Hepatitis B Virus Surface Antigen Activates Myeloid Dendritic Cells via a Soluble CD14-Dependent Mechanism

ABSTRACT

Hepatitis B virus (HBV) infection can cause chronic liver disease, which is associated with increased risk of liver cirrhosis, liver failure, and liver cancer. Clearance of HBV infection requires effective HBV-specific immunity; however, the immunological mechanisms that determine the development of effective HBV-specific immunity are poorly understood. Dendritic cells (DC) play a pivotal role in the regulation of antiviral immunity. Here, we investigated the interaction between HBV surface antigen (HBsAg), the main envelope glycoprotein of HBV, and BDCA1⁺ myeloid dendritic cells (mDC). Exposure of peripheral blood-derived BDCA1⁺ mDC to HBsAg resulted in strong DC maturation, cytokine production, and enhanced capacity to activate antigen-specific cytotoxic T cells (CTLs). By using neutralizing antibodies, crucial roles for CD14 and Toll-like receptor 4 (TLR4) in HBsAg-mediated BDCA1⁺ mDC maturation were identified. Concordantly, HBsAg-mediated DC maturation required fetal calf serum (FCS) or human plasma, naturally containing soluble CD14 (sCD14). Intriguingly, HBsAg-induced DC maturation was significantly reduced in umbilical cord blood plasma, which contained less sCD14 than adult plasma, indicating that sCD14 is an important host factor for recognition of HBsAg by DC and subsequent DC activation. A direct interaction between sCD14 and HBsAg was demonstrated by using enzyme-linked immunosorbent assay (ELISA). Moreover, sCD14-HBsAg complexes were detected both in vitro and in sera of HBV-infected patients. The abundance of sCD14-HBsAg complexes varied between chronic HBV disease stages and correlated with activation of BDCA1⁺ mDC in vivo. We conclude that HBsAg activates BDCA1⁺ DC via an sCD14-dependent mechanism. These findings provide important novel insights into the initiation of HBV-specific immunity and facilitate development of effective immunotherapeutic interventions for HBV.

IMPORTANCE Hepatitis B virus (HBV) infection is a significant health problem, as it causes progressive liver injury and liver cancer in patients with chronic HBV infection, which affects approximately 250 million individuals worldwide. Some of the infected adults and the majority of neonates fail to mount an effective immune response and consequently develop chronic infection. The viral and host factors involved in the initiation of effective HBV-specific immune responses remain poorly understood. Here we identified CD14 and TLR4 as receptors for HBsAg, the main HBV envelope antigen. HBsAg induced strong maturation of dendritic cells (DC), which have a central role in regulation of virus-specific immunity. These results provide essential novel insights into the mechanisms underlying the initiation of HBV-specific immunity. Intriguingly, since neonates have naturally low sCD14, the finding that serum-derived sCD14 is a crucial host factor for recognition of HBsAg by DC may have implications for immunity of neonates to HBV infection.

INTRODUCTION

Hepatitis B virus (HBV) is a double-stranded DNA (dsDNA) virus that is transmitted via blood and specifically infects hepatocytes. It can cause chronic liver disease and progressive liver injury leading to increased risk of liver cirrhosis, liver failure, and liver cancer (1). The current estimated prevalence of HBV infection is 248 million individuals globally (2). Although the initiation of an effective antiviral immune response is paramount for resolving HBV infection (3), the early steps in the recognition of the virus by immune cells and the functional consequences of this interaction remain to be resolved.

A pivotal role for dendritic cells (DC) is anticipated, because these cells play a central role in the orchestration of antiviral immunity due to the expression of a wide variety of different pathogen recognition receptors (PRR) and their unique capacity to initiate virus-specific cytotoxic T cell (CTL) responses (4, 5). Among the different DC subsets, BDCA1⁺ myeloid DC (mDC) are of particular interest for HBV-specific immunity, since hepatitis B surface antigen (HBsAg)-positive mDC were detected in liver (6) and peripheral blood (7) of HBV patients. This suggests that the interaction between BDCA1⁺ mDC and HBsAg, the main envelope glycoprotein present on HBV infectious particles and subviral particles, occurs in vivo. Although initially it was suggested that HBsAg has immune-regulating capacities (reviewed in reference 8), more recently, also immune-stimulatory effects of HBsAg on myeloid cells, including monocytes (9), monocyte-derived DC (10), and Kupffer cells (KC) (11), have been reported. Nevertheless, whether HBsAg is able to directly activate mDC and which receptor(s) is involved are currently unknown.

In the present study, we investigated the interaction of HBsAg with BDCA1⁺ myeloid DC and its functional consequences. We demonstrated that HBsAg can induce maturation/activation and cytokine production in these mDC. By using neutralizing antibodies, we showed that HBsAg-dependent maturation of DC, but not uptake of HBsAg, is dependent on CD14 and TLR4. Since DC hardly express membrane-bound CD14, we postulated that HBsAg-induced DC maturation is mediated via serum component soluble CD14 (sCD14). Interestingly, by using serum from umbilical cord blood, which naturally contains low sCD14, we showed that sCD14 concentration is associated with the strength of HBsAg-mediated DC maturation. In conclusion, we identified CD14 and Toll-like receptor 4 (TLR4) as pattern recognition receptors (PRR) for HBsAg and sCD14 as an important host factor for activation of BDCA1⁺ DC by HBsAg.

MATERIALS AND METHODS

Isolation of monocytes and BDCA1⁺ myeloid DC.

