Chronic evolution of acute hepatitis C (aHC) occurs in more than 80% of patients but can frequently be prevented by early treatment with interferon (IFN)-α. Plasmacytoid dendritic cells (pDCs) are the major endogenous IFN-α producers, but their role in aHC is unknown. In this study, frequency, phenotype, and pDC function were analyzed in 13 patients with aHC and 32 patients with chronic hepatitis C (cHC) compared with 20 healthy controls, 33 sustained responders to antiviral treatment, 14 patients with acute hepatitis B (aHB), and 21 patients with nonviral inflammatory disease. In aHC, pDCs in the peripheral blood were significantly reduced compared with healthy controls (median, 0.1% vs. 0.36%, P < .0005) and were inversely correlated to alanine aminotransferase levels (r = −0.823; P < .005). Circulating pDCs in aHC were immature, as determined via reduced expression of HLA-DR and CCR7, and produced little amounts of IFN-α (median, 3.5 pg/50,000 peripheral blood mononuclear cells [PBMCs] vs. 498.4 pg/50,000 PBMCs in healthy controls; P < .0005). Less pronounced changes were present in cHC (median, 0.17%, 28.0 pg/50,000 PBMCs IFN-α, respectively). However, a significantly reduced frequency and IFN-α production was also found in self-limited aHB (median 0.1%, 8.6 pg/50,000 PBMCs) and in patients with nonviral inflammatory disease (median 0.19%, 7.5 pg/50,000 PBMCs). In conclusion, in aHC frequency and IFN-α–producing capacity of peripheral blood pDCs are dramatically reduced and inversely correlated with the degree of liver inflammation. In cHC there is incomplete recovery of pDC function, which, however, could be solely due to the chronic inflammatory state. (HEPATOLOGY 2005;41:643–651.)
Hepatitis C virus (HCV) infection leads to chronic viral persistence in the majority of newly infected patients. Nevertheless, approximately 15% of all patients and up to 50% of symptomatic subjects with acute hepatitis C (aHC) spontaneously achieve long-term viral clearance,1 which is attributed to a strong HCV-specific CD4+ T helper 1 and CD8+ cytotoxic T-cell response.2–5 In contrast, once chronic HCV persistence is established, spontaneous viral clearance becomes extremely rare.6 The acute and chronic phases of HCV infection also differ remarkably in their response to antiviral treatment: whereas in chronic hepatitis C (cHC), a combination treatment of pegylated interferon (IFN)-α and ribavirin is required to achieve approximately 50% of sustained virological response,7, 8 in aHC, monotherapy with conventional IFN-α leads to more than 95% sustained viral clearance.9 The reasons for the difference in treatment response between aHC and cHC are currently not understood. Nevertheless, it seems that the substitution of IFN-α in the early phase of virus–host interaction during aHC successfully interferes with a process that is required for chronic viral persistence, suggesting that type I IFNs play a central role in the initial interaction between HCV and the host.
Plasmacytoid dendritic cells (pDCs) have been identified as the major IFN-α–producing cells in the human body.10 pDCs circulate at low levels (<1%) in the peripheral blood but have been shown to migrate to lymph nodes and other tissues to produce IFN-α in situ.11 pDCs respond to viral infection and certain oligodeoxynucleotides containing unmethylated cytidine phosphat guanosine (ODN CpGs) with the production of high amounts of type I IFNs and may be an important part of the innate immune response to intracellular pathogens.12, 13 In addition, they are able to stimulate specific T cells and may thus represent a link between the innate and the specific immune response.12, 14
In cHC, an impairment of monocyte-derived dendritic cells has been suggested in several studies15–18 and could be due to direct infection of these cells by HCV.19 Along these lines, infection of pDCs,20 a reduced relative frequency of pDCs, and a reduced capacity to produce IFN-α have recently been shown in cHC21–23 and during IFN-α treatment,24 but their relevance—particularly during the acute phase of HCV infection, when the decision of viral clearance versus persistence is made—has not been studied.
In this study, we analyzed the frequency, phenotype, and function of pDCs in patients with aHC in comparison with healthy controls, patients with cHC, patients with resolved cHC, patients with acute hepatitis B (aHB), patients with nonviral liver disease, and patients with bacterial infections.
