The hepatitis C virus (HCV) infects ∼200 million people worldwide. The majority of infected individuals develop persistent infection, resulting in chronic inflammation and liver disease, including cirrhosis and hepatocellular carcinoma. The ability of HCV to establish persistent infection is partly due to its ability to evade the immune response through multiple mechanisms, including suppression of NK cells. NK cells control HCV replication during the early phase of infection and regulate the progression to chronic disease. In particular, IFN-γ produced by NK cells limits viral replication in hepatocytes and is important for the initiation of adaptive immune responses. However, NK cell function is significantly impaired in chronic HCV patients. The cellular and molecular mechanisms responsible for impaired NK cell function in HCV infection are not well defined. In this study, we analyzed the interaction of human NK cells with CD33+ PBMCs that were exposed to HCV. We found that NK cells cocultured with HCV-conditioned CD33+ PBMCs produced lower amounts of IFN-γ, with no effect on granzyme B production or cell viability. Importantly, this suppression of NK cell–derived IFN-γ production was mediated by CD33+CD11bloHLA-DRlo myeloid-derived suppressor cells (MDSCs) via an arginase-1–dependent inhibition of mammalian target of rapamycin activation. Suppression of IFN-γ production was reversed by l-arginine supplementation, consistent with increased MDSC arginase-1 activity. These novel results identify the induction of MDSCs in HCV infection as a potent immune evasion strategy that suppresses antiviral NK cell responses, further indicating that blockade of MDSCs may be a potential therapeutic approach to ameliorate chronic viral infections in the liver.
Hepatitis C virus (HCV), the causative agent of hepatitis C, infects >200 million people worldwide. The majority of infected patients are unable to clear the infection (1, 2), and consequently develop liver fibrosis, cirrhosis, and hepatocellular carcinoma (3). The discovery and use of viral protease and polymerase inhibitors have dramatically improved treatment outcomes and prognoses for HCV patients. However, the high cost of these inhibitors precludes benefit to many patients infected with HCV. Furthermore, no vaccine is available to prevent HCV infection and the subsequent spread of virus. As such, there is a continued need to find alternative, more cost-effective prophylactic and therapeutic treatments.
One of the main challenges to developing a vaccine against HCV is the ability of the virus to evade immune responses (4, 5). HCV infection dysregulates both innate and adaptive immunity by hampering IFN production, skewing the differentiation of CD4 T cells toward unfavorable Th2, Th17, and Treg subsets and impairing the function of cytotoxic CD8 T cells (6–8). HCV is also known to suppress the function of NK cells (9), which play an important role in viral clearance because they comprise 20–30% of hepatic lymphocytes in humans (10, 11). Indeed, NK cells are key players in orchestrating effective immune responses, because they can directly lyse infected cells (12) and cross-talk with Kupffer cells and dendritic cells through production of IFN-γ (13, 14), leading to regulation of T cell responses (15, 16). Notably, production of IFN-γ by NK cells correlates with their expression of CD56. In healthy individuals, CD56high NK cells produce cytokines, whereas CD56low NK cells are both cytotoxic and capable of producing cytokines (17, 18). In contrast, chronic HCV patients have a subset of CD56− NK cells that are impaired in both IFN-γ production and cytotoxicity through an unknown mechanism (16, 19). Given the central role of NK cells in regulating adaptive immune responses, it is important to understand how NK cell functions are impaired during HCV infection because it may aid in the design of vaccines against the virus.
Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population defined by their ability to suppress proinflammatory immune responses. Although the generation of MDSCs was originally described in tumor development, the immunosuppressive function of MDSCs has been reported in a variety of pathological conditions, including viral infections (20). MDSCs suppress the effector function of target cells through a number of mechanisms including the production of reactive oxygen species (ROS), inducible NO synthase, and arginase-1 (Arg-1). We previously reported that HCV infection generates CD33+CDllb+HLA-DRlo/− MDSCs, which effectively suppress T cell responses through the production of ROS (21). However, despite the pivotal role of NK cells in controlling HCV infection through inhibition of viral replication and regulation of adaptive immunity, it is not known whether MDSCs generated during HCV infection regulate the effector function of NK cells.