Monocytes and BDCA1⁺ myeloid DC (mDC) were isolated from buffy coats of healthy blood donors. First, peripheral blood mononuclear cells (PBMCs) were isolated by using Ficoll-Paque (GE Healthcare) density gradient centrifugation. Monocytes were isolated from PBMCs by positive selection using anti-CD14 microbeads (Milteny Biotec). BDCA1⁺ mDC were isolated from PBMCs by CD19⁺ cell depletion followed by positive selection using anti-BDCA1-phycoerythrin (PE) and PE-conjugated microbeads (Miltenyi Biotec). The acceptable purity was minimally 95%, as assessed by flow cytometry. The standard culture medium was RPMI 1640 (Lonza), with 8% heat-inactivated fetal calf serum (FCS; Sigma), penicillin-streptomycin (Invitrogen), l-glutamine (Lonza), and 10 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF; Leukine; Genzyme).

Hepatitis B surface antigen and control proteins.

HBsAg isolated from pooled sera of patients (pHBsAg; subtype ay; purity, 98%) was purchased from American Research Products (ARP). Recombinant HBsAg (rHBsAg) and beta nerve growth factor (beta-NGF), both produced in mammalian cells, and recombinant HBsAg produced in yeast cells (yHBsAg) were purchased from Prospec. Since many commercially obtained recombinant proteins contain endotoxin impurities (12), we carefully selected preparations with the highest purity. Furthermore, all proteins were tested at used concentrations for endotoxin impurities by Endolisa (Hyglos GmbH, Bernried, Germany), a specific and sensitive method to detect endotoxins (12), according to the manufacturer's instructions. Recombinant HBsAg and beta-NGF had endotoxin levels of <0.1 endotoxin units (EU)/ml, which is below the detection limit for BDCA1⁺ DC (12). Unexpectedly, patient-derived HBsAg used at 1 μg/ml contained low endotoxin impurities (Endolisa result, 1.4 EU/ml), yet at 100 ng/ml, which was used in most experiments, pHBsAg endotoxin levels were <0.1 EU/ml, which is below the reported detection limit for BDCA1⁺ DC (12). Heat treatment of 3× concentrated preparations of pHBsAg and lipopolysaccharide (LPS; Ultrapure, from Salmonella enterica serovar Minnesota mutant R595; Invivogen) was performed by boiling for 30 min at 100°C in a heat block (13). Depletion of pHBsAg was performed via 2 different methods, starting with a stock of pHBsAg of 100 μg/ml diluted in medium. First, heparin depletion was performed by injecting HBsAg stock (untreated fraction) into a HiTrap heparin HP column (GE Healthcare). Flowthrough was collected (depleted fraction), the column was washed with phosphate-buffered saline (PBS), and bound HBsAg was eluted with Tris buffer (350 mM NaCl, 20 mM Tris, pH 7.4). The efficiency of HBsAg depletion, as measured by HBsAg enzyme-linked immunosorbent assay (ELISA), was 95.6%, whereas recovery after elution was 52.2%. For specific immune depletion, pHBsAg was incubated with protein G Sepharose beads (GE Healthcare) precoated with either Hepatect (Biotest), a pool of human IgG specific for HBsAg, or mouse IgG1 (Abcam) for 2 h at 4 degrees followed by centrifugation to obtain an HBsAg-depleted supernatant. Delipidation of HBsAg was performed as described previously (14). In brief, 1 mg/ml of HBsAg was incubated for 2 h at room temperature (RT) with the nondenaturing detergent β-d-octyl glucoside (OG; Sigma-Aldrich) at 0% (control delipidation) or 2% (wt/vol) in PBS and subsequently dialyzed twice against PBS.

Exposure of mDC to HBsAg.

Isolated mDC were incubated with medium or HBsAg at indicated concentrations. To assess the role of CD14 or TLR4, cells were preincubated with 0.2 μg/ml neutralizing antibodies to CD14 (clone MEM-18, mouse IgG1; Abcam), TLR4 (clone HT52, mouse IgG1; eBioscience), or mouse IgG1 isotype control (clone MG1-45; Abcam) for 1 h at 37°C. After 20 h of exposure at 37°C, mDC supernatants were harvested and stored at −20°C, and cell surface markers were analyzed by flow cytometry. Antibodies used were mouse anti-BDCA1-PE (Miltenyi Biotec), mouse anti-CD14-eFluor450 (clone 61D3), mouse anti-CD83-Fi (clone HB15e), and mouse anti-CD40-APC (clone 5C3, all eBioscience). Samples were stained for 30 min at 4°C in flow buffer (PBS containing 1% bovine serum albumin [BSA], 1% heat-inactivated human serum, and 0.02% NaN₃). Samples were acquired on a FACSCanto II cytometer (BD Biosciences), and data were analyzed by FlowJo software (Treestar). Concentrations of human interleukin-6 (IL-6) and IL-12p40 were measured in culture supernatants by ELISA (eBioscience).

For experiments assessing the role of serum, standard medium was replaced by X-Vivo (Lonza) with 10 ng/ml GM-CSF, supplemented or not with 8% heat-inactivated FCS or 1% or 2% human serum/plasma, as indicated. Human plasma, from adult peripheral blood or umbilical cord blood, was collected after Ficoll-Paque density gradient centrifugation and stored at −80°C. Plasma was centrifuged for 1 min at 3,000 rpm before use in experiments.

Exposure of monocytes to HBsAg.