The following patients were included: 13 patients with aHC, 32 patients with cHC (based on anti-HCV enzyme-linked immunosorbent assay positivity and HCV RNA in serum), 33 individuals with sustained viral clearance, 14 patients with acute symptomatic hepatitis B infection, 10 patients with nonviral hepatic disease (5 primary sclerosing cholangitis, 2 autoimmune hepatitis, 1 primary biliary cirrhosis, 1 drug-induced, and 1 steatohepatitis), 11 patients with bacterial infections (4 pneumonia, 3 urinary tract infection, 1 diverticulitis, 3 cholangitis), and 20 healthy controls (Table 1). Acute HC was diagnosed as a result of documented seroconversion to anti-HCV antibodies or all of the following: acute onset of hepatitis in a previously healthy individual; aminotransferases at least 10 times the upper limit of normal; exclusion of other infectious, metabolic, or toxic causes of hepatitis; recent exposure; or source of infection identified. Acute HB was diagnosed as a result of hepatitis B surface antigen, hepatitis B e antigen, and anti–hepatitis B core immunoglobulin M positivity and acute onset of hepatitis in a previously healthy individual. All patients with acute viral hepatitis were symptomatic and were studied within 1 month after onset of clinical symptoms. Of the 32 cHC patients, 2 had received IFN/ribavirin combination treatment previously (2 and 9 years ago, respectively). No patient with cHC received antiviral therapy during the study period. The genotype of patients within the cHC group was determined in 26 of the 32 cases. Nineteen patients were infected with genotype 1 (7 with 1a, 9 with 1b, 3 not subtyped), 3 patients were infected with genotype 2 (1 with 2a and 2 with 2b), and 4 patients were infected with genotype 3a. Liver histology was available in 4 patients, showing mild fibrosis in 3 and moderate fibrosis in 1. For the other patients, the APRIscore (aspartate aminotransferase to platelet ratio index) as a noninvasive test to predict cirrhosis in cHC was applied.25 Only 2 of 32 patients had a value of more than 2.0, which is predictive of cirrhosis, and 3 patients had a score of more than 1.5, predicting significant fibrosis. Thus, the majority of our patients probably had early-stage liver disease.
Within the group of patients with viral clearance, 27 had cleared HCV through IFN-based antiviral therapy 12 to 82 months before entry in the study (mean: 39 months). Twenty-three patients had received IFN/ribavirin combination treatment, 4 had received IFN monotherapy for 6 to 12 months according to genotype, and 6 had cleared the virus spontaneously. All patients gave informed consent to participate in the study, and the protocol and the procedures of the study were conducted in conformity with the ethical guidelines of the Declaration of Helsinki.
Preparation of Peripheral Blood Mononuclear Cells.
Peripheral blood mononuclear cells (PBMCs) were isolated via Ficoll-Hypaque density centrifugation (Biochrom, Berlin, Germany) of fresh heparinized peripheral blood and were resuspended in tissue culture medium (RPMI 1640 medium; Gibco, Grand Island, NY) containing 2 mmol/L L-glutamine, 1 mmol/L sodium pyruvate, 100 units penicillin per mL, 100 μg streptomycin per mL, and 5% human AB serum.
FACS Analysis of pDCs.
PBMCs were stained with the following antibodies: BDCA-2 phycoerythrin (PE), BDCA2-Biotin (Miltenyi Biotec, Bergisch Gladbach, Germany), CD3 FITC, CD14 FITC, CD16 FITC, CD19 FITC, CD80 PE, CD83 PE, CD86 PE, immunoglobulin G1 PE, immunoglobulin G2a PE, immunoglobulin G2b PE (Immunotech, Marseille, France), human leukocyte antigen (HLA)-DR PE (PharMingen, San Diego, CA), CCR1 PE, CCR5 PE, CCR7 PE (R&D Systems, Wiesbaden, Germany), CD4 PC5, and streptavidin PC5 (Becton Dickinson Biosciences, Heidelberg, Germany). Stained cells were analyzed using a FACScan and Cellquest software (Becton Dickinson Biosciences).
After excluding dead cells based on light scatter analysis, 100,000 to 300,000 events per run were acquired. Lineage-negative cells (CD3−, CD14−, CD16−, and CD19−) were gated and analyzed for the expression of CD4+ and BDCA2+ (relative frequency of pDCs). The absolute number of circulating blood pDCs was calculated using the percentage of cells with respect to the mononuclear cell counts, as determined by an automated differential blood count (absolute frequency of pDCs).
Type I IFN Production.
PBMCs (5 × 104/well) from controls and patients were incubated in 96-well U-bottom (TPP, Trasadingen, Switzerland) with CpG for 24 hours (ODN 2216: 5′-ggGGGACGATCGTCgggggG-3′ (small letters indicate phosphorothioate linkage; capital letters indicate phosphodiester linkage 3′ of the base; boldface letters indicate CpG dinucleotides synthesized by Metabion, Munich, Germany). Cell-free supernatants were harvested and tested via commercial enzyme-linked immunosorbent assay using a matched antibody pair for IFN-α (PBL Biomedical Laboratories, Piscataway, NJ) and performed according to the manufacturer's instructions. The sensitivity of the assay was 10 pg/mL.
