Polarization of Monocytic Myeloid-Derived Suppressor Cells by Hepatitis B Surface Antigen Is Mediated via ERK/IL-6/STAT3 Signaling Feedback and Restrains the Activation of T Cells in Chronic Hepatitis B Virus Infection

Chronic hepatitis caused by hepatitis B virus (HBV) is a global problem with no effective treatment. According to data from the World Health Organization, >240 million people have chronic (long-term) HBV infections, and >780,000 people die every year due to the consequences of hepatitis B infections (updated July 2014). The available data have indicated that although a multispecific and vigorous HBV-specific T cell response is detectable in acute, self-limited hepatitis, a weaker response is always associated with chronic HBV infection (1, 2). Individuals with chronic HBV infection exhibit significantly diminished HBV-specific CD4⁺ and CD8⁺ T cell responses (3). In particular, HBV core epitope-specific CD8⁺ T cells are almost undetectable and have a reduced ability to produce IFN-γ (2). However, the precise mechanisms underlying HBV-mediated T cell immune attenuation are incompletely understood.

Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid precursor cells that function as negative regulators of immune responses by inhibiting T cell proliferation, Ab and IFN-γ production, and CTL induction and have been reported to play important roles in the development of microbial inflammation and infection (4, 5). Recently, the enhanced accumulation of MDSCs has been observed in mice with chronic but not acute infections with different strains of lymphocytic choriomeningitis virus (6). An increased frequency of MDSCs has also been described in the context of chronic hepatitis C virus (HCV) infections, and these cells have been shown to suppress the antiviral activities of CD8⁺ T cells (7). HCV-associated impairment of T cell activation has also been correlated with the accumulation of MDSCs, which have been implicated in stimulation of the HCV structural core protein (5). Recent studies have provided evidence that MDSCs accumulate significantly in liver of HBV-transgenic (tg) mouse models (8). A higher percentage of MDSCs, defined as CD14⁺HLA-DR^−/low, has been detected in the peripheral blood of chronic hepatitis B (CHB) patients compared with healthy control subjects. These cells have been shown to suppress HBV-specific CD8⁺ T cell responses (9). However, the mechanisms involved in the expansion of MDSCs in patients with CHB have not been identified.

Hepatitis B surface Ag (HBsAg), which is secreted by HBV-infected hepatocytes, is the most abundant HBV protein in the peripheral blood of patients with CHB (10, 11). In a previous study, we demonstrated that the direct interaction of HBsAg with human monocytes and macrophages interferes with the activation of intracellular signaling pathways regulating the production of IL-12 (12). Because the induction of HBV core-specific T cell responses is inversely correlated with the serum HBsAg concentration (13), we hypothesized that the high concentration of circulating HBsAg that is found during HBV infection may facilitate the development of MDSCs and may be responsible for the attenuation of HBV-specific T cell responses, leading to the persistence of HBV.

The aim of this study was therefore to examine whether and how circulating HBsAg is involved in MDSC expansion and contributes to T cell suppression in patients with CHB. We observed a correlation between the frequencies of monocytic MDSCs (mMDSCs) and the level of HBsAg in the sera of patients with CHB, and this correlation was confirmed in HBsAg-stimulated human PBMCs and in HBV mouse models. We then found that HBsAg could impair T cell activation by polarizing monocytes toward mMDSCs in an ERK/IL-6/STAT3 signaling–dependent manner. Moreover, we demonstrated that a drug that targets MDSC could restore the responses of Ag-specific T cells from patients with CHB ex vivo and could prevent increase of viral load in an HBV mouse model. In summary, our findings reveal a novel mechanism responsible for mMDSC expansion in patients with CHB and suggest that the abrogation of MDSCs may complement existing therapies aimed at disrupting immune suppression in patients with CHB.

Acute and CHB peripheral blood samples (in EDTA) were obtained from patients who were not receiving any antiviral or immune therapy in the recent 6 mo. Patients provided written, informed consent before inclusion in the study. Plasma was collected, and PBMCs were isolated using a Ficoll Paque density gradient. Serovirological markers (HBsAg, hepatitis B extracellular Ag [HBeAg], viral DNA, and ALT) were quantified by the institutional inspection department of the Shanghai Public Health Clinical Center. PBMCs from anonymous healthy donors were provided by the Red Cross Blood Center of Shanghai. Use of peripheral blood samples from human subjects was reviewed and approved by the Ethics Committee of the Shanghai Public Health Clinical Center.