Monocytes were preincubated with αCD14 or isotype control for 1 h and subsequently exposed to medium, 1 μg/ml pHBsAg or rHBsAg for 18 h at 37°C. During the last 16 h of culture, brefeldin A (10 μg/ml; Sigma-Aldrich) was added. The cells were harvested, fixed with 2% formaldehyde (Merck), permeabilized with 0.5% saponin (VWR) in flow buffer, and stained for CD14, IL-6, and tumor necrosis factor alpha (TNF-α) (all from eBioscience).

In vitro uptake/binding experiments.

pHBsAg was conjugated to DyLight 650 sulfhydryl-reactive dye (FL-HBsAg; Thermo Fisher Scientific). After conjugation, unbound conjugate was removed by 2 rounds of dialysis against PBS. For confocal microscopy analysis, mDC were seeded in 35-mm poly-d-lysine-coated glass bottom petri dishes (MatTek Corporation) and exposed for 2 or 20 h to 5 μg/ml FL-HBsAg. After exposure, DC were washed with PBS, stained for HLA-DR (clone LN3; eBioscience) and goat anti-mouse Dylight 488 (clone Poly4053; Biolegend) as a secondary antibody (Ab), and fixed with 2% formaldehyde. DC were acquired on a Zeiss LSM 510 inverted confocal microscope with Argon (488 nm) and HeNe (633 nm) lasers and a 63× oil-immersed objective with a 3× digital zoom. Image J software was used to merge images. For flow cytometry analysis, PBMCs depleted for CD14⁺ cells by magnetically activated cell sorting (MACS) were pretreated with medium, 1 μg/ml αCD14, or isotype control Ab and subsequently exposed to 1 μg/ml FL-HBsAg for 2 h, either at 4°C or at 37°C. Cells were washed with cold PBS, stained with mouse anti-BDCA1-PE (Miltenyi Biotec) and mouse anti-CD20-eFluor450 (eBioscience), fixed with 2% formaldehyde, and analyzed by flow cytometry.

T cell activation and cross-presentation experiments.

To examine the potency of HBsAg to enhance T cell-stimulatory activity by DC, 25,000/well freshly isolated HLA-A2⁺ BDCA1⁺ mDC were preincubated for 30 min at 37°C with 1 μg/ml pHBsAg or medium as control and subsequently incubated with 100 ng/ml HBV Core_18–27 peptide, for 20 h at 37°C. Subsequently, DC were washed 2 times and cocultured with 50,000 HLA-A2-restricted HBV Core_18–27-specific CD8⁺ T cells (15). To examine cross-presentation of HBsAg by DC, 25,000/well freshly isolated HLA-A2⁺ BDCA1⁺ mDC were preincubated for 1 h at 37°C with 5 μg/ml rHBsAg or medium as a control. After 2 washes, DC were cocultured with 50,000 HLA-A2-restricted HBsAg_183–192-specific CD8⁺ T cells (15, 16). HBV-specific CTLs were kindly provided by A. Bertoletti, Emerging Infectious Diseases, Duke-Nus Graduate Medical School, Singapore. After 20 h, supernatants were collected and gamma interferon (IFN-γ) was measured by ELISA (eBioscience).

Analysis of sCD14 concentration and sCD14-HBsAg complexes by ELISA.

CD14 concentration and sCD14-HBsAg complexes were analyzed by sandwich ELISA. For sCD14 and sCD14-HBsAg complexes in serum, samples were diluted 1,000 times and 100 times, respectively. The coating antibody was rat anti-human sCD14 (clone 55-3; BD Biosciences). Detection antibodies were biotinylated anti-human sCD14 (clone 3-C39; BD Biosciences) or biotinylated anti-HBsAg (clone 9H9) (17). Assay diluent, used for blocking and sample dilution, streptavidin-horseradish peroxidase (HRP), and TMB (3,3′,5,5′-tetramethylbenzidine) solution were all from eBioscience. Wash buffer was PBS with 0.05% Tween 20. The reaction was stopped with 1:3 (vol/vol) addition of H₂SO₄. Optical density (OD) at 450 nm was measured on a Bio-Rad imager.

Patients.

The serum samples of 60 patients with HBV infection who visited our outpatient clinic (Erasmus MC) and 15 age- and sex-matched healthy control individuals were collected and stored at −80°C. The 60 patients were all seropositive for HBsAg and represented a well-characterized cohort belonging to different clinical phases of chronic HBV infection (CHB) according to standardized criteria (18). Serum HBV DNA was determined by real-time PCR using a Cobas 48 (Roche). HBsAg levels were determined on a Cobas 411 analyzer (Roche). Serum ALT was measured on an automated analyzer. Qualitative HBeAg and anti-HBeAg were measured on an Abbot Architect analyzer. For a subgroup of these patients, PBMCs were also collected and stored at −150°. The institutional ethical review board of the Erasmus MC, Rotterdam, Netherlands, approved the clinical protocols, and written informed consent was obtained from all individuals prior to their donation of blood.

Analysis of DC phenotype ex vivo.

One million thawed PBMCs from patients with CHB (n = 4/group) and healthy controls (n = 7) were analyzed for mDC-specific activation markers by flow cytometry. Samples were stained for 30 min at 4°C in flow buffer. The antibodies used were mouse anti-BDCA1-PE (Miltenyi Biotec), mouse anti-CD20-eFluor450, anti-CD80-FITC (mIgG1) or isotype control, and anti-CD83-APC (mIgG1) or isotype control (all from eBioscience). Dead cells were excluded with Aqua dead cell stain (eBioscience). mDC were gated as BDCA1⁺/CD20⁻. Isotype controls were used to discriminate positive cells from nonspecific background staining.