Nonparametric statistical analysis was performed using the Mann-Whitney U test; Spearman rank correlation was used for correlation analysis. A P value of less than .05 was considered statistically significant.
Frequencies of pDCs in Viral Hepatitis
pDCs were identified among lineage marker–negative cells by staining with BDCA-2 and CD4 antibodies (Fig. 1A-B). All BDCA2+ cells were also positive for CD123, which is typical for pDCs (Fig. 1C) and BDCA4 (data not shown) The staining was of similar quality in all groups of patients and controls as shown in Fig. 2, where the patient closest to the median of the group is shown as a representative example. The percentage of pDCs in PBMCs from patients with aHC infection (median, 0.1%; range, 0.02%-0.23%) were significantly reduced compared with the control group (median, 0.36%; range, 0.07%-1.08%; P < .0001). Within the aHC group, there was no obvious difference between patients with spontaneous viral clearance and patients with a chronic course of HCV infection (Fig. 3A). There was also a decrease of pDCs in cHC patients (median, 0.17%; range, 0.04%-0.60%; P < .005). No significant difference could be observed between the sustained responder group (median, 0.29%; range, 0.08%-0.57%) and the control group (Figs. 2, 3A). The frequency of pDCs in patients with aHB, nonviral liver disease, and bacterial infections was also significantly reduced (aHB virus: median, 0.09%, range, 0.02%-0.19%; nonviral liver disease: median, 0.24%, range, 0.07%-0.32%; bacterial infection: median, 0.17%, range, 0.01%-0.37%).
Based on the differential blood count, the absolute numbers of pDCs per μL peripheral blood were calculated (Fig. 3B). Similar to the relative frequencies, the absolute counts of pDCs were significantly reduced in patients with aHC and to a lesser extent in patients with cHC compared with the pDC counts of healthy controls (P < .0005 and P < .05, respectively). However, no significant differences were found between patients with aHC and patients with aHB, patients with nonviral liver disease, or patients with bacterial infections.
IFN-α Production in Acute Viral Hepatitis
To obtain a better understanding of the effector functions of pDCs, their ability to produce IFN-α in response to ODN CpG was analyzed within the different patient groups. As previously described, pDCs have been shown to be the main source of IFN-α in the peripheral blood of humans.10 To confirm this finding in patients with HCV infection, we separated PBMCs with magnetic beads coupled to BDCA-4 antibodies and stimulated both fractions with ODN CPG, which showed 50 times the amount of IFN-α in the supernatant of the BDCA-4–positive fraction compared with the BDCA-4–negative fraction. To elucidate whether the residual IFN-α production in the BDCA-4–negative fraction is due to contaminating pDCs or another subset of PBMCs, we repeated the separation experiment with BDCA-2, which is also exclusively expressed on pDCs and which very efficiently and specifically turns off IFN-α production in pDCs.26 Both the BDCA-2–positive and the BDCA-2–negative fraction were completely unresponsive with respect to their ability to produce IFN-α after stimulation with ODN CPG, confirming that IFN-α production in this experimental setting is exclusively due to BDCA-2–positive cells, namely pDCs (data not shown). Based on these results, we chose to measure IFN-α secretion in unseparated PBMCs to avoid the loss of some pDCs or any potential change in phenotype caused by the isolation procedure.
These experiments showed that in addition to the decrease of pDCs in acute viral hepatitis, the capacity of PBMCs to produce IFN-α was dramatically reduced in patients with aHC (median, 3.5 pg/5 × 104 PBMCs; range, 0.1–67.4 pg/5 × 104 PBMCs) and acute hepatitis B virus (median, 8.6 pg/5 × 104 PBMCs; range, 1.1–86.7 pg/5 × 104 PBMCs) compared with healthy controls (median, 498.4 pg/5 × 104 PBMCs; range, 11.3–2399.3 pg/5 × 104 PBMCs; P < .0005). Furthermore, in the group of patients with chronic HCV infection (median, 28.0 pg/5 × 104 PBMCs; range, 1.0–396.0 pg/5 × 104 PBMCs) and in the group of sustained responders to antiviral therapy (median, 90.0 pg/5×104 PBMCs; range, 1.0–770.0 pg/5 × 104 PBMCs), a statistically significant decrease in the capacity of PBMCs to produce IFN-α could be detected (P < .0005 and P < .005, respectively) (Fig. 4A). Similar to the reductions in pDC frequency, significant reductions of IFN-α were also observed in patients with nonviral liver disease and bacterial infections. The calculation of IFN-α production per pDC shows that the reduction of IFN-α synthesis in acute viral hepatitis and also in cHC as well as in control patients is not only due to the reduced frequency but also to the reduced capacity of a single pDC to produce IFN-α in acute viral hepatitis (P < .005) (Fig. 4B). Because strong activation of pDCs in vivo could also explain the decreased IFN-α–producing capacity in vitro, we measured IFN-α serum levels of patients with acute viral hepatitis which, however, were not significantly different from healthy controls (data not shown).
