Sustained Hyperresponsiveness of Dendritic Cells Is Associated with Spontaneous Resolution of Acute Hepatitis C

Abstract

Some studies have reported that dendritic cells (DCs) may be dysfunctional in a subset of patients with chronic hepatitis C virus (HCV) infection. However, the function of DCs during acute HCV infection and their role in determining infectious outcome remain elusive. Here, we examined the phenotype and function of myeloid DCs (mDCs) and plasmacytoid DCs (pDCs) during acute HCV infection. Three groups of injection drug users (IDUs) at high risk of HCV infection were studied: an uninfected group, a group with acute HCV infection with spontaneous resolution, and a group with acute infection with chronic evolution. We examined the frequency, maturation status, and cytokine production capacity of DCs in response to the Toll-like receptor 4 (TLR4) and TLR7/8 ligands lipopolysaccharide (LPS) and single-stranded RNA (ssRNA), respectively. Several observations could distinguish HCV-negative IDUs and acute HCV resolvers from patients with acute infection with chronic evolution. First, we observed a decrease in the frequency of mature CD86⁺, programmed death-1 receptor ligand-positive (PDL1⁺), and PDL2⁺ pDCs. This phenotype was associated with the increased sensitivity of pDCs from resolvers and HCV-negative IDUs versus the group with acute infection with chronic evolution to ssRNA stimulation in vitro. Second, LPS-stimulated mDCs from resolvers and HCV-negative IDUs produced higher levels of cytokines than mDCs from the group with acute infection with chronic evolution. Third, mDCs from all patients with acute HCV infection, irrespective of their outcomes, produced higher levels of cytokines during the early acute phase in response to ssRNA than mDCs from healthy controls. However, this hyperresponsiveness was sustained only in spontaneous resolvers. Altogether, our results suggest that the immature pDC phenotype and sustained pDC and mDC hyperresponsiveness are associated with spontaneous resolution of acute HCV infection.

INTRODUCTION

A minor fraction of individuals infected with the hepatitis C virus (HCV) eliminate the virus spontaneously, while the majority develop persistent infection, which may lead to chronic liver diseases (1). Spontaneous resolution of acute HCV infection is associated with the development of a robust and sustained HCV-specific CD4 and CD8 T cell response (2). In contrast, such responses are weak, inefficient, or transient in individuals who develop chronic infection. Several mechanisms underlying this immune failure have been previously described. First, escape mutations in epitopes targeted by virus-specific CD8⁺ T cells can occur early during acute HCV infection, and this correlates with virus persistence (3–6). Second, failure of HCV-specific CD4 helper T cells to proliferate and produce cytokines was associated with viral recurrence and persistence. This loss of CD4 T cell help may compromise the function of virus-specific CD8 T cells and contribute to the development of suboptimal or exhausted CD8⁺ T cells (7–13).

Dendritic cells (DCs) are the most potent antigen-presenting cells in the body, and these could be part of one of the limiting steps in priming of an efficient anti-HCV CD4 and CD8 T cell response. Two major subsets of peripheral DCs exist in humans: CD11c⁺ myeloid DCs (mDCs) and CD123⁺ plasmacytoid DCs (pDCs). Both DC subsets can sense pathogens through pattern recognition receptors, such as Toll-like receptors (TLRs). mDCs express several TLRs, including TLR3 and TLR7/8, which allow them to recognize double-stranded RNA (dsRNA) and single-stranded RNA (ssRNA), respectively. pDCs also express several TLRs, including TLR7, which recognize ssRNA viruses. Under basal conditions, DCs are in an immature state and they express lower levels of major histocompatibility complex (MHC) classes I and II as well as costimulatory molecules but have high phagocytic capacity. Upon antigen uptake and the recognition of pathogen-associated molecular patterns (PAMPs) by TLRs, DCs upregulate the expression of MHC and costimulatory molecules and acquire an activated mature phenotype. This maturation process allows DCs to prime naive T cells through multiple signals (14). The first signal is antigen recognition by its specific T cell receptor as antigen-derived peptides presented by MHC class I and II molecules. The second signal is mediated by costimulatory molecules such as CD80 and CD86 which stabilize DC interactions with T cells. The third signal is provided by cytokines that can polarize the developing antigen-specific T cells toward a particular lineage. For example, interleukin-12 (IL-12), IL-4, and IL-23 polarize T cells toward the Th1, Th2, and Th17 lineages, respectively (15).

Earlier studies have reported that monocyte-derived DCs (MDDCs) from patients with chronic HCV infection exhibit an immature or normal phenotype and impaired allostimulatory capacity (reviewed in reference 57). Analyses performed directly ex vivo have demonstrated that the frequencies of mDCs and pDCs are reduced during acute and chronic HCV infections (18, 21–23) and that mDCs from patients with chronic HCV infection are defective in their IL-12-producing capacity (24–26). However, other functional data were inconsistent. Several studies have reported that mDCs from individuals with chronic HCV infection are defective in inducing the proliferation of allogeneic or antigen-specific T cells (23, 25, 26, 57), while others could not confirm these observations (18, 60). Similarly, alpha interferon (IFN-α) production by pDCs was reported to be impaired in some (23, 25, 28) but not all (27, 29) studies. Such inconsistencies may be due to variable patient demographics or experimental conditions, such as the type of stimuli used to activate DCs (cytokines or TLR agonists), the HCV genotype, the duration of infection, the patient's age and gender, and/or coinfection with other viruses. Notably, DCs were not extensively studied during acute HCV infection, where any defects in their function may impact priming of adaptive T cell responses. In particular, it is unclear whether defects in DC function, if any, occur during acute infection and whether viral persistence is the cause or the consequence of this dysfunction.

In this study, we used multiparametric flow cytometry to longitudinally monitor the phenotypic and functional changes in mDCs and pDCs from a unique cohort of injection drug users (IDUs) during acute HCV infections that progressed to spontaneous resolution or viral persistence. We report that spontaneous resolution of HCV infection is associated with an immature phenotype of pDCs and a sustained hyperresponsiveness of pDCs and mDCs to TLR7/8 stimulation.

