Article Text

Original article
Effects of HDV infection and pegylated interferon α treatment on the natural killer cell compartment in chronically infected individuals
  1. Sebastian Lunemann1,
  2. David F G Malone2,
  3. Jan Grabowski1,
  4. Kerstin Port1,
  5. Vivien Béziat2,
  6. Birgit Bremer1,
  7. Karl-Johan Malmberg2,3,4,
  8. Michael P Manns1,
  9. Johan K Sandberg2,
  10. Markus Cornberg1,
  11. Hans-Gustaf Ljunggren2,
  12. Heiner Wedemeyer1,
  13. Niklas K Björkström2,5
  1. 1Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
  2. 2Center for Infectious Medicine, Department of Medicine Huddinge, Karolinska Institutet, Stockholm, Sweden
  3. 3Institute for Cancer Research, Oslo University Hospital, Oslo, Norway
  4. 4Institute for Cancer Research, Institute of Clinical Medicine, University of Oslo, Oslo, Norway
  5. 5Liver Immunology Laboratory, Department of Medicine, Karolinska Institutet, Stockholm, Sweden
  1. Correspondence to Dr Niklas K Björkström, Center for Infectious Medicine, F59, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm S-14186, Sweden; niklas.bjorkstrom{at}ki.se

Abstract

Objective Although hepatitis delta is considered an immune-mediated disease, adaptive immune responses to hepatitis delta virus (HDV) are hardly detectable. Thus, the role of other immune responses, including those mediated by natural killer (NK) cells, must be considered in HDV pathogenesis and in treatments with immune-stimulating agents such as interferon (IFN)α. However, the phenotype and function of NK cells in chronic HDV infection, or in HDV-infected individuals undergoing IFNα treatment, have not been extensively studied.

Design We performed an extensive analysis of NK cells in chronically HDV-infected patients before and during treatment with IFNα, and compared the results with those for patients with HBV mono-infection as well as healthy controls.

Results In untreated HDV-infected patients, a higher than normal frequency of NK cells was observed in peripheral blood with unaltered phenotypic NK cell differentiation status. In contrast, long-term IFNα treatment of HDV-infected patients caused a significant change in NK cell differentiation status, with selective loss of terminally differentiated NK cells and, in parallel, a relative enrichment in immature NK cell subsets. Treatment was associated with marked functional impairment of the NK cells, which was independent of the changes in NK cell differentiation status. Furthermore, treatment polarised NK cell IFN signalling from STAT4 towards STAT1 dependency. Strikingly, a high frequency of CD56dim NK cells at baseline was positively associated with IFNα treatment outcome in the patients.

Conclusions We describe in detail how HDV infection, and IFNα treatment of this infection, affects the NK cell compartment and what consequences this has for the functional capacity of NK cells.

  • Hepatitis D
  • Interferon-Alpha
  • Immunology

Statistics from Altmetric.com

Significance of this study

What is already known about this subject?

  • Chronic hepatitis delta virus (HDV) infection is the most severe form of viral hepatitis.

  • Although hepatitis delta is considered an immune-mediated disease, adaptive immune responses to HDV are hardly detectable.

  • The role of innate immune cells such as NK cells, known to be important in other viral hepatitis infections, has not been studied extensively in chronic HDV.

What are the new findings?

  • Chronic HDV infection is associated with elevated levels of peripheral blood NK cells with reduced functional capacity in terms of ability to respond to IFNα.

  • IFNα treatment of patients with HDV infection causes a selective loss of terminally differentiated NK cells, a marked functional impairment of NK cells, and a reduced capacity to signal via STAT4.

  • A high frequency of NK cells before IFNα treatment, and retained high numbers of NK cells during treatment, are positively associated with IFNα treatment outcome.

How might this impact on clinical practice in the foreseeable future?

  • The finding that the levels of NK cells before treatment are associated with treatment outcome identifies a possible cellular immunological biomarker of HDV treatment outcome that merits further investigation in larger prospective clinical trials.

Introduction

Hepatitis delta virus (HDV), which chronically infects up to 20 million individuals worldwide, is a defective RNA virus that requires co-infection with HBV for virion packaging and propagation.1 ,2 Similarly to other viral hepatitis infections, chronic hepatitis delta progresses to liver cirrhosis and hepatic decompensation.1 ,2 However, patients with chronic HDV infection advance more rapidly in their disease than those with either hepatitis B or C.1 ,2

HDV, similarly to other viral hepatitis infections in humans, is not directly cytopathic to infected hepatocytes, and it is believed that both innate and adaptive immune responses may contribute to the immunopathogenesis in the liver.3 Adaptive immune responses against HDV are, however, weak in subjects with chronic infection, suggesting that innate immune cell subsets might play a more central role than previously thought.4–6 Natural killer (NK) cells are innate effector cells with an important role in the host defence against many virus infections and represent 30–40% of all intrahepatic lymphocytes.7 ,8 It has become clear that NK cells are pivotal for antiviral defence against hepatitis B and C.9–11 On the other hand, NK cells have also been implicated as drivers of liver pathology, especially during chronic HBV infection.9 ,12

To date, there is no approved treatment for hepatitis delta; however, we recently showed in a randomised controlled clinical trial that pegylated interferon-α-2a (peg-IFNα) cleared chronic infection in about 25% of infected patients, and that the majority of treated patients responded with a decline in HDV RNA viral load.13 Previous studies have explored the effects of peg-IFNα on NK cells in HCV and HBV infections, with an altered functional phenotype and increase in the more immature CD56bright subset observed in treatment of both infections,14 ,15 possibly due to the effects of IFNα signalling on the signal transducer and activator of transcription (STAT) pathways in the NK cells.16 However, except for a few early reports published in the mid-1980s,17–19 the possible role of NK cells in the course of chronic HDV infection and during peg-IFNα therapy has not been studied in detail.