RESULTS

HBsAg induces maturation, cytokine production, and enhanced T cell activation capacity in myeloid DC.

The interaction between BDCA1⁺ mDC and patient-derived HBsAg (pHBsAg) was studied in vitro by exposing freshly isolated mDC to fluorescent pHBsAg (FL-HBsAg). After 2 h of incubation with FL-HBsAg and subsequent costaining for HLA-DR to visualize the cell membrane, confocal microscopy showed intracellular HBsAg positivity in the majority of DC, indicating that mDC can efficiently take up HBsAg (Fig. 1A). After 20 h of incubation with FL-HBsAg, HBsAg+ DC formed clusters, indicating that HBsAg induced activation of DC.

FIG 1.

Interaction with HBsAg induces mDC maturation and function. (A) BDCA1⁺ mDC were incubated with medium (control) or fluorescently conjugated HBsAg for 2 or 20 h, and binding/uptake was measured by confocal microscopy after costaining for HLA-DR. Representative merged confocal images of 2 independent experiments are shown. (B and C) BDCA1⁺ mDC were incubated for 20 h with medium or 100 ng/ml or 1,000 ng/ml patient-derived HBsAg (pHBsAg). (B) CD83 and CD40 expression was determined by flow cytometry and is displayed as the percentage of positive DC or mean fluorescence intensity (MFI), respectively. Representative FACS (fluorescence-activated cell sorter) histograms show mDC incubated with medium (left) or 1,000 ng/ml pHBsAg (right) and stained for CD83 (gray) or isotype control (white). (C) Concentrations of IL-6 and IL-12p40 in mDC supernatants were determined by ELISA. Paired Student's t test was used to compare the means from 6 to 15 experiments with different donors. *, P < 0.05; **, P < 0.01; ***, P < 0.001. (D) IFN-γ production by HBV Core-specific CTL after 20 h of coculture with HLA-A2⁺ BDCA1⁺ mDC preincubated with pHBsAg or medium and subsequently loaded with peptide HBV Core_18–27. Summary of results from five experiments with DC from different subjects. (E) IFN-γ production by HBsAg-specific CTL after 20 h of coculture with HLA-A2⁺ BDCA1⁺ mDC incubated with rHBsAg or medium control (−). Shown are means ± standard deviations (SD) of IFN-γ production from 2 experiments with DC from different subjects. (F) BDCA1⁺ mDC were exposed to increasing concentrations of pHBsAg (left graph) or LPS (right graph), either untreated (normal, black bars) or heated at 100°C (boiled, gray bars). Shown are means ± SD of percent CD83⁺ DC from 2 experiments with DC from different subjects. (G to I) BDCA1⁺ mDC were exposed to untreated pHBsAg (control) or a preparation depleted for pHBsAg via heparin affinity chromatography (depleted) or the depleted preparation supplemented with normal pHBsAg (depleted + add back) (G); pHBsAg after immunoprecipitation with protG-isotype (isotype) or protG-Hepatect (depleted) or the latter with added normal pHBsAg (depleted + add back) (H); pHBsAg that was treated with 2% OG (delipidation) or 0% OG (control) and pHBsAg treated with 2% OG and complemented with normal HBsAg (delipidation + add back) (I). Means ± standard errors of the means (SEM) of relative percent CD83⁺ mDC (G to I) and IL-6 concentration in mDC supernatant (G) based on three to five pooled experiments with DC from different subjects. The percent decrease of the second bar compared to the first bar is indicated. Paired Student's t test was used to compare means of IL-6 production (G). **, P < 0.01.

Therefore, the functional consequence of this interaction was further investigated by exposing BDCA1⁺ mDC isolated from healthy donors to different concentrations of HBsAg. Exposure of mDC to pHBsAg resulted in dose-dependent upregulation of the typical DC maturation marker CD83 and the costimulatory molecule CD40 (Fig. 1B) and dose-dependent secretion of IL-6 and IL-12 (Fig. 1C). In addition to DC maturation and cytokine production, preincubation of mDC with pHBsAg enhanced peptide-specific IFN-γ production by HBV core-specific CTL (Fig. 1D). Interestingly, HBsAg-positive DC were able to cross-present HBsAg-derived epitope to the cognate CTLs, in the absence of any additional maturation stimulus (Fig. 1E).

Since antibodies capable of neutralizing the immune-stimulatory effect of HBsAg are not available, we addressed the specific role of HBsAg by several approaches. Heat treatment at 100°C completely abrogated the stimulatory activity of LPS, whereas the stimulatory activity of HBsAg was maintained (Fig. 1F). Depletion of HBsAg from the preparation via either heparin chromatography or specific immune precipitation significantly reduced upregulation of CD83 and secretion of IL-6 (Fig. 1G and H). When HBsAg was added back to the depleted preparations, DC maturation and secretion of IL-6 were restored (Fig. 1 G and H), showing that the treatment by itself did not cause the reduced DC maturation and function. Heparin depletion of LPS did not reduce its ability to stimulate BDCA1⁺ DC (data not shown). Together, these experimental approaches demonstrated that the observed DC maturation and function were specifically dependent on HBsAg.