In recent years, there have been a number of reports demonstrating the importance of metabolic pathways to immune cell function (22–24). The mammalian target of rapamycin (mTOR) pathway is central to both cellular metabolism and immune activation because it integrates environmental cues such as nutrient availability with the growth, proliferation, and production of effector cytokines in immune cells (25–27). Signaling through the PI3K-Akt axis stimulates the serine-threonine kinase activity of the mTOR complex, which activates protein production via phosphorylation of 4EBP1 (28). Upon phosphorylation, 4EBP1 is inactivated and releases eukaryotic translation initiation factor 4E (29), allowing recruitment of ribosomes to the 5′ cap of mRNAs to initiate protein translation (30). Amino acids play a major role in triggering mTOR activation (31–33) given that protein synthesis is one of the primary outcomes of mTOR signaling (34). In particular, the amino acid l-arginine can induce the phosphorylation and activation of mTOR (31, 35). Because l-arginine availability is reduced by the production of Arg-1 by MDSCs, we hypothesized that inhibition of mTOR activation may be a key mechanism by which MDSCs regulate NK cell function.
In this article, we show that HCV-induced MDSCs suppress NK cell IFN-γ production by reducing the bioavailability of l-arginine via Arg-1. The suppression of NK cell IFN-γ production is due to a block in protein translation because there was no difference in the ability of NK cells to transcribe IFNG gene. The defect in translation of IFN-γ transcript appears likely caused by a deficiency in mTOR activation, because NK cells exposed to HCV-induced MDSCs displayed decreased phosphorylation of mTOR and its substrates.
Materials and Methods
Cell lines and virus
Huh7.5.1 were grown in DMEM containing 10% FBS, penicillin/streptomycin (100 μg/ml), l-glutamine (2 mM), and 1× nonessential amino acids and infected with the JFH-1 strain of HCV at a multiplicity of infection (m.o.i.) of 0.1 for 5 d. JFH-1 was kindly provided by Dr. T. Wakita (Tokyo Metropolitan Institute) and grown as previously described (6).
CD33+ cells and NK cell cocultures
Human PBMCs were isolated from healthy donors (Virginia Blood Services, Richmond, VA) using Sepmate-50 (Stemcell Technologies) and frozen in 90% FBS/10% DMSO. CD45+, CD33+, or NK cells were purified from cell mixtures using EasySep selection kits (Stemcell Technologies). CD45+ cells were purified from coculture of PBMCs with uninfected/infected Huh7.5.1 cells after 7 d and stained for MDSC markers by flow cytometry. In parallel experiments, CD33+ cells were obtained from coculture of PBMCs and uninfected/infected Huh7.5.1 cells and were subsequently cocultured for 2 d with autologous NK cells in RPMI 1640 containing 10% FBS, penicillin/streptomycin (10 μg/ml), and l-glutamine (2 mM) at a ratio of 1:2. Purity of autologous NK cells was confirmed via flow cytometry as >82% CD56+ cells and <2.5% CD3+ cells. NK cells were stimulated with IL-12 (10 ng/ml; PeproTech), IL-18 (10 ng/ml; R&D Systems), and IL-2 (4 μg/ml; eBioscience). The ROS scavenger catalase (100 U/ml; Sigma-Aldrich, St. Louis, MO), l-NG-monomethyl l-arginineacetate (l-NMMA; 500 μM; Sigma-Aldrich), or Nω-hydroxy-nor-l-arginine (Nor-NOHA; 500 μM; Cayman Chemicals, Ann Arbor, MI) was added during the 2-d coculture of CD33+ cells and NK cells.
IFN-γ and granzyme B in culture supernatants were measured using IFN-γ Ready-Set-Go ELISA kit (eBioscience) and Granzyme B Platinum ELISA kit (eBioscience), respectively.
Flow cytometry for MDSCs
For identifying MDSCs, CD45+ cells magnetically sorted from the coculture of PBMCs with uninfected/infected Huh7.5.1 cells were blocked with FcR blocking reagent (Miltenyi) and stained with the Live/Dead marker DAPI (Life Technologies), anti-CD33, -CD11b, and –HLA-DR (all from BD Pharmingen). For detecting intracellular Arg-1 production, CD33+ cells were magnetically sorted from cocultures with NK cells and stained for MDSC surface markers. The cells were then fixed and permeabilized by Cytofix/Cytoperm (BD Biosciences) and stained with the MDSC markers described earlier and anti–Arg-1 (R&D Systems). Aqua Live/Dead stain (Life Technologies) was included to analyze cell viability. All stained cells were run on BD FACSCantoII (BD Biosciences) and analyzed using FlowJo software.