In this study, we characterised, in detail, NK cell response to HDV infection and throughout treatment with peg-IFNα. We report that prolonged antiviral treatment of chronic untreated hepatitis delta with peg-IFNα has a significant effect on NK cell frequency, differentiation status, and function. Furthermore, high levels of CD56dim NK cells at baseline are positively associated with treatment outcome. These results are discussed in relation to current knowledge on human NK cell responses to other viral hepatitis infections and the possible role of NK cells in hepatitis delta.

Materials and methods

Patients

Subjects included in this study were seen at the outpatient clinic of the Department of Gastroenterology, Hepatology, and Endocrinology at Hannover Medical School, Hannover, Germany. They gave written informed consent for the investigation of immunological variables as part of protocols approved by the ethics committee of Hannover Medical School. Thirty-one subjects with chronic untreated HDV infection were analysed (table 1). Sixteen of these were part of an ongoing randomised clinical trial in which they were treated with peg-IFNα. Peripheral blood from the latter group was sampled longitudinally at baseline and after 12 weeks of treatment (table 2). For direct comparisons with the subjects with chronic hepatitis delta, healthy donors as well as 15 patients with chronic HBV infection were recruited. For assessment of IFNα responsiveness in relation to NK cell differentiation, 20 healthy donors—10 with immature and 10 with terminal NK cell phenotypes—were selected from an already existing set of more than 200 healthy donors who had been previously characterised.20 Peripheral blood mononuclear cells (PBMCs) were isolated by standard density-gradient separation, and cryopreserved for later analysis.

Table 1

Clinical characteristics of study subjects

Table 2

Clinical characteristics of responders and non-responders after interferon α treatment

Flow cytometry

Flow cytometry staining was performed as previously described.21 Briefly, PBMCs were washed twice with phosphate-buffered saline (PBS) containing 2% fetal calf serum and 2 mM EDTA, followed by 30 min incubation in the dark at room temperature with the desired antibody combination. They were then washed once and incubated for 20 min in the dark at room temperature with streptavidin conjugates. Next, they were washed twice and fixed in 1% paraformaldehyde for 10 min in the dark at 4°C and finally acquired on the flow cytometer. For intracellular cytokine staining, cells were then incubated in Cytofix/Cytoperm (BD Biosciences) for 20 min in the dark at 4°C, washed twice with Perm/Wash buffer (BD Biosciences), stained for cytokines for 30 min in the dark at 4°C, washed twice with Perm/Wash buffer, fixed in 1% paraformaldehyde for 10 min in the dark at 4°C, and washed once with PBS containing 2% fetal calf serum and 2 mM EDTA. Data were acquired on an LSR Fortessa cell analyser (BD Biosciences) and analysed with FlowJo software V.9.5 (Treestar). The gating strategy used to identify NK cells is shown in figure 1. Detailed information on antibodies used can be found in the online supplementary information.

Figure 1

Patients with chronic hepatitis delta virus (HDV) infection have increased levels of natural killer (NK) cells. (A) Representative staining for the gating strategy used to identify total, CD56dim and CD56bright NK cells. (B–D) Frequencies of total (B), CD56dim (C) and CD56bright (D) NK cells in 49 healthy controls (open squares) and 31 patients with chronic hepatitis delta (black squares). *p<0.05; ***p<0.001 (Mann–Whitney U test); horizontal bars represent mean. FSC, forward scatter; SSC, side scatter; DCM, dead cell marker.

Functional NK cell assays

Functional NK cell assays were performed as previously described.22 Briefly, PBMCs were incubated overnight in the absence or presence of 100 ng/mL IFNα followed by a 6 h coculture with K562 target cells. After coculture, PBMCs were washed with PBS and then stained for flow cytometry to measure expression of CD107a, IFNγ, tumour necrosis factor (TNF)α and macrophage inflammatory protein (MIP)-1β . TNF-related apoptosis-inducing ligand (TRAIL) expression was assessed by incubating PBMCs in the presence or absence of IFNα for 6 h. After 6 h of stimulation, PBMCs were washed and stained for flow cytometry. To analyse phosphorylated (p)STAT, PBMCs were rested overnight in medium with 50 U/mL interleukin (IL)-2. After surface staining for exclusion of non-NK cells, cells were incubated with IFNα at 37°C for 30 min, then fixed with BD phosflow fix buffer I, washed, and permeabilised with BD phosflow perm buffer III. Finally, PBMCs were washed again and stained with phospho-specific as well as cell-specific monoclonal antibodies. Detailed information on the functional protocols can be found in the online supplementary information.

