Human cytomegalovirus inhibits maturation and impairs function of monocyte-derived dendritic cells

Dendritic cells (DCs) are the most potent professional antigen-presenting cells (APCs) playing a central and unique role in the generation of primary T-cell responses.

Whereas virus infections in general stimulate DC maturation and lead to the generation of efficient antiviral T-cell responses, some viruses have developed mechanisms to interfere with the normal functions of DCs. Human immunodeficiency virus (HIV), measles virus, herpes simplex virus-1 (HSV-1), and vaccinia virus, for example, are all known to cause defects in DC maturation and function which consequently lead to impaired antiviral T-cell responses.1-4 

The major risk factor for HCMV infection in bone marrow transplant patients is their pre-existing seropositive status.5 Delayed reconstitution of the antiviral cellular immune response in these patients is often associated with HCMV infection and progressive disease with poor prognosis. CD8⁺T cells are the main cell type controlling HCMV infection as proven by their successful adoptive transfer preventing viraemia and HCMV disease in bone marrow transplant recipients.6 HCMV infection is also associated with serious secondary bacterial or fungal infections in immunosuppressed patients,7,8 although during the early phase of acute HCMV infection, transiently suppressed cellular immune responses9,10 and immunologic abnormalities of healthy individuals11 were also reported. HCMV has been shown to interfere with antigen presentation in fibroblasts and endothelial cells12 by (1) expressing viral proteins which interfere with the MHC class I antigen–presenting pathways (US3, US2, US11); (2) modulating antigen procession of the HCMV immediate early 1 (IE1) protein (pp65); and (3) preventing peptide transport through TAP proteins (US6). However, little is known about whether the induction phase of HCMV-specific T-cell responses is also affected. From the few studies that looked at HCMV infection of monocytes, macrophages, and DCs,13-15 only one seemed to establish a link between HCMV infection of macrophages and impaired proliferation of CD4⁺T cells upon stimulation with these macrophages.15 

We hypothesized that similar to its effects in fibroblasts and endothelial cells, HCMV also inhibits the expression of MHC class I molecules in DCs. Although this inhibitory effect alone would probably be sufficient to prevent or impair the generation of virus-specific primary T-cell responses, other important features such as the expression of costimulatory molecules, the production of proinflammatory cytokines, and the antigen-presenting function of monocyte-derived DCs were also studied following HCMV infection. We show that HCMV impairs all these properties of DCs. The viral inhibition may contribute not only to the successful establishment of latency by HCMV but also to the immunosuppression observed in association with the infection.

HCMV seropositive and seronegative healthy laboratory volunteers were included in this study. Ethical approval was obtained and informed consent was provided according to the Declaration of Helsinki. HCMV serostatus was determined by a high sensitivity IgG enzyme-linked immunosorbent assay (ELISA) (Department of Medical Microbiology and Public Health Laboratory, University Hospital of Wales, Cardiff, United Kingdom). The HLA types for T-cell assays were determined by microlymphocyte cytotoxicity assay (Welsh Blood Transfusion Service, Cardiff, United Kingdom) or by polymerase chain reaction–sequence-specific primers (PCR-SSP).16 

Human foreskin fibroblasts (HFFs) were cultured in minimum essential medium (MEM; Sigma, St Louis, MO) supplemented with 10% fetal calf serum (FCS; PAA Laboratories, Linz, Austria), 100 IU/mL penicillin, 100 μg/mL streptomycin, 2 mM L-glutamine, 25 mM Hepes buffer, sodium pyruvate, and nonessential amino acids. HFFs were used for experiments between passages 10 and 30. B lymphoblastoid cell lines (BLCLs) were prepared according to the standard method by infecting peripheral blood mononuclear cells (PBMCs) with Epstein-Barr virus (EBV)–containing B95.8 cell supernatant. The tCD40L cell line, which is a murine L fibroblast line transfected with human CD40L, was kindly provided by M. Rowe (University of Wales, Cardiff, United Kingdom). The BLCLs and tCD40L cells were cultured in complete RPMI (RPMI-1640; Sigma; supplemented as above except for nonessential amino acids) containing 10% FCS.

