Engulfment of apoptotic cells expressing HCV proteins leads to differential chemokine expression and STAT signaling in human dendritic cells


  • Anne M. Wertheimer,

    1. Department of Medicine, Portland VAMC and Oregon Health & Science University, Portland, OR
    2. Department of Molecular Microbiology and Immunology, Oregon Health & Science University, Portland, OR
    3. Vaccine and Gene Therapy Institute, Beaverton, OR
    Search for more papers by this author
  • Stephen J. Polyak,

    1. Department of Laboratory Medicine, University of Washington, Seattle, WA
    2. Department of Microbiology, University of Washington, Seattle, WA
    3. Department of Pathobiology, University of Washington, Seattle, WA
    Search for more papers by this author
  • Rachel Leistikow,

    1. Graduate Program in Microbiology, University of Colorado, Boulder, CO
    Search for more papers by this author
  • Hugo R. Rosen

    Corresponding author
    1. Division of Gastroenterology & Hepatology, University of Colorado, Boulder, CO
    2. Integrated Program in Immunology, University of Colorado, Denver, CO
    3. Hepatitis C Center, University of Colorado, Denver, CO
    • University of Colorado UCHSC GI Division, 4200 East Ninth Avenue #B-158, Denver, CO 80262. Express Mail: 4200 East Ninth Avenue, Research Bridge–Room 6412, Denver, CO 80202
    Search for more papers by this author
    • fax: 303-315-5711

  • Potential conflict of interest: Nothing to report.


In the majority of cases, infection with hepatitis C virus (HCV) becomes chronic and is often associated with impaired innate and adaptive immune responses. The mechanisms underlying viral persistence and lack of protective immunity are poorly understood. Considering that dendritic cells (DCs) play critical roles in initiating and modulating immune responses, we explored the effect of HCV proteins on DC gene and protein expression, phenotype, and function. Human DCs were generated following plastic adherence of monocytes and culture with granulocyte-macrophage colony-stimulating factor and interleukin-4 (IL-4) from normal subjects. Autologous nonadherent peripheral blood mononuclear cells were infected with vaccinia constructs expressing various HCV proteins (core-E1, NS5A, NS5B) or an irrelevant protein β-galactosidase (β-gal) as the control, induced to undergo apoptosis, then co-cultured with DCs. Between 2% and 10% of the genes probed in a cDNA nylon array were differentially regulated within DCs that had engulfed HCV proteins. In particular, the presence of intracellular NS5A led to increased transcriptional and protein expression of IL-8 (CXCL-8), a chemokine with proinflammatory and anti-interferon properties, and impaired interferon induction of signal transducers and activators of transcription 1 (STAT1) serine and tyrosine and STAT2 tyrosine phosphorylation. Conclusion: These data provide novel mechanisms by which HCV subverts antiviral host immunity. (HEPATOLOGY 2007;45:1422–1432.)

Hepatitis C virus (HCV) is a positive-stranded RNA virus within the genus Hepacivirus of the Flaviviridae family,1 and is the leading causative agent of chronic hepatitis.2 Nearly 4 million Americans (2% of the population) have antibodies to HCV.3, 4 The National Health and Nutrition Examination Survey indicates that the vast majority (85%) of individuals acutely infected with HCV develop persistent infection and only 15% spontaneously recover.5 Thus, recovery is a relatively rare event. The high proportion of chronic HCV infections may be attributed to active escape mechanisms counteracting the protective immune response or to HCV not inducing an efficient immune response.6, 7

The dendritic cell (DC) is the central player in both innate and adaptive immunity8; circulating DC precursors play roles in the immediate reaction to pathogens and in the shaping of immune response, including interactions with T cells, natural killer (NK) T cells (NKT cells), and NK cells. Following tissue damage, immature DCs capture antigen or dying cells that contain antigen and migrate to the lymphoid organs, where they select rare antigen-specific T cells, thereby initiating immune responses.8 DCs present antigen to CD4+ T-helper cells, which in turn regulate the immune effectors, including antigen-specific CD8+ cytotoxic T cells and B cells, as well as non–antigen-specific macrophages, eosinophils, and NK cells.8

Through co–evolution with their human host, many viruses have developed mechanisms that subvert their hosts' immune system, i.e., make infected cells less “visible” to immune recognition.9 Some viruses directly alter the function of antigen presenting cells, specifically impairing antigen processing and presentation at several steps.10–15

