Elevated Hepatic CD1d Levels Coincide with Invariant NKT Cell Defects in Chronic Hepatitis B Virus Infection Free

Chronic hepatitis B (CHB) caused by hepatitis B virus (HBV) infection is associated with significant mortality and morbidity (1). Approximately 15–40% of patients with CHB infection develop fibrotic reaction and cirrhosis, which are associated with over 50% of the total cases of hepatocellular carcinoma (HCC) worldwide (2). Most HBV-infected adults recover after viral clearance by the development of a vigorous immune response associated with acute inflammatory liver diseases (3). However, 90% of neonates and 30–50% of children under 5 y of age exposed to HBV become chronic carriers, resulting in over 240 million people being chronically infected (4, 5). Besides aiding in transformation, HBV plays a crucial role in enhancing immune suppression (6). During chronic HBV infection, T cells lost effector function after exposure over decades to high levels of the virus (7). The failure of the immune system to eradicate the virus and to halt progression of HCC is closely linked with pathogenesis and the survival of patients with HBV infection (8, 9). Therefore, reconstitution of an efficient anti-HBV immune response is still a promising approach for treatment of hepatitis B (10).

CD1d-restricted invariant NKT (iNKT) cells that recognize lipid Ags are characterized by expression of an invariant TCR α-chain (Vα14Jα18 in mice; Vα24Jα18 in human) paired with a limited array of TCR β-chains (Vβ8.2, Vβ7, or Vβ2 in mice; Vβ11 in human) (11). As one of the predominant lymphocyte populations in liver, iNKT cells are potent regulators of the local immune system (11). Upon activation, iNKT cells respond with robust Th1 and Th2 cytokine production, helping to modulate both innate and adaptive immune responses (12). Despite low frequency in humans, iNKT cell activity is now recognized to play important roles in infectious disease (13). In humans with X-linked lymphoproliferative syndrome, a selective defect in iNKT cells is associated with lethal EBV infection and increased lymphoma (14, 15). Although studies on transgenic mouse models of HBV infection have suggested iNKT cell control of HBV replication (16), less is known about the role of iNKT cells in chronic HBV infection in humans. Current descriptions of frequency and functional status of iNKT cells are highly inconsistent in chronic HBV-infected patients. Both increased circulating iNKT cell frequency with elevated IFN-γ production (17) and reduced circulating iNKT cell frequency with preserved function (18) have been reported in CHB patients. iNKT cells have also been implicated in liver injury and HCC through secreting IL-4 and IL-13 (19). Of note, characterization of iNKT cells and CD1d expression in HBV-infected liver tissue is poorly elucidated.

Marine sponge–derived agent α-galactosylceramide (α-GalCer) is a specific agonist of iNKT cells and capable of expanding iNKT cells in vivo and in vitro (20). Activation of iNKT cells by α-GalCer inhibits HBV replication and promotes the breakage of CD8⁺ T cell tolerance in HBV transgenic animals (21). Furthermore, increased iNKT cell ratio is associated with enhanced antitumor effect against hepatitis B surface Ag (HBsAg)–expressing human hepatoma cells (22). Findings from these preclinical studies raise the rationale of exploring α-GalCer–mediated and iNKT cell–based immunotherapies for the treatment of HBV infection. Nonetheless, a recent finding from a randomized placebo-controlled phase I/II trial indicates that α-GalCer administration failed to promote sustained antiviral activity in CHB patients (23). Given that iNKT cell roles during the development of CHB remain controversial, several concerns should be addressed regarding iNKT cell functional status and their therapeutic potential in these patients.

In this study, we evaluated hepatic CD1d levels as well as iNKT cell frequency and functional status in chronic HBV infection, aiming to investigate the factors related to iNKT cell changes and explore iNKT cell–based therapeutic potential for CHB.

Liver and tumor specimens were collected during hepatic resection from 30 HBV-infected HCC, 3 HBV-negative HCC, and 10 HBV-negative hepatic carcinoid patients in Tongji Hospital, Wuhan, China. Blood samples of 206 patients with chronic HBV infection (HBsAg-positive >6 mo) and 279 healthy donor (HD) samples were collected from Tongji Hospital and Wuhan Blood Center, respectively. The HBV-infected patients were classified into four groups according to the disease stages as follows. 1) The immune tolerance (IT) group included patients with hepatitis B e Ag (HBeAg)–positive, evaluated viral load, and a normal level of aminotransferase. 2) Inactive carriers (IC) were defined by absence of HBeAg, presence of hepatitis B e Ab (HBeAb), persistently normal aminotransferase levels, and low levels of HBV DNA (<1 × 10⁴ copies/ml). 3) CHB patients included individuals in either immune-active phase or reactivation phase, confirmed by elevated aminotransferase or active inflammation on liver biopsy. Cirrhosis was diagnosed by imaging, biochemical, or histological examination. 4) HCC patients were documented by magnetic resonance imaging or computerized tomography scan and, in surgical patients, by the histological analysis of the removed mass. Individuals with other concurrent types of viral hepatitis, HIV, autoimmune liver disease, or alcoholic liver disease, were excluded. Patient characteristics are listed in Tables I and II.