Correlation of Clinical Parameters With Frequency and Function of pDCs in aHC Infection
We were able to analyze pDC functions in 9 patients with aHC. Of these, 2 patients cleared the virus spontaneously, 4 were treated with combination antiviral therapy and achieved a sustained virological response, 1 developed cHC, and 2 were lost to follow-up. Interestingly, the 2 patients with spontaneous control of HCV infection could mount the highest IFN-α response (Fig. 4A-B) among the aHC group, although this subgroup is too small for a statistical analysis.
To test the hypothesis that pDCs are attracted to the site of inflammation, we looked for a correlation of alanine aminotransferase (ALT) levels and pDC frequency in aHC. Indeed, a significant inverse correlation of ALT levels and the relative frequency of pDCs within the first 4 weeks of aHC (P < .005) was noted (Fig. 5).
Frequency and Function of pDCs in the Course of aHC.
Six patients with acute symptomatic HCV infection, 3 with self-limited disease, 1 evolving into cHC, and 2 treated with IFN/ribavirin 3 and 5 months after disease onset, respectively, were followed over a period of 7 to 17 months (Fig. 6A-F).
All patients had reduced pDC numbers and IFN-α production during the first 4 weeks after disease onset. In 2 patients with self-limited aHC, the pDC number and function remained low for more than 6 months; this is notable because HCV RNA was undetectable and ALT had been normal for several months. One patient transiently controlled HCV but eventually relapsed and developed cHC. Similar to the patients with self-limited disease, pDC number and function remained low throughout the 4 months of viral control. Two patients were treated 3 and 5 months after the onset of symptoms, respectively, when spontaneous recovery seemed unlikely. In agreement with previous studies, pDC number and IFN-α remained low during exogenous IFN-α treatment but recovered to normal levels 3 to 6 months after successful treatment was stopped. The patient who started treatment 5 months after disease onset (Fig. 6F) is of particular interest: this patient went through a phase of transient viral control with HCV RNA levels as low as 4,000 IU/mL with normal or slightly elevated ALT levels. Although pDC number and function were very low during the first few weeks of aHC, both parameters returned to normal levels during the phase of relative viral control. Relapse occurred nonetheless and was again accompanied by reduced pDC number and IFN-α secretion.
Activation and Homing Receptors in Acute Viral Hepatitis
To determine the activation level and the expression of homing receptors in pDCs during aHC, we measured the expression levels of HLA-DR; important costimulatory molecules such as CD80 and CD86; the maturation marker CD83; and the chemokine receptors CCR1, CCR5, and CCR7 within the different groups. A significantly lower expression of CCR7 and HLA-DR (P < .05) was seen in patients with acute and chronic viral hepatitis compared with healthy donors (Fig. 7). No significant difference was detected between the sustained responder group and the healthy donors. The expression levels in a patient with aHC who cleared the virus after antiviral therapy were measured during acute disease and approximately 4 months after termination of treatment; there was a marked increase of the expression levels of HLA-DR and CCR7. In contrast, only a slight increase of the expression levels (time points: 1 month and 3 months after disease onset) was seen in a patient who had a chronic course of the disease.
Type I IFNs are crucial components of the innate immune response against viral infections; consequently, many viruses have developed strategies to evade their effects.27 For HCV, interference with the IFN system has been described at different levels: the HCV protease NS3/NS4a may inhibit the activation of IFN regulatory factor 3, which is a key factor in the initiation of type I IFN gene transcription in virus-infected cells.28 In addition, downstream events of IFN signaling and antiviral effector mechanisms may also be inhibited by, for example, inhibition of Jak-STAT–mediated signaling29 or inhibition of double-stranded RNA-activated protein kinase by HCV-NS5A,30 respectively. The extent to which each of these mechanisms—which have all been described in vitro—contribute to IFN resistance in vivo is currently unknown; that notwithstanding, IFN resistance is not complete, given the clinical efficacy of IFN-α treatment for cHC where viral kinetics suggest that the initial decline in viral load is in fact due to the direct antiviral effects of IFN-α (as opposed to indirect immunomodulatory mechanisms).31 The most intriguing clinical situation, however, is aHCV infection: first, because it is virtually the only phase of HCV infection in which spontaneous viral clearance can be observed; and second, because more than 90% of patients can be successfully treated by a relatively short course of IFN-α monotherapy.1, 9 This finding implies an inadequacy of endogenous IFN-α secretion in aHC, which could either be due to viral interference with IFN-secreting cells or to relative viral IFN resistance in the face of an optimally stimulated endogenous IFN secretion.