HBV-tg mice were generated at the Shanghai Research Center for Model Organisms by routine microinjection of the linearized HBV DNA of clone 25-8 (GenBank identification number AF461363) into fertilized eggs of C57BL/6J mice. HBsAg-tg mice (C57BL/6J) were purchased from The Jackson Laboratory (Bar Harbor, ME). A volume of 10 μg pAAV-HBV1.2 plasmid in 2 ml PBS was hydrodynamically injected into the tail vein of C3H/HeN mice to establish an HBV replication model (14). No mice died before the end point of our experiment. All of the mice were sacrificed with CO₂ inhalation, and all efforts were made to minimize suffering. Use of mice was reviewed and approved by the Ethics Committee of the Shanghai Public Health Clinical Center.

For human MDSC staining, cells were incubated with FcγR binding inhibitor for 10 min at 4°C and further incubated with anti-human CD14 or CD15 PE, anti-human CD33 allophycocyanin, anti-human CD11b FITC, and anti-human HLA-DR PE-Cy7 Abs (eBioscience, San Diego, CA) for another 30 min at 4°C. For murine MDSC staining, cells were incubated with anti-mouse CD11b PerCP, anti-mouse Gr-1 allophycocyanin, and anti-mouse Ly6c PE Abs (eBioscience). For T cell staining, cells were stimulated with PMA and ionomycin (500 ng/ml each; Sigma-Aldrich) or restimulated with HBV core-derived overlapping peptides (OLPs) for 4 h and then stained with anti-human CD3 allophycocyanin-Cy7, anti-human CD4 PE, or anti-human CD8 allophycocyanin Abs (eBioscience). Then, the cells were resuspended in Fixation/Permeabilization Solution (BD Biosciences) and stained with anti-human IFN-γ PE-Cy7 (BioLegend, San Diego, CA). Anti-human p-STAT3 allophycocyanin and p-ERK1/2 allophycocyanin Abs (eBioscience) were used to detect the signaling activation in CD14⁺ cells from patients with CHB and healthy donors. The stained cells were analyzed with an FACSCalibur flow cytometer (BD Biosciences). The data were analyzed with FlowJo software (Tree Star).

Monocytes were separated from PBMCs by density-gradient centrifugation with an isopycnic gradient of Percoll as described previously (15). Human CD8⁺ T cells and CD4⁺ T cells were purified using a total CD8⁺ or CD4⁺ T Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). All cells were cultured in RPMI 1640 (HyClone) with 10% FBS (Life Technologies) and 1% penicillin/streptomycin (Life Technologies).

PBMCs (2 × 10⁶ per ml) from healthy donors were cultured in the presence of HBsAg (vaccine quality, Chinese hamster ovary [CHO]-derived, adjuvant-free) or BSA (Life Technologies) for 7 d. A total of 1 × 10⁶ monocytes/ml was cultured in the presence of HBsAg or BSA in RPMI 1640 complete media supplemented with GM-CSF and IL-4 (50 ng/ml each; R&D Systems) and containing either HBsAg or BSA for 7 d. The supernatants were collected on days 2 and 7. For some tests, a functional recombinant human (rh)IL-6, rhIL-10, anti–IL-6 Ab, or anti–IL-10 Ab (eBioscience) was included in the culture media. CD14⁺ mMDSCs, which were used for coculture experiments, were purified using CD14 Positive Selection Beads (Miltenyi Biotec). The remaining CD14⁻ cells served as controls.

The supernatants from the cultured cells were analyzed with a Human IL-6 ELISA Ready-SET-Go Kit or a Human 11plex FlowCytomix Multiplex Kit (eBioscience) according to the manufacturer’s instructions. The acquired data were analyzed using FlowCytomix Pro 2.4 software. IL-6 in the patients’ plasma was detected using a Human IL-6 High Sensitivity ELISA Kit (eBioscience). Intracellular IFN-γ was evaluated by flow cytometry as described.

