Restoration of TLR3-Activated Myeloid Dendritic Cell Activity Leads to Improved Natural Killer Cell Function in Chronic Hepatitis B Virus Infection

Abstract

There is increasing evidence that the function of NK cells in patients with chronic hepatitis B (CHB) infection is impaired. The underlying mechanism for the impaired NK cell function is still unknown. Since myeloid dendritic cells (mDC) are potent inducers of NK cells, we investigated the functional interaction of mDC and NK cells in CHB and the influence of antiviral therapy. Blood BDCA1⁺ mDC and NK cells were isolated from 16 healthy controls or 39 CHB patients at baseline and during 6 months of antiviral therapy. After activation of mDC with poly(I · C) and gamma interferon (IFN-γ), mDC were cocultured with NK cells. Phenotype and function were analyzed in detail by flow cytometry and enzyme-linked immunosorbent assay. Our findings demonstrate that on poly(I · C)/IFN-γ-stimulated mDC from CHB patients, the expression of costimulatory molecules was enhanced, while cytokine production was reduced. In cocultures of poly(I · C)/IFN-γ-stimulated mDC and NK cells obtained from CHB patients, reduced mDC-induced NK cell activation (i.e., CD69 expression) and IFN-γ production compared to those in healthy individuals was observed. Antiviral therapy normalized mDC activity, since decreased expression of CD80 and CD86 on DC and of HLA-E on NK cells was observed, while poly(I · C)/IFN-γ-induced cytokine production by mDC was enhanced. In parallel, successful antiviral therapy resulted in improved mDC-induced NK cell activation and IFN-γ production. These data demonstrate that CHB patients display a diminished functional interaction between poly(I · C)/IFN-γ activated mDC and NK cells due to impaired mDC function, which can be partially restored by antiviral therapy. Enhancing this reciprocal interaction could reinforce the innate and thus the adaptive T cell response, and this may be an important step in achieving effective antiviral immunity.

INTRODUCTION

Chronic hepatitis B (CHB) is the result of an inadequate immune response against a persistent hepatotropic virus (30). Research has traditionally been directed toward the ineffective hepatitis B virus (HBV)-specific T-cell response, while in recent years innate immunity and its cellular components, such as natural killer (NK) cells and dendritic cells (DC), also have received increasing attention (5). It has been reported that both NK cells and DC are affected in patients with CHB. NK cells are important during chronic HBV infection because they cause hepatocyte damage through apoptosis (12). We and others recently reported that NK cell activation, as demonstrated by CD69 expression, and production of proinflammatory cytokines such as gamma interferon (IFN-γ) by NK cells are decreased in CHB patients, whereas direct cytotoxicity remains intact (25, 32). This functional dichotomy may likely contribute to viral persistence. A role for interleukin-10 (IL-10) in causing the NK cell impairment has been suggested (26), although negative regulation through enhanced expression of the inhibitory receptor NKG2A also has been described (32).

NK cells are in close proximity with DC, and the outcome of NK cell function is dependent on reciprocal interaction with DC (9, 16, 28). There is evidence that the function of myeloid DC (mDC) in blood is impaired in patients with CHB (36). Stimulated mDC secrete several cytokines, such as IL-6 and IL-12 (14), that act as potent inducers of NK cell activation and proliferation. In turn, activated NK cells produce IFN-γ and tumor necrosis factor (TNF), which are directly involved in completing the maturation program of mDC (1). NK cell activation is driven by a direct interaction between activated NK cells and mature mDC through attenuated expression of HLA-E, the ligand for NKG2A (10), and major histocompatibility complex class I (MHC-I) molecules, the natural ligands for inhibitory NK receptors (22), and also through expression of activating receptor NKp30 on mDC (15). In HIV infection, defective interaction through impaired function of NKp30 on mDC leads to secretion of inadequate amounts of IFN-γ by NK cells (20). Even though altered NKp30 expression is observed in acute liver failure (23, 37), little attention has been given to this interaction between mDC and NK cells in CHB. In this study, we aimed to characterize the functional interaction between mDC and NK cells in CHB.

