sole exception of human immunodeficiency virus (HIV), no other in vitro studies have addressed the antiviral effects of CpG ODNs on human viruses.

In fact, it has been reported that CpG 2006 can suppress HIV replication in productively infected cultured tonsillar tissue, as well as in peripheral blood mononuclear cells, primary CD4⁺ T cells, and T-cell lines (23, 24). Consistent with our observations, in both of the earlier studies the inhibitory effect was found to be independent of the presence of the CpG motif, as it was also observed with control ODNs in which the immunomodulatory motif was reversed to GpC or changed to TpG. Similar to what we observed with the 2006 PD ODN, the in vitro anti-HIV activity of CpG 2006 has been shown to be related to its PTO backbone, since a control ODN with a phosphodiester modification, even with the CpG motif, has only marginal antiviral activity (24). Furthermore, when CpG 2006 and its non-CpG control were tested in a cell-based fusion assay to investigate their effects on HIV-cell fusion, both were shown to block syncytium formation. These results suggest that the primary CpG 2006 mechanism of action against HIV depends on inhibition of virus entry, and thus it is independent of TLR9 triggering (24). It was therefore hypothesized that CpG ODNs, through their PTO-modified backbone, specifically interact with the virus envelope to block fusion with the cell membrane. This hypothesis is supported by other studies that have shown that PTO-modified ODNs, such as the 28-mer deoxycytidine SdC28, bind to the positively charged V3 loop of gp120, thus interfering with virus binding to the target cells (29, 30). The options for preventing HIV attachment by ODNs has been further exploited by Horvath et al. (7), who reported the ability of a 35-mer 4-thio-deoxyuridylate to bind to CD4 receptors, thus preventing virus attachment. More recently, it has also been reported that long hydrophobic PTO-ODNs (>30-mer) have the ability to inhibit HIV fusion with the cell membrane by interacting with the N-terminal heptad repeat region of gp41 and blocking its six-helix bundle formation (32).

Similarly, in this study, we observed that CpG 2006 and its control ODN 2137 effectively inhibited HCMV entry. In fact, the binding of a reporter HCMV strain was unaffected by the presence of ODNs (Fig. 6), which suggests that they act as fusion inhibitors rather than blockers of HCMV attachment. Furthermore, the preincubation of virions with CpG 2006 did not result in a significant loss of infectivity (Fig. 9), which indicates that it inhibits HCMV infection only if present at the time of virus entry into the cells. Thus, it is possible that the targets of CpG 2006 are virion components not in their native conformation but rather undergoing conformational modifications toward a fusion-active conformation.

However, the molecular interactions that likely occur between CpG ODNs and the viral proteins involved in the HCMV fusion process and that lead to entry blocking remain to be established. In this regard, it has recently been reported that the antiviral activity of a GT-rich phosphorothioate-modified ODN (ISIS 5652) against HSV-1 is mediated by a conformational change in glycoprotein B (gB) that results in loss of infectivity of viral particles (26). Furthermore, it has also been shown that β-peptides designed to bind to the heptad repeat segment of HCMV gB inhibit HCMV infection by blocking those homo-gB and/or -gB-gH protein-protein associations that are thought to be needed for fusion between the viral envelope and target cell membrane (4). Thus, similar to ISIS 5652 and gB β-peptides, CpG 2006 may represent an interesting candidate for further development as an inhibitor of HCMV entry.

CpG ODN signaling through TLR9 results in the activation of nuclear factor-κB (NF-κB) and AP-1 transcription factors, which in turn directly upregulate cytokine and chemokine gene expression (9). Interestingly, it has been reported that when transiently introduced in the murine macrophage cell line RAW 264.7, the transcriptional activity of the HCMV major immediate-early enhancer promoter (MIEP) element can be significantly stimulated by CpG 1826, through a mechanism that depends on NF-κB activation (13). The HCMV MIEP regulates expression of critical IE gene products during both productive viral replication and reactivation from latency (12, 16, 27). Several types of cellular transcription factors bind to multiple sites within the MIEP and modulate (both activating and repressing) its activity (15). Among these, activation of the NF-κB pathway and binding to the four NF-κB sites of MIEP have recently been observed to be required for productive HCMV replication in fibroblasts, as well as in endothelial cells (2).

Thus, CpG ODNs may exert dual activity on HCMV. In the present study, we have shown that they can potently suppress HCMV infection by blocking virus entry independently from activation of the TLR9 pathway. However, it can be hypothesized that in TLR9-expressing cell types, such as dendritic cells, the CpG ODN-induced signaling can stimulate MIEP activity and subsequent IE gene expression. Therefore, this dual CpG activity could be considered in the development of new options to more effectively control virus infection.

HCMV remains the most important pathogen in transplant recipients (12, 16, 17), and it has also been implicated in vascular disorders, such as transplant vasculopathy, restenosis, and atherosclerosis, characterized by endothelial cell activation, inflammatory cell infiltration, and smooth muscle cell proliferation (28). Standard therapy for HCMV disease is associated with considerable adverse effects, and prolonged treatment may lead to the emergence of drug-resistant strains. In addition, the currently used antivirals cannot prevent reactivation of latent HCMV infection or the expression of IE proteins which play crucial roles in viral pathogenesis and immunomodulation. The importance of IE functions and the inability of currently available antiviral therapies to prevent their expression have led to the suggestion that prevention of their expression and/or functions may provide an alternative strategy to inhibit HCMV reactivation, replication, and immunopathogenesis (25, 33). What is needed, therefore, is the identification of novel anticytomegaloviral agents that can block either HCMV entry and/or gene expression at very early stages, without causing major adverse effects.

Our results indicate that CpG ODNs (e.g., CpG 2006) may be attractive candidates for such a new class of antiviral drugs which exert their effects via a novel pathway that targets virus entry. Their potent anticytomegaloviral activity in vitro warrants further studies to evaluate whether CpG ODN treatment may result in antiviral activity in vivo, in animal models of acute infection, as well as reactivation from latency. Since in vitro experiments have demonstrated that CpG ODNs inhibit MCMV replication (Fig. 3), in vivo studies may be helpful in validating their potential in the control of HCMV infection.