Compounds that target host cell proteins prevent varicella-zoster virus replication in culture, ex vivo, and in SCID-Hu mice

Abstract

Varicella-zoster virus (VZV) replicates in quiescent T cells, neurons, and skin cells. In cultured fibroblasts (HFFs), VZV induces host cyclin expression and cyclin-dependent kinase (CDK) activity without causing cell cycle progression. CDK1/cyclin B1 phosphorylates the major viral transactivator, and the CDK inhibitor roscovitine prevents VZV mRNA transcription. We investigated the antiviral effects of additional compounds that target CDKs or other cell cycle enzymes in culture, ex vivo, and in vivo. Cytotoxicity and cell growth arrest doses were determined by Neutral Red assay. Antiviral effects were evaluated in HFFs by plaque assay, genome copy number, and bioluminescence. Positive controls were acyclovir (400 microM) and phosphonoacetic acid (PAA, 1 mM). Test compounds were roscovitine, aloisine A, and purvalanol A (CDK inhibitors), aphidicolin (inhibits human and herpesvirus DNA polymerase), l-mimosine (indirectly inhibits human DNA polymerase), and DRB (inhibits casein kinase 2). All had antiviral effects below the concentrations required for cell growth arrest. Compounds were tested in skin organ culture at EC(99) doses; all prevented VZV replication in skin, except for aloisine A and purvalanol A. In SCID mice with skin xenografts, roscovitine (0.7 mg/kg/day) was as effective as PAA (36 mg/kg/day). The screening systems described here are useful models for evaluating novel antiviral drugs for VZV.

Figure 1. Neutral Red cytotoxicity assays

Subconfluent HFFs were treated in sets of 6 replicates for 48 h with compounds at the indicated concentrations. The concentration of diluent in each dose was equalized to the amount at the highest dose. Staurosporine (35nM) was used to induce apoptosis and served as a positive control for cell death. Average absorbance (540 nm) of each group at 48 h was divided by the average absorbance at time zero and recorded as fold change. The dashed line indicates fold change equal to 1, which corresponds to no change in cell number. A fold change less than 1 indicates cytotoxicity. Cytotoxic and cell growth arrest doses are listed in Table 1.

Figure 2. Effects of drug treatment on VZV yield by infectious focus assay and qPCR

Subconfluent HFFs were treated with the compounds in duplicate for 48 h. Virus yield was determined by infectious focus assay and plaque-forming units (pfu) per mL were calculated (left axis, points are average ± standard deviation). The number of virus genomes was determined in quadruplicate by real-time TaqMan quantitative PCR (qPCR) and compared to the β-globin gene copy number (right axis).

Figure 3

(A) Comparison of VZV-BAC-Luc growth curves by IVIS and infectious focus assay. HFFs were infected with VZV-BAC-Luc and bioluminescence signal was measured daily in 6-well plates by IVIS (total flux = photons/sec). Each day cells from duplicate wells were collected and then titered by infectious focus assay (pfu/mL). Results were graphed as VZV (pfu/mL) × 10³ on the x-axis and total flux (photons/sec) × 10⁹ on the y-axis. A line of best fit was plotted (solid line) with ±95% confidence intervals (dashed lines). R² = 0.920, indicating that 92% of the variance is shared between IVIS and infectious focus assay. Data obtained from the average of 2 wells. (B) Antiviral effects in culture and ex vivo measured by IVIS. HFFs or SOC infected with VZV-BAC-Luc were treated with the inhibitor panel. Bioluminescent images were taken daily using the IVIS and total flux was measured. Percent of total flux was calculated from the average of 3 wells or 3 pieces of skin per drug at 2 dpi: average total flux of drug was divided by the average total flux of VZV (DMSO) and multiplied by 100. All drugs tested exhibited antiviral activity. Mock samples were uninfected wells or DMSO treated skin. Input VZV was inoculum at Day 0. (C) Aloisine A and purvalanol A did not inhibit VZV replication in SOC. Growth curves in the SOC model were generated plotting total flux over 5 days. The trend of the resulting curves was typical of VZV growth (treated samples, dashed lines; controls, solid lines). Neither aloisine A (black squares) nor purvalanol A (black diamonds) reduced bioluminescent signals compared to the DMSO-treated controls (aloisine A control, open diamonds; purvalanol A control, open squares). (D) Representative bioluminescence images of HFFs and SOC. HFFs (top row) were infected with VZV-BAC-Luc and treated with DMSO (b) or the inhibitor panel (c–j) for 2 days. High total flux values (yellow and red) were observed in the control, indicating VZV growth, and low total flux values (blue and purple) were observed in all treated samples, indicating antiviral efficacy in culture. Skin (lower rows) was mock inoculated (a) or inoculated with VZV-BAC-Luc (b–j) and total flux was approximately equal in all samples on Day 0. By 2 dpi (bottom row), total flux increased in skin treated with DMSO (b), aloisine A (h), and purvalanol A (i), and decreased in all others. Treatments: DMSO (a, b), 400 μM acyclovir (c), 1 mM PAA (d), 1 mM L-mimosine (e), 50 μM DRB (f), 20 μM aphidicolin (g), 20 μM aloisine A (h), 20 μM purvalanol A (i), or 25 μM roscovitine (j). Images acquired using the IVIS-50 instrument.

Figure 4

(A) Representative bioluminescence images of SCID-Hu mice infected with VZV-BAC-Luc. A representative mouse from each of the 3 groups [diluent 1 (50% DMSO), PAA, and roscovitine] is shown at imaging Day 5, which corresponds to treatment Day 3. Bioluminescence signals were located on the left flank directly above the skin implant. Signal intensity was visibly reduced in both drug-treated mice compared to the diluent. Images were acquired with the IVIS-200 instrument; a pseudocolored image of photon emissions was overlaid on a grayscale photographic image of the mice. Scale: minimum = 2×10⁴ total flux, maximum = 1.3×10⁶ total flux. (B) Representative IVIS-200 results. Total flux was determined by drawing a ROI over the skin implant of the 3 mice shown in (A) and then plotted versus the imaging day. Although the bioluminescence signal for each of these mice began at approximately the same value, ~2×10⁴ total flux, the virus growth rate was slower in mice treated with PAA and roscovitine compared to the diluent. (C) PAA and roscovitine reduced VZV growth rate in vivo. The VZV growth rate was calculated for each mouse as Log₁₀ photons/sec/day and plotted by treatment group (individual symbols). The average growth rate was calculated for each group (solid lines). Combining diluent-treated mice from all experiments (n = 20) produced an average growth rate of 0.46 ± 0.12, which is typical of VZV in cultured cells. A range of doses of PAA (n=10 combined) and roscovitine (n=21 combined) significantly reduced VZV growth rates. Purvalanol A (n=10 combined) was not effective at reducing the rate of VZV growth. For details see Table 2. Data from 3 separate experiments are shown. Mann-Whitney U two-tailed test: p<0.01 = **, p<0.001 = ***. Diluent 1, 50% DMSO; Diluent 2, 50% DMSO and 25% EtOH in PBS.