Intrahepatic mononuclear cells (MNCs) were first isolated by modified enzymatic dispersal protocol as previously described (24), followed by Percoll density purification. Intrahepatic MNCs were collected from the layer between 27% Percoll and 50% Percoll. PBMCs were isolated by Ficoll density gradient centrifugation as described before (25).

1 × 10⁵ intrahepatic MNCs or 3 × 10⁵ PBMCs were simulated with 200 ng/ml α-GalCer (KRN7000; Avanti). Recombinant IL-2 and IL-15 (PeproTech) were used at 50 U/ml and 2.5 ng/ml, respectively. Purified mouse anti-human CD1d (51.1; BioLegend) was used as a blocking Ab at 5 μg/ml. Supernatants and cells were harvested on day 3 and day 7, respectively.

Fluorescence-conjugated mAbs were purchased from BD Biosciences, eBioscience, and BioLegend: CD3 (UCHT1), CD14 (M5E2), CD19 (HIB19), CD38 (HB-7), CD45 (HI30), CD69 (FN50), CD95 (DX2), CD95L (NOK-1), CD1d (51.1), PD-1 (MIH4), and HLA-DR (L243).

PBS57/CD1d tetramer was kindly gifted by the National Institutes of Health Tetramer Core Facility. Cells were stained with PBS57/CD1d tetramer, corresponding Abs, and/or CFSE (Thermo Fisher). The IFN-γ in-culture supernatant was determined by using Cytometric Bead Array Flex Sets (BD Biosciences). Data were collected using FACSVerse or LSR II cytometer (BD Biosciences) and analyzed by FlowJo software (Tree Star). CFSE proliferation assay was analyzed by ModFit LT V3.2 (Verity Software House).

Liver and hepatoma specimens were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned. Fluorescent double staining of CD1d and HBsAg in sections was performed with FITC-conjugated anti-HBsAg (Bioss) and anti-CD1d (NOR3.2; AbD Serotec). Whole sections were scanned by Pannoramic MIDI (3D HISTECH). More than 60 images without overlap were taken from each section, followed by integrated OD analysis for HBsAg and CD1d levels by ImageJ software.

To determine the mRNA levels of CD1d and HBsAg, total RNA was extracted from liver tissue using RNAfast200 (Fastagen, Shanghai, China) according to the manufacturer’s instructions; RNA was reverse-transcribed with a PrimeScript RT Reagent Kit (RR037A; Takara) to generate cDNA. The PCR reaction mixture was prepared using a SYBR Green Premix Ex Taq II Kit (RR820A; Takara) with the primers as follows: CD1d (forward: 5′-TCTCGTCCTTCGCCAATAGC-3′; reverse: 5′-GGGACCAAGGCTTCAGAGAG-3′); HBsAg (forward: 5′-ACAGGCGGGGTTTTTCTTGT-3′; reverse: 5′-CGCAGACACATCCAGCGATA-3′); and GAPDH (forward: 5′-CAGTCCATGCCATCACTGCCACCCAG-3′; reverse: 5′-CAGTGTAGCCCAGGATGCCCTTGAG-3′). The relative mRNA levels of CD1d and HBsAg were normalized to GAPDH mRNA.

Graphs were generated and analyzed by GraphPad Prism 5.0 (GraphPad Software). Data were analyzed by Student t test for comparisons of groups with normal distribution and equal variance. For nonnormal distributed variables, Mann–Whitney U test or Wilcoxon matched-pairs signed rank test was performed. Correlations between HBsAg and CD1d levels were assessed by Spearman rank correlation. The p values < 0.05 were considered statistically significant.

The ethics committee of Tongji Medicine College, Huazhong University of Science and Technology, granted approval for all aspects of this study. All patients gave written informed consent. All human subjects were adults.