MATERIALS AND METHODS

Study subjects and clinical follow-up.

Twenty-five patients with acute HCV infection, 10 HCV-negative IDUs, and 15 healthy donors were included in this study. Subjects with acute HCV infection and HCV-negative subjects were recruited among high-risk IDUs participating in the Montreal Acute Hep C cohort study (HEPCO), the methadone maintenance program, and the Hepatology Clinic at St-Luc Hospital of the Centre Hospitalier de l'Université de Montréal (CHUM). This study was approved by the institutional ethics committee (protocols SL05.014 and SL05.025) and conducted according to the Declaration of Helsinki. All participants signed informed consent upon enrollment.

Acute HCV infection (n = 25) was defined as the detection of HCV RNA and/or HCV antibodies following a previous negative test available in the 6 months prior to the positive result or positivity for HCV RNA with concomitant negative HCV antibody tests. The mean follow-up interval between the last aviremic time point and the first viremic time point was 3 months (range, 1 to 25 weeks), and the estimated time of infection was defined as the median between the last aviremic and the first viremic time points. The duration of infection was defined as the time (in months) after the estimated time of infection. Spontaneous viral resolution (n = 10) or persistent infection (n = 15) was defined as the absence or presence of HCV RNA, respectively, at 6 months postenrollment. In this study, three time points representing the three phases of HCV infection were analyzed for each patient: early acute phase, late acute phase, and follow-up. The early acute phase of HCV infection was defined as 2 ± 1 months after the estimated time of infection, the late acute phase/early chronic phase/early resolution phase (referred to here as the late acute phase for simplicity) was defined as 8 ± 2 months after the estimated time of infection, and the follow-up time point was defined as 17 ± 6 months after the estimated time of infection. All patients tested negative for human immunodeficiency virus (HIV) and hepatitis B virus (HBV) and were either medically ineligible for or declined IFN therapy.

HCV RNA testing and quantification.

Qualitative HCV RNA tests were performed using an automated Cobas AmpliPrep/Cobas Amplicor HCV test, version 2.0 (sensitivity, 50 IU/ml; Roche Molecular Systems, Inc., Branchburg, NJ). HCV genotyping was done using standard sequencing for the NS5B region and was performed by the Laboratoire de Santé Publique du Québec (Ste-Anne-de-Bellevue, QC, Canada) as part of the clinical follow-up of patients. Additional HCV RNA quantification was performed using an in-house quantitative real-time reverse transcription-PCR (qRT-PCR) assay as previously described (30). For the purpose of statistical analysis, plasma samples with undetectable HCV RNA were given a value of 100 IU/ml, which is the limit of detection.

Flow cytometry antibodies and reagents.

Directly conjugated antibodies against the following molecules were used: CD3 Alexa-700 (clone UCHT1), CD14 Alexa-700 (clone M5E2), CD19 Alexa-700 (clone HIB19), CD56 Alexa-700 (clone B159), CD86 phycoerythrin (PE)-Cy5 (clone IT2.2), CD123 PE (clone 7G3), HLA-DR allophycocyanin (APC)-H7 (clone L243), IL-6 fluorescein isothiocyanate (FITC; clone MQ2-6A3), IL-12p70 APC (clone C11.5), programmed death-1 receptor ligand (PDL1) PE-Cy7 (clone MIH1), PDL2 APC (MIH18), and tumor necrosis factor alpha (TNF-α) peridinin chlorophyll protein (PerCP)-Cy5.5 (clone MAb11) (all from BD Biosciences, San Jose, CA) and CD11c Pacific Blue (clone Bu15) and CD83 PerCP-Cy5.5 (clone HB15e) (both from BioLegend, San Diego, CA). Live cells were identified using an Aqua Live/Dead fixable dead cell stain kit (Life Technologies, Burlington, ON, Canada) according to the manufacturer's protocol. Fluorescence-minus-one control stains were used to determine the background levels of staining. Multiparameter flow cytometry was performed using a standard BD LSR II instrument equipped with blue (488 nm), red (640 nm), violet (405 nm), and yellow-green (561 nm) lasers (BD Biosciences) to systematically perform nine-color staining using FACSDiva software (version 5; BD Biosciences). Compensation was performed with single fluorochromes and BD CompBeads (BD Biosciences). Data files were analyzed using FlowJo software, version 9.4.10 for Mac (Tree Star, Inc., Ashland, OR). Polyfunctional data were exported using Boolean gates in FlowJo and further analyzed using PESTLE (version 1.6.2) and SPICE (version 5.22) software, obtained from M. Roederer, National Institutes of Health, Bethesda, MD (31).

Antigen uptake assay.

All flow cytometry assays were performed on cryopreserved samples. Peripheral blood mononuclear cells (PBMCs) were thawed and rested for 16 h at 5 × 10⁶ cells/ml in RPMI (Life Technologies), 2% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO), and 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM glutamine, and nonessential amino acids (all from Life Technologies). The FBS lot used was tested and shown not to induce nonspecific DC maturation. After resting, 2 × 10⁶ to 4 × 10⁶ PBMCs were left unstimulated for 24 h and for simplicity are referred to here as ex vivo PBMCs. To evaluate endocytosis, unstimulated cells were then incubated with FITC-dextran as previously described (32). Briefly, PBMCs were washed, resuspended at 20 × 10⁶ cells/ml in phosphate-buffered saline (PBS), 1% FBS, and distributed in two 96-well U-bottom plates before adding 1 mg/ml FITC-dextran (molecular weight, 40,000; Life Technologies). The FITC-dextran solution was vortexed for 30 s and sonicated for an additional 30 s immediately before use. One plate was incubated at 37°C (specific FITC-dextran uptake), and the second was incubated at 4°C (baseline uptake) for 30 min. Cells were washed twice in fluorescence-activated cell sorting (FACS) buffer (1× PBS, 1% FBS, 0.02% NaN₃) and surface stained as described below.