Monocytes were cultured at 1 × 10⁶ cells/ ml in the presence of HBsAg. ERK1/2 phosphorylation and IκBα degradation were examined by Western blot analysis using anti-p44/42 MAPK (ERK1/2), p-p44/42 MAPK (ERK1/2), or IκBα (44D4) Abs (Cell Signaling Technology, Beverly, MA) following 20 min of treatment with HBsAg. STAT3 phosphorylation was determined by Western blot analysis using an anti-STAT3 or anti–p-STAT3 Ab (Cell Signaling Technology) after 4 h of treatment with HBsAg. Stattic, which is a STAT3-specific inhibitor (Tocris), was used to inhibit STAT3 activation. PD98059, which is an ERK-specific inhibitor (Medchem Express), was used to inhibit ERK1/2 activation.

mMDSCs were cocultured with autologous CD8⁺ and CD4⁺ T cells at a 1:2 (mMDSCs/T cells) ratio. CD8⁺ and CD4⁺ T cells were activated using plate-bound anti-CD3 and anti-CD28 Abs (eBioscience). The proliferation of CFSE-labeled CD4/CD8 T cells and IFN-γ production were measured by flow cytometry or ELISA after 3 d of coculture.

Monocytes were treated with 5 μg/ml HBsAg or BSA overnight, and total RNA was extracted using TRIzol (Invitrogen) and quantified using an ND-1000 spectrophotometer (NanoDrop). cDNA was generated using a PrimeScript RT Reagent Kit (Takara). Semiquantitative real-time PCR was performed using a SYBR Premix Ex Taq Kit (Takara). GAPDH mRNA was used as an internal control. Specific primers were designed based on the reported sequences of proteins involved in MDSC suppression and as follows: arginase-1, 5′-TTTCTCAAAGGGACAGCCAC-3′ and 5′-TGTTCTTTAAGTTTCTCAAGCAGAC-3′; inducible NO synthase (iNOS), 5′-GAGCCTGACCACTGCGTG-3′ and 5′-CAGCGAGTCCTCACACGTG-3′; NOX (p22phox), 5′-CGCTGGGGACAGAAGTACATG-3′ and 5′-GCACCGAGAGCAGGAGATG-3′; NOX (p47phox), 5′-GCCAGCACTATGTGTACATGTTC-3′ and 5′-TTTATGGAACTCGTAGATCTCGG-3′; NOX1, 5′-CCTTGGGTCAACATTGGCC-3′ and 5′-GAAGGACAGCAGATTGCGAC-3′; and NOX2 (gp91phox), 5′-CCAGTGCGTGCTGCTCAAC-3′ and 5′-GGTGTGAATCGCAGAGTGAAG-3′.

HBsAg-polarized CD14⁺ cells and PBMCs from chronic patients were treated with all-trans retinoic acid (ATRA; Sigma-Aldrich) overnight. The cells were washed twice, cultured for 2 d, and then analyzed by flow cytometry. For T cell experiments, ATRA-treated PBMCs were stimulated with anti-CD3/CD28 Abs for 3 d or with OLPs (5 μg/ml) for 9 d. T cell activation was performed as described above. For the in vivo mouse treatment, ATRA was dissolved in the dark in Cremophor EL (Sigma-Aldrich). Eight weeks after hydrodynamic injection (HDI), C3H/HeN mice were injected i.p. with 40 mg/kg body weight ATRA for 3 consecutive wk (Monday through Friday of each week).

The p values were determined using a nonparametric one-way ANOVA and parametric unpaired or paired two-tailed t tests. The null hypothesis was rejected when p was <0.05 or 0.01. The error bars in the figures represent the means ± SEMs. Correlations were evaluated by goodness-of-fit tests. Linear correlations between two continuous variables were tested using the coefficient of determination, R.