As liver inflammation and HBV DNA load closely determine the immunoactive phase of CHB (30), the effect of the continuous presence of HBV DNA on the immune system can be studied through antiviral therapy. The cytokine-producing capacities of both mDC and NK cells are partially restored upon antiviral therapy with nucleoside analogues (26, 34). We recently demonstrated that restoration of IFN-γ production was paralleled by enhanced activation of NK cells and a decrease of NKG2A expression (32). We therefore hypothesized that antiviral therapy results in improved mDC-NK cell interaction through restoration of cytokine-mediated pathways of both mDC and NK cells and through direct receptor-ligand expression. To study this, we examine the functional interaction of mDC and NK cells derived from a cohort of CHB patients treated with antiviral therapy.

MATERIALS AND METHODS

Patients and healthy subjects.

Heparinized peripheral blood samples were obtained from 39 patients with CHB for longitudinal analysis during antiviral therapy. Blood samples were obtained at baseline (t = 0) and after 6 months of antiviral therapy (t = 6). Patients (HBeAg positive, n = 10; HBeAg negative, n = 29) were treated with the nucleoside or nucleotide analogue entecavir (0.5 mg once a day [o.i.d.]; n = 22) or tenofovirdisoproxil (245 mg o.i.d.; n = 17), and all of them met the most recent EASL guideline criteria for treatment of CHB. Patients receiving antiviral therapy within 6 months prior to treatment with a nucleoside or nucleotide analogue were excluded. All patients were negative for antibodies against hepatitis C virus, hepatitis D virus, and human immunodeficiency virus. An age- and sex-matched control group comprised 16 healthy controls (HC). The study was approved by the local ethics committee, and all patients and healthy individuals gave informed consent before blood donation.

Expression of intracellular and cell surface molecules on mDC and isolated NK cells determined by flow cytometry.

Peripheral blood mononuclear cells (PBMC) were obtained by Ficoll-Isopaque (GE Healthcare, Finland) gradient centrifugation. mDC were isolated from fresh PBMC by positive immunomagnetic selection using the mini-MACS system (Miltenyi Biotec, Germany) according to the manufacturer's instructions. NK cells from PBMC were isolated by negative selection (purity and viability were >95%) with an NK cell isolation kit (Miltenyi Biotec, Germany). For phenotypic analysis, incubation with a cocktail of fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)-, peridinin chlorophyll protein (PerCP)-, and allophycocyanin (APC)-conjugated monoclonal antibodies (MAbs) was performed. MAbs directed against the following molecules were used: CD3 (UCHT1), CD69 (L78), and CD86 (2331) from BD Bioscience, Belgium; CD56 (N901), CD80 (mAB104), CD244 (C1.7), and CD48 (J4-57) from Beckman Coulter; BDCA1 (AD5-8E7) from Miltenyi Biotec, Germany; MHC class ABC (W6/32) and MIC-A (6D4) from Biolegend; and HLA-E (3D12) from eBioscience. As controls, cells were stained with corresponding isotype-matched control antibodies. Stained cells were analyzed using a 4-color flow cytometer (FACSCalibur) and CellQuest software (both from BD Bioscience, Belgium). The mean fluorescence intensity (MFI) and/or the percentage of positive cells was determined.

Coculture of mDC and NK cells.

Freshly isolated mDC from CHB patients and HC were cultured at a concentration of 1 × 10⁵ in 200 μl culture medium (RPMI-1640 [BioWhittaker/Cambrex, Belgium], 100 U/ml penicillin and 100 μg/ml streptomycin [Invitrogen], 10% fetal bovine serum [Thermo Fisher Scientific, The Netherlands], and 2 mM l-glutamine and 1 M HEPES [BioWhittaker/Cambrex, Belgium]) with 50 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF) (Berlex, Germany), 20 μg/ml poly(I · C) (Sigma-Aldrich, Germany), and 250 U/ml IFN-γ (Strathman, Germany) in a round-bottom 96-wells plate. Isolated mDC were also monocultured with 50 ng/ml GM-CSF as a positive control. After 18 h, a small proportion of mDC were analyzed by flow cytometry; the remainder of the cells were washed. Fresh autologous NK cells and isolated synchronous (t = 0 weeks or t = 6 months) or thawed NK cells isolated 2 weeks prior to baseline were added at a 1:5 mDC/NK cell ratio (>20 × 10³ mDC/well) in a round-bottom 96-wells plate containing 200 μl per well. Isolated NK cells were also monocultured with 800 U/ml IL-2 (Strathman, Germany) as a positive control. After coculture for 48 h, cells were washed and processed for analysis. At all time points, supernatants were collected.