Flow cytometry for NK cells
After coculture with uninfected/HCV-conditioned CD33+ cells, NK cells were magnetically sorted and replated in fresh media containing IL-12 (10 ng/ml) and IL-18 (10 ng/ml) in the presence of GolgiPlug (eBioscience) for 5 h. After blocking Fc receptor using the FcR blocking reagent (Miltenyi), the cells were stained with Aqua Live/Dead (Life Technologies), anti-CD56, -CD16, and -CD33 (all from BD Pharmingen). The cells were then permeabilized with Cytofix/Cytoperm (BD Biosciences) and stained with anti–IFN-γ (BD Pharmingen). For intracellular mTOR staining, NK cells were recovered after coculture with uninfected or HCV-conditioned CD33+ cells separated by a 0.45-μm transwell insert and restimulated with IL-12 (10 ng/ml) and IL-18 (10 ng/ml) for 2 d. The recovered cells were fixed in Cytofix (BD Biosciences), permeabilized using BD Phosflow Perm Buffer (III), and stained with rat anti-mTOR (R&D Systems) and mouse anti–phospho-mTOR (BD Phosflow), or mouse anti–phospho-4EBP1 (pT69) (BD Phosflow). All cells were run on BD FACSCantoII (BD Biosciences) and analyzed using FlowJo software.
Quantitative real-time PCR
RNA was extracted from magnetically sorted NK cells using GenElute Mammalian Total RNA Miniprep Kit (Sigma-Aldrich). cDNA was made using the High Capacity RNA-to-cDNA kit (Applied Biosystems), and quantitative real-time PCR (qRT-PCR) was performed using Fast SYBR Green master mix (Applied Biosystems). Gene expression was quantified on the StepOne Real Time PCR system (Applied Biosystems). Results were first normalized to GAPDH and then set relative to uninfected conditioned controls. The following primers were purchased from Eurofins MWG Operon: IFNG forward 5′-TCGGTAACTGACTTGAATGTCCA-3′ and reverse 5′-TCGCTTCCCTGTTTTAGCTGC-3′, GAPDH forward 5′-TGCACCACCAACTGCTTAGC-3′ and reverse 5′-GCATGGACTGTGGTCATGAG-3′.
CD33+ cells were cocultured with autologous NK cells separated by a 0.45-μm transwell insert in complete media in the absence of phenol red. The cells were supplemented with IL-2 (4 μg/ml) and stimulated with IL-12 (10 ng/ml) and IL-18 (10 ng/ml) for 48 h. The CD33+ cells were recovered by magnetic sorting, and the arginase activity assay (Sigma-Aldrich) was performed according to manufacturer’s instructions.
Sucrose purified JFH-1 virus
Sucrose purified JFH-1 virus was obtained from Dr. Lucy Golden-Mason. Sucrose density-gradient ultracentrifugation purification and concentration of virus was performed on pooled supernatants of JFH-1 (Takaji Wakita, National Institute of Infectious Diseases, Tokyo, Japan)–infected Huh7.5.1 cells (Francis Chisari, Scripps Research Institute, La Jolla, CA), and m.o.i. was determined by titration on Huh7.5.1 cells as previously described (36). Purified virions were used to infect Huh7.5.1 cells at m.o.i. of 0.01, 0.1, and 1.0 for 5 d before the addition of PBMCs. After 7 d, CD45+ cells were obtained by magnetic selection (Stemcell Technology) and stained for MDSC markers as shown earlier. In addition, CD33+ cells were selected from the coculture and cultured with autologous NK cells and stimulated for 2 d. The resulting cell culture media was obtained and tested for IFN-γ by ELISA (eBioscience).
Experimental results were analyzed for statistical significance using Wilcoxon matched pairs test, two-tailed paired t test, Mann–Whitney U test, or Kruskal–Wallis test (one-way ANOVA) with Dunn’s posttest, as appropriate. The p values <0.05 were considered significant and are indicated in the figures.