The expansion of MDSCs in HBV patients and mouse models has been reported (8, 9). However, little information exists regarding the mechanism underlying the effects of HBV on MDSCs in patients with CHB. Because human MDSCs are usually defined as CD14⁺/CD15⁺CD11b⁺CD33⁺HLA-DR^−/low (16, 17), we first confirmed published results by analyzing HLA-DR^-/lowCD33⁺CD11b⁺ population in PBMCs from healthy donors and patients with acute or CHB. The frequency of HLA-DR^−/lowCD33⁺CD11b⁺ cells in the chronic patients, but not in the acute patients, was significantly higher than that in the healthy donors (Fig. 1A). In addition, HLA-DR^−/lowCD33⁺CD11b⁺ cells from the chronic patients were all CD14⁺ mMDSCs rather than the CD15⁺ granulocytic MDSCs (gMDSCs) subtype (Fig. 1B). To further substantiate these results, we analyzed MDSCs in an HBV mouse model. Two distinct subsets of MDSCs have been identified in mice, as gMDSCs (CD11b⁺Gr1⁺Ly6C^low) and mMDSCs (CD11b⁺Gr1⁺Ly6C^hi) (16, 17). Thus, spleen and liver leukocytes were isolated and first gated by CD11b, then Gr-1 and Ly6c were used to define the MDSC population. The frequency of Gr1⁺Ly6c^hi mMDSCs in the HBV-tg mice was significantly higher than that in the background C57BL/6 mice, but the frequency of Gr1⁺Ly6c^low gMDSCs did not differ from those observed in control mice (Fig. 1C).

The presence of HBV DNA and HBV Ags, such as HBsAg and HBeAg, in the plasma, together with elevated levels of ALT, are clinical indicators of HBV infection. We therefore examined the associations between the expression of these disease markers and the frequency of mMDSCs. We found the levels of ALT, viral DNA, and HBeAg were not correlated with the frequency of mMDSCs (Fig. 1D). In contrast, the level of HBsAg were positively correlated with the frequency of mMDSCs (Fig. 1D), suggesting that HBsAg may play a role in the expansion of mMDSCs. Because it has been reported that circulating HBsAg decreases sharply in patients with acute hepatitis B (18), we evaluated the HBsAg levels in patients with acute and chronic HBV. We found that the level of serum HBsAg in patients with CHB was much higher than that in acute donors (Fig. 1E), and this finding paralleled the higher frequency of mMDSCs found in the PBMCs from the patients with CHB. Taken together, these results suggest that the accumulation of serum HBsAg in patients with CHB may contribute to the expansion of mMDSCs.

To determine the effects of HBsAg on the differentiation of mMDSCs, PBMCs obtained from healthy donors were cultured with exogenous HBsAg (vaccine quality, CHO-derived). The frequency of mMDSCs in the PBMCs cultured with HBsAg was significantly increased compared with untreated and BSA-treated control subjects (Fig. 2A). Moreover, the expansion of mMDSCs occurred in an HBsAg dose-dependent manner (Fig. 2C). In contrast, neither exogenous HBeAg nor hepatitis B core Ag promoted the polarization of mMDSCs from healthy PBMCs (Fig. 2B). Additionally, same as vaccine-quality CHO-derived HBsAg, plasma-derived HBsAg (pHBsAg) also promoted induction of mMDSCs (Fig. 2B). Because of poor purity, pHBsAg was not used in following experiments.

In view of these findings, we supposed that HBsAg induces the development of CD14⁺ mMDSCs from monocytes. To test this hypothesis, we depleted monocytes from PBMCs and cultured the remaining cells with HBsAg. A dramatically lower frequency of HLA-DR^−/lowCD11b⁺CD33⁺CD14⁺ mMDSCs was observed in the depleted cultures (Fig. 2D). Then, we purified monocytes from healthy PBMCs using Percoll centrifugation to avoid the influence of Abs used for bead selection (15). The purity of the isolated monocytes was >85%, as confirmed by flow cytometry (Supplemental Fig. 1A). Purified monocytes were then cultured with HBsAg or were left untreated or treated with BSA in the presence of GM-CSF/IL-4. We found a significantly higher frequency of CD14⁺ mMDSCs in HBsAg-treated group compared with the controls, which exhibited the CD14⁻ phenotype characteristic of dendritic cells (Fig. 2E). Costimulatory molecules were also detected by flow cytometry, which showed no difference in expression of CD40, CD80, CD83, and CD86 (Supplemental Fig. 1B). We further confirmed these results in HBsAg-tg mice. The frequency of mMDSCs, but not of gMDSCs, was statistically increased in both spleen and liver of HBsAg-tg mice compared with control mice (Fig. 2F). Taken together, these findings suggest that HBsAg promotes the differentiation of mMDSCs from monocytes.

