ABSTRACT
Cytotoxic CD8+ T lymphocytes (CTL) directed against the matrix protein pp65 are major effectors in controlling infection against human cytomegalovirus (HCMV), a persistent virus of the Betaherpesvirus family. We previously suggested that cross-presentation of pp65 by nonpermissive dendritic cells (DCs) could overcome viral strategies that interfere with activation of CTL (G. Arrode, C. Boccaccio, J. Lule, S. Allart, N. Moinard, J. Abastado, A. Alam, and C. Davrinche, J. Virol. 74:10018–10024, 2000). It is well established that mature DCs are very potent in initiating T-cell-mediated immunity. Consequently, the DC maturation process is a key step targeted by viruses in order to avoid an immune response. Here, we report that immature DCs maintained in coculture with infected human (MRC5) fibroblasts acquired pp65 from early-infected cells for cross-presentation to specific HLA-A2-restricted CTL. In contrast, coculture of DCs in the presence of late-infected cells decreased their capacity to stimulate CTL. Analyses of DC maturation after either coculture with infected MRC5 cells or incubation with infected-cell-conditioned medium revealed that acquisition of a mature phenotype was a prerequisite for efficient stimulation of CTL and that soluble factors secreted by infected cells were responsible for both up and down regulation of CD83 expression on DCs. We identified transforming growth factor β1 secreted by late HCMV-infected cells as one of these down regulating mediators. These findings suggest that HCMV has devised another means to compromise immune surveillance mechanisms. Together, our data indicate that recognition of HCMV-infected cells by DCs has to occur early after infection to avoid immune evasion and to allow generation of anti-HCMV CTL.
Infection by human cytomegalovirus (HCMV), a member of the Betaherpesvirus family, is common and usually well controlled in healthy people, in whom the virus establishes latency and persistency. In contrast, patients whose immune systems are compromised, such as those undergoing bone marrow transplantation and newborns who are infected in utero, are especially susceptible to HCMV disease (for reviews, see references 6 and 23). Persistency of the virus is associated with a high frequency of cytotoxic CD8+ T lymphocytes (CTL) directed against the matrix protein pp65 (UL83) as detected in the blood of immunocompetent individuals (35). This is surprising, owing to the numerous escape mechanisms developed by the virus to prevent assembly and transport of HLA class I peptide complexes (1). To explain how a CD8+-T-cell response develops under these unfavorable conditions, we could suggest a role for dendritic cells (DCs) because of their unique ability to initiate CD8+-T-cell immune responses through unusual antigen uptake mechanisms. Indeed, it has been shown that DCs are able to capture antigens through many different pathways, including phagocytosis of apoptotic and necrotic cells and transfer from live cells, with subsequent cross-presentation to CTL (3). In this context, we demonstrated that immature DCs derived from peripheral blood mononuclear cells (PBMC) that were not susceptible to HCMV infection acquired pp65 through phagocytosis of infected apoptotic and necrotic bodies (2), providing antigenic epitopes for cross-presentation to CD8+ T cells. More recently, Tabi and coworkers confirmed our data, although they suggested that cross-presentation occurred through an unidentified mechanism (34).
Recruitment and localization of DCs at sites of inflammation and infection and migration to lymphoid organs are essential steps in the immunobiology of DCs. It is generally accepted that upon exposure to inflammatory stimuli secreted at the site of pathogen invasion, DCs acquire a maturation signal and migrate to regional lymph nodes. Obviously, the DC maturation process is a key step targeted by viruses in order to avoid an immune response (16). Throughout the infection process, HCMV can affect the functions of host cells as well as neighboring cells in particular through deregulation of cytokine production (1), which can disrupt DC maturation and subsequently the normal progress of the specific immune response. In this study, we examined whether HCMV could interfere with cross-presentation to anti-pp65 CTL.
Since we previously used artificially killed infected cells in experiments with cross-presentation by immature DCs, we first investigated whether virus-mediated events could induce activation of anti-pp65 CTL. Our second aim was to determine whether cross-presentation by DCs may be temporally regulated in coculture with HCMV-infected fibroblasts. To this end, immature DCs were added to fibroblasts infected by HCMV for various periods of time and then cocultured. We showed that in cocultures, DCs acquired pp65 from infected fibroblasts through a cell-to-cell contact-dependent mechanism and that cross-presentation was more efficient in the presence of early-infected cells than with late-infected cells. This time-dependent modulation of CTL activation was correlated with the regulation of DC maturation as assessed by the expression of CD83. We further demonstrated that up and down regulation of DC maturation and cross-presentation depended on soluble mediators contained in the supernatants of infected fibroblasts. Furthermore, we showed that transforming growth factor β1 (TGF-β1) secreted by late HCMV-infected fibroblasts was in part responsible for down regulation of DC maturation and T-cell activation. This finding reinforces our previous report speculating that recognition of infected cells by DCs at the beginning of the HCMV infectious cycle was responsible for the observed frequency of CTL against pp65 (24). Moreover, our data suggest that the regulation of DC maturation and subsequent cross-presentation may depend on the stage of infection at the time of the DC encounter at sites of virus entry.