Statistical analysis

Data were analysed using GraphPad Prism software. Statistical tests used are outlined in the respective figure legends. p<0.05 was considered significant.

Results

Chronic untreated HDV infection is associated with elevated levels of peripheral blood NK cells exhibiting reduced functional capacity to respond to IFNα

Dynamic changes within the peripheral blood NK cell compartment have been reported for many chronic viral infections in humans including chronic HBV, HCV and HIV-1.9 ,10 ,23 ,24 No such studies have to date been performed in patients with chronic hepatitis delta. Here, multicolour flow cytometry was used to identify the major NK cell subsets in PBMCs from 31 treatment-naïve subjects chronically infected with HDV and from 49 healthy controls (table 1). The gating strategy first excluded CD14 monocytes, CD19 B cells, CD3 T cells and dead cells. CD56bright and CD56dim NK cells were identified on the basis of their surface expression levels of CD56 and CD16 (figure 1A). The percentage of total NK cells was significantly higher in treatment-naïve HDV-infected subjects (13.8%) than in healthy controls (9.8%) (figure 1B). In particular, the frequency of CD56bright NK cells was more than twofold higher in patients, whereas a modest, yet significant, increase was noted for the CD56dim NK cell subset compared with healthy controls (figure 1C,D). CD56CD16+ NK cells, a subset that was increased in size during chronic HIV-1 and HCV infection, was also studied.24 No differences in frequency of CD56CD16+ NK cells was detected between healthy controls and patients with either HBV mono-infection or HDV co-infection (see online supplementary figure S1A). The increase in CD56bright NK cells appeared to be specific for HDV and did not seem to be a consequence of the underlying HBV infection, since previous studies had found no differences in total NK cell and CD56dim NK cell frequencies between healthy individuals and patients with chronic HBV infection.25

As a next step, we evaluated the capacity of NK cells to respond functionally against target cells with or without additional IFNα stimulation. We simultaneously assessed their capacity to degranulate (CD107a upregulation) and produce cytokines (IFNγ and TNFα) and chemokines (MIP-1β) (figure 2A). No major differences in functionality were observed when untreated NK cells from patients with chronic hepatitis delta were compared with healthy controls (figure 2B–E). NK cells from both healthy donors and infected subjects exhibited increased functional responses after target cell stimulation after IFNα-mediated priming (figure 2B–F). However, this increase was generally lower for the patients with hepatitis delta than for healthy donors, and it was also in line with NK cell responses from patients with chronic HBV infection (figure 2B–F).25 These data indicate that chronic untreated hepatitis delta is associated with a relative accumulation of CD56dim and CD56bright NK cells and that these cells exhibit a relatively reduced capacity to respond to IFNα. However, this decrease in function appears not to be specific for HDV, as NK cells from patients with HBV mono-infection also exhibit reduced function.25

Figure 2

Natural killer (NK) cell functional responses to interferon (IFN)α are decreased in chronic hepatitis delta virus (HDV) infection. (A) Representative staining of NK cells from one healthy donor for CD107a, IFNγ, macrophage inflammatory protein (MIP)-1β, and tumour necrosis factor (TNF)α. (B–E) Frequencies of total NK cells positive for CD107a (B), IFNγ (C), MIP-1β (D) and TNFα (E) in 45 healthy controls (open squares) and 18 patients with chronic hepatitis delta (black squares). (F) NK cells responding against K562 cells with CD107a, IFNγ, MIP-1β and TNFα in healthy controls and subjects with chronic untreated hepatitis delta comparing the fold change of the responses between non-activated and short-term (12 h) IFNα-activated NK cells. Forty-five healthy controls (open squares) and 18 patients with chronic hepatitis delta are shown (black squares). *p<0.05; **p<0.01; ***p<0.001 (Mann–Whitney U test); horizontal bars represent mean.

Characterisation of NK cell subsets after peg-IFNα treatment of chronic HDV

After characterising the effects of chronic untreated HDV infection on the NK cell compartment, we examined how treatment of hepatitis delta affected the subjects’ NK cells. Sixteen previously untreated patients with chronic HDV infection received weekly injections of 180 μg peg-IFNα, and their PBMCs were sampled at baseline and after 12 weeks of therapy (table 1). Results were compared with healthy controls and patients with chronic HBV. At 12 weeks of peg-IFNα therapy, the percentage of total NK cells and CD56dim NK cells—the dominating NK cell subpopulation in peripheral blood—was not significantly different from the pretreatment level (figure 3A–C). However, the treatment induced a threefold increase in the frequency of CD56bright NK cells (figure 3D). Strikingly, this increase represented a sixfold increase in CD56bright NK cells over that in healthy controls—that is, CD56bright NK cells made up 1.8% of total lymphocytes in the infected subjects compared with only 0.3% in the controls (figure 3D). Furthermore, the treatment caused a decrease in the frequency of CD56CD16+ cells (see online supplementary figure S1). These relative changes in the NK cell compartment were further corroborated by observed dynamics in absolute numbers of NK cells. In the 16 patients tested, total lymphocytes decreased by 46% during treatment (data not shown). Total NK cells followed a similar pattern, decreasing from ∼220 cells/μL to ∼120 cells/μL (figure 3B). This decrease was primarily attributed to a reduction in CD56dim NK cells. In contrast, the CD56bright compartment exhibited an almost twofold increase in absolute numbers (figure 3C,D).