Cytokines have been shown to play a role in the expansion of MDSCs in a variety of cancers (19). We therefore examined the profile of cytokines produced by HBsAg-treated monocytes. Three- to 6-fold higher levels of IL-8 and IL-10, respectively, and an ∼10-fold higher level of IL-6 (4000 pg/ml) were detected in HBsAg-treated monocytes compared with controls by day 2 (Fig. 3A). Because IL-8 is a chemokine that is primarily involved in cell recruitment, we focused our efforts on examining the involvement of IL-6 and IL-10 in the HBsAg-induced polarization of mMDSCs.

To determine the role of IL-6 in the polarization of mMDSCs by HBsAg, PBMCs from healthy donors were cultured in the presence of HBsAg supplemented with an anti–IL-6 Ab. Blocking IL-6 interrupted the HBsAg-mediated development of mMDSCs (Fig. 3B). In contrast, rhIL-6 was able to efficiently promote mMDSC development in BSA-treated PBMCs (Supplemental Fig. 2A). These findings were further confirmed by culturing purified monocytes in the presence of GM-CSF/IL-4 in media supplemented with either an anti–IL-6 Ab and HBsAg (Fig. 3C) or rhIL-6 and BSA (Supplemental Fig. 2B). A dose-dependent effect of IL-6 on the induction of mMDSCs in BSA-treated monocytes was also observed (Supplemental Fig. 2C). Likewise, the depletion of monocytes from PBMCs treated with HBsAg led to dramatically decreased IL-6 production (Fig. 3D). Moreover, we also found a positive correlation between the level of serum IL-6 and the percentage of cells that were mMDSCs in patients with CHB (Fig. 3E). In contrast, neither anti–IL-10 Abs nor rhIL-10 affected the polarization of mMDSCs (Supplemental Fig. 3). Our results clearly demonstrate that IL-6 induced by HBsAg contributes in an autocrine manner to the induction of mMDSCs.

MAPKs and NF-κB have been shown to regulate cytokine production in monocytes (20–22). To investigate the molecular mechanism of IL-6 production in HBsAg-treated monocytes, we evaluated the phosphorylation of ERK1/2 MAPK and the degradation of IκBα. The stimulation of monocytes with HBsAg induced ERK1/2 phosphorylation in a dose-dependent manner, whereas there was no obvious degradation of IκBα (Fig. 4A), suggesting lack of NF-κB activation. Moreover, the ERK1/2-specific inhibitor PD98059 completely inhibited the polarization of mMDSCs by HBsAg (Fig. 4B) and attenuated IL-6 production to basal levels (Fig. 4C). In contrast, inhibitors of NF-κB and P38 MAPK did not alter the HBsAg-mediated differentiation of mMDSCs (data not shown). These results show that the HBsAg-mediated activation of ERK1/2 plays a role in the induction of IL-6 and in the associated differentiation of MDSCs.

To further delineate the mechanism by which IL-6 induces the differentiation of mMDSCs, we examined the induction of STAT3 activation in HBsAg-treated purified monocytes. Stimulation with HBsAg induced a dose-dependent increase in STAT3 phosphorylation, and the addition of anti–IL-6 Abs effectively prevented STAT3 phosphorylation, suggesting that STAT3 phosphorylation was mediated in an autocrine manner by HBsAg-induced IL-6 (Fig. 4D). In addition, a STAT3 specific inhibitor, stattic (23, 24), dramatically decreased the number of mMDSCs in HBsAg-treated monocyte cultures (Fig. 4E). In summary, these results suggest that the HBsAg-mediated development of mMDSCs depends on ERK/IL-6/STAT3 signaling feedback.

To confirm these ERK1/2 and STAT3 results, we further compared the levels of p-ERK1/2 and p-STAT3 in PBMCs from patients with CHB with those in healthy donors. Because there were few mMDSCs in healthy donors, phosphorylation levels of ERK1/2 and STAT3 were analyzed in CD14⁺ cells gated from PBMCs. Through intracellular staining with phosflow Abs, we found both phosphorylation levels of ERK1/2 and STAT3 were significantly increased in patients with CHB (Fig. 4F).

