Eur J Clin Invest 2010; 40 (9): 851–863
Background Hepatitis B virus (HBV) and hepatitis C virus (HCV) are major human hepatotropic pathogens responsible for a large number of chronic infections worldwide. Their persistence is thought to result from inefficiencies of innate and adaptive immune responses; however, very little information is available on the former. Natural killer (NK) cells are a major component of innate immunity and their activity is tightly regulated by several inhibitory and activating receptors.
Design In this review, we examine controversial findings regarding the role of NK cells in the pathogenesis of acute and chronic liver disease caused by HCV and HBV.
Results Recent studies built up on technical advances to identify NK receptors and their functional correlates in this setting. While NK cells seem to behave correctly during acute hepatitis, it would appear that the NK cytotoxic potential is generally conserved in chronic hepatitis, if not increased in the case of HCV. In contrast, their ability to secrete antiviral cytokines such as interferon ex vivo or after cytokine stimulation is severely impaired.
Conclusions Current evidence suggests the existence of an NK cell functional dichotomy, which may contribute to virus persistence, while maintaining low-level chronic liver inflammation. The study of liver-infiltrating NK cells is still at the very beginning, but it is likely that it will shed more light on the role of this simple and at the same time complex innate immune cell in liver disease.
Hepatitis B virus (HBV) and hepatitis C virus (HCV) are responsible for a large number of chronic infections worldwide, and their control and eradication are considered as major public health challenges of the 21st century. A chronic infection occurs in approximately 70% of individuals exposed to HCV, whereas HBV persistence is directly related to age, being maximal at birth and negligible in adult life in immunocompetent persons . A variable proportion of patients with chronic viral hepatitis B and C subsequently develop severe complications including cirrhosis, liver failure and hepatocellular carcinoma . The pathogenetic mechanisms responsible for liver disease progression are still poorly understood, although much has been learned during the last few years, thanks to the development of animal models of infection and liver disease and the discovery of new molecular tools to finely dissect the capacity of the immune response to induce both tissue injury and viral clearance.
Similarly to all viruses, HBV and HCV are obligatory intracellular pathogens that depend upon the host cell synthetic machinery for their replication and growth. Resolution of acute viral hepatitis is thought to depend upon a complex interplay between innate and adaptive immunity. The latter is fundamental for complete control and eradication of viral infections with antibodies being responsible for neutralisation of circulating virus and T cells being responsible for the elimination of intracellular virus. This requires not only the immune destruction of infected cells by cytotoxic T lymphocytes (CTL), but also the non-cytolytic elimination of the virus from infected cells mediated by the effect of antiviral cytokines that inhibit viral gene expression and replication, thereby curing infections without destroying the infected organ or killing the host [3,4]. Innate immune cells and molecules play a fundamental role in controlling pathogen invasion and replication during the early infectious process, while priming and regulating adaptive immune responses later in the course of acute infections and in persistently infected hosts. Innate immune cells such as stromal cells, monocytes, neutrophils, natural killer (NK) cells and dendritic cells are rapidly activated by a wide array of pattern recognition receptors (PRRs), such as Toll-like receptors (TLRs), or the intracellular nuclear oligomerisation domain (NODs)-like receptors and RIG-I-like RNA helicases (RLH/RLRs) . All three classes of PRRs are relevant to viruses. TLRs identify infectious signals derived by molecular patterns common to all pathogens (pathogen-associated molecular patterns or PAMP), such as lipopolysaccharide, bacterial DNA or viral RNA; NOD recognise both bacteria and virus PAMP, whereas RLRs are thought to recognise only PAMP expressed by viruses . Following pathogen interaction with PRRs, complex and distinct signalling pathways are generated that ultimately lead to the activation of interferon-stimulated genes, which represent the very first cellular response to microbes and, indeed, type I interferons are generally involved in the initial containment of viral infections. In general, hepatocytes make no exception to this rule when exposed to hepatotropic viruses; however, HBV and HCV behave differently. Thus, studies on early immune defence mechanisms in the chimpanzee model showed that HBV infection does not induce appreciable changes in innate immune response genes in the first weeks of infection, even though the paradigm of HBV being a stealth virus has recently been challenged in vitro  and in vivo . HCV does induce interferon (IFN)α/β-response genes , but this is counteracted by the ability of HCV to interfere with IFN signalling pathways [10–13], effectively thwarting cellular responses. The behaviour of innate cytokines in the early phase of HCV and, particularly in HBV, infections contrasts with that observed in acute HIV-1 infection, in which a real cytokine storm takes place peaking 6 days from onset [14,15]. Activation of type I IFN genes is followed by the secretion of cytokines by dendritic cells (IL12) and stromal cells (IL15), which are essential for the activation and expansion of natural killer (NK) cells .
