Myeloid Dendritic Cells Repress Human Cytomegalovirus Gene Expression and Spread by Releasing Interferon-Unrelated Soluble Antiviral Factors

Abstract

Cytomegalovirus (CMV) is a betaherpesvirus that latently infects most adult humans worldwide and is a major cause of morbidity and mortality in immunocompromised hosts. Latent human CMV (HCMV) is believed to reside in precursors of myeloid-lineage leukocytes and monocytes, which give rise to macrophages and dendritic cells (DC). We report here that human monocyte-derived DC (mo-DC) suppress HCMV infection in coculture with infected fibroblast target cells in a manner dependent on the effector-to-target ratio. Intriguingly, optimal activation of mo-DC was achieved under coculture conditions and not by direct infection with HCMV, implying that mo-DC may recognize unique molecular patterns on, or within, infected fibroblasts. We show that HCMV is controlled by secreted factors that act by priming defenses in target cells rather than by direct viral neutralization, but we excluded a role for interferons (IFNs) in this control. The expression of lytic viral genes in infected cells and the progression of infection were significantly slowed, but this effect was reversible, indicating that the control of infection depended on the transient induction of antiviral effector molecules in target cells. Using immediate early or late-phase reporter HCMVs, we show that soluble factors secreted in the cocultures suppress HCMV replication at both stages of the infection and that their antiviral effects are robust and comparable in numerous batches of mo-DC as well as in primary fibroblasts and stromal cells.IMPORTANCE Human cytomegalovirus is a widespread opportunistic pathogen that can cause severe disease and complications in vulnerable individuals. This includes newborn children, HIV AIDS patients, and transplant recipients. Although the majority of healthy humans carry this virus throughout their lives without symptoms, it is not exactly clear which tissues in the body are the main reservoirs of latent virus infection or how the delicate balance between the virus and the immune system is maintained over an individual's lifetime. Here, for the first time, we provide evidence for a novel mechanism of direct virus control by a subset of human innate immune cells called dendritic cells, which are regarded as a major site of virus latency and reactivation. Our findings may have important implications in HCMV disease prevention as well as in development of novel therapeutic approaches.

Keywords: antiviral agents; cytomegalovirus; dendritic cells; fluorescent image analysis; video microscopy.

FIG 1

Mo-DC suppress spread of the dual-late-reporter HCMV^dLr virus in coculture with infected human MRC-5 fibroblasts. Human mo-DC were cocultured at the indicated E/T ratios with MRC-5 cells infected with 375 PFU of HCMV^dLr (MOI of 0.0125). Mo-DC were added immediately upon removal of the virus. The numbers of GFP/mCherry-expressing cells at 5 to 6 days postinfection in coculture with conventionally generated (A) or rapidly generated (B) mo-DC are shown. Dot plots depict combined data from at least three independent experiments with mo-DC generated from 6 (A) or 17 (B) PBMC donors. Horizontal lines denote group means, and error bars indicate the standard deviations. (C) Cocultures of infected MRC-5 cells and mo-DC at the indicated E/T ratios in six-well replicates were infected with 1,050 PFU of HCMV^IEr (MOI of 0.035). Virus titer in the SN at 6 dpi was established by plaque assay. Symbols indicate individual replicates; bars indicate mean values. Each group was statistically analyzed by nonparametric two-way ANOVA (Kruskal-Wallis) with Dunn's multiple-comparison test against the control (without [w/o] mo-DC) group (*, P < 0.05; ***, P < 0.001; ****, P < 0.0001; ns, not significant). (D) GFP/mCherry-expressing cells were quantified in MRC-5 cells infected with 45 PFU (MOI of 0.0015) of HCMV^dLr in the presence of mo-DC at 6, 9, 12, and 15 dpi (+ mo-DCs), and levels were compared to those of controls infected in the absence of mo-DC (w/o DCs) or those where mo-DC were removed (+/− mo-DCs) at 6 dpi. Plotted are combined data from two independent experiments (means with standard deviations). (E) Cells were infected as described for panel D, but only 6 PFU per well (MOI of 0.0002) was added to MRC-5 cells in 12-well replicates. A well was reported as positive if ≥1 fluorescent cell was observed. The plot depicts combined data from four independent experiments.

FIG 2

Mo-DC suppress the spread of the immediate early reporter HCMV^IEr virus in coculture with infected human fibroblasts. (A) Mo-DC were cocultured at increasing E/T ratios with HCMV^IEr-infected MRC-5 cells at 150 PFU/well (MOI of 0.005). mNeonGreen-expressing infection foci were manually quantified in each well at 2 to 3 dpi. The dot plot depicts combined data from four independent experiments with mo-DC generated from five donors; values are means with standard deviations. Kruskal-Wallis one-way ANOVA with Dunn's multiple-comparison test for each data set against the control data set (without mo-DC) was used (***, P < 0.001; ****, P < 0.0001). Representative images are shown from cocultures infected at an MOI of 0.035 (1,050 PFU/well) at 8 (B) and 23 (C) dpi. (D) As described in the legend of Fig. 1E, 6 PFU/well (MOI of 0.0002) of HCMV^IEr was added to MRC-5 cells in 12-well replicates. A well was reported as positive if ≥1 fluorescent cell was observed. The plot depicts combined data from five independent experiments (n = 2 for day 21).