Next, we determined whether HBsAg-polarized mMDSCs have the ability to suppress T cell activation. For this purpose, CD14⁺ mMDSCs were isolated by positive selection from HBsAg-treated monocytes on day 7, as described above. Because there were low levels of CD14⁺ cells in the BSA-treated and untreated monocytes (Fig. 2E), the remaining CD14⁻ cells from the HBsAg-treated monocytes were used as controls. CD14⁺ mMDSCs and CD14⁻ cells were then separately cocultured with autologous CD4⁺ or CD8⁺ T cells in the presence of plate-coated anti-CD3/CD28 Abs for 3 d. We found that the HBsAg-polarized mMDSCs effectively suppressed the proliferation of CD4⁺ and CD8⁺ T cells (Fig. 5A). Moreover, the HBsAg-polarized mMDSCs had the ability to inhibit IFN-γ production by CD4⁺ and CD8⁺ T cells, as determined by intracellular cytokine staining (Fig. 5B). These findings were further confirmed by coculture experiments of BSA/IL-6–polarized mMDSCs with CD4⁺ and CD8⁺ T cells, which also exhibited the suppression of T cell proliferation and IFN-γ production (Supplemental Fig. 2D, 2E, respectively).

It has been reported that the metabolism of arginine by ARG-1, increased production of iNOS and the generation of reactive oxygen species by NOX regulate the suppressive functions of MDSCs in both cancer and HCV infection (5, 7). We therefore examined the induction of mRNA encoding ARG-1, iNOS, and members of the NOX complex by real-time PCR. ARG-1 and NOX complex members (p22^phox, p47^phox, NOX1, and NOX2/gp91^phox) were upregulated in HBsAg-treated monocytes (Fig. 5C). Taken together, these findings indicate that HBsAg-polarized mMDSCs could functionally suppress anti-CD3/28 Ab-activated T cell responses.

Although the latest paper has shown a significant accumulation of CD14⁺HLA-DR^−/low MDSCs in patients with CHB, and these cells were demonstrated to suppress HBV-specific CD8⁺ T cell response (9), whether HBV viral load or HBV-specific T cell responses could be modulated by inhibition of MDSCs has not been fully clarified. Because ATRA has been known to target MDSCs (25), we therefore investigated whether the inhibition of mMDSCs could restore T cell responses and control HBV infection using ATRA. After a 7-d stimulation of monocytes with HBsAg, ATRA was added for a following 2-d culture. We found a dramatic decrease of mMDSCs, which gated as CD14⁺ cells from HLA-DR^−/lowCD11b⁺CD33⁺ population (Fig. 6A) and decreases in the expression of CD11b, CD33, and CD14 (Supplemental Fig. 4C). More importantly, ATRA treatment dramatically downregulated the expression of CD14 and CD11b in PBMCs from the patients with CHB (Supplemental Fig. 4D). By contrast, ATRA treatment of PBMCs from healthy donors did not result in the decreased expression of CD11b, CD33, CD14, or HLA-DR, but rather led to increased expression of these molecules (Supplemental Fig. 4A).

Then, we further examined the activation of T cells in PBMCs from the patients with CHB following treatment with ATRA. Although abrogation MDSCs with ATRA enhanced the proliferation of and IFN-γ production by anti-CD3/CD28 Ab-stimulated T cells, only the proliferation of CD4⁺ T cells showed a significant difference compared with the controls (Supplemental Fig. 4E, 4F). However, the stimulation of ATRA-treated PBMCs in patients with CHB with HBV core-derived OLPs resulted in the statistical recovery of Ag-specific proliferation and IFN-γ production by CD4⁺ and CD8⁺ T cells (Fig. 6B, 6C). Further, in PBMCs from healthy donors, ATRA treatment did not affect the proliferation and IFN-γ production of CD4⁺ and CD8⁺ T cells (Supplemental Fig. 4B).