NK cells exhibit impaired IFN-γ production upon coculture with HCV-conditioned myeloid cells
NK cells play a pivotal role in limiting virus replication via direct killing of infected cells and production of IFN-γ, which, in turn, augments antiviral adaptive immunity. Although NK cells from patients with acute HCV infection have intact effector function, those from chronically infected patients produce significantly lower amounts of IFN-γ (37). MDSCs are known to dampen immune responses of lymphocytes in acute and chronic viral infections (20). We therefore investigated whether myeloid cells exposed to HCV infection suppress NK effector function. To this end, we cultured PBMCs with the hepatocyte cell line Huh7.5.1 uninfected or infected with HCV. After 7 d of coculture, we magnetically sorted CD33+ (myeloid) cells with several wash steps (hereafter referred to as uninfected conditioned CD33+ cells or HCV-conditioned CD33+ cells) and added them to NK cells from autologous donors (see Fig. 1A). After 48 h of stimulation with IL-12 and IL-18, NK cells cocultured with HCV-conditioned CD33+ cells produced less IFN-γ (Fig. 1B, 1C). In contrast, there was no difference in granzyme B production between NK cells cocultured with HCV-conditioned CD33+ cells and those cultured with uninfected conditioned CD33+ cells (Fig. 1D, 1E). The decrease in IFN-γ production was not due to a loss of cell viability because the number of NK cells was comparable upon coculture with uninfected or HCV-conditioned CD33+ cells (Fig. 1F). Lastly, we confirmed that NK cells were the primary source of IFN-γ in the coculture because only a negligible proportion of CD33+ myeloid cells stained for IFN-γ (Supplemental Fig. 1A). To verify that the JFH-1 virus is responsible for the MDSC-mediated suppression, we used sucrose gradient-purified virions to infect Huh cells. This too yielded the suppression of NK cell IFN-γ production (Fig. 1G). Together, these results show that HCV-conditioned myeloid cells specifically interfere with IFN-γ production by NK cells.
Reduction in NK cell IFN-γ production is posttranscriptionally regulated
Because the defect in IFN-γ production was not due to differences in the number of total NK cells, the reduction in NK cell IFN-γ secretion after coculture with HCV-conditioned myeloid cells likely reflected a defect in overall NK cell activation or a selective deficiency in IFN-γ production. To explore these possibilities, we first assessed the ability of NK cells to produce IFN-γ at the single-cell level by intracellular cytokine staining. As shown in Fig. 2A and 2B, the frequency of IFN-γ–producing NK cells was comparable upon coculture with uninfected or HCV-conditioned CD33+ cells. However, the mean fluorescence intensity (MFI) of the IFN-γ–producing NK cells was reduced in NK cells that encountered HCV-conditioned CD33+ cells, indicating that these cells were producing less IFN-γ on a per-cell basis (Fig. 2C). However, there was no significant difference in the expression of the human NK differentiation marker CD56 in total NK cells when cultured with HCV-conditioned myeloid cells (Fig. 2A, 2D, 2E). These results suggest that the differentiation status of NK cells was not altered in the presence of HCV-conditioned CD33+ cells. To further investigate the molecular basis of the decrease in IFN-γ production by NK cells, we assessed the transcriptional status of the IFNG gene. Quantitative PCR analysis revealed that IFNG mRNA was found in similar levels in NK cells cocultured with uninfected or HCV-conditioned CD33+ cells (Fig. 2F). Thus, the deficit in IFN-γ production by NK cells was likely due to regulation of posttranscriptional events by HCV-conditioned CD33+ cells.
HCV-conditioned MDSCs suppress NK cell IFN-γ production via Arg-1
We next sought to identify the molecular determinants originating from HCV-conditioned CD33+ cells that enabled regulation of IFN-γ production by NK cells. We have previously demonstrated that PBMCs cocultured with HCV-infected hepatocytes exhibit immunosuppressive functions characteristic of MDSCs (21). Given that HCV-conditioned CD33+ cells potently inhibited NK cell effector function, we investigated whether the CD33+ population included MDSCs with immunosuppressive capabilities. Indeed, coculture of PBMCs with HCV-infected hepatocytes produced a distinct population of CD33+CD11bloHLA-DRlo cells (Fig. 3A, 3B), as we described previously (4). Importantly, this population of MDSCs was minimally represented among PBMCs cultured with uninfected hepatocytes (Fig. 3C, 3D). To verify HCV infection in hepatocytes, we determined the quantity of HCV RNA and the level of core protein by qRT-PCR and Western blot analysis, respectively (Supplemental Fig. 2A, 2B), as well as the ability of sucrose purified virus to infect hepatocytes (Supplemental Fig. 2C). We further examined the relationship between viral dose and the frequency of MDSCs by assessing MDSC accumulation after the coculture of PBMCs with various m.o.i. of JFH-1 virus (Supplemental Fig. 2D), suggesting the increased frequency of MDSCs detectable in high virus dose. In addition, there is a trend where chronic HCV patients with high virus titer (>800,000 IU/ml) have more MDSCs than patients with low virus titer (Supplemental Fig. 2E), whereas there is no correlation between the frequency of MDSC and alanine transaminase level (data not shown). These results suggest that HCV is capable of inducing MDSC, which plays a role in controlling infection rather than hepatic inflammation.