NK: a smart, non-educated cell programmed to respond to danger signals
NK cells represent the principal effector cell population involved in innate immune responses to viral infections and are particularly enriched in the liver where they may account for about 40 to 60% of the intrahepatic (IH) lymphocyte pool compared to the peripheral blood (PB) compartment, where they represent 5-15% of the total lymphocyte population [16,17]. Morphologically, they are defined as large granular lymphocytes, being rich in perforin/granzyme granules, and exert their antiviral activity through a direct cytotoxic effect, which destroys virus-infected target cells or through the production of immunoregulatory cytokines (IFN-γ, TNF-α, TGF-β, GM-CSF, IL-10, etc.) and may influence adaptive immune responses as well as directly inhibiting virus replication. NK cells can be easily identified by flow cytometry by the expression of CD56 in the absence of typical markers of T-cell and B-cell lineages. The majority of CD56+ NK cells also express the FcγIII receptor CD16, which confers the capacity to mediate antibody-dependent cell-mediated cytotoxicity. NK cells can be further divided in two subsets on the basis of intensity of expression of the CD56 molecule. Thus, CD56dim cells represent the majority of circulating NK cells and are considered developmentally mature, whereas CD56bright cells are the minority and thought to be at an earlier stage of maturation [18–20], although viral infections may occasionally alter these proportions . Moreover, CD56dim cells have been shown to predominantly mediate cytotoxicity, whereas CD56bright cells appear to principally secrete cytokines [22–24]. This subdivision of tasks is not as rigid in humans as it is often in mice  and can be significantly influenced by persistent viral infection . Contrary to T and B cells that display sophisticated rearranged antigen-specific receptors, NK cell receptors (NKRs) are in germ-line configuration. Moreover, NK effector function is controlled by a complex network of signals, which interact with membrane-expressed inhibitory and activating receptors . The latter allows the recognition of altered self via binding to ligands, which are expressed by stressed cells, effectively functioning as danger signals . It is generally believed that NK cells do not require priming, which would confer early protection against neoplastic transformation and intracellular pathogens. However, recent work suggests that prior exposure to pathogens significantly shorten the time required for activation upon rechallenge , casting doubts on the current dogma that NK cells do not have memory.
The nuts and bolts of NK cell function: activating and inhibitory receptors
As mentioned above, NK function is tightly regulated by activating and inhibitory receptors, which bind to a variety of ligands, some being classical MHC class I (HLA-B and HLA-C in man) or non-classical (HLA-E) molecules, as well as other molecules, such as MHC class I-like stress-induced self ligands (MICA/B) and the UL16 binding proteins (ULBPs) 1-5 recognised by NKG2D in humans [30,31]. A list of activating and inhibitory NKRs and their ligands is provided in Table 1. The critical role of inhibition in maintaining immunological homeostasis has been emphasised in a landmark study, which proposed the ‘missing self’ hypothesis to explain why healthy cells normally expressing MHC class are not killed by NK cells . Down-regulation of MHC class I or loss of its expression during viral infection or carcinogenesis lifts the inhibitory signal to NK cells that can be activated by danger signals. Among inhibitory receptors, killer inhibitory receptors (KIRs) have been the topic of intense investigations over the last few years, as KIR-ligand interactions may influence the outcome of viral infections, including HCV. Indeed, preferential expression of the inhibitory receptor KIR2DL3 in association with HLA C1C1 has been reported in patients with a self-limited outcome of acute HCV infection characterised by a low-viral burden . Interestingly, donor-recipient KIR-HLA-C mismatching has also been associated with recurrent hepatitis C after liver transplantation and faster fibrosis progression, but only when KIR2DL3 was present . KIR-HLA class I interaction and its relative avidity are increasingly generating interest, because the ligands HLA-B and C are polymorphic, which would pave the way to future studies aimed at predicting infection outcome on the basis of HLA-B and HLA-C alleles. KIRs also play an important role in the NK maturation process, as it has been shown that self MHC class I-reactive KIRs are required to ‘license’ NK cells to exert their functions . According to this concept, in the absence of licensing, NK cells are hyporesponsive. However, recent evidence indicates that ‘unlicensed’ NK cells are important in controlling mouse cytomegalovirus infection in vivo , challenging the licensing hypothesis and suggesting that unlicensed NK cells are critical for protection against viral infections. Other important NK inhibitory receptors are members of the C-type lectin-like receptor family. Each heterodimer consists of an invariant CD94 molecule covalently associated with a molecule of the NKG2 group, some of which are activating and some inhibitory (Table 1). NKG2A, a prominent member of this group, binds to the non-classical MHC class I molecule HLA-E which, unlike classical HLA-A, B, C, is relatively non-polymorphic and, therefore, its inhibitory function is less dependent upon correct receptor-ligand match.