HCMV_TB40/E is an endothelial cell–adapted HCMV strain isolated from a bone marrow transplant recipient. It was kindly provided by Dr Christian Sinzger (Tübingen, Germany).17 HFFs were grown to about 90% confluence and infected with a low multiplicity of infection (MOI) (0.1-1 plaque-forming unit [PFU]/cell) of HCMV_TB40/E in 4 mL of MEM for 1 to 2 hours at 37°C. The virus was then removed and the cells cultured in MEM plus 10% FCS until the first cytopathic effects appeared (4-7 days). Supernatants of infected HFFs were collected daily, or every second day, until the cytopathic effects became advanced. Supernatants were centrifuged at 1200g for 10 minutes to remove cells and debris and the supernatant was stored at −70°C in small aliquots. In some experiments a concentrated virus stock was used which was obtained from cell-free supernatant by ultracentrifugation at 80 000g for 70 minutes at room temperature in a Beckman L8-M ultracentrifuge using an SW41 rotor (Beckman Instruments, Fullerton, CA). The pellet containing viral particles from 20 mL supernatant was resuspended in 1 mL complete RPMI. Viral titres, determined from the TCID₅₀values (dilution of virus required to infect 50% of cultures) on HFF cells, grown in 48-well plates for 2 to 3 weeks, varied between 5 × 10⁶/mL and 5 × 10⁸/mL. HCMV_TB40/E was inactivated at room temperature by a 2-minute UV irradiation (253 nm, 30 W/G30 T8 tube by Philips, Almelo, The Netherlands).

Influenza A (AX/31) virus was kindly provided by Dr D. B. Thomas and Dr J. Skehel (National Institute for Medical Research, London, United Kingdom). The virus has been grown in the allantoic cavity of embryonated chicken eggs. The viral hemagglutination titer, determined by chicken red cell agglutination, was 1000 hemagglutinating units (HAU)/50 μL. The cell lines and the virus stocks were negative for contamination with mycoplasma as determined by a mycoplasma PCR kit (VenorGem; Minerva Biolabs GmbH, Berlin, Germany).

PBMCs were isolated by Ficoll-Histopaque (Sigma) density gradient centrifugation of heparinized peripheral blood of healthy donors. DCs were generated from blood monocytes according to standard methods.18 Briefly, the adherent fraction of PBMCs was obtained after a 1 hour plastic adherence step (6-well plates, 15 × 10⁶ PBMCs/well) (Nunc, Naperville, IL) in complete RPMI containing 1% FCS. The nonadherent cells were removed and the adherent cells were cultured in complete RPMI containing 10% FCS, 50 ng/mL human recombinant granulocyte macrophage–colony-stimulating factor (GM-CSF) (Leucomax; Novartis Pharmaceuticals, East Hanover, NJ) and 500 U/mL human recombinant IL-4 (BD Pharmingen, San Diego, CA). Nonadherent and loosely adherent DCs were collected and used in the experiments after 4 to 7 days in culture. More than 90% of the cells were of DC phenotype (CD1a⁺, HLA-DR⁺, CD14⁻, CD80⁺) after gating on size and side scatter parameters by fluorescence activated cell sorting (FACS) to exclude lymphocytes. Maturation of DCs was induced by 1 μg/mL LPS (Sigma) in the presence of 100 ng/mL interferon gamma (IFN-γ) (Peprotech, Rocky Hill, NJ) and 50 ng/mL TNF-α (R&D, Minneapolis, MN) at 0.5 × 10⁶ to 1 × 10⁶ DCs/mL in complete RPMI containing 10% FCS.

DCs were counted and either infected with HCMV_TB40/Eat MOI = 1-10 PFU/cell or mock infected with MEM plus 10% FCS for 3 hours or overnight as indicated in the figure legends. After removal of the virus or MEM, DCs were washed and cultured in complete RPMI containing 10% FCS and 50 ng/mL GM-CSF for a total of 24 or 48 hours. For infection with influenza, 10⁵ DCs were incubated with 5 HAU influenza A (AX/31) in a total volume of 200 μL for 10 minutes on ice followed by 30 minutes at 37°C. DCs were washed once in serum-free complete RPMI and cultured in complete RPMI containing 10% AB serum (Sigma) for 6 hours before using them in the experiments.