MATERIALS AND METHODS
Virus and cells.HCMV strain AD169 was propagated in MRC5 human fibroblasts (BioMérieux, Marcy l’Etoile, France). Virus was collected when cytopathic effects were >90%. Supernatants were clarified by centrifugation at 1,500 × g for 10 min at 4°C and were stored at −80°C until they were used. The virus titer was determined by PFU titration in human foreskin fibroblasts (American Type Culture Collection) according to standard procedures. MRC5 (HLA-A2) cells and PBMC from blood donors were phenotyped for major histocompatibility complex (MHC) class I and class II by the Laboratoire Central d’Immunologie-Rangueil (E. Ohayon and M. Abbal, Rangueil Hospital, Toulouse, France). Infections with HCMV were performed at a multiplicity of infection (MOI) of 1 unless otherwise stated.
Generation of DCs.DCs were prepared from PBMC using a VacCell processor as described by Goxe et al. (13). Briefly, PBMC obtained from leukapheresis were cultured for 7 days in hydrophobic bags (Stedim, Marseille, France) in AIMV serum-free medium (Life Technologies, Cergy Pontoise, France) supplemented with granulocyte-macrophage colony-stimulating factor (GM-CSF; 500 U/ml; purchased from Novartis, Ruel Malmaison, France) and interleukin-13 (IL-13; 50 ng/ml; purchased from Sanofi-Synthelabo, Labège, France). Fresh IL-13 was added again after 4 days of culture. DCs were isolated on day 7 by elutriation. The isolation procedure gave rise to immature DCs with the phenotype CD1a+ MHC-I+ MHC-II+ CD64− CD83− CD80low CD86low, as described by Arrode et al. (2). DC purity was around 90%, and viability was >95%. Cytofluorimetric analysis of purified DCs revealed the absence of B lymphocytes (CD19+), macrophages (CD84+), and NK cells (CD56+) but contamination with red blood cells.
Assessment of cell death.Mycoplasma-free cells were cultured in either 24- or 6-well culture plates in RPMI 1640 (RPMI medium; Gibco, Cergy Pontoise, France) containing 10% fetal calf serum and supplemented with Glutamax-I, sodium pyruvate, and antibiotics, including the antimycoplasma antibiotic ciprofloxacine chlorhydrate (CIPRO; Bayer Pharma, Puteaux, France) (referred to as complete medium) unless otherwise stated. MRC5 cells were either mock infected or infected with HCMV AD169 for 6, 24, 48, or 72 h as described below. After incubation at 37°C under the conditions specified above, cells were detached with trypsin and added to those recovered from culture supernatant for washing with phosphate-buffered saline (PBS). The presence of dying cells was detected by multiparameter flow cytometry (Epics; Coulter Immunotech, Marseille, France) using propidium iodide (PI) according to the manufacturer’s instructions.
Phagocytosis assay.MRC5 cells were dyed red with PKH26 (Sigma) according to the manufacturer’s procedure and then infected with HCMV for various periods of time as indicated above or not infected. Red cells were then cocultured with immature DCs (one DC for one red cell) in complete medium supplemented with GM-CSF and IL-4 (a gift from J.-P. Marolleau, Paris, France) for 10 h at either 4 or 37°C. DCs were labeled with a fluorescein isothiocyanate-labeled mouse anti-HLA-DR antibody (L243; Coulter Immunotech) or with isotype control (immunoglobulin G2a-fluorescein isothiocyanate; Coulter Immunotech). Qualitative and quantitative phagocytosis of red cells by DCs was determined by flow cytometry and fluorescence microscopy.
Expansion of anti-pp65 CD8+ T cells from HCMV-seropositive donor PBMC.PBMC (2 × 106/ml) from an HCMV-seropositive healthy HLA-A2 donor were cultured in 24-well plates in RPMI medium containing 10% human AB serum, 1% modified Eagle medium nonessential amino acids (Gibco), and 10 mM HEPES (Gibco). pp65-derived peptide corresponding to a known CTL HLA-A2 binding epitope (NLVPMVATV [N9V]) was obtained from Neosystem (Strasbourg, France). The restimulation procedure for PBMC has previously been described in detail (2). On days 3 and 7, recombinant human IL-7 (a gift from Sanofi-Synthelabo) was added at a final concentration of 25 ng/ml, and CD8+ T lymphocytes were obtained by positive selection with anti-CD8 magnetic beads (Miltenyi Biotec, Paris, France) according to the manufacturer’s procedure. The purity (>90%) of the isolated CD8+-cell subset was determined by dual staining with anti-CD4 and anti-CD8 antibodies (Coulter-Immunotech) and flow cytometry analyses. Activation of anti-pp65 CTL was assessed through quantitation of secreted IFN-γ by enzyme-linked immunosorbent assay (ELISA) as described below.