|Natural cytotoxicity receptors (NCR)||NKp30||Activating||BAT-3, pp65|
|C-type lectin-like receptors||CD94/NKG2C||Activating||HLA-E|
|CD2 family||CD244 (2B4)||Activating||CD48|
|Receptors for nectin or nectin-like molecules||CD226 (DNAM-1)||Activating||Nectin-2, CD155|
|Killer cell immunoglobulin-like receptor (KIR) family||2DS1||Activating||Group 2 HLA-C|
|2DS2||Activating||Group 1 HLA-C|
|2DL1||Inhibitory||Group 2 HLA-C|
|2DL2/3||Inhibitory||Group 1 HLA-C|
|Leukocyte immunoglobulin-like receptor (LILR) family||LILRB1/ILT2/LIR1||Inhibitory||HLA-A, -B, -C|
|Sialic acid binding||Siglec-7||Inhibitory||Sialic acid|
|Ig-like lectin family||Siglec-9||Inhibitory||Sialic acid|
|Killer cell lectin-||KLRG-1||Inhibitory||Cadherins|
|like receptor family||KLRF-1 (NKp80)||Activating||Unknown|
|β1 integrins||α4β1 integrins||Activating||VCAM-1, Fibronectin|
|β2 integrins||LFA-1 (CD11a/CD18)||Activating||ICAM1, ICAM2|
|CR3 (CD11b/CD18)||Activating||LPS, ICAM1, Fibrinogen, β-glucan|
|CR4||Activating||LPS, ICAM1, Fibrinogen, CD23|
It is currently believed that NK activation occurs when a critical threshold of activating signal exceeds inhibition. The major activating receptors lack an intracellular signalling domain and transduce signals via adaptor molecules, such as DAP10 for NKG2D  and DAP12,CD3ζ and FcRIγ for natural cytotoxicity receptors (NCRs) [38–40]. NKG2D is a prominent member of this group, also belonging to the C-type lectin-like family of receptors, which unlike other NKG2 receptors does not associate with CD94. NKG2D binds to ligands (MICA/B, ULBPs) that are poorly expressed on healthy cells, but they are up-regulated by stress, viral infection or DNA damage . Up-regulation of these ligands may tip the balance of NK cells from inhibition to activation (‘induced self’ recognition). Other important activating receptors are the NCRs NKp30, NKp44 and NKp46, which are exclusively expressed by NK cells and for which no cellular ligands have been formally identified, although there is evidence that NKp44 and NKp46 may bind to viral haemagglutinins [42,43] and that NKp30 recognises the cytomegalovirus tegument protein pp65 . Moreover, human leukocyte antigen-B-associated transcript 3 (BAT3), an intracellular protein involved in apoptosis and proliferation, has also been identified as ligand for NKp30 .
Normal or dysfunctional NK cells in viral hepatitis?