Multiparametric phenotypic characterization of DCs.

PBMCs (1 × 10⁶ to 2 × 10⁶) were stimulated with 1 μg/ml lipopolysaccharide (LPS; Escherichia coli O26:B6; Sigma-Aldrich) or with 5 μg/ml ssRNA40/LyoVec TLR7/8 agonist (Invivogen, San Diego, CA) for 24 h. For phenotypic characterization purposes, unstimulated (cells used for the antigen uptake assay described above) and stimulated PBMCs were then stained with surface antibodies for 30 min at 4°C, washed twice in FACS buffer, and then fixed in FACS fix buffer (1× PBS, 1% formaldehyde). The percent specific expression of FITC-dextran is calculated as the background-adjusted uptake at 37°C or 4°C. Unstimulated PBMCs incubated with FITC-dextran at 4°C served as a negative control to calculate the percent specific upregulation of phenotypic markers after TLR ligand stimulation.

Intracellular cytokine staining.

PBMCs were thawed and rested as described above. PBMCs (1 × 10⁶ to 2 × 10⁶) remained unstimulated or were stimulated with 1 μg/ml LPS or 5 μg/ml ssRNA40/LyoVec at 37°C. Following 1 h of stimulation, 10 μg/ml of brefeldin A (Sigma-Aldrich) was added and cells were incubated for a total of 6 h. Cells were washed with FACS buffer, stained for viability and cell surface antigens, and then permeabilized using BD Cytofix/Cytoperm solution (BD Bioscience). Cells were then stained with anti-IL-6, -IL-12, and -TNF-α antibodies for 30 min, washed twice in BD Perm/Wash buffer (BD Biosciences), and fixed in FACS fix buffer.

IFN-γ enzyme-linked immunosorbent (ELISPOT) assay.

HCV-specific T cell responses were measured for the majority of patients at most time points (75% of samples) where the DC phenotype and function were measured. The magnitude of the HCV-specific T cell response was determined by IFN-γ secretion upon recognition of HCV peptides. As previously described (33), PBMCs from HCV-infected patients were stimulated with 11 peptide pools spanning the entire HCV polyprotein and corresponding to HCV genotype 1a (H77 sequence) or genotype 3a (K3a/650 sequence), according to the patient's infecting HCV genotype. PBMCs from patients infected with other HCV genotypes, patients with undetermined genotypes, as well as HCV-negative IDUs were stimulated with peptides corresponding to HCV genotype 1a (H77 sequence).

IL-28B genotype screening by PCR.

IL-28B genotyping was performed using a rapid PCR-based assay developed by our laboratory (34). Briefly, the isolated genomic DNA was PCR amplified with standard Taq polymerase (Invitrogen) using appropriate oligonucleotide primer pairs in a reaction volume totaling 25 μl. PCR conditions in the thermocycler (Biometra, Goettingen, Germany) were as follows: initial denaturation at 94°C for 2 min, followed by 35 cycles of 94°C for 30 s, 65°C for 15 s, and 72°C for 1 min for genotype rs12979860 or 30 cycles of 94°C for 30 s, 54.5°C for 15 s, and 72°C for 1 min for genotype rs8099917. A final extension step at 72°C for 7 min was applied in each case. Appropriate volumes of the PCR mixtures were resolved on a 2% agarose gel, and the genotype was determined by the presence of amplified bands of interest.

Statistical analysis.

Comparisons between patient groups with HCV chronic evolution and spontaneous resolution during the early acute, late acute, and follow-up phases of infection were evaluated by two-way analysis of variance (ANOVA) (repeated measures). The data obtained were analyzed with SigmaPlot software, version 11.0 for Windows (Systat Software, Inc., Chicago, IL). Comparisons between HCV chronic evolution, spontaneous resolution, healthy donors, and HCV-negative IDUs were evaluated by one-way ANOVA. Correlations were examined by Pearson's test if data passed the normality test or by Spearman's test if the data did not pass normality. Data were analyzed with GraphPad Prism software, version 5.02 for Windows (GraphPad Software, San Diego, CA).

RESULTS

Reduced frequency of pDCs during acute HCV infection with chronic evolution.

We longitudinally examined the phenotype and function of DCs during acute HCV infection leading to patients with persistent viremia (n = 15) or spontaneous viral clearance (n = 10), here referred to as chronics and resolvers, respectively. Fifteen healthy donors and 10 HCV-negative IDUs were also included. The longitudinal follow-up of HCV-infected patients included evaluations of patients at 3 time points after the estimated date of infection: early acute phase (2 ± 1 months), late acute phase (8 ± 2 months), and long-term follow-up (17 ± 6 months). The patients' demographics and characteristics are summarized in Table 1. The individual patients' viral loads and alanine aminotransferase (ALT) values are listed in Table S1 in the supplemental material. First, we longitudinally monitored the frequency of mDCs (lineage negative [Lin⁻] HLA-DR⁺ CD11c⁺) and pDCs (Lin⁻ HLA-DR⁺ CD11c⁻ CD123⁺). The DC gating strategy by flow cytometry is shown in Fig. 1A. The frequency of mDCs did not change significantly between patient groups or infection time points (Fig. 1B). However, the frequency of pDCs was generally lower in HCV-negative IDUs and HCV-infected individuals, irrespective of the outcome or the phase of infection (Fig. 1C). This reduced frequency of pDCs was statistically significant only in individuals with persistent viremia compared to that in healthy donors during the long-term follow-up (P = 0.027).

Table 1.