To assess expression of surface molecules on DCs, 1 × 10⁵ to 2 × 10⁵ DCs were labeled for 40 minutes at 4°C with antibodies at 1 μg per sample or as recommended by the manufacturer. In order to detect viral antigens, intracellular staining of DCs, labeled as above for surface molecules or unlabeled, was carried out. DCs were fixed with 4% paraformaldehyde for 20 minutes at room temperature and permeabilized with 0.025% Triton X-100 for 20 minutes at room temperature. Nonspecific binding was blocked with 2% mouse serum for 10 minutes at room temperature followed by labeling with fluorescein isothiocyanate (FITC)– or phycoerythrin (PE)–conjugated antibodies for 1 hour at 37°C. Cytokine production by LPS-stimulated or infected DCs was detected by adding 1 μL/mL GolgiPlug (BD Pharmingen) to DCs 2 hours following infection or LPS treatment and culturing the cells for a further 18 to 20 hours. DCs were fixed, permeabilized, and blocked as above and HCMV pp52-FITC antibody was added together with either IL-12–PE or TNF-α–PE antibody for 1 hour at 37°C. The cells were analyzed on a FACSCalibur (BD Pharmingen) using CellQuest 3.1 software. The following antibodies were used for labeling: HCMV p52 early antigen-specific FITC-conjugated antibody (CCH2; Dako, Carpinteria, CA), anti-MHC class I/CyChrome, anti–HLA-DR/CyChrome, anti–CD80/CyChrome, anti–CD83/PE, anti–IL-12/PE and anti–TNF-α/PE (all from BD Pharmingen), anti–CD86/TriColour and anti-CD120b (TNFRII, 75 kd)/RPE (Caltag, Burlingame, CA), anti-CD40 (Diaclone, Besançon, France), PE-conjugated mouse IgG₁ (Immunotech, Luminy, France), and PE-conjugated rabbit anti–mouse IgG F(ab')₂ (STAR 12A; Serotec, Oxford, United Kingdom).

Immature DCs were either mock infected, infected with HCMV_TB40E, or UV-irradiated HCMV_TB40E was added. After 24 hours, the cells were either left unstimulated or were stimulated with LPS in the presence of IFN-γ and TNF-α, as above, for 24 hours. Cells were washed in phosphate-buffered saline. For nuclear extracts the cell pellet of 3 × 10⁵ to 5 × 10⁵ DCs per group was lysed in nuclear extraction buffer and an equal volume of 2x gel sample buffer was added to the clarified lysate. The proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidenefluoride (PVDF) for immunoblotting using an alkaline phosphatase chemiluminescent detection protocol. RelB and p50 were detected using rabbit polyclonal antibodies, kindly provided by Dr Nancy Rice (Frederick, MD). The polyclonal actin-specific antibody was purchased from Sigma.

Nonadherent PBMCs were kept frozen in liquid N₂ and later used as a T-cell source. T-cell stimulation was carried out in complete RPMI containing 10% AB serum (Sigma). DCs were either (1) infected with influenza for 6 hours; (2) infected with both HCMV_TB40/E (total of 48 hours) and influenza (6 hours); or (3) mock infected as described earlier. T-cells were added to DCs at a ratio of 10:1 at the concentration of 1.5 × 10⁶nonadherent PBMCs plus 1.5 × 10⁵ DCs per mL and cultured in 12-well trays at 3 mL/well for 7 days prior to testing the cytotoxic T lymphocyte (CTL) activity in a standard⁵¹Cr-release assay. Autologous BLCs were used as target cells. They were labeled with 135 μCi (5 MBq)⁵¹Cr/10⁶ cells for 1 hour at 37°C. BLCs were washed twice and incubated either with 50 μM HLA-A2–restricted influenza M1 58-66 peptide in 5 μL dimethyl sulfoxide (DMSO) or with 5 μL/mL DMSO without peptide for another hour. Targets were plated out at 3 × 10³ BLCs/well. T cells, cultured for 7 days with DCs, were washed, counted, and added at gradually decreasing numbers to target cells, starting at a 40:1 effector-target (E/T) ratio, in triplicates. The cells were incubated for 4 hours at 37°C in 200 μL final volume per well. A 25-μL aliquot of the supernatant from each well was transferred into wells of 96-well soft plates and mixed with 130 μL Betaplate scintillation fluid. Radioactivity was measured on a β-plate counter (1450 Microbeta; Wallac, Milton-Keynes, United Kingdom). The lysis of targets from each well was calculated as follows: (experimental release − spontaneous release) / (maximum release − spontaneous release) × 100, and expressed as percent specific lysis. Spontaneous release and maximum release of⁵¹Cr were determined by incubating target cells either with medium only or with 5% Triton X-100. T-cell proliferation was measured after stimulation of the nonadherent PBMCs with DCs generated and infected as described above. A quantity of 10⁵ autologous nonadherent PBMCs per well was coincubated with 5 × 10³, 2.5 × 10³, 1.25 × 10³, 6.25 × 10², or 3.1 × 10² DCs/well in triplicates. ³H-thymidine (Amersham Pharmacia, Piscataway, NJ) was added at 0.5 μCi/well (0.018 MBq) on day 5 of the culture for 16 hours. The cells were harvested onto fibroglass filtermats and ³H-thymidine incorporation was measured by a Wallac 1450 Microbeta β-plate counter.