What is the role of NK cells in viral hepatitis? Several years ago when very little was known about NK cell biology and pathobiology and when its function was simply defined by the ability to kill tumour targets, we showed that PBMC from patients in the early phase of acute HBV infection displayed significantly enhanced NK cell cytotoxic activity for the susceptible K562 cell line, returning to normal in the convalescent phase . In contrast, no NK cell cytotoxicity was detected in acute non-A, non-B hepatitis, as it was then defined a rather heterogeneous group of patients, which also included hepatitis C. Cytotoxicity correlated with serum levels of IFN-α, providing evidence in favour of a role for type I IFN in the early phase of infection. The approach was then limited to acute hepatitis, as it was hypothesised that NK activity was confined to early acute infection, while an efficient cytotoxic T cell (CTL) response was thought to be instrumental in viral clearance or in maintaining low level chronic inflammation during persistent infection. This assumption proved to be too restrictive, as it became clear that NK plays a central role in the crosstalk between innate and adaptive immune responses. Much has been learned since those early days, and NK cells have become fashionable again when their sophisticated regulatory system had been unravelled. In more recent years, several studies looked at phenotypic and functional features in a more systematic way taking advantage of the several monoclonal antibodies (mAb) that specifically identify the large number of NKRs. Very few studies have looked at phenotypic and functional features of NK cells during acute viral hepatitis. An early rise in circulating NK cells has been documented in the incubation phase of HBV infection, suggesting that they may contribute to the initial viral containment in this setting . More recently, exciting data suggested that innate cytokines such as IFNα, IL15 and IFNλ1 remained barely detectable throughout the observation period in acute hepatitis B, and NK production of IFNγ was also defective particularly at peak viraemia . This was due to a burst of the immune suppressive IL10 production in the early stages of HBV infection and would be compatible with the deficient type I IFN response documented in the chimpanzee model of acute HBV infection . These findings are somehow at variance with those of others who showed a transient increase in NK cytolytic activity and IFNγ production associated with increased expression of CD69+/NKG2D+ NK cells, soon after peak HBV viraemia in two subjects followed prospectively during an atypical acute HBV infection with persistently normal serum ALT levels .
Studies in acute hepatitis C are extremely scarce due to the difficulty in collecting patients with ‘true’ acute HCV infection, which is very often asymptomatic and is sometimes confused with reactivation of chronic hepatitis C . A recent prospective study has been published, which reaffirmed the importance of the KIR2DL3-expressing NK cells that produced a higher amount of IFNγ and lysed target cells more efficiently compared to total NK cells, although no clear differences were noted between patients who resolved and those who went on to develop persistent infection . This study also emphasised a possible role for the NKG2D activating receptor, as it was up-regulated in this setting and it has been convincingly shown to play a role also in chronic HCV infection [26,52].
Owing to the widespread availability of patients with chronic hepatitis and because of the interest to broaden our understanding of the role of NK cells in this context, several studies focused in this area. A virtually complete list of publications on PB and IH NK cells in chronic HCV and HBV infections is reported in Tables 2 and 3, respectively. Despite the availability of standardised reagents, in the case of chronic HCV infection, simple questions such as quantifying the number of circulating NK cells, examining their phenotype and correlating those parameters to NK cell function yielded diverging data in many cases, with some studies suggesting that reduced NK cell frequencies did not affect spontaneous or cytokine-induced cytolytic effector function [21,26,53–56], while others showed instead deficient NK cytolytic activity [57–61]. The reasons for such discrepancies are not immediately apparent, although they may be due, in part, to clinical, biochemical and virological heterogeneity of patients, for which some investigators attempted to control by performing an extended phenotypic and functional analysis in a substantial number of unselected patients and healthy donors. Moreover, it has been shown that even under physiological conditions, there are considerable donor variations in the magnitude and the kinetics of NK cell responses that may also be affected by different stimulation protocols, which may significantly influence the data obtained in different laboratories . Unfortunately, most studies are biased by a small sample size. In our own comprehensive study involving a sizeable number of patients with chronic HCV and chronic HBV infections , we have shown significant differences in NK cell phenotype and function between chronic HBV and HCV infections. In the latter, there was an increased proportion of NKG2D-expressing NK cell and a concomitant decrease in those expressing inhibitory receptors, supporting the concept of a phenotype skewed toward activation, predominantly involving the NKG2D pathway. Instead, in the former, there was an increased proportion of NKG2C-expressing cells with normal inhibitory receptor expression, suggesting a more balanced activating/inhibitory receptor expression. Differences in NKR expression were mirrored by clear functional differences. Thus, NK cells from HCV positive patients responded well to cytokine stimulation displaying normal or increased cytolytic activity, at variance with data suggesting that HCV-envelope proteins impair NK cell activation and function [57,58] and consistent with recently published findings indicating that culture-derived infectious HCV particles do not affect NK function . This apparent discrepancy is probably due to the fact that the former studies were performed using recombinant E2 proteins, while the latter used HCV virions derived from supernatants of cell cultures productively infected with HCV [64–67]. A recently published study apparently solved this issue by showing that inhibition of NK secretion of IFNγ was only possible with immobilised, but not with soluble, culture-derived virus . Additional experiments clearly showed that persistent HCV infection is able to enhance NK cytotoxicity mainly through the NKG2D pathway and that NCRs, such as NKp30 and NKp46, were also involved despite apparently not being expressed at levels higher than healthy donors. Instead, patients with chronic HBV infection displayed normal or even lower cytotoxic potential upon cytokine stimulation compared with patients with chronic HCV infection and healthy subjects. However, there was a major functional NK cell defect in both settings, which was more pronounced in chronic HBV infection. Indeed, cytokine (IFNγ and TNFα) production following cytokine stimulation of PB NK cells was significantly reduced compared to healthy donors, suggesting the existence of a functional dichotomy, featuring a conserved or enhanced cytolytic activity and a reduced cytokine production. The mechanisms responsible for defective NK function have not been clarified yet, but recent evidence suggests that a co-inhibitory molecule, the T cell immunoglobulin- and mucin-domain-containing molecule-3 (Tim-3), plays a significant role in the setting of chronic HBV infection , and similar events may occur during chronic HCV infection.