MDSCs are known to inhibit immune cells through contact-dependent and contact-independent mechanisms (38). Therefore, we determined whether the cross-talk between CD33+ cells and NK cells required cell–cell contact by using 0.45-μm transwell inserts to physically separate the two populations during coculture (Fig. 4A). HCV-dependent suppression of NK cell IFN-γ production occurred even in the absence of cell contact between myeloid and NK cells (Fig. 4B). Given these findings, we sought to identify soluble immunosuppressive mediators produced by HCV-conditioned MDSCs.
MDSCs use numerous soluble factors to inhibit host immune responses, including the production of ROS, NO synthase (NOS), and Arg-1. ROS regulates immune responses through activation of apoptosis in immune cells (39), whereas NO produced by NOS nitrosylates and dissociates protein complexes involved in immune activation (40, 41). Arg-1 depletes local supplies of l-arginine and can cause inefficient proliferation of activated lymphocytes (42). Considering their potent immunosuppressive effects, we investigated whether ROS, NOS, or Arg-1 may be responsible for dampening NK cell IFN-γ production using pharmacologic inhibitors of each of these factors during coculture of NK cells with CD33+ cells: catalase scavenges ROS, l-NMMA inhibits NOS, and Nor-NOHA inhibits Arg-1. Addition of catalase and l-NMMA failed to reverse the suppressive effect of HCV-conditioned CD33+ cells on NK cell IFN-γ production (Fig. 4C, 4D). In contrast, Nor-NOHA restored IFN-γ production in NK cells exposed to HCV-conditioned CD33+ cells (Fig. 4E). Importantly, Nor-NOHA did not change IFN-γ production when added to NK cells cocultured with uninfected conditioned CD33+ cells (Supplemental Fig. 1B). HCV-induced CD33+ cells thus appear to be MDSCs that suppress NK cell IFN-γ production via Arg-1.
Inhibition of NK cell IFN-γ production by HCV-conditioned myeloid cells is reversed by l-arginine supplementation
Arg-1 catabolizes l-arginine into urea and ornithine. Although l-arginine is nonessential in most individuals, it is possible that an acute loss of l-arginine due to Arg-1 activity could affect the function of local immune cells. Therefore, we cultured NK cells in complete media or media depleted of l-arginine. Consistent with the increase in IFN-γ production upon inhibition of Arg-1 (Fig. 4E), NK cells grown in complete media produced more IFN-γ than those grown in l-arginine–deficient conditions (Fig. 5A). We further confirmed the requirement for Arg-1 in suppressing NK cell IFN-γ production by replenishing l-arginine in l-arginine–depleted cocultures of CD33+ cells and NK cells. Addition of 1 mM l-arginine, which approximates l-arginine levels found in standard RPMI formulations, reversed the inhibition of NK cell IFN-γ production (Fig. 5B). Notably, supplementing l-arginine in cocultures of NK cells and uninfected conditioned CD33+ cells did not change IFN-γ production (Supplemental Fig. 1C), verifying that l-arginine was specifically counteracting the enhanced Arg-1 activity of HCV-conditioned CD33+ cells. Indeed, direct assessment of Arg-1 function revealed that HCV-conditioned CD33+ cells had increased Arg-1 activity when compared with uninfected conditioned CD33+ cells (Fig. 5C).