Statistical analysis was performed using the Studentt test. Differences were considered statistically significant withP < .05.

HCMV infection does not induce maturation of DCs: the expression of cell-surface molecules, especially that of MHC class I, CD40, and CD80, is decreased on DCs following infection. Viral infections can either induce or suppress the maturation and functional activation of DCs. To study the effect of HCMV on DCs, we first compared the expression of surface markers on DCs infected with HCMV_TB40/E (Figure 1A, second row) with that on mock-infected DCs (Figure 1A, first row). The infection rate of DCs varied between 30% and 70% in the experiments and thus it was necessary to analyze HCMV Ag-positive DCs (Figure 1A, second row, right side of dot plots) separately from Ag-negative DCs (Figure 1A, second row, left side of dot plots). The summary of results from 3 to 4 repeated experiments (Figure 1B) shows that the relative fluorescence intensity (RFI) of surface MHC class I molecules on Ag-positive DCs is only 20% of that on mock-infected DCs. The RFI of the expression of CD40 and CD80 molecules (Figure 1B) was also less than 50%. The expression of MHC class II and CD86 molecules was also weaker on Ag-positive than on mock-infected DCs, whereas CD83 expression was low on mock-infected DCs and remained low following infection (Figure 1A). The level of expression of all the molecules mentioned above was at least as high or higher on the Ag-negative DCs (Figure 1A, second row, left side of dot plots) than on mock-infected DCs (Figure 1A, first row), indicating that in the absence of viral infection bystander DC activation by binding of viral products or soluble factors may occur. Higher level of TNFRII expression on Ag-positive than on Ag-negative DCs indicated that the down-regulation of surface molecules was not due to a general shut-down of protein synthesis by the virus (data not shown).

HCMV infection of DCs inhibits LPS-induced DC maturation. As shown in Figure 1, HCMV does not induce maturation of immature DCs, unlike, for example, adenoviruses or dengue virus. The ability of immature DCs to begin the differentiation process into mature DCs in response to certain proinflammatory cytokines (TNF-α, IFN-γ), or to bacterial products (LPS) and probably to other stimuli is crucial in the generation of antiviral immune responses and in combating bacterial infections. Therefore, we investigated the effect of HCMV infection on the maturation of DCs by first infecting DCs with HCMV_TB40/E and then inducing maturation with LPS in the presence of IFN-γ and TNF-α. Despite the strong activation signal, HCMV Ag-positive DCs failed to up-regulate MHC class I, MHC class II, costimulatory, and CD83 molecules (Figure2A, second row, right side of dot plots) to the same level that mock-infected (Figure 2A, first row) or Ag-negative DCs did (Figure 2A, second row, left side of dot plots). The greatest inhibitory effect on up-regulation can be observed with MHC class I, CD80, CD86, and CD83 molecules (Figure 2B; RFI of these molecules on HCMV Ag-positive DCs was 31%, 30%, 53%, and 17%, respectively) whereas MHC class II expression was inhibited to a lesser extent (RFI 77%, Figure 2B). Down-regulation of the CD40 molecules upon infection seemed to vary more among individuals. As UV-inactivated HCMV did not inhibit LPS-induced DC maturation, measured by the surface expression of MHC class I, class II, CD40, CD86, and CD83 molecules (data not shown), our results indicate that active viral infection by HCMV of immature DCs is inhibiting DC maturation.