Figure 3

Interferon (IFN)α treatment of chronic hepatitis delta virus (HDV) results in an increase in CD56bright natural killer (NK) cells, a decrease in CD56dim NK cells, and altered expression of activating receptors. (A) Representative staining of CD56 and CD57 with gates identifying CD56dim and CD56bright NK cells from two patients before and during treatment. (B–D) Frequencies (upper plots) and absolute numbers (lower plots) of total (B), CD56dim (C) and CD56bright (D) NK cells summarised for 16 patients with chronic hepatitis delta before and after 12 weeks of peg-IFNα treatment. The mean values of healthy controls are represented for 30 individuals and indicated by a dashed line (upper plots). Absolute counts were not available for the healthy controls. (E) Representative staining for CD16, NKp46, Siglec-7, NKG2D, NKp30, Eomes and T-bet from one patient with chronic HDV infection. (F) Frequencies (CD16, NKp46 and Siglec-7) and expression levels (NKG2D, NKp30, Eomes and T-bet) of activating receptors and transcription factors on CD56dim NK cells from 15 patients with chronic hepatitis delta before and during treatment. (G) Frequencies (CD16, NKp46 and Siglec-7) and expression levels (NKG2D, NKp30, Eomes and T-bet) of activating receptors and transcriptions factors on CD56bright NK cells from 15 patients with chronic hepatitis delta before and during treatment. BL, baseline; W12, after 12 weeks of treatment. *p<0.05; **p<0.01; ***p<0.001 (paired t test).

Next, we evaluated the expression of activation receptors and key NK cell transcription factors on CD56dim and CD56bright NK cells from patients with hepatitis delta (figure 3E). No differences were observed in NK cell expression of CD16, NKp46, Siglec-7, NKG2D, NKp30, Eomes and T-bet between healthy donors or HBV-infected or HDV-infected individuals (see online supplementary figure S1). Peg-IFNα therapy caused a significant reduction in expression of CD16, Siglec-7 and NKG2D in CD56dim NK cells (figure 3F). CD56bright cells exhibited lower expression of CD16 and NKp30 in patients on treatment, whereas levels of NKp46 and Siglec-7 increased (figure 3G). Finally, T-bet expression did not change during treatment, but a significant increase in Eomes expression was noted in CD56dim cells (figure 3F).

Peg-IFNα therapy causes a selective loss of terminally differentiated CD56dim NK cells

Some infections, such as human cytomegalovirus (CMV), hantavirus and HIV-1, cause imprinting in the CD56dim NK cell population characterised by accumulation of more highly differentiated CD56dim NK cells.26–28 Thus, during the response to these viruses, the more immature NKG2A+KIRCD57 CD56dim NK cells gradually redistribute towards terminally differentiated NKG2AKIR+CD57+ CD56dim NK cells.29 To determine the effect of HDV infection and long-term in vivo peg-IFNα treatment on CD56dim NK cell differentiation, we longitudinally evaluated the expression of NKG2A, CD57 and total killer-cell immunoglobulin-like receptors (KIRs) on CD56dim NK cells from the 16 subjects undergoing peg-IFNα treatment (figure 4A). No major differences in expression of NKG2A, CD57 or KIRs were noted between infected subjects at baseline and healthy controls (figure 4B–D). These findings are in line with recent results showing that mono-infection with HBV (or HCV) has no general effect on NK-cell differentiation.25 However, 12 weeks of peg-IFNα treatment led to significantly higher expression of NKG2A on CD56dim NK cells in the patients with HDV (figure 4B). In parallel, the frequency of CD57-positive NK cells was reduced almost twofold during treatment, whereas no such changes were detected in total KIR expression (figure 4C,D). When assessing absolute numbers of CD56dim NK cells expressing NKG2A, KIRs and CD57, we observed modest, although significant, declines in numbers of NKG2A+ and KIR+ cells (figure 4B,C). In contrast, 12 weeks of peg-IFNα therapy caused a striking fivefold reduction in the absolute numbers of CD57+ CD56dim NK cells from approximately 95 cells/μL to 20 cells/μL (figure 4D). This redistribution from more terminally differentiated CD57+ cells to more immature NKG2A+ cells was further corroborated by simultaneous analysis of expression patterns of NKG2A, KIRs and CD57 on CD56dim NK cells (figure 4E).

Figure 4

Selective loss of terminally differentiated CD56dim natural killer (NK) cells after therapy. (A) Representative staining for CD57, pan-KIR and NKG2A from one patient with chronic HDV infection before and after 12 weeks of pegylated interferon α (peg-IFNα) treatment. (B–D) Frequencies (upper plots) and absolute numbers (lower plots) of CD57+ (B), pan-KIR+ (C) and NKG2A+ (D) CD56dim NK cells summarised for 16 patients with chronic hepatitis delta before and during treatment. The mean values of healthy controls are represented for 30 individuals and indicated by a dashed line (upper plots). Absolute counts were not available for the healthy controls. (E) Boolean analysis of NK cell subsets expressing combinations of NKG2A, pan-KIR and CD57 before and during treatment, frequency (upper graph) and absolute numbers (lower graph). *p<0.05; **p<0.01; ***p<0.001 (paired t test); horizontal bars represent mean. Bar graphs show mean values for 16 patients and error bars represent SEM. BL, baseline; KIR, killer-cell immunoglobulin-like receptor; W12, after 12 weeks of treatment.