We next validated the increased Arg-1 activity by intracellular staining of Arg-1. As shown in Fig. 5D–F, both the frequency and the level of Arg-1 expression were increased in HCV-conditioned CD33+ cells. Collectively, these results demonstrate that HCV-conditioned CD33+ cells suppress NK cell IFN-γ production via a contact-independent, Arg-1–dependent mechanism, which is distinct from the production of ROS that is used to suppress T cells (21). Arg-1 produced by HCV-conditioned MDSCs is thus most likely responsible for reducing IFN-γ synthesis in NK cells.
l-Arginine depletion selectively affects IFN-γ production in NK cells
To further dissect our findings on Arg-1 in MDSC-mediated suppression of NK cell IFN-γ production, we assessed the effects of l-arginine deprivation on other NK cell functions. Accordingly, we evaluated cell viability and granzyme B production in NK cells cultured in the presence of the IL-2/IL-12/IL-18 stimulatory mixture in l-arginine–deficient media. In contrast with the decrease in IFN-γ production seen in NK cells grown in l-arginine–deficient media (Fig. 5A), there was no difference in NK cell viability or granzyme B production (Supplemental Fig. 3A, 3B), suggesting that the absence of l-arginine does not result in a global insufficiency in NK activity. Instead, the availability of l-arginine regulates specific effector functions, namely, IFN-γ production, in NK cells.
MDSC-mediated suppression of IFN-γ production is mediated by reduced mTOR signaling
l-Arginine and other amino acids are a primary stimulus for activating the mTOR pathway (43, 44). The mTOR pathway integrates complex environmental cues such as nutrient availability with protein translation and higher-order cellular functions, including proliferation and cytokine production (45). Considering that the defect in NK cell IFN-γ production in our system was posttranscriptional, we reasoned that the mTOR pathway would be inefficiently activated in NK cells cultured with HCV-conditioned CD33+ cells. To test this hypothesis, we analyzed activation of the mTOR pathway in NK cells grown in l-arginine–deficient media. As shown in Fig. 6A, NK cells that were cultured with HCV-conditioned CD33+ cells expressed reduced levels of phosphorylated mTOR when compared with those cultured with uninfected conditioned CD33+ cells. Similarly, NK cells grown in l-arginine–free media also expressed reduced levels of phosphorylated mTOR compared with their counterparts grown in complete media (data not shown).
4EBP1, a downstream target of the mTOR complex, is a translational repressor that is inactivated upon phosphorylation. To further examine the effect of Arg-1–producing cells on the mTOR pathway in NK cells, we examined the phosphorylation status of 4EBP1 in NK cells cocultured with HCV-conditioned CD33+ cells. Not surprisingly, a lower percentage of NK cells recovered after coculture with HCV-conditioned CD33+ cells expressed phosphorylated 4EBP1 compared with those isolated after coculture with uninfected conditioned CD33+ cells (Fig. 6B). Complementing these results, phosphorylation of 4EBP1 was also decreased in NK cells grown in l-arginine–deficient media (data not shown). Lastly, we treated NK cells with the mTOR inhibitor, rapamycin, to verify the importance of the mTOR pathway to NK cell IFN-γ production. Indeed, IFN-γ production by NK cells treated with rapamycin was comparable with IFN-γ produced by NK cells grown in l-arginine–depleted media (Fig. 6C). Taken together, our results suggest that HCV-induced CD33+ MDSCs exhaust local supplies of l-arginine, causing insufficient mTOR activation, which likely decreases translation of IFN-γ transcript in NK cells.
As we have established that NK cells deprived of arginine as a result of exposure to MDSCs have impaired IFN-γ production caused by impairment in mTOR signaling, we sought to determine whether NK cells from chronic HCV patients exhibited similar traits. NK cells obtained from chronic HCV patients exhibited impaired IFN-γ response compared with NK cells from healthy donors upon stimulation (Fig. 7A). Further analysis on mTOR activation by NK cells from chronic HCV patients revealed that there was a slight decrease in mTOR activation by chronic HCV patients’ NK cells compared with healthy individuals with no statistical significance (Fig. 7B). Moreover, NK cells from chronic HCV patients tended to display lower levels of 4EBP1 activation (Fig. 7C), suggesting that the mTOR signaling pathway may be impaired in NK cells during chronic HCV.
In this report, we demonstrate that HCV infection induces CD33+CD11bloHLA-DRlo MDSCs, which suppress NK cell IFN-γ production by depleting l-arginine via an Arg-1–dependent mechanism. The loss of l-arginine subsequently fails to drive efficient activation of the mTOR pathway, which is necessary for translating IFN-γ mRNA into secreted protein. In contrast, l-arginine deprivation or the presence of HCV-conditioned CD33+ cells does not affect granzyme B production or NK cell viability. These results underscore the multitude of immune-evasion strategies used by HCV and identify a specific target that can be manipulated in chronic viral infections in the liver.