Early studies from Kanto et al.16 showed that DCs cultured from patients with HCV had impaired allostimulatory capacity, i.e., were less able to recognize foreign T cells, and showed reduced production of cytokines important for establishment of T cell responses and NK cell–mediated cytotoxicity (specifically, interleukin [IL]-12). These findings have been extended and confirmed by the demonstration that transduction of HCV core and E1 proteins into DCs results in reduced allostimulatory capacity.17, 18 Moreover, recent studies19–24 focusing on direct ex vivo analysis of circulating DCs demonstrate that hepatitis C and chronic liver disease23per se leads to depletion of circulating myeloid DCs and plasmacytoid DCs. In contrast, subjects with resolved HCV infection have myeloid DC and plasmacytoid DC frequencies comparable to normal subjects.23 These findings imply that antigen presentation function may be altered. Further, the finding of negative-strand HCV RNA within DCs21 suggests that impairment of DC activity may in part be related to direct HCV infection.

In this study, we characterized the effect of engulfment by DCs of apoptotic cells containing HCV proteins, since this is likely relevant in vivo. We propose that our in vitro model system recapitulates critical events in the initial host immune response to this virus. We defined the effect of HCV proteins on phenotype, cellular gene and protein expression, function, and response to interferon (IFN) stimulation of human monocyte-derived DCs. Of particular interest, we find that intracellular (but not exogenous) uptake of HCV NS5A protein by DCs leads to increased transcriptional expression and secretion of CXCL-8 and impaired IFN induction of signal transducers and activators of transcription 1 (Stat1) serine and tyrosine and Stat2 tyrosine phosphorylation.


β-gal; β-galactosidase; CMV, cytomegalovirus; DC, dendritic cell; DiD, 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine, 4-chlorobenzenesulfonate salt; HLA, human leukocyte antigen; IFN, interferon; IL, interleukin; MHC, major histocompatibility complex; NK, natural killer; OHSU, Oregon Health & Science University; RT-PCR, real-time PCR; STAT, signal transducers and activators of transcription.

Patients and Methods

Human Subjects.

A total of 8 normal donors were used for these experiments. The subjects were recruited from the Red Cross (without identifiable markers), Oregon Health Sciences University (OHSU), and the Portland Veterans Administration medical center (PVAMC). We obtained informed consent using Institutional Review Board protocols at both OHSU and PVAMC. The health status of the donors was confirmed by screening for universal viral markers.

Generation of Monocyte-Derived DCs and Characterization of Nonadherent Cells.

The initial phenotypic characterization is illustrated in Supplemental Fig. 2 where 2 patients were cultured in duplicate under identical conditions. Nonadherent cells were removed after the plastic adherence step at day 0 of culture and stained for CD3, CD4, CD8, CD56, Lineage (Lin) cocktail (CD3, CD4, CD14, CD19, CD56), HLA-DR, CD11c. The population was 90% lymphocyte based on size and granularity. Of these cells 96% were Lin+. DC purity was confirmed by flow cytometry on 5 day cultured DCs based on size and granularity. Of these cells 100% were HLA-DR+, CD11c+, CD3−, and 97% were Lin(−). Of interest, up to 26% of these monocyte derived cultured dendritic cells were Lin, HLA-DR+, CD123+23 (Supplemental Fig. 2).


Human monocyte–derived DCs were cultured for 5 days25 (see Supplementary Data for details on phenotypic characterization Supplemental Fig. 2). On the 4th day of culture, autologous nonadherent peripheral blood mononuclear cells were infected with vaccinia virus for 12 to 18 hours. Four recombinant vaccinia constructs were used throughout most of these studies: HCV core, HCV NS5A, and HCV NS5B (gifts from Chiron, see Fig. 1); the control for all experiments was a vaccinia construct expressing β-galactosidase (β-gal). Specifically, using a 24-well plate, 1 × 106 autologous nonadherent cells were plated in 1 ml of RPMI-1640 media supplemented with 10% human serum, then infected with 1 of the 4 recombinant vaccinia constructs per well for 16 to 18 hours. The infected autologous cells, each containing 1 of the 4 recombinant vaccinia constructs (including the β-gal control), were simultaneously irradiated with ultraviolet light to inactivate viral replication and induce apoptosis. All experiments were repeated at least in duplicate per human subject tested, according to these parameters, and at least 2 subjects were evaluated. Cells were then washed with sterile phosphate buffered saline (PBS) and added at a 1:1 ratio to DCs at day 5. Co-culture continued at 37°C, 6% CO2 for 8 hours for RNA harvest or a specified time course for fluorescence-activated cell sorting analysis of phenotype and functional assays. For some studies, a cytokine cocktail for in vitro maturation stimulus was added after 8 hours of co-culture and cultured an additional 48 hours. The maturation cocktail as originally described by Frank et al.26 was modified as follows: 10 ng/ml [IL-6, IL-1B, tumor necrosis factor α (R&D Systems)], 10−6 M PGE2 (Sigma), and 1 μg/ml CD40L (R&D Systems). Gamma radiation–inactivated cytomegalovirus (CMV) antigen (containing all parts of the CMV viral life cycle) that had been cultured in MRC-5 monolayers was purchased from Microbix Biosystems (Encinitas, CA; 1:32 titer).