These results reveal that CD56dim NK cell differentiation remains intact during the course of chronic untreated hepatitis delta infection. In contrast, long-term treatment with peg-IFNα yields not only a significant increase in CD56bright NK cells (figure 3), which are considered to be predecessors of CD56dim NK cells, but also causes redistribution within the CD56dim subset towards less mature cells (figure 4).

NK cells exhibit reduced functional responses after long-term peg-IFNα therapy

After the phenotypic assessment of NK cells during peg-IFNα therapy, we next set out to examine potential alterations in their functional capacity. PBMCs from 11 subjects with chronic hepatitis delta infection, sampled at baseline and after 12 weeks of therapy, were rested or short-term-activated with IFNα followed by coculture with target cells and evaluated for multifunctional responses (figure 2A). Twelve weeks of peg-IFNα therapy led to dampened NK cell responses, with significantly lowered levels of degranulation (CD107a) and production of MIP-1β and TNFα. A similar trend, although not statistically significant, was observed for IFNγ production (figure 5A–D). This difference in functionality was even clearer when the NK cells were preactivated ex vivo with IFNα. At baseline, NK cells from untreated patients showed strongly increased responses, whereas NK cells from the same subjects sampled 12 weeks later were almost completely refractory to ex vivo IFNα stimulation (figure 5A–D). The total response strength decreased from 34% in the patients before therapy to 16% at week 12 after therapy for resting NK cells stimulated with K562 cells and from 49% to 20% for IFNα-preactivated cells. The multifunctional response profile further corroborated this difference, with significantly fewer multifunctional NK cells present at week 12, both after resting before stimulation with target cells and after IFNα stimulation (figure 5E). On assessment of the functional responses of NK cell subsets, CD56dim and CD56bright NK cells exhibited similarly affected profiles at week 12. These differences were also present when the absolute level (mean fluorescent intensity) of the response was evaluated (data not shown).

Figure 5

Altered natural killer (NK) cell function after interferon (IFN)α treatment of chronic hepatitis delta virus (HDV). (A–D) Frequencies of total NK cells positive for CD107a (A), IFNγ (B), macrophage inflammatory protein (MIP)-1β (C) and tumour necrosis factor (TNF)α (D) in 11 patients with chronic hepatitis delta before (circles) and during (squares) treatment, with (black) or without (open) IFNα prestimulation followed by cocultre with K562 cells. (E) Multifunctional NK cell responses grouped on the basis of numbers of functions executed. (F) Frequencies of CD56dim (left plot) and CD56bright (right plot) NK cells positive for TNF-related apoptosis-inducing ligand (TRAIL) in 12 patients with chronic hepatitis delta before (circles) and during (squares) treatment, with (black) or without (open) IFNα stimulation. *p<0.05; **p<0.01; ***p<0.001 (paired t test). BL, baseline; W12, after 12 weeks of treatment.

As a final functional parameter, we evaluated the capacity of NK cells to express TRAIL, both directly ex vivo and in response to in vitro IFNα stimulation. No differences were seen when baseline ex vivo TRAIL levels of NK cells from healthy donors, patients with HBV and patients with HDV were compared. Furthermore, NK cells from all three groups were equally efficient in upregulating TRAIL in response to IFNα stimulation (see online supplementary figure S2). Twelve weeks of peg-IFNα treatment led to higher ex vivo levels of TRAIL on both CD56dim and CD56bright NK cells (figure 5f). However, and similarly to the reduced functional responses against K562 cells after IFNα stimulation, CD56dim NK cells were less efficient in upregulating TRAIL in response to IFNα after 12 weeks of therapy than before onset of treatment (figure 5f).

These results indicate that long-term peg-IFNα therapy for chronic HDV infection not only alters NK cell differentiation (figure 4), but also results in functional impairment of NK cells.

IFNα therapy affects interferon signalling in patients with hepatitis delta infection

After characterising the functional NK cell impairment resulting from long-term peg-IFNα therapy, we next sought to investigate underlying factors behind this phenomenon. We have previously shown that NK cell responses to IL-12 and IL-15 stimulation is tightly coupled to CD56dim NK cell differentiation.29 However, the functional relationship between CD56dim NK cell differentiation and responsiveness to IFNα is not known. Thus, it is plausible that the reduction in NK cell functional responses observed after 12 weeks of peg-IFNα therapy might be due at least partly to suboptimal responses of less mature NKG2A+CD57 CD56dim NK cells (figures 3 and 4). To test this, and to circumvent the potential confounding factor of long-term peg-IFNα stimulation in vivo, we used an existing cohort of more than 200 healthy donors whose CD56dim NK cell differentiation status had been predetermined.20 Based on NKG2A and CD57 expression patterns in this cohort, we selected 10 donors with the most differentiated and 10 donors with the least differentiated CD56dim NK cell phenotype (see online supplementary figure S3A–C). However, there were no appreciable differences in NK cell responsiveness to IFNα stimulation between these two groups of healthy donors (see online supplementary figure S3D–G). Corroborating these findings, equally diminished functional responses were observed in immature and mature CD56dim NK cell subsets in six HDV patients after 12 weeks of therapy (see online supplementary figure S4A–D).