In addition to patient-derived HBsAg, DC maturation and function were also induced by recombinant HBsAg (rHBsAg) produced in Chinese hamster ovary (CHO) cells but neither by a control CHO-derived glycoprotein from the same company, recombinant nerve growth factor beta (beta-NGF), nor by recombinant HBsAg produced in yeast (yHBsAg) (data not shown). The immune-stimulatory effect observed by both rHBsAg and pHBsAg, but not yHBsAg, suggests that the mammalian nature of the host-derived lipids and/or glycosylation may be important for its immune-stimulatory effect. The role of lipids in the HBsAg particle was further addressed by exposing mDC to pHBsAg that was treated with 2% β-d-octyl glucoside (OG) to extract lipids from the particle, as described previously (14). OG-treated but not sham-treated HBsAg totally lost its capacity to induce DC maturation (Fig. 1I), suggesting that the lipids play a major role in the immune-stimulatory effect of HBsAg on DC. In contrast, similar OG treatment of LPS only partly reduced its capacity to induce DC maturation and function (data not shown). Thus, exposure of freshly isolated mDC to HBsAg induces DC maturation, cytokine production, and enhanced capacity to activate virus-specific CTLs. Furthermore, the immune-stimulatory effect on myeloid DC is specific for HBsAg and is restricted to HBsAg from a mammalian host.

HBsAg-induced DC maturation is dependent on CD14 and TLR4.

Since monocytes are described to bind HBsAg via CD14 (19), we investigated the role of this molecule in cellular activation by HBsAg. First, we examined the role of CD14 in HBsAg-induced activation of monocytes. pHBsAg efficiently induced secretion of IL-6 and TNF-α in the majority of monocytes, which was largely absent when monocytes were pretreated with αCD14 but not isotype control antibodies (data not shown). Next, we examined the role of CD14 in HBsAg-dependent activation of BDCA1⁺ mDC. To exclude a possible interference of endotoxin contamination, we used HBsAg concentrations that were tested and shown to be free of endotoxin contamination by Endolisa (see Materials and Methods). Preincubation of mDC with αCD14, but not isotype control antibodies, completely abrogated pHBsAg-dependent upregulation of CD83 and CD40 (Fig. 2A) and secretion of IL-6 and IL-12 (Fig. 2B) but did not affect pHBsAg uptake (Fig. 2C), suggesting a crucial role for CD14 in HBsAg-induced maturation and function of mDC. Given the essential role of CD14 as coreceptor for TLR4 (20), we also investigated the role of TLR4 in HBsAg-induced DC maturation. Interestingly, neutralizing antibodies to TLR4 also blocked HBsAg-mediated upregulation of CD83 (Fig. 2D), suggesting that HBsAg-induced DC maturation is dependent on both CD14 and TLR4. HBsAg-induced DC maturation by rHBsAg was also dependent on CD14 and partially dependent on TLR4 (Fig. 2E). The observation that only a subpopulation of about 11% of BDCA1⁺ DC express CD14 on their cell membrane (Fig. 2F) was not in line with the crucial observed role of CD14 in HBsAg-induced DC maturation. Additional depletion of CD14⁺ cells revealed that purified CD14⁻ BDCA1⁺ DC were also activated by HBsAg in a CD14-dependent manner (Fig. 2G). Based on these results, the role of membrane-expressed CD14 (mCD14) in HBsAg-mediated DC maturation was brought into question, and the data pointed toward a role for soluble CD14 (sCD14), which is naturally present in FCS and human serum and can facilitate cellular activation of cells that do not express mCD14 (21, 22).

FIG 2.

HBsAg-induced DC maturation is dependent on CD14 and TLR4. (A and B) mDC were preincubated with αCD14 or isotype control and subsequently incubated with medium or 100 ng/ml pHBsAg for 20 h. Means ± SEM of percent CD83⁺ (%CD83⁺) DC and CD40 MFI (A) and IL-6 or IL-12p40 (B) for 4 or 5 experiments with different donors. n.d., not detected. (C) PBMCs depleted for CD14⁺ cells were preincubated with medium (−), αCD14, or isotype control and subsequently exposed to 1 μg/ml FL-HBsAg for 2 h at 37°C. mDC exposed to medium and FL-HBsAg at 4°C were used as controls. The percentage of HBsAg⁺ cells of BDCA1⁺CD19-DC was determined by flow cytometry. Representative FACS histograms and summary of means ± SD from duplicate assays showing the percentage of HBsAg⁺ DC. Data are representative of three experiments with different donors. (D and E) mDC were preincubated with medium (−), αCD14, αTLR4, a combination of αCD14 and αTLR4 (combi), or isotype control and subsequently exposed to medium or 100 ng/ml pHBsAg (D) or recombinant HBsAg (rHBsAg) (E) for 20 h. Mean ± SEM %CD83⁺ DC from four (D) or three (E) independent experiments with different donors. Paired Student's t test: *, P < 0.05; **, P < 0.01; ***, P < 0.001. (F) Example of flow cytometry analysis of CD14 and BDCA1 expression on PBMC. Indicated percentage represents the percentage of CD14⁺ BDCA1⁺ cells. The plot is representative of five experiments with different donors. (G) mDC, isolated either by standard procedure (white bars) or with additional CD14 depletion (black bars), were preincubated with αCD14 or isotype control and subsequently exposed to medium (−) or pHBsAg. Data represent mean ± SEM %CD83⁺ mDC for three experiments with different donors.

HBsAg-mediated DC maturation is dependent on sCD14 naturally present in FCS and human serum.