Assay for cross-presentation.Immature DCs (5 × 105/well) obtained from HLA-A2 donors were added in medium supplemented with GM-CSF (100 ng/ml; Novartis) and IL-4 (50 ng/ml) to HCMV-infected or uninfected MRC5 cells (1:1 ratio) for 24 h. Then, the DCs were gently pipetted to dissociate them from adherent fibroblasts. The cells were washed, plated in duplicate on a 96-well plate at 5,000/well, and incubated for 24 h in the presence of the anti-pp65 T-cell line at different effector-to-DC ratios in a final volume of 200 μl. Alternatively, DCs were used either unloaded or pulsed overnight with 1 μM N9V peptide in the presence of tumor necrosis factor alpha (TNF-α; 50 ng/ml; R&D Systems, Abingdon, United Kingdom) for 24 h. To exclude the possibility of direct stimulation of T cells by HLA-A2-positive MRC5 cells that could be recovered during pipetting, MRC5 cells, either uninfected or infected with HCMV, were used as additional controls.
For transwell studies, immature DCs were added to 3-μm-pore-size transwell chambers (Becton Dickinson, Le Pont de Claix, France) inserted into wells containing HCMV-infected or uninfected MRC5 cells (1:1 ratio) and then cultured for 24 h. Alternatively, immature DCs were cultured for 24 h with clarified supernatants that were obtained from infected or uninfected fibroblasts and then incubated overnight with 1 μM N9V peptide or not incubated.
ELISA for IFN-γ secretion.ELISA for IFN-γ secretion by activated CTL was performed as follows. Ninety-six-well plates (Nunc) were coated overnight at 4°C with primary anti-IFN-γ monoclonal antibody (MAb; Biosources), washed, and blocked for 2 h at 37°C with the buffers provided (Medgenix screening line). Diluted (1/10) supernatants from CTL expansion experiments were added to precoated wells in duplicate and supplemented with secondary antibody (biotin-conjugated anti-IFN-γ MAb; Biosources) for 2 h. After the plates were washed with PBS-fetal calf serum (1%), streptavidin-bound peroxidase (Coulter Immunotech) was added and left for 30 min at room temperature. The plates were then washed and incubated for 5 min in chromogen buffer (Biosources). The optical density was measured on an ELISA apparatus (Dynatech Laboratories).
Analysis of DC phenotype.Mature DCs used as a control were produced by incubation for 24 h with medium containing GM-CSF and IL-4 and supplemented with TNF-α (50 ng/ml). In the case of coincubation with MRC5 cells, immature DCs were directly added to HCMV-infected or uninfected cells (1:1 ratio) and cocultured for 24 h. Alternatively, coincubations were performed with clarified supernatants that were obtained from infected or uninfected MRC5 fibroblasts, as described below. After the treatments, the cells were harvested, washed in PBS supplemented with 5% fetal calf serum, and incubated for 30 min at 4°C with either phycoerythrin (PE)-conjugated anti-CD83 antibody (HB15A; Coulter Immunotech) or isotype control (immunoglobulin G-PE; Coulter Immunotech). The mature DC phenotype (CD83-positive cells) was detected by flow cytometric analysis (Coulter Epics) by gating the cell population of interest on forward scatter-side scatter.
Quantification of TGF-β1 in culture supernatants and neutralizing assays.MRC5 cells were either mock infected or infected at an MOI of 1 to 5 with HCMV AD169. The virus was adsorbed for 30 min to 24 h at 37°C, and the cells were washed twice with PBS. They were cultured with complete medium in coincubation experiments or with serum-free medium containing 100 μg of bovine serum albumin (Sigma) when TGF-β1 quantification was performed. Supernatants from infected cells were collected in the time ranges 0 to 6, 6 to 24, 24 to 48, and 24 to 72 h, and 0 to 72 h for uninfected cells. They were clarified at 900 × g for 10 min at 4°C and then were either stored at −80°C for ELISA or incubated with immature DCs.
Latent and active TGF-β1 protein was measured by specific ELISA in both serum-free medium and supernatants. To activate the latent form of TGF, samples were acidified for 1 h by dilution at 1/10 in 1 N HCl and then neutralized with 1 N NaOH and tested immediately by ELISA. To quantify TGF-β1, 96-well plates were coated overnight at 4°C with 1 μg of an anti-TGF MAb/ml (MAb 2G7, which recognizes both active and latent forms of TGF-β1, TGF-β2, and TGF-β3) (21). Serial dilutions of medium and supernatants, either acidified or not (100 μl/well), were added to precoated wells and incubated for 1 h at 37°C. After three washes with PBS-Tween (0.1%), biotinylated chicken anti-human TGF-β1 antibody (BAF240; R&D) was added to the wells for one additional hour at 37°C. Binding of biotinylated antibodies was revealed by incubation with streptavidin-conjugated alkaline phosphatase (Jackson). The plates were then washed with PBS-Tween (0.1%) and incubated for 30 min with nitrophenyl phosphate substrate (Sigma). The optical density was measured on an ELISA apparatus (Dynatech Laboratories), and the amounts of TGF-β1 (in picograms per milliliter) were derived from a standard curve constructed using human recombinant TGF (R&D). For neutralization assays of TGF-β1, MRC5 cultures were treated with either an anti-TGF-β1 neutralizing antibody (MAB1032; Chemicon) or an isotype control antibody (Dako), both at 10 μg/ml for 2 h at 37°C, before DCs were added. Then, cross-presentation assays were performed as described above.