Cytokine production.

Cytokine production by isolated mDC or NK cells from CHB patients and HC was determined after maturation for 18 h with poly(I · C) and IFN-γ and after coculture with isolated NK cells. Levels of IL-6, IL-12p40, IL-12p70, and IFN-γ present in culture supernatants were determined by standard enzyme-linked immunosorbent assay (ELISA) according to the manufacturer's instructions (eBioscience). Levels of IL-18 in culture supernatants were determined by flow cytometry using labeled beads (human IL-18 Flowcytomix Simplex; Bender MedSystems).

Statistical analysis.

Data are expressed as means ± standard errors of the means (SEM) unless indicated otherwise. Data were analyzed with Prism 5.0 (GraphPad software) using the Mann-Whitney U test to compare variables between 2 independent groups, the Wilcoxon matched-pairs test for comparisons between paired variables, and the Spearman rank correlation coefficient test for nonparametric correlations. In all analyses, a two-tailed P value of less than 0.05 was considered statistically significant.

RESULTS

Diminished mDC-induced activation and function of NK cells in CHB.

We recently showed that CD69 expression and IFN-γ production by NK cells are impaired in CHB patients (32). Since the interplay between NK cells and mDC is crucial for their function, we decided to investigate the reciprocal functional interaction between mDC and NK cells of CHB patients and HC. For this purpose, we performed coculture assays as described by Gerosa et al. (16) for monocyte-derived DC and adjusted for ex vivo experiments as described in Materials and Methods. Group characteristics are shown in Table 1. For mDC-NK cell cocultures, we activated the isolated mDC with poly(I · C) and IFN-γ prior to coculture to prevent NK cell-mediated killing of immature mDC (16). As a result of the mDC-NK cell interaction, NK cells became activated, as evidenced by upregulation of expression of the early activation marker CD69 on NK cells. After coculture, NK cell viability was >90% whereas mDC viability was around 60%. mDC of CHB patients showed a 5.6-fold-lower CD69 upregulation than those of HC (Fig. 1A). This corroborates our previous findings that in CHB, baseline NK cells in unfractionated PBMC are less activated (32). Besides lower activation levels, NK cell-derived IFN-γ levels detected in mDC-NK cell cocultures of patients also were strongly diminished compared to those of controls (18-fold less) (Fig. 1B). There was no correlation between CD69 upregulation and IFN-γ production. Thus, in contrast to the case for cells from HC, coculture of poly(I · C)/IFN-γ-activated mDC and NK cells from CHB patients results in only weak activation of NK cells and poor cytokine production.

Table 1.

Characteristics of the study population

Characteristic	Value for:
	Healthy controls (n = 16)	CHB cohort (n = 39)
Mean age, yr (mean ± SEM)	41.9 ± 2.2	42.5 ± 1.7
Sex (no. female/male)	4/12	4/35
Race (no.)
Caucasian	7	14
Asian	3	5
Mediterranean	4	9
Other	2	11
ALT, IU/ml [median (interquartile range)]	NAa	74 (50-101)
HBV DNA, log10 IU/ml (mean ± SEM)	NA	7.3 ± 0.3
HBeAg (no. positive/negative)	NA	10/29
HBV genotype (no. A/B/C/D/E)	NA	13/3/4/14/2
Fibrosis score, Metavir (no. F0/F1/F2/F3/F4)	NA	2/14/11/9/3
Prior antiviral therapy
None	16	15
Nucleoside or nucleotide analogue		2
(PEG)IFNb		15
(PEG)IFN + nucleoside or nucleotide analogue		7

^aNA, not applicable.

^b(PEG)IFN; (pegylated) alpha interferon.

Fig 1.

Diminished mDC-induced activation and function of NK cells in CHB. Purified fresh NK cells were cocultured with stimulated BDCA1⁺ mDC as described in Materials and Methods. Cells were isolated from 11 healthy controls (HC) and 22 CHB patients (HBV) (HBV DNA, log₁₀ 8.1 ± 0.4 IU/ml; ALT, 77 ± 7 IU/ml [mean ± SEM]). After 48 h, the early activation marker CD69 on NK cells was determined by fluorescence-activated cell sorter (FACS) analysis, and supernatants of these cultures were evaluated for IFN-γ content by ELISA. Data are expressed as means ± SEM and represent CD69-expressing NK cells within the total NK cell population (A) or IFN-γ levels in supernatant (B). *, P < 0.05; **, P < 0.01.