Mature DCs are relatively resistant to infection by HCMV. Immature DCs capture antigens in the periphery, and during maturation DCs gradually lose their ability to take up antigens by phagocytosis. However, they still may be susceptible to viral infections which could impair DC function. We mimicked this scenario by exposing DCs first to maturation signals provided by LPS, TNF-α and IFN-γ and then infecting them with HCMV 24 hours later. The control immature DCs were infected with HCMV in a dose-dependent manner, the proportion of HCMV early antigen-positive DCs reaching 38% when the virus was added at 10 MOI (Figure 3A). By contrast, only 1.4% of LPS-matured DCs were susceptible to infection at the same MOI of the virus (Figure 3B), and further increase of infectious virus up to MOI = 50 and MOI = 100 also failed to significantly increase the proportion of infected mature DCs (data not shown). Although the proportion of infected DCs was low following LPS treatment, FACS analysis of these DCs revealed lower levels of cell-surface MHC class I, costimulatory, and CD83 molecules than in mock-infected LPS-treated DCs (not shown). As the phenotype of DCs which became infected after treatment with LPS resembled the phenotype of Ag-positive immature DCs, it would be difficult to determine whether (1) up-regulation of maturation-associated molecules was inhibited due to the infection with the virus or (2) these DCs represent a small population of cells that remained immature with immature phenotypes and were more susceptible to infection with HCMV.

DCs produce no IL-12 and only low levels of TNF-α following infection with HCMV. An important function of DCs during viral and bacterial infections is the production of proinflammatory cytokines, such as IL-12 and TNF-α. IL-12 enhances the proliferation and cytolytic activity of NK cells and memory T cells at the site of infection and regulates Th1 cell development in lymphoid organs.19TNF-α mediates IL-12 production, DC maturation, and migration.20 Some pathogens are known to be efficient inducers of IL-12 and TNF-α production by DCs, but the effect of HCMV on the production of these cytokines has not been studied. Here we show that Ag-positive DCs do not produce IL-12 (0.2% vs 0.3% by mock-infected DCs) (Figure 4A, first and second panel; Figure 4C) within the first 24 hours following infection, whereas 18.9 ± 6.6% LPS-treated DCs produced IL-12 (Figure 4C, n = 3). UV-inactivated HCMV was also ineffective in the stimulation of IL-12 production (Figure 4A, third panel). The proportion of DCs producing TNF-α was much higher (67.6 ± 29%, n = 4, Figure 4D; also Figure 4B) than that of DCs producing IL-12 following LPS treatment. A low but statistically significant proportion (14 ± 6.7%, Figure 4D) of the Ag-positive DCs produced TNF-α following HCMV infection. However, viral replication did not seem to be important for the stimulation of TNF-α production, as Ag-negative DCs (8.1 ± 3.3%, Figure 4D) and those that were exposed to UV-inactivated HCMV (26.5%, Figure 4B, third panel) also produced TNF-α. To exclude the possibility that the production of IL-12 and TNF-α peaked later than the first 24 hours following infection with HCMV, intracellular cytokine staining experiments, identical to those shown in Figure 4A-B, were carried out 2 and 3 days after infection. No IL-12 production was detected by HCMV-infected Ag-positive or Ag-negative cells on either day, and the proportion of TNF-α–producing DCs decreased to background level by day 2 and day 3 after infection (data not shown).

HCMV-infected DCs remain unresponsive to further stimulation to produce cytokines. Some pathogens stimulate DCs to produce IL-12 with early kinetics followed by a refractory period when DCs become unresponsive to further stimulation to produce cytokines.21 This can lead either to some degree of immunosuppression, similar to that observed in IL-12R–deficient patients,22 or to protection of the host against the pathogenic events of prolonged inflammation.21 On the other hand, interaction with certain pathogens or with their products “conditions” DCs to be susceptible to second signals such as T-cell encounter (DC40-CD40L interaction) which then leads to the production of high levels of IL-12.23 HCMV did not induce an early peak of IL-12 production and only a low level of TNF-α was produced following infection (Figure 4). We studied whether the interaction of DCs with HCMV influenced the ability of these cells to produce IL-12 and TNF-α upon a secondary stimulation. DCs were mock infected or infected with HCMV_TB40/E for 48 hours before stimulation with LPS or with CD40L, and intracellular cytokine production was detected by FACS analysis. A representative experiment showing the original data (Figure5A) and results calculated from repeated experiments (Figure 5B; n = 3) show that HCMV-infected DCs, 48 hours after infection, produced significantly reduced levels of IL-12 and TNF-α upon stimulation with LPS (Figure 5B), compared with mock-infected DCs. Cytokine production was inhibited both in Ag-positive and in Ag-negative DCs, suggesting that this inhibitory effect does not require active viral replication.