As CD1d-restricted self-lipids modified by HBV contribute to activation of iNKT cells to control HBV infection (26), we first sought to evaluate whether HBV infection affected hepatocyte CD1d expression. Double staining of HBsAg and CD1d was performed in liver tissue from HBV-infected HCC patients. Normal liver sections from HBV-negative hepatic carcinoid patients served as controls (Table I). As shown in Fig. 1A and 1B, liver sections from HBV-negative individuals were weakly stained with CD1d on the cell surface (upper panel), whereas HBV-infected tissue had significantly higher CD1d levels (lower panel). In addition, CD1d-expressing lymphocytes were frequently found in HBV-infected tissue (Fig. 1A, arrowhead). In magnified images, it was clear that hepatocytes in HBV-infected liver had elevated CD1d levels both on cell surface and in cytoplasm (Fig. 1B, asterisk). The higher CD1d staining was not likely due to nonspecific staining, because the staining of CD1d isotype control in HBsAg-positive areas gave only trace background levels (Fig. 1C).

After quantification by counting pixels, we found liver sections from HBV-infected HCC patients had higher CD1d levels than those from HBV-negative individuals (Fig. 1D). Intrahepatic accumulation of HBsAg was positively correlated with CD1d staining, as indicated by correlation analysis of double-stained nonmalignant liver sections from three HBV-infected HCC patients (Fig. 1E). Meanwhile, HBsAg-positive hepatoma sections had significantly higher CD1d levels than their negative counterparts (Fig. 1F, 1G), further supporting the positive relationship between HBV infection and upregulation of CD1d. What’s more, tissue from HBV-infected patients expressed higher mRNA levels of CD1d than those from HBV-uninfected individuals (Fig. 1H). A positive correlation was found between hepatic CD1d and HBsAg transcripts in HBV-infected patients (Fig. 1I). Collectively, our results unveiled elevated CD1d expression in HBV-infected liver and hepatoma tissue.

The elevated CD1d in HBV-infected liver and hepatoma tissue is expected to facilitate an iNKT cell–based immunosurveillance effect during chronic HBV infection. However, we found a remarkable reduction of iNKT cell proportion in the liver and hepatoma tissue from HBV-infected HCC patients (Fig. 2A–C). The percentage of CD3 and PBS57/CD1d tetramer double-positive (CD3⁺CD1d Tet⁺) iNKT cells ranged from 0.26 to 1.04% in hepatic T cells from HBV-uninfected patients. This ratio decreased significantly in liver and hepatoma tissue from HBV-infected HCC patients (Fig. 2B, 2C). Furthermore, intrahepatic MNCs in hepatoma and adjacent liver tissue from HBV-infected HCC lost α-GalCer–induced IFN-γ production (Fig. 2D, middle and right panel), whereas those from HBV-negative liver had significant IFN-γ secretion (Fig. 2D, left panel). The impaired responsiveness to α-GalCer was most likely due to reduced number and impaired function of iNKT cells in the HBV-infected HCC patients.

To assess the reciprocal effects between iNKT cells and HBV infection, peripheral blood samples from 268 HDs and 191 HBV-infected patients in different disease stages were analyzed for changes of iNKT cell proportion (Table II). Prevalence of circulating iNKT cells was considerably low, ranging from 0.0002 to 1.08% among all detected subjects (Fig. 3A, 3B). Unlike the Gaussian distribution of conventional T cell frequency in the peripheral blood, the ratios of circulating iNKT cells from both HDs and patients exhibited skewed distribution with a long right tail (Supplemental Fig. 1A), which could fit a normal distribution only after logarithmic transformation (Supplemental Fig. 1B). The individual differences revealed by the asymmetric distribution suggested that a large sample size would be beneficial to characterize the frequency changes of iNKT cells in chronic HBV infection. In this study, a reduction of iNKT cell proportion in PBMCs was observed in HBV-infected patients (Fig. 3B). T cell ratios in PBMCs from the patients also decreased as compared with HDs (Supplemental Fig. 1C). Further analyzing the changes of circulating iNKT cells in different disease stages during chronic HBV infection, we found iNKT cell proportion among peripheral T cells significantly decreased in IC, CHB, and HCC patients, but not in IT individuals (Fig. 3C).