FIG 3

The antiviral factors from mo-DC delay the onset and repress HCMV IE gene expression in fibroblasts. MRC-5 fibroblasts were infected with HCMV^IEr at an MOI of 0.035 using centrifugal enhancement, and medium (blue line) or the antiviral supernatant (from coculture of infected MRC-5 cells and mo-DC, referred to as SN°) (red line) was immediately added to them; infection was followed by epifluorescence time lapse imaging in 24 fields per condition at the rate of one frame per 20 min. (A) Solid lines show three-point smoothing of the average signal ratio (ratio of signal from infected cells to the background signal of the field) from 24 fields, with standard deviations of the average up to 35 h postinfection (p.i.). A representative plot from one of three independent experiments is shown. (B) The signal ratio in untreated cells deviated from the baseline signal ratio (dashed line) at around 3 hpi (blue arrow) while this occurred at >5 h for SN°-treated cells (red arrow). (C) Scatter dot plots show the distribution of signal onset in each field of untreated (blue) and SN°-treated (red) conditions. An unpaired nonparametric t test (Mann-Whitney test) was used (****, P < 0.0001). (D) Representative time series image montages from panel A up to 174 hpi. IE1/2-associated mNeonGreen signal in the GFP channel is depicted in false colors (Fire Blue Green look-up table) for better visibility. Scale bar, 100 μm.

FIG 4

Blocking IFN-α/β receptor increases IFN-β availability in coculture medium yet does not negatively affect the antiviral function of mo-DC. (A) IFN-β concentration was measured by ELISA in coculture SNs from MRC-5 cells infected with 1,050 PFU of HCMV^IEr (MOI of 0.035) and cocultured with or without mo-DC in the presence or absence of 10 μg/ml anti-IFNAR2 for 14 days. Bars are representative of one of two independent experiments. (B) Infectious virus titers were measured in SNs from the experiment shown in panel A. Representative data from one of three independent experiments show means with standard deviations. (C) MRC-5 cells in eight-well replicates were infected with 150 PFU (MOI of 0.005) of HCMV^IEr and cocultured with mo-DC in the presence or absence of 10 μg/ml anti-IFNAR2, and the number of infection foci was quantified at 3 dpi. Data are from one of three independent experiments showing means with standard deviations. Two-way repeated-measures ANOVA with Sidak's multiple-comparison test was used (ns, P ≥ 0.05; **, P < 0.01). (D) Representative images from the experiment shown in panel C at 8 dpi.

FIG 5

Pretreatment with the antiviral SN improves control of infection. (A) Antiviral SN°s from two donors (H. BS22 and H. BS23) were added to MRC-5 cells at 24 h before infection and then replaced with medium 21 h later (pretreatment; white bars). Synchronized infection of cells with HCMV^IEr at an MOI of 0.035 in medium or SN° (added at infection; gray bars) was carried out, and SN° was added to all conditions after the virus was removed (posttreatment; black bars). (B) Gating strategy for flow cytometric analysis. Bars show the percentage of infected cells in four replicates (C) and the mean fluorescence intensity of the infected cells (D) for each condition. Regular two-way ANOVA with Tukey's multiple comparisons between all data sets was used in one representative of two independent experiments (***, P < 0.001; ****, P < 0.0001). FSC, forward scatter; SSC, side scatter; H, height; A, area; H. BS, human blood sample.

FIG 6

Mo-DC contact with infected MRC-5 cells is required for stimulation of mo-DC antiviral function and does not involve IFN. (A) Representative images at 5 dpi (A) and quantification at 5 to 6 dpi of the spread of infection HCMV^dLr (MOI of 0.0125) (B) in MRC-5 cells treated, from left to right, with culture medium as a control, with SN of an mo-DC/MRC-5 coculture at an E/T ratio of 10:1 infected with HCMV^dLr (MOI of 0.001) or uninfected, or with SN of a culture of 1 × 10⁶ mo-DC infected at an MOI of 1 collected at 6 to 8 dpi (the experimental design is depicted in Fig. S1B in the supplemental material). Data represent at least three independent experiments with mo-DC generated from six different PBMC donors, showing means with standard deviations. Representative images at 8 dpi (C) and quantification at 3 dpi of the spread of HCMV^IEr (MOI of 0.005) infection (D) in cells treated as described for panel B. (E) Concentrations of IFN-α2, -β, -γ, -λ1, and -λ2/3 were measured in the SNs using a multianalyte flow assay (LEGENDplex), with n ≥ 9 samples for antiviral SN and n ≥ 3 for other conditions. Differences were not statistically significant. Kruskal-Wallis ANOVA with Dunn's multiple-comparison test for each column against the control column (culture medium) was used (ns, P ≥ 0.05; ****, P < 0.0001).

FIG 7

Mo-DC cocultures and their coculture supernatants suppress HCMV spread in primary human cells. Primary human cells were infected with 300 PFU (MOI of 0.01) and cocultured with mo-DC (A); alternatively, SNs were added to the infected cells, and the spread of infection was quantified at 5 dpi (B) (as described in the legend of Fig. 6). Combined data from two independent experiments showing means with standard deviations are shown. Two-way ANOVA with Dunnett's multiple-comparison test was used (ns, P ≥ 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).