CMV drives clonal expansion of NKG2C+ NK cells expressing self-specific KIRs in chronic hepatitis patients

Authors

  • Vivien Béziat,

    1. INSERM UMR-S 945, Hôpital Pitié Salpêtrière, Paris, France
    2. Université Pierre et Marie Curie, Paris, Paris, France
    3. Center for Infectious Medicine, Department of Medicine, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm, Sweden
    Search for more papers by this author
  • Olav Dalgard,

    1. Service d'Hépatologie, Centre de Recherche INSERM CRB3, Hôpital Beaujon, Clichy, France
    2. Medical Department, Rikshospitalet University Hospital, Oslo, Norway
    Search for more papers by this author
  • Tarik Asselah,

    1. Service d'Hépatologie, Centre de Recherche INSERM CRB3, Hôpital Beaujon, Clichy, France
    Search for more papers by this author
  • Philippe Halfon,

    1. Laboratoire ALPHABIO, Hôpital Ambroise Paré, Marseille, France
    Search for more papers by this author
  • Pierre Bedossa,

    1. Service d'Anatomie Pathologique, Centre de Recherche INSERM CRB3, Hôpital Beaujon, Clichy, France
    Search for more papers by this author
  • Ali Boudifa,

    1. Laboratoire d'Immunologie Cellulaire et Tissulaire, Paris, France
    Search for more papers by this author
  • Baptiste Hervier,

    1. INSERM UMR-S 945, Hôpital Pitié Salpêtrière, Paris, France
    2. Université Pierre et Marie Curie, Paris, Paris, France
    Search for more papers by this author
  • Ioannis Theodorou,

    1. Université Pierre et Marie Curie, Paris, Paris, France
    2. Laboratoire d'Immunologie Cellulaire et Tissulaire, Paris, France
    Search for more papers by this author
  • Michelle Martinot,

    1. Service d'Hépatologie, Centre de Recherche INSERM CRB3, Hôpital Beaujon, Clichy, France
    Search for more papers by this author
  • Patrice Debré,

    1. INSERM UMR-S 945, Hôpital Pitié Salpêtrière, Paris, France
    2. Université Pierre et Marie Curie, Paris, Paris, France
    3. Laboratoire d'Immunologie Cellulaire et Tissulaire, Paris, France
    Search for more papers by this author
  • Niklas K. Björkström,

    1. Center for Infectious Medicine, Department of Medicine, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm, Sweden
    Search for more papers by this author
  • Karl-Johan Malmberg,

    1. Center for Infectious Medicine, Department of Medicine, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm, Sweden
    Search for more papers by this author
  • Patrick Marcellin,

    1. Service d'Hépatologie, Centre de Recherche INSERM CRB3, Hôpital Beaujon, Clichy, France
    Search for more papers by this author
    • These authors are senior co-authors.

  • Vincent Vieillard

    Corresponding author
    1. INSERM UMR-S 945, Hôpital Pitié Salpêtrière, Paris, France
    2. Université Pierre et Marie Curie, Paris, Paris, France
    • INSERM UMR-S 945, Laboratoire Immunité et Infection, Hôpital Pitié-Salpêtrière, Paris, France Fax: +33-142177490===

    Search for more papers by this author
    • These authors are senior co-authors.


Abstract

Natural killer (NK) cells are affected by infection with human cytomegalovirus (HCMV) manifested by increased expression of the HLA-E binding activating receptor NKG2C. We here show that HCMV seropositivity was associated with a profound expansion of NKG2C+CD56dim NK cells in patients with chronic hepatitis B virus (HBV) or hepatitis C virus (HCV) infection. Multi-color flow cytometry revealed that the expanded NKG2C+CD56dim NK cells displayed a highly differentiated phenotype, expressed high amounts of granzyme B and exhibited polyfunctional responses (CD107a, IFN-γ, and TNF-α) to stimulation with antibody-coated as well as HLA-E expressing target cells but not when stimulated with IL-12/IL-18. More importantly, NKG2C+CD56dim NK cells had a clonal expression pattern of inhibitory killer cell immunoglobulin-like receptors (KIRs) specific for self-HLA class I molecules, with predominant usage of KIR2DL2/3. KIR engagement dampened NKG2C-mediated activation suggesting that such biased expression of self-specific KIRs may preserve self-tolerance and limit immune-pathology during viral infection. Together, these findings shed new light on how the human NK-cell compartment adjusts to HCMV infection resulting in clonal expansion and differentiation of educated and polyfunctional NK cells.