The role of iNKT cells in infection remains controversial, as both protective and pathogenic effects of iNKT cells have been proposed (19, 26, 27). To evaluate the relationship between the number or functional status of iNKT cell and liver injury, CHB patients were selected for further analysis because this group of patients showed a wide range of alanine aminotransferase (ALT)/aspartate aminotransferase (AST) levels. Notably, the CHB patients had an even lower circulating iNKT cell proportion as compared with IC groups (Fig. 3C). The number of circulating iNKT cells in CHB patients significantly reduced (Fig. 3D). Moreover, the frequency of iNKT cells reduced to a greater extent in CHB patients with cirrhosis or high HBV DNA levels (Fig. 4A, 4B). The iNKT cell frequencies were negatively associated with the degree of liver injury because patients with higher serum AST/ALT tended to have fewer iNKT cells (Fig. 4C).

To assess the functional status of residual iNKT cells in CHB patients, we stimulated PBMCs with α-GalCer to check iNKT cell expansion and IFN-γ production. As expected, culturing purified PBMCs from HDs with α-GalCer led to CD1d-dependent IFN-γ production and significant expansion of iNKT cells (Fig. 3E–G, HDs). However, in PBMCs from CHB patients, impaired IFN-γ production as well as reduced expansion rate of iNKT cells were observed (Fig. 3E–G, CHB). The iNKT cell expansion generated by α-GalCer stimulation among HD and CHB patients was actually the result of proliferation, as indicated by CFSE staining (Supplemental Fig. 2). Although the lower iNKT cell ratio in patients could partially account for lower IFN-γ production, it is worthy to note that the increased folds of ratio and number were much less in iNKT cells from CHB patients than those from HDs (Fig. 3G). What’s more, lower expansion rates were found among iNKT cells from CHB patients with high AST levels (Fig. 4D), and there was no CD1d-dependent IFN-γ production by iNKT cells from those with more severe liver injury (Fig. 4E). The in vitro hyporesponsiveness of iNKT cells was unlikely to be the result of impairment of CD1d presentation because there was no decrease in CD1d-expressing APCs in CHB patients (Supplemental Fig. 3). Altogether, these results unveiled distinct iNKT cell defects in the patients at different disease statuses. The defects associated with disease progression suggested iNKT cells attenuated their effects in CHB patients with cirrhosis, high virus load, and severe liver injury.

It is reported that iNKT cells are rapidly eliminated by apoptosis mediated by Fas/Fas ligand (FasL) interaction after activation (28). It could therefore be speculated that this mechanism was also responsible for the reduction of iNKT cells in CHB patients. Consistent with this notion, we observed upregulation of both Fas (CD95) and FasL (CD95L) in iNKT cells from CHB patients as compared with those from HDs (Fig. 5A, 5B). Further, CD69, CD38, and HLA-DR levels also increased on iNKT cells from the patients (Fig. 5A, 5B). Despite the PD-1:PD-L pathway playing a critical role in iNKT cell anergy after antigenic stimulation (29), there was no upregulated PD-1 expression on circulating iNKT cells of CHB patients (Fig. 5C). Furthermore, hepatic iNKT cells from HCC patients showed no increase in PD-1 level either, although hepatic CD8⁺ T cells from hepatoma tissue had elevated expression of PD-1 (Fig. 5D). Altogether, our results suggested aberrant activation, but not PD-1 upregulation, was associated with iNKT cell defects in CHB patients.

It is widely accepted that IL-2 significantly promotes iNKT cell proliferation and overcomes the hyporesponsive phenotype of α-GalCer–pretreated iNKT cells (30). Moreover, IL-15 is reported to play a critical role in iNKT cell homeostasis and expansion via promoting production of IL-12 by monocytes (31). As lower expansion rate and less CD1d-dependent IFN-γ production were found among CHB patients with severe liver injury (AST >3× upper limit of normal), we tested the capacity of IL-2 and IL-15 to rescue proliferation and cytokine production of the iNKT cells. As expected, IL-2 and IL-15 significantly increased iNKT cell expansion and CD1d-dependent IFN-γ production of PBMCs from HDs in response to α-GalCer (Fig. 6). For the patients, culture of PBMCs with α-GalCer in the absence of the cytokines resulted in low expansion rate of iNKT cells (Fig. 6A, 6B), which was consistent with our findings above (Fig. 3). IL-2 fully rescued the α-GalCer–induced expansion of iNKT cells from the CHB patients to reach the comparable expansion rate as those from HDs (Fig. 6B).

Unexpectedly, IL-2 alone was not sufficient to recover the CD1d-dependent cytokine production of PBMCs from patients (Fig. 6C). Although IL-15 was less efficient in promoting α-GalCer–induced expansion of iNKT cells than IL-2, the synergistic effect of IL-2 and IL-15 recovered the CD1d-dependent IFN-γ production of iNKT cells from the CHB patients (Fig. 6C). This suggested that administration of α-GalCer with IL-2 and IL-15 might help to restore the Th1 response of iNKT cells in CHB patients.