|Circulating NK||Intrahepatic NK||Ref. no|
|Ahlenstiel G, 2010||↑ NKp44, NKp46, NKG2C, NKG2A, TRAIL, CD122 vs. with HD|
HCV n = 42, HD n = 12
|↑ Cytotoxic activity in HCV patients with elevated ALT vs. HCV patients with low ALT|
↓ IFNγ production vs. HD
HCV n = 15, HD n = 12
|↑ NKp46,TRAIL, CD122 vs. PB|
HCV n = 10
|Dessouki O, 2010||↓ NK cell percentage|
↓ perforin vs. with HD
HCV n = 29, HD n = 22
|↑ Cytotoxic activity|
↓ IFNγ production HCV n = 18, HD n = 21
|Bonorino P, 2009||↓ NK cell percentage|
↑ CD56bright and ↓CD56dim vs. HD
HCV n = 28, HD n = 18
|N.D.||↓ NK vs. PB|
↓CD56bright and ↑CD56dim
↑ perforin vs. hepatitis B virus (HBV) HCV n = 11, HBV n = 6
|Oliviero B, 2009||↓ NK cell percentage|
↑ NKG2D vs. HD
HCV n = 35, HD n = 30, HBV n = 22
|↑ Cytotoxic activity vs. HD and HBV|
↓ IFNγ and TNFα production vs. HD
|↑ NKG2D vs. PB HCV n = 16||N.D.||26|
|Yamagiwa S, 2008||N.D.||N.D.||↓ NK cell percentage and CD161 vs. HD|
HCV n = 21, HD n = 10
|De Maria A, 2007||↑ NKp30, NKp46|
HCV n = 15, HD n = 12
|IFNγ and IL10 production comparable to HD|
HCV n = 12, HD n = 12
|Nattermann J, 2006||↓ NKp30, NKp46|
HCV n = 30, HD n = 10
|↓ NCRs mediated target cell killing|
HCV n = 5, HD n = 3
|↓ NKp30, NKp46 vs. non-HCV related hepatic diseases|
HCV n = 5, HD n = 4
|Morishima C, 2006||↓ NK cell percentage vs. HD|
HCV n = 42, HD n = 26
|Not impaired NK cytolytic activity|
HCV n = 29, HD n = 24
|Nattermann J, 2005||N.D.||N.D.||↑ NKG2A vs. with HD|
HCV n = 3, HD n = 3
|Meier UC, 2005||↓ NK cell percentage vs. HD|
↓ CD56 dim
↓ CD56dim perforinhigh
HCV n = 36, HD n = 31
|↓ Cytotoxic activity vs. HD|
HCV n = 2, HD n = 2
|NK cell percentage comparable to PB compartment|
HCV n = 5, HD n = 5
|Kawarabayashi N, 2000||N.D.||N.D.||↓ NK cell percentage in cirrhotic pts|
HCV+ n = 12
HCV− n = 12
HCV+ cirrhosis n = 6
|↓ Cytotoxic activity|
↓IFNγ production in cirrhotic pts
|Corado J, 1997|| NK cell percentage vs. HD|
HCV n = 11, HD n = 18
|↓ Cytotoxic activity vs. HD|
HCV n = 10, HD n = 10
|Clinical setting||Circulating NK||Intrahepatic NK||Ref. no|
|Ju Y, 2010||CHB||↑ Tim-3 expression|
CHB n = 40, HD n = 18
|↑cytotoxicity by Tim-3 blocking|
CHB n = 6
|↑ Tim-3 expression|
CHB n = 16, HD n = 6
|Zou Y, 2010||ALCF and CHB||↑ NKp30 and NKp46 in ALCF|
ALCF n = 14, CHB n = 15
|N.D.||↑ NK percentage ALCF n = 20, CHB n = 10||N.D.||84|
|Bonorino P, 2009||CHB||↓ NKG2A vs. hepatitis C virus (HCV) and HD;|
↓CD158a,h vs. HCV CHB n = 19, HD n = 18, HCV n = 28
|N.D.||↑CD56bright vs. PB;|
↑CD158 a,h CD158 b,j vs. HCV CHB n = 6, HD n = 6
|Zou Z, 2009||ACLF, CHB, LC||↓NK cell percentage in ACLF and in LC|
ALCF n = 75, CHB n = 31, LC n = 36
|N.D.||↑ NK cell percentage in ACLF|
ALCF n = 9
|Oliviero B, 2009||CHB||↓NK cell percentage vs. HD;|
↑NKG2C vs. HD and HCV;
↓ CD69 vs. HCV;
↓ KIR3DL1/DS1 CHB n = 22, HD n = 30, HCV n = 35
|↓ cytotoxicity vs. HD for K562 after cytokines stimulation;|
↑ cytotoxicity vs. HD for P815 in redirecting experiments;
↓ cytotoxicity vs. HCV in redirecting experiments after cytokines stimulation;
↓IFNγ and TNFα production vs. HD after cytokines stimulation.
|Dunn C, 2007||CHB||↑ TRAIL concurrent with hepatic flares CHB n = 72, HD n = 13||NK induce TRAIL-mediate hepatocyte apoptosis after stimulation.||↑CD69 vs. PB|