To investigate the role of sCD14, HBsAg-dependent DC maturation was compared between serum-free cultures and cultures supplemented with FCS (Fig. 3A and B). Serum-free cultures neither impaired cell viability nor induced a general defect in DC maturation, as TNF-α/IL-1β-induced DC maturation was still intact (data not shown). HBsAg-induced DC maturation and cytokine production were absent in serum-free cultures but were restored by addition of FCS (Fig. 3A and B), suggesting that a serum component is needed to support HBsAg-induced DC maturation. Similar to what was observed with FCS, HBsAg-induced DC maturation and cytokine production were restored by adding human serum (Fig. 3C and D). Furthermore, in the presence of human serum, HBsAg-dependent DC maturation was totally blocked by preincubation with CD14 neutralizing antibodies (Fig. 3C and D), suggesting a crucial role for serum factor sCD14 in facilitating HBsAg-dependent DC maturation.

FIG 3.

HBsAg-mediated DC maturation is dependent on sCD14 naturally present in FCS and human serum. (A to D) mDC were preincubated with αCD14 or isotype control and subsequently incubated with 100 ng/ml pHBsAg or medium in Xvivo serum-free medium with or without 8% FCS (A and B) or 1% human serum (C and D) for 20 h. Mean ± SEM %CD83⁺ mDC or CD40 MFI (A and C) and mean ± SEM of IL-6 and IL-12p40 (B and D) are displayed. Paired Student's t test was used to compare means of five (A and B) and three (C and D) experiments with different donors. Paired Student's t test: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Low CD14 in umbilical cord blood-derived plasma correlates with reduced HBsAg-induced DC maturation.

Interestingly, plasma from umbilical cord blood contains significantly lower sCD14 levels than does plasma from adults (Fig. 4A) (23). To further establish the role of sCD14 in BDCA1⁺ DC activation by HBsAg, we tested if neonatal serum consequently has a reduced capacity to facilitate HBsAg-induced DC maturation compared to adult serum. HBsAg-dependent maturation of healthy control DC was compared in serum-free medium supplemented with plasma samples derived either from different adult individuals or from different cord blood plasma samples. Upregulation of CD83 (Fig. 4B), CD40 (data not shown), and production of IL-6 and IL-12 (Fig. 4C) in response to HBsAg was indeed significantly reduced when cultures were supplemented with cord blood plasma samples compared to adult plasma samples. In both cultures, HBsAg-induced maturation and function were completely abrogated in the presence of CD14 neutralizing antibodies. Furthermore, the level of sCD14 in each individual plasma sample highly correlated with its capacity to mediate HBsAg-induced DC maturation and cytokine production (Fig. 4B and C). Increasing concentrations of cord blood plasma enhanced HBsAg-induced DC maturation (Fig. 4D); however, HBsAg-induced DC maturation never reached the level observed with adult plasma (data not shown). Furthermore, addition of adult plasma together with neonatal plasma in a 1:1 ratio resulted in HBsAg-induced DC maturation similar to the level induced by adult plasma alone (Fig. 4D). These data suggest that the reduced capacity of cord blood plasma to facilitate HBsAg-dependent DC activation is not caused by the presence of a dominant negative/inhibitory factor in cord blood plasma but can be attributed mainly to a suboptimal concentration of sCD14. Together these results indicate that the strength of HBsAg-induced maturation and function of DC is regulated by the concentration of sCD14.

FIG 4.

Low CD14 in neonates correlates with reduced HBsAg-induced DC maturation. (A) Means ± SEM of sCD14 concentrations in plasma obtained from peripheral blood of healthy adults (n = 17) or umbilical cord blood (n = 20). Student's t test: ***, P < 0.001. (B and C) mDC obtained from healthy adult donor were pretreated with isotype control or αCD14 and subsequently exposed to 100 ng/ml pHBsAg in the presence of plasma from adult blood (n = 17) or cord blood (n = 20) for 20 h. (B) Summary of mean ± SEM %CD83⁺ mDC (left) and correlation between sCD14 concentration and %CD83⁺ mDC (right) after exposure to HBsAg for each individual plasma sample. (C) Summary of means ± SEM of IL-6 or IL-12p40 in mDC supernatants and correlation between sCD14 concentration and IL-6 concentration after exposure to HBsAg for each individual plasma sample. Student's t test: **, P < 0.01; ***, P < 0.001. Spearman's correlation coefficient and P value are shown. Data shown in panels B and C are representative of 3 independent experiments with DC of different donors. (D) mDC obtained from healthy adult donors were exposed to medium (−) or 100 ng/ml pHBsAg (+) in the presence of X-Vivo medium supplemented with 1 or 2% plasma from adult blood, cord blood, or both. Mean ± SD %CD83⁺ mDC from 2 independent experiments with different mDC donors in which 3 different cord blood samples were separately tested. Paired Student's t test: **, P < 0.01.

sCD14-HBsAg complexes are detected in vitro and in sera of HBV-infected patients.

Next, it was important to know whether HBsAg and sCD14 directly interact. To investigate this, we developed a sandwich ELISA to detect complexes of HBsAg and sCD14 (Fig. 5A). The ELISA specifically detected in vitro-formed complexes of pHBsAg and recombinant sCD14 but not pHBsAg or sCD14 alone (Fig. 5B). The detection of sCD14-HBsAg complexes demonstrated that HBsAg and CD14 can directly interact. Furthermore, the dose-dependent detection of sCD14-HBsAg complexes (Fig. 5B) indicated that the ELISA is also suitable for semiquantification of these complexes. To investigate whether these complexes are also present in vivo, we tested serum samples from HBV-infected patients and healthy control subjects for the presence and abundance of sCD14-HBsAg complexes. The patients represented different clinical disease stages, based on viral load, serum HBsAg (Fig. 5C), and alanine aminotransferase (ALT) (Table 1). The HBsAg concentrations measured in the sera of these patients could reach more than 5 × 10e5 IU/ml, which corresponds to approximately 1,000 μg/ml (Table 1). This level exceeds more than 500 times the concentrations used in our in vitro experiments, which were thus well within the range found in patients. sCD14-HBsAg complexes were readily detected in sera of HBV-infected patients but were nondetectable in sera of healthy control individuals (Fig. 5D). In addition, the sCD14-HBsAg complexes varied substantially in different HBV clinical disease phases; soluble CD14-HBsAg complexes were detected in sera of patients in immune-tolerant (IT) and immune-active (IA) stages but were low in sera of patients in inactive carrier (IC) and E-negative CHB stages (P < 0.001) (Fig. 5D). The abundance of sCD14-HBsAg complexes correlated with the level of HBsAg in serum (Fig. 5E) and the viral load but not with ALT (data not shown).