MDSCs have been reported to suppress other immune cells through a variety of mechanisms. We previously reported that extracellular HCV core protein triggers the generation of MDSCs (21). These MDSCs upregulated NADPH oxidase to increase ROS production, which subsequently suppressed CD4 and CD8 T cell IFN-γ production. Because NK cells are also a significant source of IFN-γ during the early phase of viral infections, it was important to identify the influence of HCV-induced MDSCs on NK cell effector function. Indeed, the decrease in NK cell IFN-γ production by HCV-induced CD33+ cells (Fig. 1B) reiterates the potent immunoregulatory role of MDSCs in inhibiting IFN-γ production during HCV infection. In contrast with ROS-dependent suppression of T cells, however, the mechanism of suppression of NK cells required increased levels of Arg-1 in HCV-conditioned MDSCs (Fig. 4). It is worthwhile to point out that Arg-1 activity is increased in HCV-infected hepatocytes (46); however, in our system, the contribution of hepatocyte Arg-1 has been eliminated because they have been eliminated from the culture with NK cells. Furthermore, it has been reported that the presence of Arg-1–producing MDSCs in HCV patients is associated with a poor prognosis, and patients undergoing antiviral therapy have fewer circulating Arg-1–producing myeloid cells (47). These observations suggest that l-arginine availability likely decreases over the course of infection. It is therefore tempting to speculate whether the difference in mechanism of suppression between T cells and NK cells reflects differences in nutrient utilization by responding immune cells over the duration of HCV infection; perhaps early responders such as NK cells use l-arginine to a greater extent than late-arriving T cells. Conversely, a recent study reported the presence of Arg-1–producing MDSCs that dampened T cell responses in hepatitis B virus infection (48). Differences in pathogens may thus also regulate the mechanisms and target cells of MDSC-mediated suppression in chronic viral infections. Consequently, MDSCs generated in vivo during chronic viral infections probably change their suppression mechanisms to target various immune populations at specific stages of infection with distinct pathogens. Nevertheless, HCV-induced MDSCs must play an important role in disease progression because two reports have shown increased MDSC frequencies in treatment-naive HCV patients compared with healthy individuals and patients undergoing antiviral treatment, and the MDSC frequencies correlated positively to HCV RNA loads (47, 49).
Arg-1 converts l-arginine into ornithine and urea, thus reducing local levels of l-arginine. Because l-arginine can directly regulate NK cell phenotype and function, the acute loss of l-arginine can profoundly affect NK cell responses (50). The effect on NK cell responses likely corresponds to the number of MDSCs and hence Arg-1 in the system (Supplemental Fig. 4). Specifically, granulocyte-derived arginase was shown to decrease NK cell IFN-γ production with little to no effect on IFN-γ transcription, NK cell viability, or granule release (51). Our results expand on these findings and identify a defect in mTOR activation as a potential mechanism by which Arg-1 activity impairs NK cell IFN-γ production. It is interesting that the percentages of cells expressing phospho-mTOR are not significantly different, but the MFIs are; in contrast, the percentages of cells expressing phospho-4EBP1 are different, but the MFI is not. We attribute these observations to the fact that most cells would express some level of phospho-mTOR when stimulated by IL-12/IL-18, but when the NK cells have sufficient arginine, more mTOR molecules become activated. We propose two reasons why there is no difference in MFI of phospho-4EBP1. On the other hand, activated mTOR can be complexed into mTORC1 and mTORC2, and mTORC2 does not activate 4EBP1. Another possibility would be because of the effects seen at various time points. At this 48-h time point, we observed no difference in MFI; however, it does not exclude the possibility that the MFI would be different over a longer suppression of the mTOR pathway. The importance of the mTOR pathway to NK cell activation was also recently demonstrated in mice, where the absence of mTOR signaling impaired nutrient uptake and acquisition of effector function, particularly IFN-γ, in NK cells (26). Given that the liver is essential for amino acid metabolism, it is intriguing to contemplate how the effect of MDSCs on NK cells would be further compounded by the metabolic status of the liver during HCV infection. In fact, postprandial increases in viral titers are well documented in HCV patients (52). However, the specific contribution of amino acids to virus production and metabolic regulation of hepatic immune responses is largely unknown and may prove useful in preventing liver damage and promoting recovery after tissue injury.