Expression of markers of mDC maturation, but not of NK cell-associated ligand expression, are enhanced on mDC of CHB patients.

To investigate whether the activation and function of NK cells are impaired as a result of a disturbed interaction with mDC, we examined the response of mDC obtained from HC and CHB patients to stimulation with poly(I · C) and IFN-γ. As presented in Fig. 2A, directly after isolation, the expression of the costimulatory molecules CD80 and CD86 on mDC was significantly higher in patients than in controls, indicating a higher maturation state of mDC ex vivo. Following stimulation, the expression of CD80 and CD86 weakly decreased in patients but remained higher than the expression levels observed in HC. No correlation between CD80/CD86 expression and HBV DNA or alanine aminotransferase (ALT) levels could be determined (data not shown).

Fig 2.

Expression of markers of maturation is enhanced on stimulated mDC of CHB patients. Isolated BDCA1⁺ mDC were stimulated with poly(I · C) and IFN-γ for 18 h. Cells were isolated from 11 healthy controls (HC) and 22 CHB patients (HBV). Expression of CD80, CD86, MHC class I molecules (HLA-A, HLA-B, and HLA-C), MIC-A, and HLA-E was determined without stimulation (no stim) and after 18 h of stimulation with poly(I · C) and IFN-γ (stim). Expression levels of maturation markers CD80 and CD86 (A), expression levels and percentages of cells positive for ligands MHC-I, MIC-A, and HLA-E (B), and representative dot plots and histograms (C) as determined by flow cytometry are presented as means ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.0001.

To further characterize membrane-bound ligands on mDC for which cognate receptors are expressed on NK cells, we evaluated the expression of HLA-A/HLA-B/HLA-C (indicated as “MHC-I”), MIC-A, and HLA-E prior to and after stimulation with poly(I · C) and IFN-γ. As shown in Fig. 2B, stimulation of mDC resulted in upregulation of the expression of MHC-I, MIC-A, and HLA-E, which has been reported before (20). Interestingly, the percentage of MIC-A-expressing mDC upon stimulation was significantly higher in CHB patients than in HC. However, no difference in the percentage or MFI levels between patients and controls could be demonstrated for MHC-I or HLA-E on stimulated mDC. Representative dot plots and/or histograms from the flow cytometric analysis are shown in Fig. 2C. Thus, mDC from CHB patients showed a more activated phenotype accompanied by higher MIC-A expression, while the MHC-I and HLA-E molecules were not affected, which suggests limited involvement of these ligands in causing the impaired NK cell function in patients.

Another molecule pair that may affect the outcome of the interaction between mDC and NK cells is 2B4-CD48 (2, 7, 33). We evaluated the expression of CD48 and 2B4 on unstimulated and stimulated mDC (Fig. 3A). No changes in the expression of CD48 and 2B4 on mDC were observed as a consequence of activation, and the levels also were similar for CHB patients and HC. In contrast, the expression of CD48 as well as 2B4 was significantly enhanced on NK cells freshly obtained from patients compared to those from controls (Fig. 3B). The upregulation of 2B4-CD48 on NK cells is not paralleled by upregulation on mDC, which suggests that a major role for this receptor-ligand pair in mDC-NK cell interaction is unlikely, since IFN-γ production by NK cells would require contact-dependent signals through 2B4 interaction (3).

Fig 3.

Expression of CD48 and 2B4 on mDC and NK cells of CHB patients. NK cells and mDC were purified from 5 healthy controls (HC) and 17 CHB patients (HBV) (HBV-DNA, log₁₀ 6.3 ± 0.5 IU/ml; ALT, 96 ± 17 IU/ml), and mDC were stimulated with poly(I · C) and IFN-γ for 18 h. Expression levels of CD48 and 2B4 were determined on mDC without stimulation (unstim) and after 18 h of stimulation with poly(I · C) and IFN-γ (stim) and on NK cells. Expression levels of CD48 and 2B4 on mDC (A) and NK cells (B), determined by FACS, are presented as means ± SEM. *, P < 0.05; ***, P < 0.001.