Demographics and characteristics of patients and donors^a

Group	Total no. of subjects	Gender (no. of M/no. of F)	Median age (yr)	No. of individuals with the following genotype:
				HCV genotype 1/2/3/ND	IL-28B genotype rs12979860 C/C-*/T- ND	IL-28B genotype rs8099917 T/T-G/*-ND
HCV chronic evolution	15	14/1	35	4/0/6/5	6-8-1	7-7-1
HCV spontaneous resolution	10	7/3	30	1/2/3/4	5-5-0	5-5-0
Healthy donors	15	10/5	32	NA	0-0-15	0-0-15
HCV-negative IDUs	10	9/1	39	NA	5-1-4	6-2-2

^aM, males; F, females; ND, not determined; NA, not applicable.

Fig 1.

Decreased pDC frequency during acute HCV infection with chronic evolution. (A) Strategy for gating on the two DC subsets: Lin⁻ HLA-DR⁺ CD11c⁺ mDCs and Lin⁻ HLA-DR⁺ CD11c⁻ CD123⁺ pDCs (Lin⁻ DCs are CD3⁻, CD14⁻, CD19⁻, and CD56⁻). (B and C) The frequency of (B) CD11c⁺ mDCs and (C) CD123⁺ pDCs was determined ex vivo in patients with HCV chronic evolution, patients with HCV spontaneous resolution, healthy donors, and HCV-negative IDUs. The mean is represented by a horizontal bar. *, P < 0.05. P values were determined by two-way ANOVA (repeated measures) or one-way ANOVA (comparison with healthy donors and HCV-negative IDUs). (D) Spearman correlation between plasma viral load and frequency of CD123⁺ pDCs in patients with HCV chronic evolution and HCV spontaneous resolution during the early acute, late acute, and follow-up phases. Plasma viral load was measured by an in-house qRT-PCR. Dashed line, correlation with patients who have detectable HCV RNA only (>100 IU/ml); solid line, correlation with all patients.

Furthermore, we evaluated whether the level of viremia correlated with pDC frequency. Two analyses were performed: (i) on samples from all patients whether viral RNA was detectable or not and (ii) on samples from patients with detectable plasma viral loads (>100 IU/ml). We observed a negative correlation between viral load and the frequency of pDCs in all samples tested (Fig. 1D; r = −0.25, P = 0.051). This correlation became more statistically significant when only patients with detectable HCV RNA were considered in the analysis (r = −0.44, P = 0.009). Our findings suggest that the frequency of pDCs is reduced with HCV persistence and that this might be associated with high levels of viral replication.

mDCs exhibit an immature phenotype during acute HCV infection, irrespective of infection outcome.

We examined the maturation status of peripheral blood mDCs by monitoring the expression of the costimulatory molecule CD86, the maturation marker CD83, and the inhibitory receptors PDL1 and PDL2 directly ex vivo and following in vitro stimulation with different TLR ligands. Representative flow cytometry plots are shown in Fig. 2A. We observed reduced frequencies of CD86⁺ mDCs in all HCV-infected individuals, irrespective of infection outcome, compared to those in HCV-negative IDUs (for resolvers during early acute phase, P = 0.032; for chronics during late acute phase, P = 0.03; for chronics and resolvers during follow-up, P < 0.001). In addition, we observed lower frequencies of CD86⁺ and PDL1⁺ mDCs in all HCV-infected individuals (Fig. 2B and C) than healthy donors. However, this difference became statistically significant only in patients with chronic infection at the long-term follow-up time point as they entered the chronic phase of the infection (for CD86⁺ mDCs, P = 0.023; for PDL1⁺ mDCs, P = 0.02). Similarly, we demonstrated that the frequency of PDL2⁺ mDCs was reduced in all patients. However, in contrast to the PDL1 expression, this difference became statistically significant only at the long-term follow-up in resolver patients following viral clearance (Fig. 2D) (P = 0.04). Finally, no difference in the expression of the maturation marker CD83 was observed between the patient groups at the time points analyzed.

Fig 2.

The mDC immature phenotype is associated with evolution of infection, irrespective of outcome. (A) Representative dot plot for phenotypic characterization of DCs. (Top) Unstimulated DCs; (bottom) DCs stimulated with ssRNA for 24 h. Surface expression of maturation markers was studied in patients with HCV chronic evolution, patients with HCV spontaneous resolution, healthy donors, and HCV-negative IDUs. (B to D) Frequency of CD86⁺ (B), PDL1⁺ (C), and PDL2⁺ (D) unstimulated mDCs. (E) Frequency of ΔPDL1⁺ ssRNA-stimulated mDCs (baseline expression was subtracted from expression in the presence of TLR ligands). The mean is represented by a horizontal bar. *, P < 0.05; ***, P < 0.001. P values were determined by two-way ANOVA (repeated measures) or one-way ANOVA (comparison with healthy donors and HCV-negative IDUs).

In order to assess the function of DCs, we examined the activation and maturation of mDCs, measured by the upregulation of CD86, CD83, PDL1, and PDL2, following stimulation by TLR ligands. We chose to analyze the DC response to ssRNA, a ligand for TLR7/8, due to the fact that HCV is an ssRNA virus. We also studied the DC response to LPS, a ligand for TLR4, because the mDC response to LPS was previously shown to be affected during chronic HCV infection (35). Furthermore, by examining two distinct TLR signaling pathways, TLR7/8 signaling, which is solely MyD88 dependent, and TLR4 signaling, which can be MyD88 dependent or independent, we could distinguish if DCs have a selective or a general defect in the two TLR sensing pathways. We observed no difference in the frequency or mean fluorescence intensity (MFI) of most maturation markers in response to ssRNA or LPS (data not shown). The only statistically significant difference was observed following ssRNA stimulation, where mDCs from resolvers upregulated PDL1 expression to a higher level during the early acute and follow-up phases than mDCs from healthy donors (Fig. 2E). In summary, although not always statistically significant due to the low patient number, our data suggest that mDCs are immature in HCV-infected individuals and that viral clearance or persistence may have a differential influence on the expression of the inhibitory receptors PDL1 and PDL2 and their upregulation upon stimulation, as observed in resolvers versus chronics.