Next, we examined the effect of abrogation of MDSCs with ATRA in a HBV replication mouse model. To imitate chronic HBV infection, mice were used at 8 wk post-HDI with pAAV-HBV1.2 plasmid (14). ATRA was injected into mice via i.p. for 3 wk. As expected, the frequency of mMDSCs in the ATRA-treated group was dramatically decreased (Fig. 6D). Moreover, whereas HBV DNA levels increased sharply in the control mice, the viral load only slightly increased and in some cases even decreased after ATRA treatment (Fig. 6E). Importantly, the fold change in the viral load was significantly different in these two groups (Fig. 6E). In summary, these results suggest that the abrogation of mMDSCs by this targeted drug leads to the enhancement of the T cell response in patients with CHB and prevents the increase of HBV load in mice.

MDSCs are pivotal in the immunotolerance of cancer and the response to chronic infection because they suppress host immune responses. In this study, we revealed a novel mechanism that underlies the HBsAg-mediated expansion of mMDSCs that contributes to immune suppression in patients with CHB. We demonstrated a new role that HBsAg plays in promoting the polarization of MDSCs from monocytes as follows: 1) by correlating the frequency of circulating mMDSCs with the level of HBsAg in the serum of patients with CHB; and 2) by inducing the development of mMDSCs from healthy PBMCs or monocytes by the addition of exogenous HBsAg. We also revealed a mechanism by which HBsAg mediates mMDSC expansion; the activation of ERK MAPK results in IL-6 production and autocrine IL-6–activated STAT3 signaling. HBsAg-induced mMDSCs suppress autologous T cell proliferation and IFN-γ production following TCR stimulation. Importantly, the inhibition of mMDSCs by the MDSC-targeted drug restored the HBV core-specific responses of T cells in PBMCs from the patients with CHB and restrained the HBV load in the HBV mouse model.

Under specific conditions, peripheral blood monocytes can serve as precursors to mMDSCs. For example, PGE₂ produced by cancer cells drives the differentiation of human monocytes away from dendritic cells and toward mMDSCs in the presence of GM-CSF and IL-4 in vitro (26). In this study, we showed that the blood level of MDSCs was positively correlated with the concentration of serum HBsAg in the patients with CHB, suggesting that circulating HBsAg plays a role in the development of mMDSCs and may mediate immune suppression. In support of this finding, it has been demonstrated that the level of circulating HBsAg is inversely correlated with the extent of HBV core-specific responses in patients with CHB (2). Furthermore, our results show that exogenous HBsAg also induces mMDSC development from healthy PBMCs or monocytes. Comparing the levels of HBsAg in patients with CHB and acute hepatitis B, we further demonstrated the role of HBsAg in the expansion of mMDSC differentiation. Dramatically low levels of HBsAg were associated with low frequency of mMDSCs in acute patients. This finding is consistent with the fact that the initially elevated concentration of HBsAg has been shown to decrease sharply during the recovery phase in acute patients (18), whereas in patients with CHB, a persistently substantial level of HBsAg can accumulate (up to 100 μg/ml) (3, 4). Hence, low levels of HBsAg in acute patients may not drive the accumulation of mMDSCs. In this study, we demonstrate that high levels of circulating HBsAg are required to induce the expansion of mMDSCs in CHB.