Introduction

Natural killer (NK) cells have the ability to kill targets without prior sensitization and their involvement in antiviral and antitumor immunity is well established 1, 2. Recent studies have demonstrated a high degree of functional heterogeneity in the NK-cell compartment attributable to a vast network of inhibitory or activating receptors that allow these cells to recognize target cells 3, 4. Killer cell immunoglobulin-like receptors (KIR) and CD94/NKG2 heterodimers are two major types of HLA class I binding receptors that regulate NK cell function 5, 6. Both these receptor-families exist in activating and inhibitory forms and contribute to the functional education of human NK cells by interactions with their cognate ligands 7, whereas KIR are expressed in a stochastic manner with a variegated distribution in the NK cell population 8, 9, NKG2A is expressed on all CD56bright NK cells and disappears gradually during differentiation of CD56dim NK cells 10, 11.

NKG2C and NKG2A are covalently associated with CD94 12. Both NKG2A and NKG2C specifically interact with the non-classical MHC class-Ib molecule HLA-E, which is expressed at low levels on almost all nucleated cells, and presents peptides derived from signal sequences of other HLA class-I molecules 13. The affinity of their interaction depends on the sequence of the HLA-E-bound nonamers and is higher for NKG2A than for NKG2C 14, 15. In the CD56dim subset, NKG2C expression largely excludes NKG2A expression 10, 16. Expression of NKG2C is induced by co-culture with HCMV-infected fibroblasts and correlates with HCMV seropositivity in healthy donors 16, 17. Recently, NKG2C+ NK cells were shown to expand during HIV and hantavirus infections in HCMV-seropositive patients, suggesting that HCMV may prime the NK-cell compartment for specific expansion of the NKG2C+ subset upon additional viral encounters 18, 19.

Two recent papers have demonstrated increased expression of NKG2C on NK cells in patients with chronic HBV and HCV infection 20, 21. Therefore, we choose this clinical setting to perform an in-depth characterization of the NKG2C+ NK-cell subset. We show that NKG2C+CD56dim NK cells are terminally differentiated, highly polyfunctional and display a clonal expression of inhibitory KIRs with specificity for self-HLA class-I molecules. Although such biased expression of self-specific receptors confers functional education, it may also serve to dampen autoreactivity and tissue damage during chronic viral infection.

Results

NKG2C expression on CD56equation image NK cells from HBV and HCV patients is associated with HCMV seropositivity

We monitored the frequency of NKG2C+ NK cells in 32 patients with HBV infection and 36 with HCV infection during the chronic phase of their disease (Table 1 and Fig. 1A). Similar to the previous reports in patients with HIV and acute hantavirus infection 18, 19, the NKG2C expression level in this study was associated with HCMV seropositivity in patients with chronic HBV and HCV (Fig. 1B). Consistent with previous studies, expansion of NKG2C+ NK cells does not seem to occur in all HCMV seropositive individuals 16, 22, 23. The reason for this is unknown. We speculated that one possibility could be HCMV reactivation, since this has been reported to be common in patients with HCV and HBV 24, 25. However, using a highly sensitive PCR method, we could not detect any evidence for undergoing viral reactivation in blood or liver (data not shown). Interestingly, anti-HCMV IgG were found in 96 and 81% of HBV- and HCV-infected patients respectively. Given the median age (40–50 years) of the studied cohorts, a seropositivity of >80% is high compared with the prevalence of HCMV that has been reported for large cohorts of age-matched European populations 26, 27. One possible explanation for the unusually high frequency of HCMV seropositivity seen here is the diverse ethnicity in the studied cohorts. Furthermore, viral co-infections might be more common in risk groups, such as intravenous drug users, prone to acquire HBV and/or HCV 28.

Figure 1.

NKG2C+CD56dim NK-cell expansion during chronic HBV or HCV infections is linked to HCMV seropositivity. (A) Frequency of NKG2C+ cells within the CD56dim NK-cell population in HBV- (n=32) or HCV-infected (n=36) patients. (B) NKG2C expression on CD56dim NK cells in HCMV seronegative and seropositive HBV- (open triangles) or HCV- (closed triangles) infected patients. (C) Box-and-whiskers plots summarizing surface marker expression on NKG2C and NKG2C+ CD56dim NK cells from HBV and HCV-infected patients with high NKG2C expression. No difference in phenotype was noted when comparing HBV- and HCV-infected patients (n=20). High expression of NKG2C was defined as greater than 20%, corresponding to the mean±standard deviation of uninfected controls. (D) Representative example of co-staining for NKG2C and cytotoxic molecules (granzyme-A, -B, and -K, or perforin) in NK cells from one HBV-infected patient, representative for six investigated HBV- and HCV-infected patients. Mann–Whitney tests were performed for individual comparisons of two independent groups. Wilcoxon's tests were performed for individual comparisons of paired groups. *p<0.05, **p<0.01, ***p<0.001.

Table 1. Patient characteristics
 HCVHBV
  • a

    NA, not applicable.

  • a)

    a) Median value is shown when more than one assessment was available.