Impaired cytokine production by mDC of CHB patients.

Besides contact-dependent signals, mDC-derived soluble factors, such as IL-12 and IL-18, also are involved in mDC-NK cell interaction and may activate NK cells (24). After 18 h of activation with poly(I · C) and IFN-γ, supernatants of mDC of CHB patients contained less IL-6, IL-12 (IL-12p40 and IL-12p70), and IL-18 than those of HC (Fig. 4). Therefore, the reduced capacity of mDC to produce cytokines may have consequences for the ability to activate NK cells in CHB patients.

Fig 4.

Impaired cytokine production upon stimulation of mDC of CHB patients. Purified mDC isolated from 5 healthy controls (HC) and 17 CHB patients (HBV) were stimulated with poly(I · C) and IFN-γ for 18 h. Supernatants of these cultures were evaluated by ELISA or FACS (labeled beads). Data represent mean concentrations (pg/ml) ± SEM of IL-6, IL-12p40, IL-12p70, and IL-18. **, P < 0.01; ***, P < 0.001.

Antiviral therapy improves NK cell function as a result of improved mDC-NK cell interaction.

The effect of antiviral therapy on mDC-NK cell interaction and the capacity to produce antiviral cytokines was examined in CHB patients treated with entecavir, which is a nucleoside analogue that inhibits viral replication without direct effects on the immune system. All patients responded to therapy, resulting in a mean reduction in HBV DNA of log₁₀ 4.2 IU/ml and mean lowering of ALT levels of 1.8-fold after 6 months of treatment. After 6 months of antiviral therapy, the mean viral load ± SEM was log₁₀ 3.1 ± 0.3 IU/ml, the mean ALT level was 36 IU/ml (range, 25 to 56 IU/ml), and three HBeAg-positive CHB patients achieved seroconversion. There was no difference in response with respect to genotype, race, or prior therapy; however, the groups were too small for significant differences to be reached (data not shown).

As shown in Fig. 5A, as a consequence of antiviral therapy, the ability of mDC to activate NK cells, as demonstrated by CD69 expression, was not affected in cocultures with synchronously isolated NK cells. In contrast, NK cell-derived IFN-γ production was elevated during antiviral therapy in these cultures, although the difference was not significant. To examine the influence of persistent viral infection on the functionality of mDC and NK cells, mDC isolated at baseline and after 6 months of antiviral therapy were cocultured with autologous NK cells that were isolated 2 weeks prior to the start of therapy (Fig. 5B). In comparison to their counterparts at baseline, mDC isolated during antiviral therapy were able to activate NK cells significantly better, which was paralleled by significantly enhanced IFN-γ production. Since purified NK cells stimulated with IL-2 showed similar CD69 expression levels and IFN-γ production at baseline and during antiviral therapy (data not shown), it is likely that impaired mDC activation rather than an intrinsic NK cell defect causes the reduced NK cell-derived IFN-γ production during the interaction.

Fig 5.

Antiviral therapy improves NK cell function as a result of improved mDC-NK cell interaction. Isolated NK cells were cocultured for 48 h with isolated BDCA1⁺ mDC after stimulation with poly(I · C) and IFN-γ for 18 h. Cells were isolated from 22 CHB patients (HBV) at baseline (t = 0) and after 6 months of treatment with entecavir 0.5 mg o.i.d. (t = 6). The frequency of fresh (A) or thawed (B) CD69-positive NK cells after 48 h of coculture with simultaneously isolated mDC was determined at t = 0 and t = 6 by FACS analysis. Fresh NK cells were isolated simultaneously with mDC at t = 0 and t = 6. Thawed NK cells were isolated 2 weeks prior to baseline and frozen at −80°C before use. Supernatants of these cultures were evaluated for IFN-γ content by ELISA. Data represent CD69-expressing NK cells within the total NK cell population or mean levels of IFN-γ ± SEM. *, P < 0.05.

Decreased expression of CD80, CD86, and HLA-E during antiviral therapy.