Figure 1.

Vaccinia constructs expressing HCV antigens. For our studies, we used construct rVV1 expressing HCV core-E1, rVV6 expressing NS5A, and rVV7 expressing NS5B. These vaccinia constructs were kindly provided by Chiron (Emeryville, CA); the names were simplified in this manuscript. The original names Chiron uses for these constructs are as follows: Sc59 6C/Ss (HCV core-E1), Sc11 NS5A, and Sc11 NS5B. Image modified from review.51

cDNA Arrays for Screening.

Total RNA was obtained following co-culture, using the Qiagen RNAeasy total RNA kit. P32 labeling of total RNA was used to generate the probes. Commercially available ATLAS membrane arrays (Atlas Human Hematology/Immunology Array, Cat#7337-1; Clontech) were probed (www.BD.com). Each array contains approximately 400 hematology/immunology–based cDNA targets. Hybridization was done at 64°C overnight, membranes and films were further processed per the manufacturers' recommendations and analyzed using Clontech array software.

Taqman Real-time PCR.

The total RNA obtained from the Qiagen RNAeasy total RNA kit was used with Perkin Elmer's Taqman IL-8 (CXCL-8) and SCYC1 kits per the manufacturer's recommendations. Reactions were run in triplicate using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the internal RNA control, for 45 cycles on an ABI RT-Thermocycler.

Macropinocytosis Assay.

Following co-culture of 8, 30, 48, and 48 hours with maturation, the co-cultures were harvested, washed, and placed into fluorescence-activated cell sorting tubes containing either 15 μg/ml dextran-fluorescein isothiocyanate (40,000 MW) (Molecular Probes) or albumin-fluorescein isothiocyanate (Sigma) for 1 hour at 37°C to measure phagocytosis and macropinocytosis, respectively. Cells were washed twice in blocking reagent then analyzed using a BD FACSCalibur instrument.

Luminex Assay.

After co-culture of 8 to 16 hours, the co-culture was harvested, and the supernatant was collected and frozen at −80°C until analysis. A 22-plex multicytokine kit (Upstate; Cat#48-011), or single cytokine (IL-8) kit (Upstate; Cat# 46-108) was used per manufacturer's instructions with the Luminex 100 IS system (www.luminexcorp.com).

Western Blot.

Monocyte-derived DC were cultured for 5 days, and treated with 1,000 U/ml IFN alpha 2b (Roferon, Roche) for 30 minutes at 37°C. Cells were washed twice in PBS and scraped. Whole cell extracts were prepared in radioimmunoprecipitation assay buffer.27 Protein lysates were quantitated by BCA Protein Assay (Pierce, Rockford, IL) and 30 μg total protein was separated on 4% to 20% SDS-PAGE gels. For detection of phosphorylated Stat proteins, Stat1 phosphotyrosine-specific (Y701), Stat1 phosphoserine-specific (S727), and Stat2 phosphotyrosine-specific (Y689) antibodies were used (Zymed-Invitrogen). Total Stat1 was detected using a polyclonal antibody (Zymed). Blots were also probed with actin and extracellular signal regulated kinase–1 specific or extracellular signal regulated kinase 2–specific antiserum as additional controls for total protein loading. Western blotting was performed as described.28


Expression of HCV Proteins from Vaccinia Vectors.

As described in the Patients and Methods section, we used vaccinia constructs (Fig. 1) to express HCV proteins in target cells, and throughout these studies we used a vaccinia construct expressing the irrelevant protein β-gal as control. It is critical to compare our finding with this type of viral vector control, as it takes into account both the viral background (vaccinia) and the possible attenuation effect of the molecular construct (expression of either the protein of interest or an irrelevant protein). We confirmed that within 18 hours after infection, proteins of the appropriate size and specificity were expressed by cells infected with the various vaccinia constructs (Supplementary Fig. 1). Viral titer was measured and multiplicity of infection (MOI) was optimized for each virus (MOI 5), so that similar levels of HCV protein were expressed from the infected cells.

Apoptotic Cells Are Successfully Engulfed by DCs.