  • b)

    b) Inactive HBsAg carrier status: HBV DNA < 2000 IU/mL and persistently normal ALT. HbeAg-negative hepatitis: HBV DNA >2000 IU/mL and persistent or intermittent elevation in ALT/AST levels (HBeAg negative). Immunotolerant HBV infection: HBeAg positive and persistently normal ALT. Immunoactive HBV infection: HBeAg positive and elevated ALT.

  • c)

    c) When data were available.

Number3632
Male, number (%)26 (72)20 (63)
Age, year, median (range)51 (33–69)45 (19–64)
ALT, U/mL, median (range)a)97 (34–361)45 (24–462)
Classification of HBV infectionb)c) number (%)NA 
Immunotolerant phase 1 (4)
Immunoactive phase 4 (17)
Inactive carrier 12 (50)
HbeAg-negative hepatitis 7 (29)
Viral load, IU/mL, median (range)0.54×106 (0.00006–40×106)1.88×103 (1×101−1.7×108)
HCV viral genotypec) (%) NA
112 (40) 
23 (10) 
37 (23) 
48 (27) 
Number of biopsies with METAVIR score3620
Fibrosis score  
F004
F1912
F2103
F3101
F470
Activity score  
A003
A12713
A284
A310

In conclusion, our results suggest that HCMV is responsible for the expansion of NKG2C+ NK cells in patients with HCV and HBV. The relatively high frequency of HCMV seropositivity and associated expansion of NKG2C+ NK cells in the studied hepatitis virus-infected cohorts prompted us to examine this subset in more detail.

Phenotypic features of NKG2Cequation imageCD56equation image NK cells

Next, we set out to determine the phenotypic characteristics of NKG2C+CD56dim NK cells from the present patient cohort. Because our data suggested that expansion of NKG2C+ NK cells was dependent on HCMV infection, we choose to perform an aggregate analysis of NKG2C+ NK cells in patients with HCV and HBV. Significantly fewer NKG2C+ CD56dim NK cells expressed NKG2A, CD161, Siglec-9, and NKp30 compared with NKG2C CD56dim NK cells (Fig. 1C). In contrast, NKG2C+ NK cells more commonly expressed ILT2, CD57, and CD2. Percentages or MFI of CD62L, CD8, NKG2D, CD16, and DNAM-1-positive cells were indistinguishable when comparing NKG2C+ and NKG2C NK-cell subsets (Fig. 1C and Supporting Information 2). The expression pattern of cytolytic molecules in the granules of CD56dim NK cells revealed that both Granzyme A and perforin were expressed at equivalent levels in NKG2C+ and NKG2C NK cell subsets. In contrast, expression of Granzyme B was higher and Granzyme K lower in NKG2C+, compared with NKG2C NK cells (Fig. 1D). Importantly, the phenotype of NKG2C+ NK cells was identical in HCV- and HBV-infected individuals (data not shown). Together, these data show that NKG2C+ NK cells have a full cytolytic arsenal and a highly differentiated phenotype, as defined by the high expression of CD57.

NKG2Cequation imageCD56equation image NK cells are polyfunctional

To examine the functionality of the NK cells and its relation to their expression of NKG2C, we separated them into three subsets: NKG2A+NKG2C, NKG2ANKG2C, and NKG2ANKG2C+ NK cells. We simultaneously assessed these subsets in the presence of various target cells for multiple functional responses. NKG2C+NKG2A NK cells derived from patients with HBV or HCV infection displayed stronger and more diverse functional responses than NKG2C NK subsets following stimulation with targets expressing HLA-E, and against RAJI cells in the presence of anti-CD20 mAb (Fig. 2A). In agreement with the prominent role for NKG2A in NK cell education 8, 29, NKG2A+ NK cells responded better than NKG2A NK cells, regardless of their NKG2C expression, against both MHC class-I-negative K562 and 721.221 target cells. Furthermore, NKG2A+ NK cells produced high levels of IFN-γ in response to stimulation with IL-12/IL-18 (Fig. 2B), while IFN-γ production was almost undetectable in the NKG2C+CD56dim subset. Together, these results demonstrate that NKG2C+ NK cells display a functional profile similar to highly differentiated NK cells, shown to have a high responsiveness via ADCC but poor ability to respond to exogenous cytokines 30, 31. Extending previous results, we here show that differentiated NKG2C+ NK cells are polyfunctional and respond strongly to specific stimulation by HLA-E expressing target cells. Of note, NKG2C+ NK cells were also present in the liver (Supporting Information 3A). NKG2C+ NK cells in the liver were mostly NKG2A and responded to stimulation with HLA-E expressing 721.221 target cells but not against control 721.221 target cells lacking HLA-E (Supporting Information 3B and 3C).

Figure 2.