IL-12 and TNF-α production upon CD40L binding following HCMV infection were also studied. Mock-infected, unstimulated DCs or DCs prestimulated with LPS produced IL-12 and TNF-α upon binding CD40L on transfected fibroblasts (Figure 5C, first and second columns), which confirms that IL-12 production is higher if DCs encounter both a microbial product and CD40L, whereas TNF-α is produced mainly upon stimulation of DCs with pathogens. However, HCMV-infected DCs produced neither IL-12 nor TNF-α upon stimulation with CD40L (Figure 5C, last 2 columns). Cytokine production was inhibited in both the Ag-positive and Ag-negative DC populations, suggesting again that the viral inhibition of cytokine production did not depend upon the active replication of the virus.

HCMV induces changes in the level of nuclear NF-κB proteins in DCs. Cytokine production by DCs is controlled by NF-κB nuclear factors which, with the exception of p65, are markedly up-regulated in uninfected DCs upon LPS treatment.24 We examined whether the inhibition of cytokine production was possibly due to a viral interference of the nuclear accumulation of NF-κB. The expression of RelB and p50 proteins in the nuclear fraction of DCs, which were mock infected or infected with HCMV or to which UV-inactivated HCMV was added, was studied before or after LPS treatment. Both infectious and UV-inactivated HCMV caused increased levels of RelB in the nucleus of immature DCs (Figure 5D, left panel, lanes “+” and “UV” vs “–”). LPS treatment of DCs resulted in the accumulation of nuclear RelB in mock-infected, HCMV-infected, and in UV-inactivated virus-treated DCs (Figure 5D, right panel). Similar results were obtained with p50 antibody (not shown). The results show that although HCMV and its viral products cause an early nuclear translocation of NF-κB factors in immature DCs which is followed by further NF-κB accumulation upon LPS treatment, the increase in the level of NF-κB proteins is not sufficient to result in cytokine production by these DCs.

T-cell proliferation and cytotoxicity are impaired following stimulation with HCMV-infected DCs. The consequences of DC infection by HCMV, such as interference with the expression of DC surface molecules and inhibition of maturation and cytokine production, probably have the main impact on the generation of primary antiviral T-cell responses. In the absence of an in vitro model to prime naı̈ve T cells, we investigated whether HCMV-infected DCs were impaired in their ability to induce MHC class I–mediated T-cell responses to a recall antigen in vitro (Figure 6). Secondary CD8⁺ T-cell responses do not require costimulation and thus the outcome of such an experiment would mainly be influenced by the down-regulation of MHC class I molecules on infected DCs. Immature DCs from an HCMV seronegative donor were infected with HCMV_TB40/E (MOI = 10) for 48 hours and then incubated with influenza virus for 6 hours prior to adding autologous T cells. Influenza-specific T-cell proliferation at day 5 following stimulation was reduced when HCMV-infected DCs were used as APCs, compared with proliferation induced with mock infected DCs (Figure 6A). A CTL assay, with T cells from day 7 cultures against M1 peptide–pulsed autologous BLCL targets revealed that HCMV-infected DCs had a much-reduced ability to stimulate cytotoxic T-cell responses to influenza virus (Figure 6B, triangles) than did mock-infected DCs (Figure 6B, circles), confirming that antigen presentation by DCs and their ability to stimulate a secondary T-cell response are impaired following HCMV infection. The results for (A) and (B) are representative of 3 similar experiments.

In this paper, we demonstrate that HCMV interferes with the maturation and antigen-presenting function of DCs. Following HCMV infection of DCs in vitro, we observed down-regulation of MHC class I molecules on immature and on LPS-treated DCs; inhibition of DC maturation, as indicated by the lack of up-regulation of MHC class II and costimulatory molecules (CD40, CD80, CD86) and CD83 on DCs; impaired proinflammatory cytokine production (IL-12 and TNF-α) and impaired antigen-presenting function, as measured by decreased CTL activity and T-cell proliferation following stimulation of T cells with HCMV-infected DCs.