To demonstrate that autologous nonadherent apoptotic cells are engulfed by monocyte-derived DCs, we conducted feeding experiments using fluorescently-stained 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine, 4-chlorobenzenesulfonate salt (DiD) nonadherent cells. DCs were cultured for 5 days, then an equal number of fluorescently-labeled, autologous nonadherent apoptotic cells were added to the cultures and co-cultured for 1, 3, 5, or 18 hours, or harvested immediately (null). Cultures were harvested and analyzed using flow cytometry, as shown in Fig. 2. Gating the DCs based on their size and granularity (R1) revealed that by 5 hours, 43% of the DCs had encountered the fluorescently-labeled cells and that additional time in culture did not significantly increase that percentage. Thus, we chose to conduct co-culture experiments for a minimum of 5 hours.

Figure 2.

Time course of feeding identifies optimal time for engulfment. Day 5 autologous myeloid DCs were co-cultured with fluorescent (DiD) apoptotic cells at 37°C over a specified time course. The panel at top left illustrates the forward scatter versus side scatter profile of the co-cultured population after 1 hour incubation. Gate R1 is drawn around the DCs (large granular cells). The remaining 5 panels illustrate the amount of fluorescence (FL2) found within the DCs after various incubation periods (labeled above each panel). Null indicates cells (DC and fluorescent apoptotic cells) that were simply centrifuged together and not co-cultured.

No Significant Alterations in DC Phenotype After Engulfment of Cells Expressing Virally-Produced Core-E1, NS5A, or NS5B.

There is a precedent for viral proteins down-regulating cell-surface molecules essential in antigen presentation. As an example, it has recently been demonstrated that the Nef protein of HIV-1 efficiently removes major histocompatibility complex (MHC) class I and II and proteins of the B7 family (CD80 and CD86) from the cell surface of antigen presenting cells, likely contributing to immune evasion.29 To analyze the effect of HCV proteins on cell-surface expression, we stained monocyte-derived DCs after engulfment of apoptotic cells (see Supplementary Data, Methods section, for details). There were no significant alterations in the expression of various cell-surface molecules involved in antigen presentation (CD80, CD86, CD40, human leukocyte antigen (HLA)-ABC, and HLA-DR) in DC exposed to HCV-containing vaccinia constructs as compared to control (β-gal) (Table 1). Slight decreases in both HLA-ABC and HLA-DR expression as measured by median fluorescent intensity (MFI) were detected after 8 and 30 hours of co-culture with NS5A compared with the β-gal control, but these values were not statistically significant. As expected, expression of costimulatory molecules (CD40, CD80, and CD86) increased after maturation stimulus; the presence of HCV proteins did not affect their expression. Of interest, HLA-ABC and HLA-DR were elevated (as compared to basal “unfed” levels, data not shown) at 8 hours postculture in all conditions and remained at similar levels after maturation stimulus was administered, suggesting that the engulfment of the apoptotic cells alone increases surface expression of these molecules. At least 3 subjects were examined for these phenotypic assays.

Table 1. Microarray Results of DC Engulfment Experiments: Phenotypic Analysis of DCs Co-cultured in the Presence of Control Construct (β-gal) or HCV Proteins (Core, NS5A, and NS5B) for 8, 30, and 48 hours with and without Maturation Stimulus (N = 2)
8 hours30 hours48 hours48 hours matured8 hours30 hours48 hours48 hours matured
  1. Table illustrates median fluorescent intensity (MFI) of various cell surface stains.

 CD11c (DC Specific Marker)CD14 (Negative Control-Monocyte Marker)

Microarray Screening Analysis of Co-cultured DCs Reveals Differential Regulation.

To screen the effect of engulfment of apoptotic cells expressing HCV proteins (NS5A and core-E1) upon monocyte-derived DCs, we harvested total RNA following co-culture. The focus of these assays was on the structural protein core and nonstructural 5a, although we subsequently performed arrays with NS5B (data not shown). The total RNA was labeled and hybridized to nylon arrays as detailed in Patients and Methods. Each array was performed in duplicate. In addition, we repeated the experiment with a separate normal donor. Combining the data from both subjects, we found up-regulation of 79 genes in response to HCV NS5A and 85 genes up-regulated in response to HCV core-E1. Over 100 genes were down-regulated in response to HCV NS5A and 69 genes were down-regulated in response to HCV core-E1 as compared to β-gal. However, in response to NS5A fewer than 10 genes of the 79 genes were up-regulated and only 2 of the 100 genes were down-regulated in both subjects. Similarly, with regard to core-E1, only 2 of the 85 genes were up-regulated in both patients, whereas none of the 103 genes down-regulated occurred in both individuals (Table 2). Interestingly, within each subject there was considerable overlap of genes that were differentially regulated, specifically in response to both vaccinia-delivered HCV-NS5A and core-E1 as compared to the vaccinia-delivered β-gal control, which was also delivered to the DCs via engulfment of apoptotic cells, suggesting a stereotypic response to HCV proteins for a given individual. Based on the kinetics of hybridization to these 200-bp to 600-bp cDNA arrays, the typical range is 2-fold to 10-fold based on the manufacturer's experience. A 2-fold to 5-fold difference will result in a significant difference upon further testing, i.e., real-time PCR (RT-PCR), with 75% probability; a 5-fold to 10-fold difference increases that probability to 95% per the manufacturer's specifications.30