RESULTS
DCs cocultured with HCMV AD169-infected fibroblasts stimulate pp65-specific CTL.We have previously shown that cross-presentation of HCMV antigen pp65 by DCs occurred in the presence of AD169-infected apoptotic fibroblasts (2). Moreover, we have also speculated that cell lysis due to cytopathic effects of the virus could serve as an exogenous antigen source for cross-presentation by DCs. To assess this question, immature DCs that are not permissive to HCMV AD169 (25) were added directly to wells containing mock-infected or HCMV-infected MRC5 fibroblasts (1:1 ratio) at various times (6 to 72 h) postinfection (p.i.) and maintained in coculture for 24 h. According to our previous report showing that cross-presentation occurred under conditions where most cells were in late apoptotic and necrotic stages corresponding to PI-positive populations, we first analyzed the phenotypes of infected cells by labeling the cells with propidium iodide. As shown in the histograms (Fig. 1A), infection of fibroblasts for 24 h resulted in the production of a small but significant percentage (20%) of PI-positive cells compared to the percentage with uninfected cells (3%). Microscopic analysis of coculture wells revealed behavioral and morphological changes of immature DCs, as shown in the photomicrograph in Fig. 1A. Indeed, we noted that coculture for 24 h with uninfected monolayer fibroblasts kept immature DCs in a resting state, whereas in the presence of infected cells (24 h p.i.), DCs were mainly clustered. It has been shown that such interactions facilitate antigen transfer between live DCs (14).
DCs cocultured with HCMV-infected fibroblasts stimulate pp65-specific CTL. HLA-A2-positive DCs were added, either directly or in a pore transwell insert, to wells containing uninfected (DC+MR ni) or HCMV-infected MRC5 cells (MOI = 1) at the indicated times p.i. (DC+MR 6hpi to 72hpi), for 24 h. (A) Before DCs were added, uninfected or 24-h-p.i. MRC5 cells were labeled with PI (histograms) and detected by flow cytometry. Cocultures were also examined by optical microscopy. (B) DCs were recovered from cocultures and incubated with HLA-A2-restricted anti-pp65 CTL, raised as described in Materials and Methods, at different responder-to-stimulator ratios (R/S). DCs alone or treated with TNF-α and a relevant peptide (DC N9V TNF) were also used as stimulator cells. Additional controls consisted of uninfected or infected MRC5 cells at each time p.i. (not shown). Stimulation of CTL was assessed through quantitation of IFN-γ secretion using ELISA. The results shown are representative of two to six independent experiments.
To study whether DCs could acquire viral antigens in such cocultures, HLA-A2-positive DCs were gently pipetted to dissociate them from adherent fibroblasts. After being washed, they were incubated with HLA-A2-restricted anti-pp65 CD8+ CTL at different stimulator-to-responder ratios. Cross-presentation of pp65 was monitored by quantification of IFN-γ secretion using ELISA. Figure 1B shows that DCs cocultured for 24 h with 24-h-p.i. MRC5 cells specifically induced the activation of anti-pp65 CTL. We ruled out the possibility that few contaminating CD4+ T cells could respond, since MHC class II did not match between DCs and T-cell lines. Under these conditions, activation with 24-h-p.i MRC5 cells was higher than that following incubation with 48- or 72-h-p.i. MRC5 cells. To exclude the possibility of direct stimulation of CTL by HLA-A2-positive MRC5 cells that could have been recovered during pipetting, uninfected or HCMV-infected MRC5 cells were used as controls. Regardless of the time p.i., neither HCMV-infected fibroblasts nor DCs alone were able to induce significant CTL activation (data not shown). In contrast, TNF-α-treated DCs pulsed with N9V peptide, considered optimal for both maturation and antigen loading, led to maximal activation of CTL. These results indicate that immature DCs are able to acquire viral antigens from cocultured infected fibroblasts and to present them to CD8+ CTL. Although variations exist between independent experiments, globally this process was more efficient in the early stages of infection (6 to 48 h p.i.) and was always reduced when infection time was extended (72 h p.i.).