In the present study, we found in patients with aHC a significantly reduced frequency and a more than 100-fold reduced capacity of IFN-α production in peripheral blood pDCs, which are otherwise the most potent IFN-α–secreting cells in humans. A strong inverse correlation was present between the decrease in peripheral blood pDCs and the serum ALT level in the first 4 weeks of aHC, suggesting that the decline could be due at least in part to the compartmentalization of pDCs into the site of inflammation (i.e., the liver). In fact, pDCs have been described histologically in the liver of patients with cHC,32 and we were able to isolate BDCA-2–positive cells from liver specimens of patients with cHC at frequencies similar to peripheral blood (A. Ulsenheimer et al., unpublished data), confirming that pDCs can home to the liver. In addition, the immature phenotype as shown by the decreased expression of CCR7 and HLA-DR on peripheral blood pDCs during aHC would also be compatible with an increased turnover of pDCs and release of less mature pDCs with a lower capacity to produce IFN-α. Obviously, the critical question is whether pDCs function appropriately following localization to the liver. Because it is impossible to obtain liver biopsies from patients with aHC, this cannot be studied directly in humans. Strong evidence for efficient type I IFN secretion in the liver, however, comes from studies in acutely infected chimpanzees, in which significant secretion of type I IFNs could be demonstrated in the liver very early in the course of disease.33–35 Furthermore, in gene expression analyses of liver biopsies from acutely infected chimpanzees, many IFN-induced genes were found among the most upregulated genes. So it is conceivable that at least part of the decline of pDCs in the peripheral blood in patients with aHC is due to efficient compartmentalization of pDCs to the site of inflammation (i.e., the liver). This, however, does not explain the prolonged suppression of pDC numbers and function in patients with self-limited aHC, lasting for several months beyond the loss of HCV RNA from the serum and normalization of ALT levels. Infection of pDCs by HCV—as has been shown for a subset of patients with cHC20—and prolonged persistence of HCV RNA in the liver cannot be excluded, but these conditions seem unlikely in this group. Therefore, other mechanisms such as changes in pDC homeostasis or exhaustion following overstimulation of pDCs may be contributing.
A failure of pDCs has been suggested in other viral infections, such as human immunodeficiency virus and Dengue fever,36, 37 in which disease progression and severity have been associated with lower numbers and decreased IFN-α secretion of peripheral blood pDCs. Again, the question of what is cause and what is effect cannot be answered easily for any of these diseases. Several mechanisms—such as increased recruitment to the site of inflammation, decreased function due to viral infection, and exhaustion—could coexist. The reduced number and IFN-α production by pDCs from our patients with nonviral liver disease or bacterial infections indeed suggests that different types of inflammation can cause similar changes in peripheral blood pDCs. Still, we cannot exclude that the persistently impaired IFN-α secretion of pDCs in cHC, when frequently little inflammation is present in the liver, could be due to a more generalized dysfunction of pDC function.
Another point that deserves consideration is the mode of stimulation used to measure the IFN-α production capacity of pDCs. The available studies of pDC function in chronic hepatitis have either used attenuated viruses,24 interleukin 3 and poly-IC,23 or, in this study, CpG oligonucleotides, which activate pDCs via different cellular receptors. Future studies have to address whether the activation pathways of pDCs are affected differentially in different disease states.
In conclusion, the frequency and IFN-producing capacity of peripheral blood pDCs are dramatically reduced in aHC. Part of the changes can probably be explained by an increased turnover of pDCs and compartmentalization to the liver. Given the extent of changes and the persistence of reduced IFN-α production into the chronic phase of disease, however, we cannot exclude that HCV may interfere more generally with pDCs. Whether the changes in pDC number and function are due to HCV interference or to other factors affecting pDC homeostasis, as suggested by the findings in patients with self-limited aHC, these findings have important implications for therapeutic strategies aimed at stimulating endogenous IFN-α secretion—for example, via CpG oligonucleotides, which have just been introduced into clinical trials for the treatment of cHC.
The authors thank Gunther Hartmann and Stefan Endres for helpful discussions and Jutta Döhrmann und Carmen Amsel for excellent technical assistance.