Importantly, Arg-1–mediated inhibition of NK cell effector function was limited to IFN-γ as granzyme B production was unaffected in the presence of HCV-induced MDSCs (Fig. 1E, 1F). Differences in processing and storage of granzyme B and IFN-γ may explain the IFN-γ–specific defect in NK cells cultured with HCV-conditioned CD33+ cells. Granzyme B, a serine protease that induces apoptosis in target cells, is stored in preformed granules in human PBLs (53). In contrast, NK cells store IFN-γ as transcript, which is then translated upon stimulation (54). Our results demonstrate that IFN-γ production in NK cells is inhibited at the posttranscriptional stage, because IFN-γ mRNA was present in equal amounts in NK cells cultured with uninfected or HCV-conditioned CD33+ cells (Fig. 2C). Because granzyme B release by NK cells is not dependent on translation, it could explain why the effect of HCV-induced MDSCs on NK cells was limited to IFN-γ production.
Surprisingly, the specific defect in IFN-γ production in NK cells was not due to changes in differentiation status. As shown in Fig. 2D and 2E, expression of CD56 was not significantly altered in NK cells cultured with HCV-conditioned CD33+ cells. CD56 is a marker of differentiation in human NK cells: CD56bright cells readily produce cytokines, have minimal cytolytic activity, and are thought to give rise to the more mature CD56dim population that is both cytolytic and capable of producing cytokines (55). Although both hepatic and peripheral blood NK cells express CD56, liver-resident NK cells are defined as CD56+, whereas 90% of NK cells found in blood are CD56dim (16). Because our analysis was performed on NK cells derived from PBMCs of healthy individuals, they most likely represent the responses of mature CD56dim NK cells that would infiltrate the liver during infection (56). It would be interesting to compare our results with those of hepatic NK cells in humans, because liver-resident NK cells have been distinguished from conventional NK cells in mice both developmentally and in effector function (57).
Indeed, the importance of NK cells to hepatic inflammation cannot be understated because they are a major lymphocytic subset in the liver (10). Hepatic NK cells are essential not only for conventional immune responses against pathogens, but also for maintenance of tissue homeostasis. For example, NK cells regulate the development of liver fibrosis by killing hepatic stellate cells, which are the major source of matrix deposition during liver injury (58). NK cells also control hepatic inflammation by stimulating the production of IL-6 by Kupffer cells (59) and inducing apoptosis in activated NKT and T cells (60). These findings are particularly informative when considering that dysregulated activation of macrophages and lymphocytes drives immunopathology in chronic inflammatory diseases. Given that chronic inflammation is a hallmark of numerous liver diseases, such as viral hepatitis, alcoholic and nonalcoholic steatohepatitis, autoimmune hepatitis, and hepatocellular carcinoma, the extensive cross-talk between NK cells and other hepatic cells may play a critical role in both removing the inflammatory insult and restoring tissue homeostasis. Consequently, our findings describing a role for suppressive myeloid populations in controlling NK cell responses may be equally beneficial in shifting chronic inflammation into an anti-inflammatory response.
In conclusion, we show that HCV-induced MDSCs suppress NK cell IFN-γ production via an Arg-1–dependent loss of l-arginine, resulting in defective mTOR signaling. Given that NK cells play a crucial role in antiviral immunity via cytolysis of infected cells and cytokine production, understanding how MDSCs affect NK cells provides novel insight into mechanisms that regulate NK cell function. Moreover, these results challenge us to consider the effect of MDSCs on other cells in the liver including Kupffer cells, hepatocytes, and stellate cells, all of which are key players in the progression of chronic liver diseases. Further exploration of the interplay between myeloid cells and other hepatic immune cells may thus help identify key molecular regulators that can resolve chronic inflammation and restore immune homeostasis.
We thank the members of the Hahn laboratory for providing critical advice on this work. In particular, we specially thank Sowmya Narayanan for thoughtful discussion on our studies and critical comments on the manuscript.
This work was supported by the National Institutes of Health (Grants AI057591 and U19 AI066328 to Y.S.H.; Grants T32 GM08715-14 and T32 AI07496 to C.C.G.) and the National Institute of General Medical Sciences (Maximizing Access to Research Careers-U-STAR Grant 1T34GM105550).
The online version of this article contains supplemental material.
Abbreviations used in this article:
The authors have no financial conflicts of interest.