NKG2C+CD56dim NK cells are highly polyfunctional against cellular targets. (A) Polyfunctionality assays of NKG2A+, NKG2CNKG2A, and NKG2C+NKG2A CD56dim NK cell subsets from patients with chronic hepatitis (12 HBV and 7 HCV, n=19). Cell-surface expression of CD107a and intracellular expression of TNF-α and IFN-γ were assessed in the presence of K562 cells, RAJI cells + anti-CD20 mAb (Rituximab), and 721.221 cells that did (721.221-AEH) or did not (721.221) express HLA-E. FlowJo software and its ‘Boolean gate’ function was used to determine if cells expressed 0, 1, 2, or 3 functions simultaneously. Data are presented as pie charts created with Pestle and Spice software. Colored arcs represent the frequencies of cells producing CD107a (gray), TNF-α (blue) and IFN-γ (green). Pie fractions represent cells performing 0 (yellow), 1 (red), 2 (purple), or 3 (black) functions simultaneously. Pie comparisons were performed with the Wilcoxon signed-rank test of Spice software. *p<0.05, **p<0.01, ***p<0.001. (B) One representative experiment out of four showing co-staining for NKG2A, NKG2C, and IFN-γ in NK cells from one HBV- (top) and one HCV- (bottom) infected patient, after overnight stimulation with IL-12 and IL-18.

It has been suggested that NK cells may contribute to immunopathology during chronic hepatitis 20, 32. Both HBV and HCV appear to be involved in the modulation of HLA-E, the ligand of NKG2C, suggesting that NKG2C+ NK cells might target HLA-E expressing hepatocytes in the liver. Intriguingly, despite their cytolytic potential, we found no correlation between expansion of polyfunctional NKG2C+ CD56dim NK cells and clinical parameters including viral load and alanine transaminase (ALT) levels (Supporting Information 4).

The NKG2C+CD56equation image NK cell subset expresses inhibitory KIRs specific for self-HLA class-I molecules

KIR expression is a major event in the terminal differentiation of NK cells 10, 11. Figure 3A shows the KIR expression profile of NKG2C+ and NKG2C CD56dim NK cells in representative patients. In each patient, a fraction of the NKG2CCD56dim subset expressed KIR2DL1, KIR2DL2/DL3, KIR3DL1, and/or KIR2DS4 in agreement with a variegated distribution of KIRs. In contrast, NKG2C+CD56dim cells had a more restricted KIR expression pattern with a dominant expression of one or two inhibitory KIRs (Fig. 3A). For example, NKG2C+CD56dim cells from patient 2 exclusively expressed KIR2DL2/3, whereas those of patient 16 expressed mainly KIR3DL1. For still other patients, oligoclonal expression of KIR2DL1 and KIR2DL2/DL3 dominated the NKG2C+ NK cells, as exemplified for patient 3. KIR2DL2/DL3 was the most frequently expressed KIR (87% of donors) compared with KIR2DL1 (35%) and KIR3DL1 (30%), in NKG2C+ NK cells (Fig. 3B and Table 2).

Figure 3.

Analysis of KIR expression on NKG2C+CD56dim NK cells reveals a dominance of inhibitory self-KIRs. (A) Representative examples of KIR expression in NKG2C+ and NKG2C CD56dim NK cells from one HBV- (♯16) and two HCV- (♯2 and ♯3) infected patients. (B) Frequency of KIR selection in the NKG2C+ subset (n=23). Statistical analyses were performed using chi square tests. (C) KIR2DL1 (n=22), (D) KIR2DL2/DL3 (n=23), and (E) KIR3DL1 (n=23) expression in NKG2C+ NK cells from patients grouped based on presence or absence of the respective KIR ligands: HLA-C2, HLA-C1 and HLA-Bw4. Mann–Whitney tests were performed for individual comparisons of two independent groups. *p<0.05, **p<0.01, ***p<0.001.

Table 2. KIR clonal selection in NKG2C+CD56dim NK cells correlates with self-HLA class I
  1. a

    Note: Summary table of 10 HCV- and 13 HBV-infected patients with high NKG2C expression (greater than 20% of CD56dim NK cells). Numbers in brackets represent the ratio ((% KIR in NKG2C+)/(% KIR in NKG2C)). Ratios greater than 1 define selected KIR in the CD56dimNKG2C+ subset (colored red). Ratios <1 define non-selected KIR (colored green). HLA class I ligands of the selected KIR are colored in blue. Three exceptions with no correspondence between KIR2DL2/DL3 with this HLA-C ligand were found (colored orange). Nd, not determined.