Table 2. DC RNA Was Extracted After Co-culture (for 8 Hours) with Autologous Apoptotic Cells Infected with Recombinant Vaccinia Virus Containing HCV Core-E1 or NS5A; Apoptotic Cells Infected with Vaccinia Virus Containing β-gal Served as the Control
GenBankAverage (Range) Fold IncreaseGene
Expression increased in presence of HCV core-E1 in common between 2 normal subjects
 L137732 (2,2)AF-4 Protein; fel protein
 D439672.5 (2,3)Acute myeloid leukemia 1 protein; oncogene aml-1
Expression increased in presence of HCV NS5A-selected genes
 L042852.5 (2,3)ENL protein
 X642283 (2,4)Nuclear pore complex protein 214 (NUP214); nucleoporin NUP214; 214-kDa nucleoporin CAN protein
 Y007872.4 (3,2)IL-8 (CXCL-8)
 M739802 (2,2)Neurogenic locus notch protein homolog 1 precursor (NOTCH1); translocation-associated notch protein (TAN1)
 U237722 (2,2)Lymphotactin precursor (LTN); cytokine scm-1 alpha; lymphotaxin; small inducible cytokine C1 (SCYC1)
U201582 (2,2)SLP-76; 76-kDa tyrosine phosphoprotein
Expression decreased in presence of HCV NS5A
 X140462.5 (2,3)Leukocyte CD37 antigen
 M875032.5 (2,3)Transcriptional regulator interferon-stimulated gene factor 3 gamma subunit (ISGF3G); IFN-α–responsive transcription factor subunit (IRF-9)

In addition to this differential regulation, we detected expression of genes encoding typical DC proteins (CD 86, CD83, and HLA-DR). Although we did not enrich for DCs after the co-culture, we did not detect typical T cell genes coding for CD2 or CD3 (part of the T cell receptor complex), indicating a relatively low level of T cell transcripts present in the cultures.

RT-PCR Confirms Microarray Screen.

In order to confirm the array data, we performed Taqman RT-PCR on the same total RNA samples used in the array, in addition to new cultures, all harvested after 8 hours co-culture. Because of the previously reported observation that NS5A induced expression of CXCL-8 in nonprimary tissue culture,31–33 and elevated levels of this chemokine in circulating HCV-infected patient serum,34 we focused on the increased expression of CXCL-8 and SCYC1 expression in response to NS5A in DCs (Fig. 3A and B, respectively) from the same 2 subjects and averaged their responses. Total RNA was probed per manufacturer's recommendations using an internal housekeeping control and run on an ABI automated thermocycler. As illustrated in Fig. 3, in response to NS5A (Fig. 3A), expression of CXCL-8 was increased nearly 4-fold, while SCYC1 (Fig. 3B) expression was increased more than 5-fold (1 subject had a >10-fold increase) over β-gal control.

Figure 3.

RT-PCR confirms microarray screen (n = 2) revealing increased CXCL-8 and SCYC1 gene expression after DC Engulfment of cells expressing NS5A. Total RNA isolated after 8 hours co-culture was used in TaqMan RT-PCR. Appropriate housekeeping controls were run and each assay point was run in triplicate. Fold expression was normalized to the β-gal control culture. Error bars represent 3 SD. (A) CXCL-8 expression from DC cultures fed core-E1, NS5A, and NS5B expressing cells (P = 0.0002). (B) SCYC1 expression from DC cultures fed core-E1, NS5A, and NS5B expressing cells (P = 0.04). (C) CXCL-8 expression from DC cultures directly infected (18 hours) with vaccinia-expressing control, core-E1, NS5A, and NS5B (P = 0.004 for the entire group, P = 0.002 for NS5A versus control, and P = 0.002 for NS5B versus control).