In vitro studies showed that antigens could access the exogenous pathway for class I presentation by DCs in various forms, including apoptotic and necrotic cells; exosomes; microorganisms, such as bacteria and viruses; and proteins following contact-dependent transfer from live cells (19). In the case of virus-generated damage, such as cytopathic effects caused by HCMV infection, both dead cells or their debris and viral proteins released from infected cells could be sources of viral antigens. Since DCs were not permissive to HCMV AD169, we determined by using a transwell system whether antigen loading of DCs could occur through transfer of viral proteins from infected cells. DCs were added to 3-μm-pore-size transwell chambers inserted into wells containing uninfected or infected MRC5 cells (1:1 ratio) at various times p.i., incubated for 24 h, and recovered for assay of CTL activation. Regardless of the time p.i., activation of specific CTL by DCs was completely abrogated. This indicates that cell-to-cell contact is required for DCs to acquire viral antigens from infected fibroblasts.
DCs phagocytose PKH26+ cells after coculture with early HCMV-infected cells and cross-present incoming pp65 to CTL.Since we had previously shown that in the very early time after infection of MRC5 fibroblasts the matrix protein pp65 from the viral inoculum contained in dead cells was available for cross-presentation, we asked whether this could occur in coculture experiments. To this end, MRC5 cells were infected at a higher MOI (MOI = 3) than that used in the experiment shown in Fig. 1 and cocultured with DCs for 24 h. Figure 2A shows that under these conditions, pp65 contained in very early infected fibroblasts (6 h p.i.) was cross-presented to CTL and that, as shown above (Fig. 1), stimulation was reduced in cocultures with late-infected cells. These data are in agreement with our previous results demonstrating that incoming pp65 was available for cross-presentation to CTL. Labeling of 6-h-infected MRC5 cells with PI revealed that 15% (Fig. 2B, left) of the cells were positive compared with 3% of uninfected cells (data not shown), suggesting that dying infected cells could be a source of antigen uptake by DCs. To clarify the mechanism by which DCs could acquire viral antigen, MRC5 cells were sequentially labeled with PKH26, infected with HCMV for 6 h, and cocultured for 12 h with DCs (1:1 ratio) at either 4 or 37°C. To quantify the uptake of PKH26-labeled material by DCs, the number of HLA-DR+ PKH26+ DCs was analyzed by flow cytometry. Figure 2B (right) shows that 30% of DCs were PKH26+ after being cocultured with 6-h-p.i. MRC5 cells. Flow cytometry analysis was confirmed by fluorescence microscopy as previously described (2) (data not shown). Thus, our results suggest that very early after infection of fibroblasts, incoming pp65 could participate in antigen loading of DCs for stimulation of CTL and that phagocytosis of dying cells could take part in this process.
DCs phagocytose PKH26+ cells during coculture with early-infected MRC5 cells and cross-present pp65 to CTL. (A) HLA-A2-positive DCs were added to wells containing uninfected (DC+MR ni) or HCMV-infected MRC5 cells (MOI = 3) at the indicated times p.i. (DC+MR 6hpi to 72hpi) and were maintained in coculture for 24 h. The experiment was extended as described in the legend to Fig. 1B. (B) (Left) Mock-infected (not shown) or 6-h-p.i. MRC5 cells were labeled with PI. The indicated percentages of PI+ cells were detected by flow cytometry. (Right) PKH26-labeled MRC5 cells infected for 6 h were cocultured with immature DCs for 12 h at 4 or 37°C at a DC/MRC5 cell ratio of 1:1. The uptake of infected MRC5 cells by HLA-DR+ DCs was analyzed by flow cytometry; dot plots were gated on FL1 (HLA-DR) high-positive cells (DCs), thus excluding the uncleared material. Similar results were obtained in three independent experiments. R/S, responder-to-stimulator ratios.
DC maturation is regulated by supernatant of HCMV-infected fibroblasts.It has been shown that optimal cross-presentation of antigens from dead tumor cells required a maturation signal in DCs provided by necrotic cells (29). To demonstrate whether HCMV-infected fibroblasts can deliver a maturation signal to DCs, the expression of CD83, an adhesion receptor highly restricted to mature DCs (30), was determined. Immature DCs were added directly to HCMV-infected or uninfected fibroblasts (1:1 ratio) and cultured for 24 h. Coculture with 6- and 24-h-p.i. MRC5 cells induced maturation of DCs at a higher ratio than TNF-α-treated DCs, as revealed by the appearance of a CD83+ population. Furthermore, this maturation was reduced in coculture with fibroblasts that had been infected for 48 h and almost completely abolished in cells infected for 72 h (Fig. 3A). Uninfected fibroblasts kept DCs in an immature state. Expressions of MHC class I and class II surface molecules, which are known to be upregulated in mature DCs, were slightly regulated following coculture with infected fibroblasts (data not shown).
DC maturation following interaction with HCMV-infected fibroblasts or their supernatant. Immature DCs (alone) were added directly to uninfected (+M ni) or infected MRC5 cells at the indicated times p.i. (+M 6h pi to M 72h pi) (A) or to supernatant from uninfected (+SN ni) or infected MRC5 cells (+SN 6h pi to 72h pi) (B). TNF-α-treated DCs (TNF) were used as a positive control for maturation. All of the cells were collected 24 h after coculture and labeled with an anti-CD83-PE antibody or isotype control. CD83 expression was detected by flow cytometry analysis by gating the DCs on forward scatter-side scatter. Four experiments were performed and gave similar results.