original image

More importantly, the KIR expressed on NKG2C+CD56dim NK cells was in most cases specific for self-HLA class-I ligands (Table 2). Hence, KIR2DL1 and KIR2DL2/DL3 were significantly more expressed in the presence of two alleles of their respective ligands, HLA-C group 2 (HLA-C2) and group 1 (HLA-C1) (Fig. 3C and D). Further, KIR3DL1 expression in NKG2C+ NK cells was almost exclusively observed in donors displaying the cognate ligand, HLA-B group Bw4 (HLA-Bw4) (Fig. 3E). Intriguingly, three donors (4, 13 and 21) had NKG2C+ NK cells expressing KIR2DL2/DL3, although they were homozygous for HLA-C2 alleles. It is known, however, that KIR2DL2 has a low affinity for HLA-C2 33, 34. KIR genotyping of patients 4, 13, and 21 showed that all possessed the KIR2DL2 gene, suggesting that these too had dominant expression of self-specific receptors (Supporting Information 5). HLA-A typing for patient 16, who expressed KIR3DL1, but had no HLA-Bw4 alleles, showed an HLA-A*24 allele, which is also a ligand for KIR3DL1 34, 35. Taken together, these results unambiguously showed that clonally expanded NKG2C+CD56dim NK cells expressed a KIR that specifically recognized self-HLA class-I molecules.

Next, we examined the functional role of clonal KIR expression in the expanded NKG2C+ NK cells. In redirect killing assays using antibodies against NKG2C and KIR degranulation was induced in the presence of anti-NKG2C mAb, but was substantially decreased in the presence of antibodies that recognized the single KIR expressed by the NKG2C+ NK cells (Fig. 4). Taken together, these results reveal that NKG2C+ NK cells have a bias for expression of self-specific KIRs that may dampen their responses to normal tissues with intact HLA class I expression.

Figure 4.

Self-KIR engagement dampens NKG2C-mediated activation. Representative data from redirectional killing assays against P815 target cells from one HBV- (♯22) and one HCV- (♯7) infected patient, of which both had an expanded NKG2C+KIR+ NK cell subset. NK cells were treated with the isotype control mAb (Ig) or anti-NKG2C mAb in the presence or absence of the specific anti-KIR mAb. Shown plots are representative for three HBV- and two HCV-infected patients investigated.

Discussion

Two recent studies reported on the expansion of NKG2C+ NK cells in chronic HBV and HCV infection 20, 21. Our in-depth analysis of the expanded NKG2C+ CD56dim NK cells reveal that the presence of this subset was linked to HCMV seropositivity, but more importantly, that these cells had a highly differentiated phenotype, were polyfunctional and displayed a clonal or oligoclonal expression of inhibitory KIR specific for self-HLA class-I molecules. Intriguingly, the expansion of highly cytotoxic NKG2C+ NK cells in peripheral blood and in the liver of patients with HBV or HCV had no effect on the clinical outcome, suggesting that the biased expression of self-specific receptors may dampen potential autoreactivity and limit immunopathology.

We, and others, have recently described a process of NK-cell differentiation associated with a number of phenotypic and functional changes 10, 11, 31, 36, 37. In this context, NKG2C+ CD56dim NK cells in patients with HBV or HCV displayed a differentiated phenotype with lack of NKG2A and expression of KIR, ILT-2, and CD57. Furthermore, NKG2C+ NK cells expressed low levels of NCRs, CD161 and Siglec-9. Terminal differentiation of NK cells has also been associated with the loss of CD62L 10, 11, 36, 37. Here, highly differentiated NKG2C+ NK cells had a heterogeneous expression of CD62L that did not differ from the NKG2C subset. The terminal differentiation status of NKG2C+CD56dim NK cells is consistent with their inability to produce IFN-γ after IL12/IL18 stimulation, their high expression of perforin and granzyme, and their strong capacity to mediate ADCC 10, 11, 31, 36, 37. We recently hypothesized that the terminal stage of NKcell differentiation is linked to the ability to kill target cells expressing HLA-E 10. In line with this hypothesis, we here show that differentiated NKG2C+CD56dim NK cells, both in peripheral blood and in the liver, are polyfunctional against HLA-E expressing target cells.

To further characterize the expanded NKG2C+ NK cells, we performed an in-depth analysis of the inhibitory KIRs expressed by NKG2C+ NK cells. In contrast to the bulk NK cell KIR repertoire that display a random distribution of self and non-self inhibitory KIRs 8, NKG2C+CD56dim NK cells, in patients with HBV or HCV infection, had a clonal or oligoclonal KIR expression pattern with a striking bias for self-specific receptors. Only four exceptions were present among our 23 patients. Three of these exceptions could possibly be explained by the fact that KIR2DL2 is not exclusively specific for HLA-C group 1 33, and the fourth exception was explained by expression of KIR3DL1 in the presence of HLA-A*24, known to carry the Bw4 motif 34, 35. We have no definitive explanation for the finding that seemingly weak KIR-HLA interactions sometimes seem to dominate and select a certain KIR in favor of another KIR with a stronger motif. It is possible that it has to do with KIR polymorphisms and binding strength of specific KIR alleles to cognate HLA alleles. To date, we lack allele-level resolution of KIR-HLA interactions. Nevertheless, there are known examples in human and rhesus macaque where peptide modifications lead to altered specificity of KIRs and HLA molecules 35, 38–40. Among the studied receptors, the most commonly selected KIR was KIR2DL2/DL3, expressed at a higher frequency by NKG2C+ NK cells compared with NKG2C in 87% of the tested patients. Correspondingly, KIR2DL1 and KIR3DL1 were selected in 35 and 30% of the patients respectively. Hence, in line with recent results on five hantavirus-infected patients 19, our data from HBV- or HCV-infected patients with high NKG2C expression support the notion that NKG2C+CD56dim NK cells express self-specific receptors. Intriguingly, a recent study on NK-cell responses to acute CMV infection revealed no bias for expression of self-KIR on NKG2C+ NK-cells 41. In contrast, the authors suggested that there is a preferential expansion of NK cells lacking self-specific receptors because these are less restrained during onset of proliferation. This result aligns with their observations in a mouse model of CMV, showing that control of murine CMV is mediated by non-educated NK cells 41. Further studies are needed to explain the discrepancy between our two studies. One possible explanation might be that they did not assess KIR2DL2/DL3 expression, the most frequently selected KIR in our cohort.