Direct Infection of DCs and Other Antigen Presenting Cells.

We went on to investigate whether the increase in CXCL-8 expression would also follow direct infection of human DCs (Fig. 3C) or lymphoid cell lines (data not shown) with the vaccinia vector expressing HCV NS5A. Using the same 2 subjects as we had for the feeding experiments, we directly infected the DCs at day 5 for 18 hours, then total RNA was harvested and processed for RT-PCR analysis. RT-PCR revealed increased expression, specifically upon infection with vaccinia containing HCV NS5A. Interestingly, the level of increase in CXCL-8 expression after direct infection of DCs was about 50% less than that observed after DC engulfment of apoptotic cells containing HCV NS5A.

CXCL-8 Protein Secretion After DC Engulfment of Cells Expressing NS5A.

We evaluated whether the alteration in gene expression levels resulted in increased cytokine production. Using the Luminex assay, we simultaneously measured IL-1α, IL-β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, CXCL-8, IL-10, IL-12(p40), IL-12(p70), IL-13, IL-15, IP-10, Eotaxin, IFN-γ, granulocyte-macrophage colony-stimulating factor, MCP (macrophage chemoattractant protein)-1, macrophage inflammatory protein 1α, RANTES (regulated on activation, normal T-cell expressed and secreted), and tumor necrosis factor α in the co-culture supernatants of the same 2 subjects we had phenotyped and analyzed via the microarray and RT-PCR. Only CXCL-8 expression was significantly increased (P = 0.0004, analysis of variance) after DCs had engulfed apoptotic cells expressing NS5A, as compared to those expressing core-E1, NS5B, or the β-gal control (Fig. 4A). Following this initial multicytokine survey, a total of 8 subjects were used for measurement of CXCL-8; DCs from all subjects demonstrated elevated CXCL-8 following uptake of apoptotic cells containing NS5A.

Figure 4.

Monocyte-derived DCs express significantly increased levels of CXCL-8 (pg/ml) after ingestion of autologous cells expressing HCV NS5A. Supernatant was collected and analyzed for CXCL-8 using a Luminex multiplex instrument. (A) Human monocyte-derived DC were cultured for 5 days then co-cultured for 8 to 12 hours with autologous cells expressing β-gal (an irrelevant protein control) or HCV proteins [Core-E1 (2 normal donors), NS5A (8 normal donors), or NS5B (2 normals donors) (P = 0.0004, analysis of variance, ANOVA)]. Increased production of CXCL-8 production was not detectable when DCs had been fed exogenous recombinant NS5A either alone (B) or in the presence of noninfected apoptotic cells (C), n = 2. These data indicate that endogenously processed NS5A is required to induce CXCL-8 secretion from human DCs.

DCs Fed Exogenous Recombinant Protein Do Not Produce CXCL-8 Expression.

To determine whether intracellular expression of NS5A protein was necessary or whether addition of exogenous NS5A could stimulate CXCL-8 production from DCs, we introduced exogenous recombinant NS5A protein or the buffer control (SDS) to day 5 DCs and incubated them for 8 and 24 hours (Fig. 4B). There was no CXCL-8 secretion upon feeding exogenous NS5A after 8 or 24 hours. Thus, recombinant NS5A did not stimulate CXCL-8 secretion above basal levels. Of interest, we found that ingestion of apoptotic nonadherent cells alone (DC + APO in Fig. 4C) did stimulate CXCL-8 production, and that exogenous recombinant NS5A protein had no additive effect (Fig. 4C). Taken together, these data indicate that NS5A must be within an apoptotic cell to optimally stimulate CXCL-8 from DCs.

Antigen Presentation Function Retained for Recall Antigen Response.

We used several approaches to determine whether DC function was affected by engulfment of the apoptotic cells expressing various HCV proteins. Having selected 2 normal donors who possess a measurable CMV T cell response, we used an enzyme-linked immunosorbent spot assay to determine if the production of IFN gamma would be decreased if the DCs had previously encountered HCV antigen within an autologous apoptotic cell. There was no statistically significant difference in CMV response in cultures with DCs containing HCV proteins as compared to those containing control β-gal (Fig. 5). These data are in accord with the relatively modest effect on HLA expression noted in Table 1. Moreover, the ability of DCs to induce mixed lymphocyte reaction and to phagocytose remain unaltered after culture with autologous cells expressing HCV antigens (Supplementary Materials).

Figure 5.