These data, as well as those reported above, suggest that when DCs are added to infected cells at a given time, conditions are combined to provide signals for DC maturation and sufficient pp65-positive cells for antigen loading, which are both essential for optimal activation of specific CD8+ T cells.
We then investigated the mechanisms responsible for regulation of DC maturation by HCMV. Contrasting results have been reported concerning interference of viruses with the maturation process of DCs. For instance, ingestion of debris from canarypox virus-infected cells by DCs partially drove their maturation (15), whereas infection by vaccinia virus induced synthesis of soluble factors that affect DC maturation (32). To establish whether soluble factors released by HCMV-infected cells were involved in modulation of their maturation, DCs were cultured for 24 h with either infected MRC5 cells or their clarified culture supernatant. As shown in Fig. 3, maturation of DCs was modulated to the same extent under both conditions, with a delayed maturation in the presence of supernatant recovered from 72-h-p.i. MRC5 cells. These results show that soluble factors released by infected cells have direct but opposite effects on DC maturation if we consider medium from early- or late-infected fibroblasts.
Supernatants from infected cells are sufficient to modulate the T-cell-stimulatory capacity of DCs.The most distinctive functional characteristic of mature DCs is their ability to potently stimulate T cells (20). Therefore, we assessed whether supernatant from infected cells could modulate the T-cell-stimulatory capacity of DCs. To answer this question, DCs were either directly cocultured as described in the legend to Fig. 1B or cultured for 24 h with clarified supernatant of infected cells supplemented or not with HLA-A2 binding N9V peptide. DCs pulsed with N9V and treated with TNF-α to stimulate their maturation or not treated were used as controls. DCs were recovered and used as stimulators in coculture with specific anti-pp65 CD8+ CTL. IFN-γ secretion by activated T cells was measured by ELISA. Figure 4A indicates that DCs cocultured for 24 h with 48-h-p.i. MRC5 cells induced the highest activation of HLA-A2-restricted anti-pp65 CTL. This activation was always reduced following incubation with 72-h-p.i. MRC5 cells. In experiments using supernatants that were supplemented with N9V peptide, specific activation of CTL was modulated to the same extent as in direct coculture (Fig. 4B). No activation was observed when supernatants alone were used in these assays. Moreover, immature DCs pulsed with N9V were relatively poor stimulators of CTL. Reduction of CTL response to N9V peptide in the presence of 72-h-p.i supernatant was not due to a decrease of MHC class I expression on DCs as assessed by flow cytometry (data not shown). These results demonstrate that soluble factors released from infected cells have direct effects on the T-cell-stimulatory function of DCs.
Supernatants from infected fibroblasts are sufficient to modulate T-cell-stimulatory capacity of DCs. HLA-A2-positive DCs were directly cocultured with uninfected (DC+ MR ni) or infected fibroblasts at the indicated times p.i. (DC+MR6hpi to -72hpi) for 24 h (A) or cultured for 24 h with clarified supernatant recovered from uninfected (MR ni) or infected cells at different times of infection (MR6hpi to -72hpi) supplemented (DC+Supernatant+N9V) or not (DC+Supernatant alone) with a relevant peptide (N9V) (B). DCs pulsed with N9V and treated (DCN9V TNF) or not (DCN9V) with TNF-α were also used as controls. After coculture, recovered DCs were treated as described in the legend to Fig. 1B, and the same controls were used. One representative experiment of two is shown. R/S, responder-to-stimulator ratios.
TGF-β1 released by late-infected fibroblasts takes part in down regulation of both DC maturation and T-cell activation.The results discussed above strongly support the notion that a soluble factor contained in the supernatant of late-infected fibroblasts is responsible for the failure of CD83 induction of DCs and of their T-cell-stimulatory ability. It has been reported that several cytokines, including IL-10 and TGF-β, caused down regulation of expression of CD83 and other costimulatory molecules on DCs (33). Michelson and coworkers (22) reported induction of TGF-β1 secretion following HCMV infection of human foreskin fibroblasts, suggesting that this cytokine could modify systemic immune reactions to the advantage of the virus by both up regulating its replication and down regulating host immune responses. Accordingly, we focused our investigations on TGF-β1 as a potential mediator of DC maturation. Thus, we first examined the production of TGF-β1 in the supernatant of mock-infected and HCMV-infected MRC5 cells by ELISA. Table 1 shows that TGF-β1 was detected in increasing amounts in infected-cell supernatants. Indeed, the amount of TGF-β1 was three times higher in supernatants recovered from 24- to 72-h-p.i. cells than in supernatants from mock-infected MRC5 cells. To test whether TGF-β1 production by infected cells was at least in part responsible for impairment of DC maturation, we used neutralizing antibody against TGF-β1. Uninfected or 48- and 72-h-p.i. MRC5 cultures were incubated for 2 h at 37°C with either anti-TGF-β1 neutralizing antibody or an isotype-matched control. DCs were then added to pretreated MRC5 cells and incubated for 24 h, and their CD83 status was determined by flow cytometry. As shown in Fig. 5, down modulation of CD83 expression in DCs cultured with 72-h-p.i. MRC5 cells was abrogated by the anti-TGF-β1 neutralizing antibody but not by the isotype control antibody. These results allowed the identification of TGF-β1 as an essential mediator of the decreased expression of CD83 on DCs following incubation with late-infected fibroblasts. To further determine whether TGF-β could interfere with CTL activation, IFN-γ release by anti-pp65 CTL was assessed following neutralization of TGF-β1 in MRC5 cultures. Under these conditions, T-cell activation was not recovered to a highly significant extent (data not shown), suggesting that unknown soluble factors contained in late supernatant may down regulate costimulatory molecules other than CD83 that are critical for CTL response.