HCMV decreases the expression of MHC class I molecules on fibroblasts12 and on endothelial cells25 and partially downregulates surface expression of MHC class I molecules on human macrophages26 by multiple mechanisms. The finding of down-regulation of MHC class I molecules on DCs is in agreement with these studies. Our results indicate that MHC class I down-regulation is an important viral mechanism not only in avoiding CTL killing of infected target cells but also in inhibiting, or at least delaying, the generation of virus-specific primary T-cell responses.

Constitutive MHC class II expression is restricted to B cells, monocytes, and thymic epithelial cells, whereas IFN-γ induces MHC class II expression on some other cell types. IFN-γ–induced MHC class II expression is impaired in HCMV-infected endothelial cells due to inhibition of the Jak/STAT pathway.27 Furthermore, HCMV gpUS2 causes degradation of 2 essential MHC class II proteins (HLA-DR-α and DM-α),28 which leads to impairment of antigen-presentation to CD4⁺ T cells. We studied both the constitutive expression of MHC class II on immature DCs following HCMV infection and the effect of infection on up-regulation of MHC class II by LPS/IFN-γ/TNF-α treatment. Unlike the strong inhibitory effect of HCMV on MHC class I expression in DCs, constitutive expression of MHC class II was only slightly impaired by HCMV. Induction of MHC class II up-regulation with LPS in the presence of IFN-γ and TNF-α was also only partially inhibited in HCMV-infected DCs. One possible explanation for the moderate MHC class II down-regulation observed by us is that class II molecules may become down-regulated with different kinetics from those of MHC class I; thus, a longer incubation after infection may have revealed more profound effects.

This is the first study to examine the effect of HCMV infection on the expression of costimulatory molecules on DCs. We demonstrate here that HCMV infection causes down-regulation of CD40, CD80 and, to a lesser extent, of CD86 molecules on immature DCs. Ligation of CD40 molecules on DCs by the CD40L on T cells normally results in the terminal differentiation of DCs. The interaction strongly increases the expression of MHC class I, class II, and costimulatory molecules on DCs and the production of cytokines (IL-12, TNF-α, IL-8) that are necessary for full activation, differentiation, and cytokine production of naı̈ve CD4⁺ T cells.29,30 In the absence of CD4⁺ T cells and therefore of CD40L-mediated help, the generation of memory CTL is also often compromised.31 Lack of costimulation via the CD80 and/or CD86 and CD28 interaction after TCR triggering may result in anergy instead of activation of T cells, marked by their inability to proliferate and to produce cytokines.32 We found that both CD80 and CD86 expression remain at low levels in HCMV-infected DCs upon stimulation with LPS. This might have a profound effect on the priming of HCMV-specific T cells either by causing anergy and/or apoptosis or by delaying the development of virus-specific primary T-cell responses, allowing enough time for the virus to reach the site of latency. T-cell unresponsiveness/anergy and an increased rate of T-cell apoptosis have in fact been reported in HCMV-infected transplant patients.33 

The production of TNF-α by activated DCs plays an important role in the maturation and migration of DCs and in the induction of proliferation and cytolytic activity of NK and T cells. On the other hand, increased levels of TNF-α have been shown to up-regulate IE enhancer/promoter activity in undifferentiated monocytic cells34 and in the lungs of latently infected mice.35 In our experiments only a small proportion of DCs produced TNF-α following HCMV infection, and it was not sufficient to drive the differentiation of infected DCs. Although we have not studied the effect of TNF-α on HCMV replication in DCs, it is possible that TNF-α had a stimulatory effect on HCMV IE gene expression but not on DC differentiation. In this study we also showed that infection of DCs with HCMV does not induce IL-12 production, which may lead to the lack of stimulation of NK and memory T cells at the site of infection. DCs also remain unresponsive to subsequent stimulation with LPS or with CD40L to produce IL-12 and TNF-α. Lack of IL-12 production, especially in the presence of viral IL-10,36 could cause a Th2/Th1 shift. Our observations may help in explaining the lack of Th1-type responses and the significantly up-regulated Th2-type cytokine production in bone marrow transplant recipients with HCMV-associated fatal interstitial pneumonitis.37 