Correlation between DC maturation status and HCV-specific T cell responses.

Next, we sought to determine whether DC maturation status can affect the induction of HCV-specific T cell responses. We examined the correlation between DC maturation status, measured by the expression of CD86, PDL1, PDL2, and CD83, and the frequency of the HCV-specific T cell response, measured by IFN-γ ELISPOT assay, in response to overlapping HCV peptide pools spanning the entire HCV polyprotein. We detected HCV-specific T cell responses in several resolvers and chronics at various phases of infection. However, we could not establish any correlation with DC maturation when both patient groups were included in the analysis or when only chronics were considered (data not shown). When the resolvers group was analyzed alone, we observed a positive correlation between the frequency of PDL1⁺ unstimulated mDCs and the HCV-specific T cell response (Fig. 3A). A further increase in the PDL1 MFI following LPS stimulation led to a negative correlation with T cell responses (Fig. 3B), suggesting that HCV-specific T cells might be inhibited by PDL1-PD1 interaction. Finally, we observed a positive correlation between the increased frequency of CD83⁺ mDCs in response to LPS stimulation and the HCV-specific T cell response (Fig. 3C), suggesting that the level of mDC maturation may influence the magnitude of T cell responses in spontaneous resolvers.

Fig 3.

Positive correlation between CD83 upregulation and HCV-specific T cell response but negative correlation between PDL1 upregulation and HCV-specific T cell response. Expression of maturation markers in unstimulated or TLR ligand-stimulated mDCs was measured as described in the legend to Fig. 2. The magnitude of the HCV-specific T cell response in the same patients was determined by IFN-γ secretion in response to overlapping HCV peptides spanning the entire HCV polyprotein using the ELISPOT assay. (A) Correlation between HCV-specific T cell response measured in spontaneous resolvers and frequency of PDL1⁺ unstimulated mDCs; (B) MFI of PDL1 on mDCs in response to LPS stimulation; (C) frequency of CD83⁺ mDCs in response to LPS stimulation. The correlation was determined with Spearman's test. SFC, spot-forming cells.

The pDC immature phenotype correlates with spontaneous resolution.

The phenotype of pDCs was also examined in samples that had >90 events in the CD123⁺ gate. We observed that the frequency of CD86⁺, PDL1⁺, or PDL2⁺ pDCs ex vivo was generally lower in HCV-infected individuals and HCV-negative IDUs than in healthy donors, albeit slight differences were noted between the time points studied (Fig. 4A to C). Despite this general immature phenotype, differences were observed between HCV resolvers and chronics that became more significant at later time points postinfection. There was a general tendency for a lower frequency of CD86⁺ or PDL1⁺ pDCs in resolvers. This difference became statistically significant during the late acute phase for CD86 and PDL1 (Fig. 4A and B) (for CD86⁺ pDCs during late acute phase, P = 0.05; for CD86⁺ pDCs during follow-up, P = 0.014; for PDL1⁺ pDCs during late acute phase, P = 0.021). Similarly, the frequency of PDL2⁺ pDCs was lower in resolvers than chronics. A two-way repeated-measure ANOVA test demonstrated that the two groups were statistically significantly different overall, but the difference was never significant at any of the time points studied. This lack of significance is probably due to the limited numbers in each group (Fig. 4C) (P = 0.027). Interestingly, the frequency of CD86⁺, PDL1⁺, and PDL2⁺ was also significantly lower in HCV-negative IDUs than healthy donors and chronics, irrespective of the phase of infection. Finally, we observed no difference in the expression of the maturation marker CD83 between the different patient groups (data not shown).

Fig 4.

The pDC immature phenotype is associated with spontaneous resolution and protection from infection. Surface expression of maturation markers was studied in patients with HCV chronic evolution, patients with HCV spontaneous resolution, healthy donors, and HCV-negative IDUs. (A to C) Frequency of CD86⁺ (A), PDL1⁺ (B), and PDL2⁺ (C) unstimulated pDCs. (D) Frequency of change in CD86⁺ ssRNA-stimulated pDCs (baseline expression was subtracted from expression in the presence of TLR ligands). The mean is represented by a horizontal bar. *, P < 0.05; **, P < 0.01; ***, P < 0.001. P values were determined by two-way ANOVA (repeated measures) or one-way ANOVA (comparison with healthy donors and HCV-negative IDUs). (E) Spearman correlation between plasma viral load and change in CD86⁺ ssRNA-stimulated pDCs in patients with HCV chronic evolution and HCV spontaneous resolution during the early acute, late acute, and follow-up phases. Plasma viral load was measured by an in-house qRT-PCR. Dashed line, correlation with patients who have detectable HCV RNA only (>100 IU/ml); solid line, correlation with all patients.

Next, we evaluated the activation and maturation status of pDCs after TLR7 ligand (ssRNA) stimulation. pDCs from resolvers and HCV-negative IDUs, which exhibited the most immature phenotype and expressed the lowest levels of the maturation marker CD86 (Fig. 4A), underwent the highest upregulation of this marker following stimulation compared to that for pDCs from both healthy donors and chronics (Fig. 4D) (for resolvers versus chronics, for the change in CD86⁺ [ΔCD86⁺] during late acute, P = 0.043; for ΔCD86⁺ during follow-up, P = 0.023). Furthermore, chronics exhibited better responses, demonstrated by increased upregulation of CD86 during the early acute phase in comparison to that during the late acute phase (P = 0.028). Finally, we observed a negative correlation between the frequency of CD86⁺ pDCs after ssRNA stimulation and viral load in all samples (Fig. 4E; r = −0.37, P = 0.016).

No differences in the frequency or MFI of other phenotypic markers monitored after ssRNA stimulation were noted, and we could not establish any correlations between pDC phenotype, plasma ALT levels, and the magnitude of HCV-specific T cell responses (data not shown). In summary, pDCs from HCV-infected individuals, irrespective of disease outcome, and from HCV-negative IDUs had a more immature phenotype than those from healthy donors. This immature phenotype was more prominent in resolvers and HCV-negative IDUs and was associated with a superior ability to respond to ssRNA stimulation by upregulating the expression of costimulatory molecules.