TGF-β1 released by late-infected fibroblasts is essential in failure to induce CD83 expression on DCs. Cultures of uninfected (MR ni) or infected MRC5 cells at 48 and 72 h p.i. were incubated with a neutralizing anti-TGF-β antibody at 10 μg/ml or an isotype control for 2 h at 37°C. Immature DCs were then added to neutralized conditioned medium and incubated for 24 h. The cells were collected and labeled with an anti-CD83-PE antibody. The flow cytometry histograms show percentages (± standard deviations) of CD83-positive cells representative of three independent experiments.
Detection by ELISA of TGF-β1 induced by HCMV infection of MRC5 cells
Altogether, our data suggest that cross-presentation of pp65 to CTL is regulated by soluble-mediator-dependent maturation of DCs.
DISCUSSION
We previously reported that HCMV infection sensitized fibroblasts to TNF-α-induced apoptosis early after infection and thus provided an antigen source for cross-presentation by immature DCs to anti-pp65 CD8+ T cells (2). Since in our previous report cell death was induced by incubation of infected cells with TNF-α to provide antigen to DCs, we asked whether in coculture of infected MRC5 fibroblasts with immature DCs cross-presentation could be recovered as well. In this study, immature DCs were added to fibroblasts that had been infected with HCMV AD169 for periods ranging from 6 to 72 h. We demonstrated that DCs could acquire incoming pp65 from early-infected MRC5 cells for cross-presentation to HLA-A2-restricted CD8+ CTL but that coculture with late-infected cells decreased their capacity to stimulate CTL. Analyses of DC maturation after either coculture with infected MRC5 or incubation with infected-cell-conditioned medium revealed that acquisition of a mature phenotype was a prerequisite for efficient stimulation of CTL and that soluble factors secreted by infected cells were responsible for both up and down regulation of CD83 expression on DCs.
More and more reports provide evidence that DCs are key cells in the development and regulation of immune responses due to their capacity to cross-present antigen from various sources, as evidenced by the recent finding showing antigen acquisition through transfer from live cells (14). In our previous study, we focused on the involvment of purified dead-cell phagocytosis by DCs in cross-presentation to anti-HCMV CD8+ T cells. This mechanism requires that infected cells be sensitive to apoptosis either through a direct cytopathic effect of the virus or through a bystander effect involving proapoptotic cytokines, such as TNF-α and Trail (31). Moreover, since contrasting results have been reported concerning modulation of apoptosis in cells infected with HCMV (12), we asked whether cross-presentation could take place in coculture of infected fibroblasts with immature DCs. According to our previous observation that phagocytosis of dead infected cells could be a major component for cross-presentation, we assessed whether it was involved in the coculture model. Under conditions ensuring that all MRC5 cells were infected, we demonstrated that DCs were able to internalize PKH26+ early-infected cells that contained 15% dead cells. Since phagocytosis is a rapid process and significant amounts of neosynthesized pp65 are not available before 48 h p.i. (2), we can suggest that incoming pp65 is involved in cross-presentation to CD8+ T cells as previously demonstrated. Nevertheless, it remains to be determined how early-infected cells undergo apoptosis and necrosis. We can make the assumption that an autocrine process involving secretion of Trail by infected fibroblasts could account for their sensitivity to apoptosis as it was described previously (31). Alternatively, it has been shown that DCs were able to kill Trail-sensitive targets through IFN-induced expression of Trail at their surfaces (9). Whether HCMV infection may induce the expression of Trail on DCs through secretion of type I IFN by infected cells remains to be explored. Nevertheless, since infected cells are sensitive to Trail (31), we speculate that this mechanism could take part in the generation of apoptotic and necrotic cells. These hypotheses are being considered in our laboratory. However, this does not exclude the possibility that DCs acquired pp65 through an unidentified mechanism, as shown recently by Tabi and coworkers (34). Indeed, those authors showed that when MRC5 cells infected for 48 h were added to immature DCs and cocultured for several days with PBMC, anti-pp65, as well as anti-IE1, CD8+ T cells could be stimulated. Surprisingly no apoptosis of MRC5 cells was observed, suggesting that uptake of pp65 by DCs occurred through an apoptosis-independent but unidentified mechanism. Confirming the data of Tabi et al., we found that cross-presentation required cell-to-cell contact, suggesting that a mechanism of antigen transfer from MRC5 cells to DCs such as that described recently (14) could be involved. Altogether, our data and those of Tabi et al. suggest that multiple processes that may depend on cell type and culture conditions could be involved in the acquisition of HCMV proteins by DCs.