The mechanism behind the expansion of NKG2C+ NK cells bearing self-specific KIR remains elusive. Given the evidence that NKG2C+ NK cells only expand in individuals positive for HCMV it is tempting to speculate that this virus, rather than HBV and HCV, is directly involved in triggering expansion and differentiation of NKG2C+ NK cells in patients with hepatitis virus infection. HCMV-infected cells express HLA-E but downregulate classical HLA class I 42, 43. In line with the rheostat model of NK-cell education 44, one may speculate that HCMV-induced loss of classical HLA class I with intact levels of HLA-E may shift the threshold for activation of NKG2C+ NK cells bearing self-specific inhibitory receptors. It is possible that non-self receptor expressing NKG2C+ NK cells are less capable of sensing dynamic changes in HLA class I induced by the virus, and, therefore do not respond with expansion. The need for persistent positive signals through ligand interactions appears crucial since education does not provide any proliferative advantage in response to cytokine stimulation alone 11. Instead, NKG2C+ NK cells do expand when stimulated by IL-15 in conjunction with HLA-E expressing target cells, supporting the notion that cellular interactions are involved in selecting the NKG2C+ repertoire 19.

Another unresolved question is whether the expanding cells retain their phenotype during expansion. In a setting of HCMV reactivation following solid organ transplantation, Lopez-Verges et al. recently described that NK cells change their phenotype and undergo differentiation during expansion as illustrated by the expression of CD57 23. Hence, although NKG2C and KIRs are likely expressed from the start it cannot be excluded that cells are further shaped during the immune response.

The comparison of NK-cell expansion with the clonal expansion of T cells is interesting and was recently reviewed 45. Although we have borrowed the term ‘clonal expansion’ from the expansion of T cells following antigen stimulation in the lymph node, there are several major differences between the two processes and we do not infer similar mechanisms. In fact, we cannot formally prove that cells have expanded clonally. It is possible that distinct NK cells, expressing the same advantageous KIR, expand in parallel. However, we favor the interpretation that there is a clonal expansion of NK cells having a particular setup of KIRs.

NKG2C expression in healthy donors has been detected only in relation to HCMV, but not EBV or HSV seropositivity 16, 46. Similarly, during both acute hantavirus and HIV-1 infection, NKG2C increases only in patients that are seropositive for HCMV 18, 19. Previous studies, reporting on the increase of NKG2C+ NK cells in chronic HBV or HCV infections, have not taken HCMV serostatus into account 20, 21. Here, we show that high NKG2C expression was associated with HCMV seropositivity also in these two chronic liver infections. Because of the unusually high frequency of HCMV positive in the studied cohorts, the role of HCV and HBV infection alone on NKG2C expression was somewhat difficult to evaluate. Nevertheless, none of the HCMV-negative hepatitis virus-infected patients (n=6) displayed significant levels of NKG2C+ NK cells suggesting that the expansion of this subset is dependent on HCMV.

Our data prompt for further studies to delineate the role of chronic HCV/HBV infection per se, on the expansion of NKG2C+ NK cells. It has been observed, both in vitro and in vivo, that hepatocytes are permissive for HCMV infection 47. Other studies suggest that chronic HBV and HCV infections might be associated with frequent HCMV reactivation in the liver 24, and that liver cirrhosis induced by HCV infection is associated with HCMV reactivation in peripheral blood 25. In the present study, quantitative PCR did not show HCMV reactivation in either peripheral blood or in the liver of HBV- or HCV-infected patients. Moreover, the frequencies of NKG2C+ NK cells in our cohorts does not seem to differ significantly from those of previous investigations in healthy controls 16, 18, 22. Hence, whether or not HCMV induced expansion of NKG2C+ NK cells was further driven by co-infection with HCV or HBV could not be examined here, since it would require comparison with a very large cohort of age-matched healthy donors of similar ethnicity. However, the high prevalence of HCMV seropositivity in hepatitis virus-infected patients and the associated expansion of NKGC+ NK cells highlight the relevance of studying NKG2C+ NK cells in this disease setting.