Next, we sought to investigate if long-term peg-IFNα therapy caused changes in NK cell interferon signalling. To this end, we measured total levels of STAT1 as well as pSTAT1 and pSTAT4 after IFNα stimulation in healthy controls, HBV-infected patients and HDV-infected patients before onset of treatment and after 12 weeks of therapy. Total levels of STAT1 were significantly higher in both HBV- and HDV-infected patients compared with healthy controls (see online supplementary figure S5A). However, no differences in interferon signalling was observed between these groups (see online supplementary figure S5B,C). Instead, a significant increase in total STAT1 and basal levels of pSTAT1 and pSTAT4 was observed in NK cells from patients with hepatitis delta after 12 weeks of therapy (figure 6B,C). The observed increase in pSTAT1 after in vitro stimulation with IFNα was consistent between baseline and treatment samples (figure 6D); in contrast, the induction of pSTAT4 was significantly inhibited after therapy (figure 6D).

Figure 6

Long-term interferon (IFN)α therapy polarises natural killer (NK) cell IFN signalling from signal transducer and activator of transcription (STAT)4 towards STAT1. (A) Representative staining for expression of total STAT1, phosphorylated (p)STAT1 and pSTAT4 in NK cells from one patient with chronic hepatitis delta virus (HDV) before and after IFNα stimulation. (B) Expression intensity (mean fluorescent intensity (MFI)) of STAT1 in NK cells (n=15) from patients with HDV. (C) Baseline levels of pSTAT1 (n=12) and pSTAT4 (n=9) as well as the log of pSTAT1 over pSTAT4 ratio (n=9). (D) Induction of pSTAT1 and pSTAT4 after 30 min of IFNα treatment (n=12) as well as the log of the ratio of the induction of pSTAT1 over pSTAT4. *p<0.05; **p<0.01; ***p<0.001 (Wilcoxon test). BL, baseline; W12, after 12 weeks of treatment.

Collectively, these data suggest that immature and mature CD56dim NK cells respond equally well to IFNα stimulation, with degranulation and cytokine production. Instead, long-term peg-IFNα therapy affects NK cell interferon signalling with increases in basal as well as pSTAT1 levels, but with a reduced capacity to signal via STAT4.

Frequency of CD56dim NK cells is positively associated with peg-IFNα treatment outcome

Finally, we investigated whether changes in NK cell numbers, differentiation or function among infected patients were associated with severity of disease, control of infection in untreated individuals, or treatment outcome after peg-IFNα therapy. NK cell phenotype and function did not correlate with severity of disease or with viral load (data not shown). In addition, no difference was detected in NK cell phenotype and function between cirrhotic and non-cirrhotic patients (data not shown).

Next, of the 16 patients followed during treatment, seven responded to therapy at week 12: three became HDV RNA negative and four had detectable but non-quantifiable (<300 copies/mL) levels of virus (table 2). The seven responders had a significantly higher frequency of CD56dim NK cells at baseline than the non-responders (15.1±1.9% vs 9.2±1.6%, p=0.017) (figure 7A). This difference was also present after 12 weeks of peg-IFNα treatment, and, strikingly, only non-responding patients exhibited a significant decline in CD56dim NK cell frequencies and numbers during treatment (figure 7A). The relative redistribution from mature CD57+ CD56dim NK cells towards more immature NKG2A+ CD56dim NK cells occurred equally in responders and non-responders during treatment (data not shown). However, responders retained higher numbers of NKG2A+ and pan-KIR+ NK cells during treatment and had significantly more NKG2A+ and CD57+ NK cells at week 12 than non-responders (figure 7B–D). It was not possible to evaluate any possible association between NK cell function and treatment response, since functional NK cell data were not available from a sufficient number of patients.

Figure 7

Frequency of CD56dim natural killer (NK) cells is positively associated with interferon (IFN)α treatment outcome. (A) Frequencies (left plot) and absolute numbers (right plot) of CD56dim NK cells at baseline (BL) and after 12 weeks of peg-IFNα treatment (W12) in responding (n=7) and non-responding (n=9) patients with hepatitis delta virus (HDV). (B–D) Absolute numbers of NKG2A+ (B), pan-KIR+ (C), and CD57+ (D) CD56dim NK cells at baseline and after 12 weeks of peg-IFNα treatment in responding and non-responding HDV patients. *p<0.05; **p<0.01; ***p<0.001 (paired t test; Mann–Whitney U test for unpaired comparisons and Wilcoxon test for paired comparisons); horizontal bars represent mean. KIR, killer-cell immunoglobulin-like receptor.

Taken together, these data indicate that HDV-infected patients that respond to peg-IFNα treatment have higher levels of NK cells in peripheral blood at the start of treatment, and that these patients retain higher numbers of NK cells during the course of treatment than non-responding patients.