Even though the mechanisms involved in loading of DCs remain to be clarified, we observed that when DCs were added to late-infected fibroblasts, cross-presentation of pp65 was less efficient. It is known that viruses may interfere directly or indirectly with DC maturation and effector functions, depending on permissivity to infection (16). For instance, measles virus suppresses cell-mediated immunity by interfering with the survival of DCs (10), whereas herpes simplex virus type 1 leads to the degradation of CD83, a cell surface molecule which is specifically up regulated during DC maturation (18). Contrary to the measles virus model, we never observed death of DCs following coculture with HCMV-infected fibroblasts (data not shown). Since it is known that optimal cross-presentation requires maturation signals for DCs, we analyzed their maturation process upon exposure to infected fibroblasts. A correlation was observed between maturation of DCs, as assessed through the appearance of a CD83+ population, and their stimulating capacity in the presence of early-infected cells. In contrast, decreased activation of CTL in the presence of late-infected fibroblasts corresponded with the disappearance of the CD83+ population, suggesting that acquisition of a mature phenotype is a key step in stimulating anti-pp65 CD8+ T cells. It has been reported that DC maturation ensued from exposure to apoptotic or secondary necrotic infected cells (15, 26, 27). Accordingly, we could speculate that in coculture, DC maturation occurred at least in part as a result of engulfment by and/or contact with HCMV-infected fibroblasts, as we demonstrated above. Moreover, we found that the maturation process was regulated by soluble factors contained in infected-cell-conditioned medium. In this case, the efficiency lower than that observed in coculture is probably due in part to a reduced incubation time, but it also suggests that the bystander effect of maturing cytokines is improved by cell-to-cell contacts.
The factors responsible for maturation remain to be identified, but we may speculate that type I IFN (IFN-α or -β) could take part in this process, since these cytokines are produced in the early stages of many viral infections (4). Nevertheless, the inability to induce DC maturation due to soluble factors, such as TGF-β1, secreted by late HCMV-infected cells can be considered as a new escape mechanism developed by the virus to prevent cross-presentation and the establishment of an anti-viral CD8+-T-cell response. TGF-β1 is well known as a critical cytokine for acquisition of the Langerhans cell (LC) phenotype. Moreover, this cytokine is responsible for prevention of noncognate maturation of LCs (11). Indeed, TGF-β1 inhibits the effects of inflammatory cytokines and those of lipopolysaccharide on LC maturation, while CD40L-induced maturation remains unaffected. Our results are in accordance with these data, since in our model, CTL activation by DCs is impaired in the absence of cognate signals such as CD40L, owing to the fact that only CD8+ T cells were used in activation tests. This finding could point out the role of cognate help provided by CD4+ T cells, which are considered major effectors in the control of HCMV replication (7). Furthermore, it has been reported that HCMV harbors its own unique IL-10 homologue (cmvIL-10) (17) and that functionally active cmvIL-10 was secreted by infected cells. Since IL-10 is able to interfere with DC maturation (8), we cannot rule out the possibility that cmvIL-10 could exert this inhibitory effect in our model. Overall, our results show that cross-presentation has to occur early after infection by HCMV to overcome viral escape mechanisms mediated by soluble factors, including TGF-β1, and may explain why the main targets for CTL are proteins, such as IE1 and pp65 (34, 35), that are available early after infection. Moreover, we suggest that the immunosuppressive effect of HCMV could be effective in the case of delayed recruitment of DCs on infection sites if we take into account that HCMV-encoded US28 protein, a viral homologue of CC chemokine receptors (5), could interfere with the chemokine-dependent recruitment of DC into inflamed tissues (28).
ACKNOWLEDGMENTS
This work was supported by institutional grants from INSERM. Géraldine Arrode was supported by an MNERT fellowship and by the Fondation pour la Recherche Médicale.
We thank Abdelhadi Saoudi for essential advice concerning TGF-β1 study and Isabelle Bernard for technical assistance with the TGF-β1 ELISA. We thank Sanofi-Synthelabo for supplying us with IL-7 and IL-13.
FOOTNOTES
- Received 13 June 2001.
- Accepted 27 September 2001.
- Copyright © 2002 American Society for Microbiology