Supporting the predominant role of HCMV, we found no correlation between expansion of polyfunctional NKG2C+CD56dim NK cells and hepatitis-related clinical parameters including viral load and ALT levels and hepatic inflammation (Supporting Information 4 and 6). HBV may induce downmodulation of HLA class-I expression, including HLA-E, on cell lines transfected with HBV 48, 49 and on infected hepatocytes positive for hepatitis B core antigen (HBcAg) and surface antigen (HBsAg) 50. Conversely, chronic HCV infection is associated with a general increase in HLA class-I molecules, including HLA-E expression in the liver 51, 52. Engagement of inhibitory KIR dampened NKG2C-mediated activation of the expanded cells suggesting that the bias for self-specific receptors may serve to limit immune pathology during chronic infection, possibly explaining the weak correlation between expansion of NKG2C+ NK cells and clinical parameters. Supporting this hypothesis, we and others have recently shown that NKG2A was able to dampen the activity of NKG2C+ NK and γδ-T cells derived large granular lymphocyte leukemia thus preventing major deleterious side effects 53, 54.

In conclusion, we show that the NKG2C+CD56dim NK cell expansion, observed in the blood and in the liver of HBV- or HCV-infected patients, is dependent on infection with HCMV. The expanded NKG2C+ NK cells displayed a terminally differentiated phenotype with strong functional responses against HLA-E expressing targets and antibody-coated targets but not to IL-12/IL-18 stimulation. Interestingly, NKG2C+ NK cells had a clonal or oligoclonal expression of self-specific KIRs that blocked NKG2C-mediated activation, possibly explaining the limited immune pathology associated with the presence/expansion of this highly cytotoxic subset. Together, these findings shed new light on how the human NK-cell compartment adjust to HCMV infection resulting in clonal expansion and differentiation of polyfunctional NK cells expressing self-specific inhibitory KIR.

Materials and methods

Patients and sample collection

Consecutive patients scheduled for liver biopsy at Beaujon Hospital (Clichy, France) were asked to participate in the study. The local ethics committee approved the study, and all patients provided written and oral informed consent. Patients were included if they had chronic HBV or HCV infection, defined by HCV RNA or seropositivity for HBsAg for at least six months. HBV/HCV co-infected patients, patients on antiviral treatment, and previously liver transplanted patients were excluded. Blood samples from patients were collected with heparin tubes. All experiments were performed on fresh whole blood or fresh isolated peripheral blood mononuclear cells (PBMCs). Phenotypes were determined on whole blood and analyzed after fixation and erythrocyte lysis (BD FACS Lysing solution; Becton Dickinson). Functional assays were performed with fresh PBMCs isolated with a Ficoll gradient. A single experienced pathologist, blinded to the clinical and laboratory data, analyzed the liver biopsy specimens. Necroinflammation and fibrosis were assessed with the METAVIR score 55. Necroinflammation activity (A) was graded as A0 (absent), A1 (mild), A2 (moderate), or A3 (severe). Fibrosis stage (F) was scored as F0 (absent), F1 (portal fibrosis), F2 (portal fibrosis with few septa), F3 (septal fibrosis), and F4 (cirrhosis). Biopsy samples were collected in RPMI containing 10% FCS (Gibco) and antibiotics (Gibco) and stored at room temperature. Biopsy samples were passed through a 70-μm cell strainer (Falcon; Becton Dickinson) and used directly for functional assays or phenotyping.

HCMV serology

HCMV IgG serology was determined with Abbott ARCHITECT Anti-Cytomegalovirus IgG Assays (Abbott). Serology for HCMV was lacking for five patients in the HBV-infected group.

Flow cytometry and monoclonal antibodies

Cell-surface staining was performed with the appropriate combinations of the following antibodies: CD2-FITC, CD3-ECD, CD8-FITC, CD16-FITC, CD56-PC7, CD56-ECD, NKG2A-allophycocyanin (Z199), NKG2D-allophycocyanin, and NKp46-PE from Beckman Coulter; CD62L-allophycocyanin, CD94-FITC, CD161-FITC, ILT-2/CD85j-FITC, DNAM-1-FITC, and CD57-FITC from Becton Dickinson; KIR2DL1-allophycocyanin, KIR3DL1-allophycocyanin, KIR2DS4-allophycocyanin, Siglec-9-allophycocyanin, and NKG2C-PE from R&D systems, and KIR2DL2/DL3-allophycocyanin and NKp30-allophycocyanin from Miltenyi Biotec. For intracellular staining, whole blood cells were fixed and the erythrocytes lysed (BD cell lysing solution; Becton Dickinson); cells were then permeabilized in PBS supplemented with 0.5% BSA and 0.1% saponin, and stained with Granzyme-K-FITC from Santa Cruz, perforin-FITC Granzyme-A-FITC, and Granzyme-B-FITC from Becton Dickinson. Depending on the experiment, cells were acquired on a FACS Navios (Beckman Coulter) or a FACS Canto (Becton Dickinson). Flow cytometry data was analyzed using FlowJo software version 9.