DCs represent an ideal target for pathogens because interference with the maturation process and function of DCs can have severe and multiple consequences for T-cell and B-cell responses. Measles virus is one of the best known examples. Although the virus induces maturation of DCs,38 it inhibits IL-12 production by activated DCs, resulting in impaired capacity to stimulate T cells and causing transient but profound immunosuppression.2Other viruses, such as HSV-1 and vaccinia,3,4 inhibit the maturation of DCs, resulting in impaired development of efficient antiviral T-cell immunity. The severe secondary bacterial or fungal infections7,8 following acute HCMV infection in immunosuppressed patients might be due to viral interference with DC function. Our results show that HCMV-infected DCs do not mature upon additional signals such as LPS and that there is a nearly complete inhibition of IL-12 and TNF-α production upon secondary stimulation with LPS or with CD40L. This could lead to impaired DC functions, resulting in a reduced ability to stimulate CTL responses, which we also found in our study. The activation of cytotoxic and helper effector memory T-cell responses, unlike the activation of naı̈ve or central memory T-cell responses, is independent from costimulation by CD80 and/or CD86 binding.39 HCMV infection of DCs in vivo is likely to result in suboptimal stimulation of the latter types of helper and cytotoxic T cells, too, and thus it would be more profound than the in vitro findings presented here.

The inhibition of DC maturation and function by HCMV and the low level of cell-surface expression of MHC class I molecules on DCs at any stage of their maturation seem to contradict the relatively high frequencies of T cells specific for HCMV antigens reported by several authors.40-43 Our previous report on cross-presentation of HCMV antigens by uninfected DCs to CD8⁺ T cells offers a likely explanation of how the immune system bypasses the inhibitory effects of HCMV.44 Our model described uninfected DCs acquiring viral antigens from infected fibroblasts, but it is probable that cross-presentation of viral antigens occurs between uninfected DCs and any infected cell, such as endothelial cells and even DCs. Cross-presentation may play a much more important role in the generation of virus-specific T-cell responses than previously thought.

Down-regulation of MHC class I molecules and the inhibition of up-regulation of maturation-associated molecules on DCs requires active viral infection because UV-irradiated HCMV did not have any inhibitory effect. Soluble inhibitory factors produced by the virus or by infected DCs can also be excluded because HCMV Ag-negative DCs from cultures containing infected DCs showed signs of bystander stimulation instead of inhibition, probably due to phagocytosed viral particles and or cellular debris. The search for viral genes responsible for the observed inhibition of cell-surface molecule expression on DCs is hampered by the fact that (1) the well-characterized laboratory strain of HCMV (AD169), which is used extensively by virologists, does not infect DCs17,44; and (2) recombinant adenovirus vectors, which would be useful for HCMV gene expression studies, cause maturation of DCs.45 The binding of HCMV or its purified ligands (gB and gH) to monocytes was shown to be sufficient in initiating nuclear translocation of NF-κB and activation of cellular factors leading to IL-1β production.46 We also observed accumulation of nuclear NF-κB factors in DCs which encountered either infectious or UV-inactivated HCMV. However, nuclear NF-κB accumulation did not lead to cytokine production in virus-treated DCs, suggesting a selective regulatory or inhibitory effect of NF-κB activity by HCMV. Finally, the straight correlation between cellular differentiation and permissiveness for productive HCMV infection, found in monocytes/macrophages,47 was not observed in DCs. We found that mature DCs are more resistant to HCMV infection than immature ones, indicating that mature DCs may acquire mechanisms that inhibit viral entry and/or replication. This increased resistance of mature DCs to infection can also be observed with some other viruses, for example, with vaccinia, HSV-1, and HIV,3,4 48but its nature is not known.

In summary, our study demonstrates the presence of an important viral evasion strategy by HCMV; namely, the impairment of the maturation and function of professional APCs. Interference with the stimulation of naı̈ve and memory T cells specific for HCMV antigens is likely to aid HCMV to escape T-cell recognition, whereas impaired DC function against other pathogens may contribute to HCMV-associated immunosuppression.

We thank C. Sinzger for the TB40E strain of HCMV and for stimulating discussions; N. Rice for the NF-κB polyclonal antibodies; G. Wilkinson and M. Rowe for helpful comments on the manuscript; and our donors for their support.