Antigen presentation function retained for CMV recall response. Day 5 monocyte-derived DCs (5000) expressing various HCV antigens were cultured with 50,000 CD4+ purified autologous T cells and stimulated with specified doses of CMV antigen (Microbix Biosystems, Encinitas, CA) in an IFN gamma enzyme-linked immunosorbent spot (ELISPOT) plate for 24 hours. All antigens were assayed in triplicate at each of the doses. This assay represents data from 1 CMV+ donor and was repeated. Total number of spots were read (y-axis) and plotted versus CMV antigen concentration (x-axis).

Basal and IFN-α-induced STAT Phosphorylation Impaired by NS5A.

The down-regulation of IFGF-3 gamma (IRF-9) observed in the microarray analyses (Table 2; last row) suggested that NS5A might impair signaling in response to IFN α. To address this question, we cultured monocyte DCs, then introduced either apoptotic cells expressing NS5A or β-gal at day 5. After 8 hours of co-culture, the cultures were stimulated for 30 minutes with IFN α. Western blot analysis revealed that the presence of NS5A within DCs decreased the basal level of phosphorylation of Stat1 serine and tyrosine phosphorylation and Stat2 tyrosine phosphorylation as compared to the β-gal control. IFN-induced phosphorylation of STAT1 and STAT2 was also impaired in DCs exposed to NS5A-containing apoptotic cells, albeit to a lesser extent than in untreated cells (Fig. 6 A, B). The total levels of STAT1 and STAT2 proteins were not differentially regulated in response to NS5A expression (data not shown). Subsequent experiments using HeLa cells with regulated expression of NS5A indicated that in this context, NS5A did not impair IFN-α induction of STAT phosphorylation (Fig. 7), consistent with previous results.31 The data suggest that NS5A impairment of IFN signaling in DCs is dependent on phagocytic engulfment of apoptotic, NS5A-expressing cells.

Figure 6.

Monocyte-derived DCs co-cultured with apoptotic cells expressing NS5A have reduced STAT1 signaling. Human monocyte-derived DC were cultured for 5 days then co-cultured for 8 hours with apoptotic cells expressing β-gal or NS5A, followed by treatment with or without 1000 units/ml IFN-α for 30 minutes. (A) Whole-cell extracts were separated on SDS-PAGE gels and probed for phosphorylated forms of Stat1 (S727 and Y701) and Stat2 (Y689). The positions of the phosphorylated Stat proteins are depicted with an arrow. The bottom panel shows a Coomassie-stained image of the membrane, and demonstrates equal loading of protein in all lanes. Blots were also probed for actin and demonstrated equal protein loading (data not shown). (B) Quantitation of pixel intensity from (A). Blots were scanned and pixel intensity over the entire area of the band was calculated using Image J software.

Figure 7.

Expression of NS5A in HeLa cells does not inhibit IFN signaling. HeLa 1b2 cells were cultured in the absence (−Tc) or presence (+Tc) of 2 μg/ml tetracycline for 48 hours. Cells were treated with or without IFN-α (100 or 500 U/ml) for 30 minutes before whole-cell protein extracts were separated by SDS-PAGE and blotted for phosphorylated forms of Stat2 tyrosine, Stat1 tyrosine and serine, total Stat1, NS5A, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control for equal loading of protein. These data suggest that NS5A expression in this cellular context does not significantly inhibit IFN signaling.


DCs are ubiquitous professional antigen presenting cells that provide a critical link between innate and adaptive immunity.8 DCs may directly encounter pathogens or pathogen byproducts via pattern-recognition receptors (toll-like receptors or TLRs), they may detect the presence of inflammatory cytokines, or alternatively, they may engulf apoptotic infected cells.35–38 Once in possession of antigen, the DCs migrate into secondary lymphoid organs where they engage naive T cells, stimulating them to begin antigen-specific proliferation.39

Consequences of HCV:DC interactions remain somewhat controversial, with some groups showing impairment of DC maturation and function, in particular diminished allostimulatory capacity of DC derived from HCV-infected patients16, 17 or following transduction with an adenovirus coding for HCV core and E1 proteins,18 whereas other groups have failed to identify phenotypic or functional defects.23, 40, 41 In addition, Dolganiuc et al.42 found that recombinant HCV protein Core-E1 and NS 3 activate monocytes and inhibit DC differentiation in the absence of the intact virus. We and others have found that HCV infection leads to a peripheral depletion in circulating DC numbers,20–23 although this may be related to liver disease per se. Combined, these studies provide evidence that HCV affects DCs, although the effects may be pleiotropic and subtle.