FIG 5.

sCD14-HBsAg complexes are detected in vitro and in sera of HBV-infected patients. (A) Schematic representation of sandwich ELISA to measure complexes of soluble CD14 (sCD14) and HBsAg. (B) Detection of sCD14-HBsAg complexes (optical density [OD]) by ELISA after 20 h of incubation of indicated concentrations of pHBsAg with medium (−sCD14) or 630 ng/ml recombinant sCD14 (+sCD14). Data are representative of 3 similar experiments. (C and D) Concentration of HBsAg (C) and abundance (OD) of sCD14-HBsAg complexes (D) in sera of healthy controls and HBV-infected patients categorized by clinical disease stages (HC, healthy controls; IT, immune tolerant; IA, immune active; IC, inactive carrier; E neg, E-negative CHB patients). Data represent individual values and means for n = 15/group. Means were compared by analysis of variance (ANOVA) followed by Dunnett's multiple-comparison test to compare means for different groups to that for the reference group, which was the IC group (C) or the HC group (D): ****, P < 0.0001; ***, P < 0.001; *, P < 0.05; n.s., not significant. (E) Correlation between OD of sCD14-HBsAg complexes and HBsAg level. Spearman's r = 0.7967, P < 0.0001.

TABLE 1.

Patient characteristics^a

Characteristic	Healthy (n = 15)	Immune tolerant (n = 15)	Immune active (n = 15)	Inactive carrier (n = 15)	E-negative hepatitis (n = 15)
Gender (F/M)	9/6	12/3	7/8	10/5	2/13
Age (yr)	34.5 (23–59)	31.4 (18–48)	31.0 (18–49)	42.1 (19–71)	38.4 (22–55)
No. with HBV genotype A/B/C/D		0/7/7/1	3/3/7/2	4/4/2/5	3/1/3/8
Log10 HBV DNA (IU/ml)		8.72 (7.90–9.25)	7.42 (2.33–9.23)	2.48 (1.30–3.35)	4.40 (2.26–7.55)
HBsAg concnb
In IU/ml		68,528 (24,450–126,070)	64,268 (865–511,700)	2,437 (0.5–14,150)	6,025 (363–19,444)
In ng/ml		130,203 (46,455–239,533)	122,109 (1,644–972,230)	4,630 (1–26,885)	11,448 (690–36,944)
HBeAg (no. pos/no. neg)		15/0	15/0	0/15	0/15
Anti-HBeAg (no. pos/no. neg)		0/15	3/12	15/0	15/0
ALT concn (IU/liter)		22.0 (14–33)	84.7 (40–229)	23.7 (13–36)	72.7 (41–236)

^aValues in parentheses represent the range; pos, positive; neg, negative.

^b1 IU/ml corresponds to 1.90 ng/ml (source, Roche Diagnostics).

sCD14-HBsAg complexes are associated with a more matured DC phenotype in vivo.

To investigate whether the immune-stimulatory effect of HBsAg on BDCA1⁺ DC is also relevant in vivo, we compared cell surface expression of maturation markers CD80 and CD83 on BDCA1⁺ DC directly ex vivo between CHB patients and healthy control individuals (Fig. 6A and B). We found that in agreement with our in vitro findings, CHB patients had higher percentages of CD80⁺ and CD83⁺ BDCA1⁺ DC than did healthy controls (Fig. 6B). Moreover, the percentages of CD80⁺ and CD83⁺ DC were positively correlated with the concentration of HBsAg in serum (Fig. 6C) and with the abundance of sCD14-HBsAg complexes (Fig. 6D), suggesting that HBsAg can activate BDCA1⁺ mDC, also in vivo.

FIG 6.

HBsAg concentrations and sCD14-HBsAg complexes are associated with matured DC phenotype in vivo. (A) Gating strategy for analysis of CD80 and CD83 expression on gated BDCA1⁺ DC in total PBMCs. Isotype control antibody was used to define background for CD80 and CD83. (B) Mean percentage of CD83⁺ (left) and CD80⁺ (right) BDCA1⁺ DC between patients with chronic HBV infection (HBV, n = 16) and healthy control individuals (HC, n = 7). Student's t test: *, P < 0.05; **, P < 0.01. (C) Correlation between HBsAg concentration and percentage of CD83⁺ (left) or CD80⁺ (right) BDCA1⁺ DC. Spearman's r and P value are indicated. (D) Correlation between sCD14-HBsAg complexes and percentage of CD83⁺ (left) or CD80⁺ (right) BDCA1⁺ DC. Spearman's r and P value are indicated.