Discussion

This study provides the first comprehensive analysis of NK cell response during chronic hepatitis delta, the most severe form of viral hepatitis affecting up to 20 million subjects worldwide.2 In untreated HDV-infected patients, NK cells increase, with no detectable change in NK cell differentiation status. Long-term peg-IFNα treatment of HDV-infected patients caused a selective loss of terminally differentiated NK cells and, subsequently, a relative enrichment of immature NK cell subsets. Treatment was associated with a marked functional impairment of peripheral blood NK cells, and this effect was independent of the observed altered NK cell differentiation. These changes may be influenced by shifts in the ratio between pSTAT1 and pSTAT4 occurring during therapy. Finally, a high frequency of CD56dim NK cells at baseline was positively associated with response to treatment in the patients.

In recent years, it has become increasingly clear that NK cells are important in chronic viral hepatitis infections, such as hepatitis B and C. In HCV infection, NK cells have been implicated in the outcome of infection according to genetic, phenotypic and functional analysis.9 ,24 For instance, in population genetic studies, the combination of the inhibitory NK cell receptor, KIR2DL3, with group 1 human leucocyte antigen (HLA)-C is associated with control of acute infection and with a beneficial response to peg-IFNα treatment.30 ,31 Furthermore, the appearance of dysfunctional CD56CD16+ NK cells correlated with an impaired ability to respond to antiviral treatment.32 On the other hand, in HBV infection, increased NK cell activity has been linked to liver injury.12 ,33 NK cells have also been suggested to negatively regulate antiviral T cell responses via targeting of HBV-specific CD8 T cells during chronic HBV infection.34 In our dataset, we observed no correlations between NK cell phenotype and function with clinical or histological parameters associated with disease severity in the chronic untreated patients. Instead, high baseline frequencies of CD56dim NK cells showed a positive association with response to peg-IFNα treatment, and responding patients, to a higher degree, retained their NK cells during the course of treatment compared with non-responding patients. This novel finding suggests that retained high numbers of NK cells is beneficial for the host during peg-IFNα treatment of chronic HDV infection.

CD56dim NK cells, maturing from CD56bright predecessors, have traditionally been considered to be terminal cells that retain fixed functional and phenotypic properties throughout their lifespan. However, we now know that NK cells, both in mouse and man, undergo a continuous differentiation process.29 ,35 This differentiation is characterised by the phenotypic progression from immature NKG2A+CD62L+CD57KIR CD56dim NK cells, via cellular intermediates, to terminal NKG2ACD62LCD57+KIR+ cells and is also associated with a profound functional reprogramming of the cells.29 ,36 Although the exact mechanisms are unknown, many viral infections, both acute (hantavirus28 and chikungunya virus37) and chronic (HIV-126 and CMV20 ,38), seem to drive NK cell differentiation, with a resulting gradual accumulation of terminally differentiated CD56dim NK cells. In this respect, the effects of chronic hepatitis virus infections, such as HBV and HCV, on NK cell differentiation is largely unknown. Here, we found that the CD56dim NK cell differentiation status was strikingly similar in the untreated subjects with chronic hepatitis delta and healthy individuals, suggesting that chronic hepatitis delta, in contrast with many other viral infections, has a limited imprinting effect on the CD56dim NK cell population. Instead, these findings are in line with previous observations in PBMCs from subjects with chronic recurrent herpes simplex virus-2 infection, where no effect on NK cell differentiation was seen.39 One possible explanation for the stable NK cell differentiation in PBMCs of patients with chronic hepatitis delta is that the virus exerts only local effects in the liver.

We performed a unique longitudinal assessment of NK cell differentiation during treatment with IFNα. From previous studies, it is evident that the NK cell differentiation status remains stable over time in healthy individuals.20 In contrast, CMV re-activation causes NK cells to undergo a rapid increase associated with a gradual terminal differentiation.38 Here, treatment of chronic hepatitis delta infection with IFNα was associated with alterations in the NK cell compartment, with higher frequencies of CD56bright NK cells and dominance of relatively immature CD56dim NK cells. Our previous studies involving transfer of NK cell subsets at discrete stages of differentiation into humanised mice suggested that NK cell differentiation is an irreversible process.29 Thus, the redistribution observed in the patients with hepatitis delta might instead result from selective proliferation of certain subsets.29 Indeed, immature CD56dim NK cells are known to proliferate more abundantly after IL-2, IL-12, IL-15 and IL-18 stimulation than are terminally differentiated CD56dim NK cells.29 ,36 ,40 NK cells express the IFNα/β receptor and readily respond to IFNα stimulation with activation.41 Plausibly, distinct NK cell subsets might respond differently to IFNα stimulation.29 ,36 However, in vitro, after long-term (up to 3 weeks) stimulation of NK cells from healthy individuals with IFNα, we observed no preferential increase in immature or terminal NK cells (data not shown). These findings were corroborated by the observation that immature and terminal NK cells from healthy individuals responded equally well with enhanced functions after short-term (overnight) IFNα stimulation. Therefore, the redistribution of CD56dim NK cell differentiation in these patients after treatment evidently was not a direct consequence of altered responsiveness to IFNα in different NK cell subsets. We were not able to compare, side-by-side, HBV/HDV co-infected with HBV mono-infected patients during treatment with IFNα. Thus, the difference in IFNα response of HBV/HDV- versus HBV-infected patients cannot be elucidated.