Increased cytokine production in response to TLR4 stimulation in acute resolving HCV infection.

DCs secrete several proinflammatory cytokines that influence the development of adaptive immune responses. We studied the capacity of mDCs to secrete IL-6, IL-12, and TNF-α in response to LPS (Fig. 5A). pDCs were not studied since previous reports have shown that their function is altered by cryopreservation (36). The ability of mDCs to produce IL-6 and IL-12 in response to LPS was similar in all groups except for the HCV-negative IDUs, whose mDCs produced significantly higher levels of cytokines than mDCs from healthy donors (Fig. 5B and C). Similarly, we observed no differences between resolvers and chronics during the early acute phase in the ability of mDCs to produce TNF-α in response to LPS stimulation. However, with the disease progressing toward resolution or persistence, we observed an increase in the frequencies of TNF-α-producing mDCs in resolvers compared to chronics during the late acute and follow-up phases (Fig. 5D). We also observed higher frequencies of TNF-α-producing mDCs in HCV-negative IDUs than healthy donors and chronics, irrespective of the phase of infection.

Fig 5.

Increased cytokine production in response to TLR4 ligand is associated with spontaneous resolution and protection from infection. (A) Representative dot plot for functional characterization of mDCs. (Top) Unstimulated mDCs; (bottom) mDCs stimulated with LPS for 6 h. (B to E) Intracellular expression of cytokines was studied in patients with HCV chronic evolution, patients with HCV spontaneous resolution, healthy donors, and HCV-negative IDUs. Baseline expression was subtracted from expression in the presence of TLR ligands. Frequency of changes in total IL-6⁺ (B), IL-12⁺ (C), and TNF-α⁺ mDCs (D) in response to LPS stimulation. (E) Frequency of change in IL-6⁺, IL-12⁺, and TNF-α⁺ mDCs in response to LPS. The mean is represented by a horizontal bar. *, P < 0.05; **, P < 0.01. P values were determined by two-way ANOVA (repeated measures) or one-way ANOVA (comparison with healthy donors and HCV-negative IDUs). (F) Spearman correlation between plasma viral load and change in IL-12⁺ LPS-stimulated mDCs in patients with HCV chronic evolution and HCV spontaneous resolution during the early acute, late acute, and follow-up phases. Plasma viral load was measured by an in-house qRT-PCR. Dashed line, correlation with patients who have detectable HCV RNA only (>100 IU/ml); solid line, correlation with all patients.

Next, we analyzed the combined cytokine production capacity of mDCs that may reflect a higher activation status. We used SPICE software (31), which allowed us to determine the frequency of cells producing one, two, or three cytokines. We did not observe any significant difference in the frequency of mDCs producing all three cytokines (IL-6 positive [IL-6⁺], IL-12 positive [IL-12⁺], TNF-α positive [TNF-α⁺]) in response to LPS stimulation (Fig. 5E). The level of secretion of most cytokines examined did not correlate with viral load. The only exception was the production of IL-12 in response to LPS, which exhibited a negative correlation with plasma viral load, suggesting that IL-12 may be implicated directly or indirectly in spontaneous viral clearance (Fig. 5F).

We then evaluated whether the capacity of mDCs to produce cytokines correlated with HCV-specific T cell responses. Similar to the results obtained for the mDC phenotype (Fig. 3), we could not establish any correlation when the analysis included both chronics and spontaneous resolvers or only chronics (data not shown). When only the resolvers group was analyzed, we observed a positive correlation between the frequency of IL-6⁺ (Fig. 6A), IL-12⁺ (Fig. 6B), TNF-α⁺ (Fig. 6C), and IL-6⁺, IL-12⁺, and TNF-α⁺ (Fig. 6D) mDCs in response to LPS stimulation and HCV-specific T cell responses.

Fig 6.

Positive correlation between mDC cytokine production in response to LPS stimulation and HCV-specific T cell responses. Intracellular expression of cytokines by mDCs in response to LPS stimulation was measured as described in the legend to Fig. 5. The magnitude of the HCV-specific T cell response in the same patients was determined by IFN-γ secretion in response to overlapping HCV peptides spanning the entire HCV polyprotein using the ELISPOT assay. (A) Correlation between HCV-specific T cell response measured in spontaneous resolvers and frequency of change in IL-6⁺ mDCs in response to LPS stimulation; (B) frequency of change in IL-12⁺ mDCs in response to LPS stimulation; (C) frequency of change in TNF-α⁺ mDCs in response to LPS stimulation; (D) frequency of change in IL-6⁺, IL-12⁺, and TNF-α⁺ mDCs in response to LPS stimulation.

Enhanced cytokine production in response to TLR7/8 stimulation in acute resolving HCV infection.