DISCUSSION

Dendritic cells (DC) are considered important players in the regulation of antiviral immunity. Proper activation of DC is a prerequisite for the induction of effective virus-specific immunity. In the present study, we demonstrated that the main HBV envelope protein HBsAg can activate BDCA1⁺ myeloid DC, leading to enhanced DC maturation and T cell-stimulatory capacity, including cross-presentation of HBsAg by DC. Patient-derived HBsAg was used at concentrations also frequently found in sera of HBV-infected individuals. A positive correlation between HBsAg levels and the percentage of activated BDCA1⁺ DC in a well-defined cohort of patients with chronic HBV infection (CHB) suggests that the observed immune-stimulatory effect of HBsAg is also present in vivo. Only a limited number of studies addressed the DC phenotype of myeloid DC in CHB patients ex vivo, as reviewed in references 8 and 24. Enhanced mDC maturation in CHB patients compared to healthy control individuals was previously reported by a single study from our group (25), whereas the majority of these studies did not find enhanced DC maturation in CHB patients (8, 24). We found a lot of variation between patients, however, and because HBsAg levels were not taken into account in these previous studies, it is possible that the effect on DC maturation may have been overlooked. In addition, our cohort consisted solely of untreated patients without certain comorbidities or advanced fibrosis, factors that could influence DC maturation.

Interestingly, the immune-stimulatory effect of HBsAg observed here was specific and restricted to patient-derived HBsAg or recombinant HBsAg produced in mammalian cells, suggesting that the type of glycosylation and/or host-derived lipids/factors may be important for its immune-stimulatory effect. A role for specific glycosylation would be in line with other viral glycoproteins that can activate DC, including the fusion protein of respiratory syncytial virus and HIV-derived gp120 as previously described (26, 27).

In addition to the immune-stimulatory effect of HBsAg, we demonstrated a novel and crucial role for CD14 in HBsAg-mediated activation of DC and monocytes, thus identifying CD14 as a receptor for HBV. CD14 is a glycoprotein expressed on the cell surface of cells of the myeloid lineage and is best known for its role as a PRR (20, 28). In contrast to HBsAg-mediated DC activation, CD14 blockade did not reduce binding of HBsAg to DC. Based on our previous work showing a role for the mannose receptor in binding of HBsAg (6), it is likely that multiple receptors can facilitate HBsAg binding to DC and CD14 is redundant for binding but not activation.

CD14 serves as a coreceptor for several TLRs, including TLR4, and contributes to ligand recognition and cellular activation (20), but an autonomous signaling function on DC has also been described (29). Since both TLR4 blockade and CD14 blockade resulted in abrogation of HBsAg-induced DC maturation, we conclude that recognition of HBsAg by mDC is mediated by the TLR4/CD14 receptor complex. In addition to the well-known role of CD14/TLR4 in recognition of bacterial products (20), this study and previous work together show that these receptors also have an emerging role in recognition of viral glycoproteins (26, 27). Similar to BDCA1⁺ DC, HBsAg can also activate other cells of myeloid origin, including monocytes and Kupffer cells (KC) (9, 11), but not BDCA3⁺ myeloid DC (data not shown) and plasmacytoid DC (30), which may relate to the absence of TLR4 on these cells (31, 32). The identification of CD14 and TLR4 as receptors for HBsAg may also have implications for HBV-associated liver pathology in the chronic phase of HBV infection. Activated hepatic stellate cells, the most fibrinogenic cell type in the liver, have a functional CD14/TLR4 response (33) and may therefore be activated by interaction with HBsAg in the liver.

In addition to membrane-bound CD14 (mCD14), CD14 exists in soluble form, which is either released from CD14⁺ cells (34, 35) or secreted by hepatocytes (36, 37) and can also be blocked by our antibody. Based on the findings that HBsAg-induced DC maturation could be abrogated by blocking CD14 and that DC maturation was not dependent on expression of mCD14 and required the presence of FCS or human serum, naturally containing soluble CD14 (sCD14), we concluded that HBsAg-dependent maturation of mDC is driven by sCD14. The pivotal role of sCD14 was further supported by the observation that sCD14 and HBsAg directly interact, both in vitro and in vivo. The abundance of sCD14-HBsAg complexes correlated with the percentage of activated DC in these patients.

In the present study, we observed that umbilical cord blood-derived plasma contained low levels of sCD14, as was previously reported in newborns (23, 38), and had reduced capacity to support HBsAg-induced DC maturation and function. This low sCD14 concentration coincides with a high risk of developing chronic hepatitis B (CHB) upon HBV infection in newborns (39). The immunological mechanisms underlying the high risk for neonates to develop CHB upon perinatal HBV infection are poorly understood. Based on our results, it is tempting to speculate that insufficient DC activation due to low sCD14 levels may be one of the factors contributing to inadequate HBV-specific immunity in HBV-infected newborns. In a seemingly paradoxical observation, Hong and colleagues recently showed that monocytes of HBV-exposed newborns had an enhanced activation state compared to those from unexposed neonates (40). Monocytes, however, in contrast to the BDCA1⁺ DC that we studied here, are not able to induce activation of naive virus-specific T cells and express membrane CD14 and thus do not depend on sCD14 for their activation. Thus, reduced sCD14 may in particular impact the initiation of HBV-specific immunity in these newborns via myeloid DC. Taken together, despite the recent work of Hong et al. and the work that we present here, further research is required to identify the definite factors that determine the high risk to develop CHB in newborns.

We conclude that HBsAg can induce maturation of BDCA1⁺ mDC via an sCD14-dependent mechanism. These findings help to comprehend the early steps in development of HBV-specific immune responses, which is essential to understand the inadequate HBV-specific immune responses in patients with chronic HBV infection and to facilitate the development of effective immunotherapies for this disease.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.