Apart from the shift within the CD56dim NK cell subset with the disappearance of terminal NK cells, we observed a concurrent increase in CD56bright NK cells from 3% to 9% of the total NK cells during treatment. Strikingly, some donors with chronic hepatitis delta had CD56bright NK cells representing up to 30% of NK cells and 5% of total lymphocytes after 12 weeks of treatment. This increase in CD56bright NK cells reiterates the NK cell responses to peg-IFNα treatment in patients with chronic HCV infection.14 ,42 Furthermore, patients with acute HCV infection had more CD56bright NK cells than healthy individuals.43 However, since IFNα does not cause an increase in specific NK cell subsets, including CD56bright NK cells, it is still unclear why the CD56bright NK cells are enriched in numbers.

Expression of CD16, NKp46 and Siglec-7 was not significantly altered when healthy controls and patients with HBV or HDV infection were compared. However, alterations in these receptors appeared during therapy, and similar patterns were also observed for NKG2D and NKp30. This is in line with the observed effect of IFNα therapy on HBV infection in NK cells.15 The transcription factor, Eomes, increased slightly in CD56dim NK cells during therapy, although this change may be too small to be considered biologically relevant.44 TRAIL expression on NK cells increased after IFNα therapy, particularly in the CD56bright subset, mirroring results previously reported for HBV15 and HCV.41

NK cells are dependent on IL-15 induced STAT4 signalling for their development.45 IL-15 also plays an important role in the transition from CD56bright to CD56dim NK cells and can drive CD56dim NK cell differentiation.29 ,46 Our knowledge on IL-15 regulation during chronic viral hepatitis infections is limited. Two reports have described elevated levels of IL-15 in patients with either chronic HBV or HCV infection.15 ,47 IFNα can also act through the STAT4 pathway, inducing the production of IFNγ,48 while simultaneously acting via STAT1 to stimulate the cytotoxic potential of NK cells by inducing perforin.49 NK cells have, relative to other immune cells, higher basal levels of STAT4 and may be predisposed to respond through the STAT4 pathway after IFNα stimulation.48 However, IFNα stimulation of NK cells alters the balance between STAT4 and STAT1, with the latter becoming the dominant signalling pathway.48 We observed elevated levels of STAT1 in NK cells from HDV-infected patients, with peg-IFNα therapy even causing a further increase in STAT1 levels, an effect also previously observed in HCV infection.16 Furthermore, we noticed that, although the total level of STAT1 was increased in chronic HDV infection and subsequent treatment, pSTAT1 was increased only after therapy. Prolonged treatment with IFNα does not appear to affect the ability to phosphorylate STAT1, but does reduce the induction of STAT4 in response to IFNα, thus shifting the balance between these two pathways. This is in line with previous findings from patients with hepatitis C as well as from murine models.16 ,48 A possible consequence of this shift might be that NK cells become less capable of differentiating in response to IL-15. Thus, the long-term effects of STAT1 polarisation might be the accumulation of CD56bright and immature CD56dim NK cells, as indeed was observed in the patients with treated chronic hepatitis delta as well as in various studies investigating NK cells during IFN therapy of hepatitis C.9 This shift might also account for the greatly decreased IFNγ and TNFα induction that we observed in the functional experiments.

In conclusion, we have performed the first comprehensive characterisation of NK cell dynamics during chronic hepatitis delta and treatment thereof. We report that chronic untreated hepatitis delta, as well as prolonged antiviral treatment with peg-IFNα, has a significant effect on NK cell frequencies, differentiation status, and function, and switches intracellular signalling from STAT4 to STAT1. Furthermore, high levels of CD56dim NK cells at baseline were positively associated with treatment outcome. Given the severity of chronic hepatitis delta compared with other chronic viral hepatitis infections, more studies on the immunopathogenesis of this disease are clearly warranted.

Acknowledgments

The authors thank all study nurses of the Department of Gastroenterology, Hepatology and Endocrinology of Hannover Medical School for support in collecting patient samples, in particular Mrs J Kirschner, Mrs J Schneider, Mrs L Sollik and Mrs C Mix. We also thank Mr H Schlums and Ms L Liu for technical assistance.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • SL and DFGM contributed equally to this work.

  • Contributors SL, DFGM: performed experiments and analysis and cowrote paper. JG, KP, VB, BB, K-JM, MPM: administrative, technical, clinical or material support. JKS, MC, H-GL: study review and analysis. HW: study supervision, design and analysis, and cowrote paper. NKB: study supervision, design and analysis, and cowrote paper.

  • Funding This work was funded by the International Research Training Group 1273 supported by the German Research Foundation (DFG), the Federal Ministry of Education and Research (BMBF) (grant 01Kl0788 to HW and MC), the Deutsches Zentrum für Infektionsforschung, the Swedish Research Council, the Swedish Cancer Society, the Karolinska Institutet, the Swedish Society for Medical Research, the Stockholm County Council, the Alex and Eva Wallström Foundation, the Swedish Society of Medicine, the Jeansson Foundation, the Bengt Ihre Foundation, the Groschinsky Foundation, the Hedlunds Foundation, Mag-tarmfonden, and the Julin Foundation.

  • Competing interests None

  • Ethics approval The study protocol conforms with the 1975 Declaration of Helsinki, and the ethics committee of Hannover Medical School reviewed the project.

  • Patient consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.