Next, we investigated whether antiviral therapy results in altered NK cell-induced mDC maturation. Prior to coculture, the expression of CD80 and CD86 on freshly isolated or activated mDC was not different, irrespective of viral load reduction (data not shown). In contrast, NK cell-induced CD80 and CD86 expression was downregulated after 6 months of antiviral therapy (Fig. 6A). No correlation between CD80/CD86 expression and HBV DNA or ALT levels could be found (data not shown). Upon antiviral therapy, the improved mDC-induced NK cell activation and production are thus paralleled by lower NK cell-induced expression levels of maturation markers on mDC.

Fig 6.

Diminished activation and decreased expression of HLA-E on mDC upon antiviral therapy. Purified fresh NK cells and mDC were isolated from 11 healthy controls (HC) and 22 CHB patients at baseline (t = 0) and after 6 months of treatment (t = 6). Expression of CD80 and CD86 on mDC upon coculture with simultaneously isolated NK cells was determined by FACS analysis. Expression of HLA-E at t = 0 and t = 6 was determined on mDC without stimulation (no stim) and after 18 h of stimulation with poly(I · C) and IFN-γ (stim). (A and B) Mean (±SEM) levels of expression of CD80 and CD86 (A) and HLA-E (B) on mDC. *, P < 0.05. (C) Spearman's rank correlation coefficient between the change in HBV DNA (log₁₀ IU/ml) and the change in mean fluorescence intensity (MFI) of HLA-E expression on mDC after stimulation at t = 0 and t = 6. Change in HBV DNA was defined as the absolute difference between levels at baseline and at 6 months. Change in MFI was defined as the percent difference between levels at baseline and at 6 months. Spearman's rank correlation coefficient between the MFI of HLA-E expression on mDC after stimulation and levels of ALT (IU/ml) at t = 6 was determined.

To investigate whether antiviral therapy also affected receptor-ligand expression, we determined the expression of MHC-I, MIC-A, HLA-E, CD48, and 2B4 on mDC and NK cells. Only the expression of HLA-E on nonstimulated and stimulated mDC was significantly downregulated (Fig. 6B), whereas the expression of all other molecules remained unchanged (data not shown). There was a strong correlation between HBV DNA load reduction and the percent downregulation of HLA-E expression on activated mDC (Fig. 6C). Furthermore, we observed an association between ALT levels during antiviral therapy and HLA-E expression (Fig. 6C). The improved mDC-NK cell interaction may thus involve, at least partly, the expression of inhibitory receptor-ligand pairs which regulate the NK cell activation state.

Antiviral therapy enhances mDC-derived cytokine production.

To examine the effect of antiviral therapy on the capacity of mDC to produce cytokines, we determined IL-6, IL-12p40, IL-12p70, and IL-18 levels in supernatants of cultures at t = 0 and t = 6. Antiviral therapy resulted in a significantly increased capacity of mDC to produce these cytokines (Fig. 7).

Fig 7.

Antiviral therapy ameliorates cytokine production upon stimulation Purified mDC isolated from 17 CHB patients (HBV) were stimulated with poly(I · C) and IFN-γ for 18 h at baseline (t = 0) and after 6 months of treatment (t = 6). Supernatants of these cultures were evaluated by immunoassay. Data represent the mean concentrations (±SEM) of IL-6, IL-12p40, IL-12p70, and IL-18. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

DISCUSSION

Bidirectional cross talk between mDC and NK cells is known to play an important role in host immunity. In the present study, we examined the functional interaction between mDC and NK cells of CHB patients. We demonstrated that mDC of CHB patients are substantially impaired in their ability to activate NK cells, which in turn fail to secrete adequate amounts of IFN-γ. Expression of costimulatory molecules and, to a lesser extent, MIC-A was enhanced on poly(I · C)-stimulated mDC of CHB patients, while Toll-like receptor (TLR)-induced cytokine production by mDC was reduced. On NK cells of CHB patients, the expression of CD48 and 2B4 was enhanced. Importantly, antiviral therapy with nucleoside or nucleotide analogues improved mDC function, and we showed that this resulted in improved mDC-induced NK cell activation and IFN-γ production. This was paralleled by decreased expression of CD80 and CD86 on mDC and of HLA-E on NK cells, while TLR-induced cytokine production by mDC was enhanced. These data not only could partially explain the impaired NK cell activity in chronic HBV infection but also may provide a mechanism for an immunomodulatory role of antiviral drugs.