↑ TRAIL at flares CHB n = 15, HD n = 4
|Yan MX, 2006||CHB||↓NK cell percentage|
CHB n = 54, HD n = 14
|Sprengers D, 2006||CHB immune-tolerance vs. immune-clearance phase||No differences CHB n = 47||N.D.||↑NK cell percentage in immune-tolerance vs. immune-clearance CHB n = 47||N.D.||87|
|Natterman J, 2006||HBV||↓ NKG2A vs. HCV|
↑ NKp30 and NKp46 vs. HCV
CHB n = 10, HD n = 9, HCV n = 30
|Duan XZ, 2004||CHB and LC||↓ NK cell percentage CHB n = 80, HD n = 25, LC n = 27||N.D.||N.D.||N.D.||88|
As the antiviral effect produced by cytokines is more efficient than single target cell lysis, the dysfunctional cytokine secretion shown here may be an important mechanism contributing to virus persistence. The fundamental importance of IFNγ in the control of viral infections has indeed been shown in several studies, which demonstrated it to be a powerful non-cytolytic mechanism of viral clearance from infected hepatocytes [3,70]. Consistent with our data, the functional NK cell defect described above for chronic hepatitis C has been interpreted as a consequence of chronic exposure to HCV-induced IFNα leading to chronic liver inflammation via cytotoxic mechanisms, but not to viral clearance because of insufficient IFNγ production .
Unlike patients with chronic hepatitis C, patients with chronic hepatitis B typically experience characteristic necroinflammatory flares that are associated with disease progression . In this context, some studies looked at changes in innate immune mechanisms during flares and found partially controversial data. Thus, in one study, there was a temporal relationship between ALT flares and increased circulating and IH-activated NK cells expressing tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) , whereas this was not confirmed in a subsequent study that failed to detect an increase in TRAIL-expressing NK cells, but instead showed an increase in soluble chemokines such as CXCL-9 and CXCL-10 .
Intrahepatic NK cells
The presence of IH NK cells in humans has been a matter of intense discussions over the past few years. Are there really resident lymphoid cells in the healthy adult human liver or are the mononuclear cells extracted from the liver simply originated from the blood flowing through the sinusoids? Whatever the answer, several studies in chronic HBV and HCV infections emphasised differences between the ‘IH’ and PB compartments [26,53,74]. Several potential problems afflict investigators wishing to concentrate in this area of organ immunology. First, only a very limited number of cells can be retrieved from a small fragment of percutaneous liver biopsy, which severely restricts the number of experiments that can be performed. Second, it is unethical to obtain liver biopsies from entire healthy controls and, therefore, alternative solutions must be sought for. That being said, the vast majority of studies omitted to address IH NK cell function and concentrated on phenotypic differences with PB NK cells. In general, a larger proportion of IH NK cells express activation molecules and TRAIL compared with the PB compartment, and this led many to advocate it as a proof of a pathogenetic role for NK in liver necroinflammation [53,72].