HLA and KIR genotyping

Genomic DNA was isolated from whole-blood samples with the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). HLA-A, HLA-B and HLA-C alleles were then typed to the intermediate resolution level with standardized luminex assays (SSO Labtype; Ingen/One Lambda). When this resolution was not sufficient to determine whether the HLA-C was from group 1 or 2, HLA-C alleles were sequenced with the SBT kit (Aria Genetics). Sequences were read with a 3100 Genetic analyzer (Applied Biosystems), with computer-assisted Conexio genomics software. KIR was genotyped with the KIR typing kit (Miltenyi Biotec).

Intracellular analysis of IFN-γ production

Freshly isolated PBMCs were incubated for 16 h in the presence of 10 ng/mL IL-12 and 100 ng/mL IL-18 at 37°C. Cells were thereafter stained for cell-surface markers including CD3, CD56, NKG2A, and NKG2C, fixed (BD Cell Fix; Becton Dickinson), permeabilized (PBS supplemented with 0.5% BSA and 0.1% saponin), and then stained for intracellular IFN-γ expression and analyzed by flow cytometry.

Polyfunctionality assay

Polyfunctionality assays simultaneously detect several markers of NK-cell functionality after the NK cells encounter target cells, as previously described 56. Briefly, 5×105 freshly isolated PBMCs were incubated with 5×105 target cells at 37°C and 5% CO2 in the presence of anti-CD107a mAb to monitor degranulation. Assays were performed against MHC class-I-deficient K562, 721.221 target cells and.221-AEH, which express the HLA-E*0101 allele 57. ADCC assays were performed against the RAJI cell line in the presence or absence of 1 μg/mL of anti-CD20 (rituximab; Roche). After 1 h of incubation, Monensin (GolgiStop; Becton Dickinson) and brefeldin A (GolgiPlug; Becton Dickinson) were added, and the incubation continued for an additional five hours. Cells were then stained for cell-surface markers, fixed (BD Cell Fix; Becton Dickinson), permeabilized (PBS with 0.5% BSA and 0.1% saponin), and stained for intracellular IFN-γ (Alexa-Fluor-700; Becton Dickinson) and TNF-α (eFluor450, ebioscience) expression. Data were analyzed with Flow Jo version 9 (TreeStar) (Supporting Information 1). Pestle software was used to remove background and generate a file compatible with Spice software, as previously described 58.

Conventional degranulation and redirect killing assays

Redirected killing assays were performed against 5×105 P815 target cells to a 1:1 effector:target ratio. Cells were incubated at 37°C in the presence of anti-CD107a-FITC (Becton Dickinson) mAb, and anti-NKG2C-PE mAb. Blockade of inhibitory KIRs was performed by adding 5 μg/mL of the indicated anti-KIR mAbs or 5 μg/mL isotypic control (R&D systems). After one hour of incubation, 2 mM monensin was added, and the cells incubated for an additional three hours. Cells were then stained for extracellular antigens and analyzed by flow cytometry. Degranulation assays of NK cells from biopsies were performed, as previously described 10.

Statistical analysis

Mann–Whitney tests were performed for individual comparisons of two independent groups. Wilcoxon's tests were performed for individual comparisons of paired groups. Statistical analysis was performed with the Prism 5 software (GraphPad Software, San Diego, CA, USA). Comparisons of group of qualitative data were performed using chi square tests. Pie comparisons were performed with the Wilcoxon signed-rank test of Spice software 58. P-Values <0.05 were considered significant. *p<0.05; **p<0.01; ***p<0.001.

Acknowledgements

The authors thank Henri Thevenet, Sabine Canivet, Sylvie Jude and Brigitte Duprey for their technical assistance and Hans-Gustaf Ljunggren for critical review of the manuscript. V. B., V. V., T. A. and O. D. are responsible for the concept and designed the study. V. B. performed cellular experiments. V. B., V. V., O. D., P. M., P. D., and B. H. analyzed data. A. B. and I. T. performed HLA typings. P. H. determined CMV serostatus and viral load. O. D., T. A., M. M., P. B., and P. M. supplied clinical material. O. D., B. H., M. M., and P. B. collected clinical data. V. V. and P. M. supervised the study. V. B., V. V., K. J. M., N. K. B., O. D., P. M., and P. D. wrote the paper. This work was supported in part by the Institut National de la Recherche Médicale (INSERM), and by the Université Pierre et Marie Curie UPMC – Paris-6.

Conflict of interest: The authors declare no financial or commercial conflict of interest.

Ancillary