Next, we evaluated the ability of mDCs to produce these cytokines in response to ssRNA stimulation. We observed a slight increase in IL-6 production by mDCs from resolvers during the early acute phase compared to that by mDCs from healthy donors (Fig. 7A) (P = 0.026). More importantly, we observed higher IL-6 production by mDCs from HCV-negative IDUs than mDCs from all HCV-infected patients and healthy donors. We also observed a significantly higher level of IL-12 production by mDCs from resolvers, chronics, and HCV-negative IDUs than those from healthy donors at all time points except during the late acute phase for the chronics (Fig. 7B) (for resolvers during early acute phase, P < 0.001; for chronics during early acute phase, P = 0.013; for resolvers during late acute phase, P = 0.005; for resolvers during follow-up, P < 0.001; for chronics during follow-up, P < 0.001). Similarly, we observed higher TNF-α responses in both resolvers and chronics during the early acute phase than healthy donors. This increased response was sustained only in resolvers and gradually diminished in chronics at the later stages of infection (Fig. 7C) (for resolvers during early acute phase, P = 0.004; for chronics during early acute phase, P = 0.02; for resolvers during late acute phase, P < 0.001; for resolvers during follow-up, P = 0.002). We also observed higher TNF-α production by mDCs from HCV-negative IDUs than mDCs from all HCV-infected patients and healthy donors. Furthermore, the overall frequency of IL-6⁺, IL-12⁺, and TNF-α⁺ mDCs was significantly higher in both resolvers and chronics during the early acute phase than healthy donors, but this difference was sustained only in resolvers (Fig. 7D) (for resolvers during early acute phase, P < 0.001; for chronics during early acute phase, P = 0.023; for resolvers during late acute phase, P = 0.046; for resolvers during follow-up, P = 0.043). We could not establish any correlation between the level of cytokines produced by mDCs in response to ssRNA stimulation and plasma ALT levels or the magnitude of HCV-specific T cell responses (data not shown). Our findings suggest that during the early acute phase both resolvers and chronics are hyperresponsive to ssRNA stimulation compared to the responsiveness of healthy donors, but this hyperresponsiveness is diminished with viral persistence. HCV-negative IDUs also exhibited a hyperresponsive profile compared to that for healthy donors, suggesting a potential confounding effect of subclinical exposure to HCV.

Fig 7.

Increased cytokine production in response to TLR7/8 ligand is associated with spontaneous resolution and protection from infection. Intracellular expression of cytokines was studied in patients with HCV chronic evolution, patients with HCV spontaneous resolution, healthy donors, and HCV-negative IDUs. Baseline expression was subtracted from expression in the presence of TLR ligands. (A to C) Frequency of changes in total IL-6⁺ (A), IL-12⁺ (B), and TNF-α⁺ (C) ssRNA-stimulated mDCs. (D) Frequency of change in IL-6⁺, IL-12⁺, and TNF-α⁺ mDCs in response to ssRNA stimulation. The mean is represented by a horizontal bar. *, P < 0.05; **, P < 0.01; ***, P < 0.001. P values were determined by two-way ANOVA (repeated measures) or one-way ANOVA (comparison with healthy donors and HCV-negative IDUs).

HCV load correlates with increased mDC antigen uptake.

In order to further assess the function of mDCs, we measured antigen uptake by using the FITC-dextran uptake assay (32). Representative flow cytometry plots are shown in Fig. 8A. We observed no differences in the frequency (data not shown) or the MFI (Fig. 8B) of cells that took up dextran beads between the different groups of patients or the different time points of infection. Nevertheless, we established a positive correlation between antigen uptake, measured by the MFI of dextran-positive cells, and the viral load in patients with detectable HCV RNA (Fig. 8C) (r = 0.44, P = 0.009). These results demonstrate that high levels of viral replication do not interfere with mDC antigen uptake and may even enhance it.

Fig 8.

Viral load correlates with increased mDC antigen uptake. (A) Representative dot plot for antigen uptake functional assay. (Left) PBMCs incubated with FITC-coupled dextran beads at 4°C (baseline spontaneous uptake); (right) PBMCs incubated at 37°C (specific uptake). (B) Change in MFI dextran-positive mDCs in patients with HCV chronic evolution, patients with HCV spontaneous resolution, healthy donors, and HCV-negative IDUs (background uptake was subtracted from specific uptake). The mean is represented by a horizontal bar. Two-way ANOVA (repeated measures) or one-way ANOVA (comparison with healthy donors and HCV-negative IDUs). (C) Spearman correlation between viral load and change in MFI of dextran-positive mDCs in patients with HCV chronic evolution and HCV spontaneous resolution during the early acute, late acute, and follow-up phases. Plasma viral load was measured by an in-house qRT-PCR. Dashed line, correlation with patients who have detectable HCV RNA only (>100 IU/ml); solid line, correlation with all patients.

DISCUSSION

We have analyzed for the first time the longitudinal evolution of mDC and pDC phenotypes, as well as the function of mDCs during acute HCV infection. We observed that HCV-infected individuals, irrespective of the infection outcome, had a reduced frequency of mature (CD86⁺, PDL1⁺, and PDL2⁺) mDCs and pDCs compared to healthy donors. HCV resolvers and high-risk uninfected IDUs demonstrated similar characteristics, but they were both distinguished from chronics by several features. First, pDCs from resolvers and HCV-negative IDUs expressed lower levels of the DC maturation markers that were tested than pDCs from chronics. This immature state was associated with higher activation and upregulation of maturation markers upon stimulation with ssRNA. Second, mDCs from resolvers and HCV-negative IDUs produced higher levels of TNF-α upon stimulation with the TLR4 ligand LPS. Third, although resolvers and chronics were hyperresponsive to stimulation with the TLR7/8 ligand ssRNA during the early acute phase, this hyperresponsiveness was sustained only in resolvers. HCV-negative IDUs were also hyperresponsive to ssRNA compared to the responsiveness of both chronics and resolvers. Collectively, these results suggest that DCs are activated in all individuals with potential exposure to HCV, including HCV-negative IDUs. They also suggest that sustaining such activity is essential for successful viral clearance.

Several key differences in DC phenotype and function in the different groups were observed. Given the limited number of patients studied, we could not assess the role of host factors like the IL28B genotype (Table 1) on the DC phenotype or function in a statistically significant manner.

DCs from resolvers and HCV-negative IDUs were similar in many phenotypic and functional aspects, and these cells differed significantly from those from chronics and healthy donors. This suggests that drug usage may influence DC function. However, we cannot rule out the possibility that DC function in HCV-negative IDUs is also affected by subclinical exposure to HCV. It is tempting to speculate that such characteristics as DC hyperresponsiveness and increased cytokine production may have a protective role against HCV infection in these highly exposed individuals. Indeed, innate immune responses were recently shown to be important for protection against HCV (37), but further studies with clearly documented HCV exposure are required to validate this hypothesis. Both HCV-negative IDUs and healthy donors are important control populations, and data interpretation may differ depending on which of the two groups is used as a reference. For simplicity and consistency with previous studies, we decided to use healthy donors as a reference to rule out any confounding factors related to subclinical HCV exposure.