However, the vast majority of studies in humans lack functional evaluation of IH NK cells (Tables 2 and 3) and, therefore, it is impossible to know whether phenotypic changes actually mirror alterations in IH NK cell cytolytic potential or cytokine production. Preliminary data from our laboratory would suggest that, using otherwise healthy subjects who agreed to donate a liver tissue fragment during laparoscopic cholecystectomy as controls, IH NK cells showed reduced cytotoxic capacity compared to controls with apparently conserved IFNγ production . The cytotoxic defect was not associated with statistically significant differences in the expression of NKRs and/or activation molecules, although there was a trend towards an increase in NKp46+ NK cells in the liver. It is difficult at this stage to draw firm conclusions on the significance of these findings, although it may be that reduced cytotoxic capacity is caused by exhaustion of IH NK cells after encounter with target cells or to anergy induced by overexposure of activating NK receptors (e.g., NKG2D) to their ligands, as shown in other systems [76,77]. A cartoon illustrating the hypothetical mechanism of IH NK cell hyporesponsiveness is shown in Fig. 1.
Can functional NK cell defects in viral hepatitis be corrected by antiviral therapy?
It is clear from the aforementioned that persistent hepatotropic virus infection modifies the phenotype and function of NK cells. Unfortunately, virtually, no data are available on the effect of viral load reduction as a consequence of antiviral treatment. One very recent study confirmed the impaired IFNγ production in patients with chronic HBV infection and suggested that this could be restored following successful response to antiviral treatment . Interestingly, expression of the NKG2A inhibitory receptor was up-regulated in these patients and the proportion of NKG2A positive NK cells correlated with serum HBV DNA levels, suggesting that exposure to HBV may result in severe functional inhibition. Similar observations were made in patients with chronic HCV infection, in whom treatment with pegylated IFNα and ribavirin increased the proportion of total circulating NK cells and up-regulated expression of NKG2D and perforin, but most importantly restored IFNγ production in sustained virological responders only .
Conclusions and outlook
NK cells, and in general innate immune responses, have for long been considered minor players in the pathogenesis of acute and chronic hepatitis B and hepatitis C, as opposed to adaptive immunity. However, the work discussed above has clearly identified some recent peculiar aspects which are worth considering. HBV is not a potent inducer of innate immune response, but at the same time allows adaptive immunity to develop progressively and in a more coordinated manner, which would result in eventual control of the infection. HCV, instead, almost immediately replicates at a very high level inducing good innate responses, which however are frustrated by the ability of this virus to interfere with IFNα signalling. The overwhelming replicative velocity of this virus literally outpaces adaptive immunity, which is very often ineffective in controlling infection. However, innate immunity still plays an important role, as self-limiting outcome appears to be associated with certain KIR-HLA matches  and specific IL28B polymorphisms , although this assumption would require formal proof. There is still limited information on the role of NK cells in acute viral hepatitis, although they seem to behave as expected in many viral infections. Although the data are still controversial, several lines of recent evidence suggest that once chronic HBV and chronic HCV infections are established, NK cells are partially dysfunctional as their cytotoxic potential appears to be conserved, whereas IFNγ secretion, an important non-cytolytic mechanism of virus control, is impaired. This would depict a scenario characterised by a continuous low level liver necroinflammation and a concomitant inability to eradicate the virus.
Several points remain to be addressed, and it is difficult to prioritise which of these are most cogently needed. Thus, for instance, despite correlations with liver enzyme elevations and NK cytotoxicity have sometimes been shown , there is no conclusive evidence supporting their role in liver disease progression. Furthermore, the fine mechanisms regulating the balance between activating and inhibitory receptors and their ligands during the course of viral hepatitis, as well as the role of antiviral therapy in reversing the alterations induced by persistent hepatitis virus infection are still unknown. An additional important function of NK cells is the crosstalk between innate and adaptive immunity, and here future research should concentrate to identify potential targets for immunotherapeutic interventions.
Molecular Infectious Disease Unit, Department of Infectious Diseases, Fondazione IRCCS Policlinico San Matteo and University of Pavia, Pavia, Italy (Mario U. Mondelli, Stefania Varchetta, Oliviero Barbara)