In unfractionated PBMC, we and others (26, 32) demonstrated improved NK cell function upon therapy. This may be attributed to the direct effect of viral load reduction on NK cells; however, our study provides evidence that mDC are at least partly responsible for the improved NK cell activation and function. NK cell activation is partly dependent on the balance between inhibitory and activating signals that are generated by an array of different cell surface receptors after engagement by their specific cellular ligands (22). We previously showed that NKp30 expression is significantly reduced in NK cells of CHB patients (32), which may compromise the mDC-NK cell interaction, in which this activating receptor plays a critical role (15). In HCV infection, HLA-E-mediated signaling partly determines activation or inhibition of NK cell function (17). Since expression of both HLA-E on mDC, as demonstrated in this study, and its receptor NKG2A on NK cells decreases after antiviral therapy in CHB (32), we speculate that increased NK cell activation may be the result of this receptor-ligand pair. In the presence of neutralizing antibodies against NKG2A, CD69 expression on NK cells as the result of coculture with mDC was indeed markedly increased (data not shown). However, the cytokine production was not affected, which indicated that NKG2A is not exclusively responsible for IFN-γ production. As previously described (31), we also observed an increase of the level of expression of 2B4 on NK cells of CHB patients. High-level 2B4 expression coincided with increased expression of PD-1 and decreased HBV-specific proliferation and cytotoxicity of CD8⁺ T cells (29). Antiviral therapy, however, did not result in decreased expression of 2B4, limiting the role of this receptor-ligand pair in improved mDC-NK cell interaction. In addition to the above-mentioned regulatory molecules, Peppa et al. recently showed that blocking IL-10 restores NK cell antiviral function in HBV infection (26), and therefore an alternative explanation for the reduced NK cell activity may be that it is mediated by DC-derived anti-inflammatory cytokines or additional negative regulation via inhibitory receptors (8).

Furthermore, we observed enhanced expression of CD80 and CD86 on mDC of CHB patients, which was reversed upon antiviral therapy. NK cell-induced mDC show a strong ability to produce IL-12p70 and thus direct the differentiation of CD4⁺ T cells and antigen-specific CD8⁺ T-cell responses (19). In unfractionated PBMC of CHB patients, in parallel with the restored allostimulatory capacity of mDC upon antiviral therapy, an increased capacity to produce IL-12 was indeed observed (34). As suggested before (4), restoration of the reciprocal interaction between NK cells and mDC upon successful antiviral therapy, exemplified by an NK cell-derived IFN-γ burst, may lead to enhanced IL-12 production by activated mDC, which may trigger the adaptive immunity to overcome viral immune evasion (6). In line with previous studies of CHB patients confirming functional impairment of mDC upon stimulation with cytokines and/or TLR ligands (11, 35, 21), this study shows lower cytokine production by mDC from patients upon stimulation with poly(I · C) and IFN-γ than by those from controls. Whether the mDC-induced NK cell function in CHB is also impaired upon the use of other stimuli remains to be resolved, since the low number of circulating mDC and the limited volume of blood that could be withdrawn from the patients did not allow additional experimental conditions to be tested.

TLR3/Mda-5 triggering with poly(I · C) may not exactly mimic the in vivo situation in CHB, but so far the physiological trigger for mDC activation in CHB has not been identified. However, this agonist is known to activate different mDC subsets, including the BDCA1⁺ mDC studied here, and has proved to be suitable to examine the functional interaction between mDC and NK cells in both humans and mice (13, 16, 27).

TLR3 expression in vivo on BDCA1⁺ mDC has not been reported to be different in CHB patients. However, decreased TLR3 protein expression in in vitro-generated monocyte-derived DC derived from CHB patients has been observed (18). Since poly(I · C)-induced TLR3 signaling in mDC has been shown to be mandatory for production of IFN-γ by NK cells through various mechanisms involving the TRIF-inducing membrane protein INAM and the RIG-1 receptor (13, 27), the reduced poly(I · C)-induced cytokine production may underlie the impaired DC-induced NK cell function.

In summary, on the basis of our findings obtained from in vitro coculture experiments, we conclude that impaired NK cell function in CHB patients may be partially the result of HLA restriction and decreased mDC-derived cytokine-mediated activation. Additional blocking experiments may provide more evidence to elucidate the exact mechanism of the defective interaction. Upon antiviral therapy, the result of improved mDC function is enhanced mDC-NK cell interaction and consequently improved NK cell function.