The development of MDSCs has been associated with chronic inflammation and the production of the IL-1β, IL-6, IL-10, TNF-α, and TGF-β cytokines in human and animal models (19, 27–30). Kong et al. (31) have found that a substantial level of IL-17a, which is a proinflammatory cytokine produced by γδT cells, is crucial for the expansion of MDSCs in an HDI HBV mouse model. Moreover, recruitment of IL-17a–produced γδT cells is crucial for production of another proinflammatory cytokine, IL-6, in hepatic regeneration of mice (32). By contrast, production of IL-17a by NK and Th17 cells is IL-6 dependent (33). In this study, we show that the induction of autocrine IL-6 by HBsAg is required for driving mMDSC polarization from monocytes. We demonstrated that IL-10 and IL-6 production was increased in the supernatants of HBsAg-polarized mMDSCs. An IL-6–blocking Ab efficiently inhibited the HBsAg-induced expansion of mMDSCs, whereas exogenous IL-6, instead of HBsAg, promoted the development of mMDSCs. In contrast, IL-10–blocking Ab, or exogenous IL-10, did not affect the differentiation of mMDSCs. IL-10 has been reported to be an effector molecule of MDSC function (9) rather than a cytokine that regulates MDSC differentiation. Our findings clearly show that HBsAg induces the development of mMDSCs via IL-6. Signaling molecules, including ERK1/2, p38 MAPK, and the STAT3, IFN regulatory factor 8, and NF-κB transcription factors, have been implicated in the regulation of MDSC differentiation in cancers (20, 21). We revealed that the HBsAg-activated ERK MAPK pathway induces IL-6 production, whereas the inhibition of ERK phosphorylation blocks the development of mMDSCs. The binding of IL-6 to its cognate receptor on monocytes usually activates the transcription factor STAT3 (34–37), which has been implicated in MDSC-mediated immune suppression (27). We showed that HBsAg induced the phosphorylation of STAT3 in an IL-6–dependent manner, whereas the inhibition of STAT3 phosphorylation blocked the HBsAg-induced differentiation of mMDSCs. To our knowledge, our results show, for the first time, a complex mechanism involving a regulatory pathway and two independent signal transduction pathways essential for the HBsAg-mediated polarization of mMDSCs.

Recent studies have shown that MDSCs suppress HBV core-specific T cell responses in patients with CHB (9) and more effectively inhibit these responses compared with unspecific T cell responses in HBV-tg mice (8). However, whether T cell responses could be recovered by the abrogation of MDSCs has not been demonstrated. ATRA has been implicated in reverting mMDSC differentiation and in the induction of gMDSC apoptosis (25, 38–41) and has been recognized as an MDSC-targeted drug (42, 43). In this study, we employed ATRA to target HBsAg-induced mMDSCs. To our knowledge, this is the first report that treatment with ATRA significantly restores HBV core-specific rather than unspecific T cell responses in PBMCs from patients with CHB and dramatically prevents the increase of HBV viral load in HDI HBV mice. We propose that inhibiting mMDSCs may be beneficial in the development of an effective immunotherapy for chronic HBV infection.

Our studies are not consistent with the observations of Huang et al. (9), who found no correlation between HBsAg levels and MDSC numbers. However, their findings may have resulted from the definition of their population of MDSCs as CD14⁺HLA-DR^−/low, whereas we defined mMDSCs by four markers as described. Moreover, to avoid potential impact caused by therapy, patients who were not receiving any antiviral or immune therapy in the prior 6 mo were chosen in our research. In addition, we further consolidated our findings by demonstrating that HBsAg could efficiently induce mMDSCs. Recently, it has been reported that gMDSCs, rather than mMDSCs, expanded in a dynamic manner during HBV infection (44). These contrasting observations may be attributed to different factors. Although Pallett et al. (44) defined gMDSC and mMDSC populations as a percentage of CD33⁺CD11b⁺ cell population rather than representation in PBMCs. The change in CD33⁺CD11b⁺ cell population, such as increase of gMDSCs, would impact the rate of mMDSC. In contrast, we defined the percentage of mMDSCs as HLA-DR^−/lowCD33⁺CD11b⁺CD14⁺ in PBMCs. Our observation demonstrating the expansion of mMDSC in CHB is consistent with several previous studies (8, 9, 31). Additionally, our method of purifying PBMCs using Ficoll gradient centrifugation tended to exclude granulocytes. Moreover, it has been observed that HBsAg internalizes predominantly in CD14⁺ cells and not in CD15⁺ cells (45). Hence, in our study, we focused on the effect of HBsAg on monocyte function and development. Although we and others have shown that MDSCs suppress HBV-specific T cells and that an MDSC-targeted drug restores HBV-specific responses of T cells, respectively, the mechanism by which HBsAg-polarized mMDSCs affect T cells in an Ag-specific way remains unknown and requires further study.

In conclusion, our findings suggest that HBsAg maintains HBV persistence and immunosuppression by promoting the ERK/IL-6/STAT3-dependent differentiation of mMDSCs from monocytes, resulting in the impaired functioning of T cells. We also suggest that therapy aimed at abrogating the expansion and functions of MDSCs may be effective in ameliorating the immune response in patients with CHB.

The authors have no financial conflicts of interest.