This report examines for the first time DCs immediately after their encounter with HCV proteins in the context of an autologous cell because it is presumed this is a physiologically relevant process. Although autologous hepatocytes are the ideal cell to study, we had to select autologous cells that we could obtain in sufficient numbers from peripheral blood of normal subjects for these studies. Using our novel feeding system, we found no measurable alterations in phenotypic markers for DCs or surface expression of various molecules involved in antigen presentation, nor any functional impact on antigen presentation itself (see also Supplementary Data Fig. 2). Interestingly, we did find a significant alteration in a variety of DC transcripts and induction in CXCL-8 production from DCs, specifically upon encounter of NS5A within autologous apoptotic cells. Particularly interesting, DCs encountering exogenous NS5A did not produce increased levels of CXCL-8, suggesting that the context in which the DC encounters the HCV proteins is significant. Additionally, we found that DCs encountering autologous apoptotic cells per se produced increased levels of CXCL-8 and had elevation in surface expression of both MHC-I and MHC-II molecules, suggesting that encounter with autologous apoptotic cells causes specific alterations within the DC itself.

Hence, even with this limitation of ideal cell type (the autologous nonadherent peripheral blood cells), we have shown that DCs do differentially regulate a significant portion of their transcripts in response to various HCV proteins within autologous cells. In addition, in our microarray screen, we found up-regulation of several transcripts previously reported to be up-regulated by directly infecting DCs with various pathogens (E. coli, C. albicans, influenza A).43 Additionally, the up-regulation of some DC maturation markers (e.g., CD83) and various cytoskeletal genes suggest DC maturation occurs in the presence of virally-derived HCV proteins even without any external maturation signals. In addition to the altered expression of chemotactic cytokines (CXCL-8), we detected differential expression of transcription activators of cytokines (IL-4 STAT), chemokine receptors (CCR1-R), and proteins integral to the DC-T cell synapse (ICAM1, CD43), as well as decreased expression of ISGF-3 gamma, a critical player in the IFN-alpha response.

Pathogenic associations of CXCL-8 with HCV infection have been reported using patient serum26, 34 and tissue,44 and in vitro using various cell lines.28, 31, 45 Moreover, we have found the basal CXCL-8 production from monocyte-derived DCs decreases in acute HCV infection that is spontaneously resolving (our unpublished observations). The current report is the first to show induction of CXCL-8 in response to HCV proteins in normal human primary cells, and importantly from professional antigen presenting cells, DCs. CXCL-8, a chemotactic cytokine expressed from numerous cell types,46 is known for its inflammatory ability to recruit lymphocytes, specifically neutrophils. However CXCL-8 serves additional proviral roles during infection by impairing the interferon alpha response during certain RNA viral infections,47 and recently CXCL-8 was shown to induce cytoskeleton reorganization conducive to permeability during dengue virus infection.48 Taken together, these proviral activities suggest that induction of CXCL-8 from DCs after engulfment of HCV-infected apoptotic cells, although not resulting in any appreciable alteration in antigen presenting function in our system, may create a general inflammatory microenvironment. Deregulation of the inflammatory response in the earliest stages of naive encounter may contribute to immune impairment, favoring virus persistence. Our finding that DCs co-cultured with HCV NS5A-containing apoptotic cells were stunted in their response to interferon alpha stimulation, specifically the impaired Stat1 serine and tyrosine and Stat2 tyrosine phosphorylation, supports our hypothesis that engulfment of apoptotic cells containing HCV proteins impairs DC viral responsiveness. These novel results extend the growing literature implicating different HCV proteins in suppressing type I IFN signal transduction pathways and contributing to viral persistence,49, 50 but are the first to specifically show this effect in DCs. Additional studies are required to determine if the NS5A-induced impairment of IFN-α signaling is mediated by the induced CXCL-8. Moreover, it is known that CXCL-8 does inhibit the antiviral actions of IFN47 and NS5A expression inhibits IFN signaling in some,50 but not all studies.28 We recognize that neutralization of CXCL-8 would enhance our results, but commercially available antibodies have not provided consistently reproducible results (data not shown).

In summary, we have developed a model system to study human DCs that have encountered apoptotic autologous cells expressing various HCV proteins. We found that following engulfment of apoptotic cells containing HCV NS5A, differential gene expression occurs within human DCs, including increased CXCL-8 secretion by DCs and impaired basal and IFN-induced Stat1 serine and tyrosine and Stat2 tyrosine phosphorylation within the DCs. This novel mechanism for subverting the antiviral response at a critical early stage of infection likely contributes to viral persistence.


We thank Sue Smyk-Pearson, Jared Klarquist, and Jessica Wagoner for technical assistance.