Previous studies display conflicting results regarding changes in circulating iNKT cells in chronic HBV-infected patients (17, 18, 27). Given the highly diverse iNKT cell frequencies in the human population, we enlarged the sample size by enrolling 268 healthy subjects and 191 chronic HBV-infected patients in this study. Changes of iNKT cell frequency and their responsiveness to α-GalCer were analyzed in patients in different phases of HBV infection. We found significant reduction in iNKT cell ratio and expansion rate in CHB patients, whereas there was no reduction of iNKT cells in IT individuals (Fig. 3). Because of the overall defects in iNKT cells from progressive CHB patients, administration of α-GalCer alone was unlikely to have a potent antiviral effect. This could partially explain the negative result of the phase I/II trail of α-GalCer treatment for chronic HBV infection.

In this study, we highlighted the elevated CD1d expression in HBV-infected hepatocytes. It is reported that HBV infection induces endoplasmic reticulum (ER)–associated lipid alterations on HBV-expressing hepatocytes, thus helping activation of NKT cells (26). Events in the ER normally regulate CD1d trafficking to plasma membranes (32). Therefore, the alteration of lipids in ER probably affects the CD1d expression on HBV-infected cells. Meanwhile, inflammation caused by infection, cytokines, and TLR signals also increases CD1d levels, which promotes NKT cell activation (33, 34). Viral danger signals such as type I IFN and viral TLR ligands are reported to enhance CD1d de novo synthesis through increasing the number of CD1d transcripts (34). In this study, it remained a question whether HBV infection itself or the subsequent inflammatory response contributed to the CD1d upregulation in HBsAg-positive livers. However, aberrant CD1d expression on nonprofessional APCs has been shown to induce overactivation and hyporesponsiveness of NKT cells (35, 36). Considering the enhanced CD69 and Fas/FasL expression in iNKT cells from CHB patients (Fig. 5), it is reasonable to speculate that iNKT cells obtained TCR-dependent hyporeponsiveness and suffered apoptosis by continuous and intensive CD1d-dependent stimulation in HBV-infected liver tissue.

IL-2 and IL-15 are reported to restore exhausted T cells among patients with chronic infection or carcinoma (37–39). A clinical research study shows that IL-2 and IL-15 in peritumoral liver tissue are significantly associated with a decreased incidence of recurrence of intrahepatic tumor and a prolonged overall survival (40). IL-2 and IL-15 receptors share two common subunits, including IL-2/15Rβ (CD122) and γ-chain (CD132), which accounts for the similar spectrum of biological activities of the cytokines (41). However, abundant evidence shows IL-2 and IL-15 would not fully replace each other. In particular, although IL-2 is well accepted to play critical roles in promoting T cell proliferation and functional differentiation, IL-15 is demonstrated to be essential for iNKT cell development, homeostasis, and functional differentiation (31). Furthermore, the loss of IL-15 also results in poor expression of key molecules such as IFN-γ as well as granzyme A and C, which are regulated by T-bet in iNKT cells (31). In this study, IL-2 was capable of recovering the α-GalCer–induced proliferation of iNKT cells from patients with HBV infection. This is consistent with the previous finding that IL-2 is helpful to restore iNKT cells in primary HIV-1 infection patients (42). However, only synergistic effects of IL-2 and IL-15 helped to partially recover the IFN-γ production of iNKT cells from CHB patients, indicating the differentiated and synergistic effects of IL-2 and IL-15 in iNKT cell functional recovery.

Although it is well accepted that virus-specific CD8⁺ T cells are the major effectors in the elimination of HBV, innate-like T cells are able to sense HBV infection and show faster kinetics than HBV-specific T cells with an earlier peak of activity (43). As a population bridges the gap between innate immunity and adaptive immunity, iNKT cells are able to license dendritic cells that then cross-present Ags to break CD8⁺ T cell tolerance, upregulate Th1 response toward HBV, and exert direct cytotoxicity against infected and transformed cells in the liver (21). Given the elevated CD1d level on HBV-infected liver and hepatoma tissue, α-GalCer administration combined with IL-2 and IL-15 that restored iNKT cell ratio and function in CHB patients might harness the antiviral and immunosurveillance effects of iNKT cells.

We thank the National Institutes of Health Tetramer Facility for CD1d tetramers. We also thank Sreya Bagchi for critical reading of the article.

The authors have no financial conflicts of interest.