We observed a general decrease in the frequencies of peripheral blood pDCs during acute HCV infection that became statistically significant only in chronics at the late follow-up phase, and this correlated with viral load. This is consistent with previous reports that demonstrated reductions in the frequencies of pDCs in patients with chronic HCV infection compared to healthy donors (21, 27, 35, 38). One possible mechanism is the increased migration of pDCs to the liver. Indeed, pDCs were shown to localize to the liver during HCV infection (38–40), where they are able to sense HCV-infected hepatocytes and produce large amounts of type I IFN (41). Another possibility could be that type I IFN induces the death of pDCs, as was recently described by Swiecki et al. (42). Type I IFNs and interferon-stimulated genes (ISGs) are induced in the liver of HCV-infected patients (43, 44) and chimpanzees (45–48). It is therefore tempting to speculate that high and persistent levels of type I IFN in blood or liver may be responsible for decreased pDC frequency during acute and chronic HCV infection, but this will require further investigation.

We have observed an immature phenotype of pDCs in resolvers, chronics, and HCV-negative IDUs compared to healthy donors. This difference was more pronounced in resolvers during the late acute and follow-up phases and in HCV-negative IDUs. Despite the prevalence of such immature phenotypes, pDCs from resolvers and HCV-negative IDUs were hyperresponsive to ssRNA stimulation compared to the responsiveness of pDCs from chronics and healthy donors and upregulated the maturation marker CD86. This may suggest a regulation mechanism used by pDCs to avoid excessive stimulation following viral clearance and to respond better to invading pathogens. This hyperresponsiveness negatively correlated with viral load, suggesting that pDCs from chronics may be overstimulated due to the high viral load and higher levels of type I IFN production and are less likely to respond to further stimulation. Similarly, mDCs were hyperresponsive to ssRNA stimulation, as demonstrated by higher cytokine production during acute HCV infection, irrespective of the infection outcome. However, this hyperresponsiveness was sustained only in resolvers. The increased mDC responsiveness to ssRNA during early acute HCV infection followed by its decline in chronics at later stages suggests that mDCs are activated early during acute infection, irrespective of infection outcome, and that reduced mDC function is not the cause but rather an effect of persistent viral replication. It is also possible that reduced cytokine production concomitant with viral persistence may be a beneficial shutoff mechanism to reduce inflammation and prevent virus-induced immunopathology by favoring the induction of regulatory T cells, which are known to play a role in the chronic evolution of HCV infection (49, 50).

Although sustained mDC hyperresponsiveness to ssRNA was associated with spontaneous resolution, it did not correlate with the frequency of HCV-specific T cells. These results suggest that cytokine production by mDCs in response to ssRNA might not directly affect priming of HCV-specific T cells but it may act indirectly on other cell types that influence T cell priming and infectious outcome. Nevertheless, we observed that high TNF-α production by mDCs in response to LPS was associated with spontaneous resolution and this correlated with the frequency of HCV-specific T cells in resolvers only. This is consistent with the findings of a previous study demonstrating that a reduced response to LPS but not ssRNA by mDCs may lead to exhaustion of CD8⁺ T cells and reduced polyfunctionality during chronic hepatitis C virus infection in a limited subset of patients who are susceptible to higher levels of functional impairment (51). Since LPS is recognized by TLR4 and signals through the TIR-domain-containing adapter-inducing IFN-β (TRIF) or MyD88 and ssRNA is recognized by TLR7/8 and signals solely through MyD88, it is possible that intact TRIF signaling in mDCs is more important for inducing HCV-specific T cell responses.

TLR desensitization may be one mechanism to explain the loss of mDC hyperresponsiveness in chronics upon viral persistence. It has been shown that HCV activates TLR2 (52, 53) and induces tolerance to further TLR2 or TLR4 stimulation (54). Furthermore, during HIV infection (55), microbial translocation leads to high plasma LPS levels and TLR4 tolerance in monocytes. Interestingly, high plasma LPS levels were also observed during chronic HCV infection (56), and this is thought to occur because of the decreased capacity of the damaged liver to filter LPS coming in from the gut. In this study, we have measured LPS in the plasma of patients with acute HCV infection, but the levels were either low or undetectable and did not correlate with TNF-α production by mDCs in response to LPS stimulation (data not shown). Since patients with acute HCV infection have minimal liver damage, low LPS levels in the plasma were expected. Nonetheless, these results suggest that TLR4 tolerance through increased excessive stimulation is not the cause of reduced mDC function in chronics, at least during the acute and early chronic phases.

Despite the reduction in mDC hyperresponsiveness with viral persistence in chronics, there was no global defect in DC function in that population compared to healthy donors. However, it is still possible that other defects restricted to liver-resident DCs may exist. If this is the case, decreased antigen uptake, migration from the liver to the lymph nodes, or antigen presentation to T cells may influence the priming of HCV-specific responses. One final possibility could be specific dysfunctions in the DCs presenting HCV antigens or those that have had prior exposure to the virus. The ability of HCV to productively infect DCs is controversial (57), but it is possible that phagocytosis of apoptotic HCV-infected hepatocytes and expression of HCV proteins are enough to disrupt DC function. In addition, the HCV NS3/4A protease has been shown to block signaling through TLR3, TLR4, and retinoic acid inducible gene I (RIG-I) and, consequently, block IFN-β production by infected cells and in infected tissues (58, 59).

In conclusion, our results suggest that DCs are activated in vivo early during acute HCV infection but persistent infection might impair their function. We propose that sustained mDC and pDC hyperresponsiveness is associated with spontaneous resolution of acute HCV infection and that this is essential in the maintenance of an efficient antiviral response and the prevention of viral recurrence.