Evasion of natural killer cells by influenza virus

Review discusses mechanisms for influenza virus evasion from NK cell surveillance.

Abstract

NK cells are important innate immune effectors during influenza virus infection. However, the influenza virus seems able to use several tactics to counter NK cell recognition for immune evasion. In this review, we will summarize and discuss recent advances regarding the understanding of NK cell evasion mechanisms manipulated by the influenza virus to facilitate its rapid replication inside the respiratory epithelial cells.

Introduction

Influenza virus is classified into types A, B, and C. Among these, influenza A remains an important threat to global health and thus, is the focus of clinical diagnosis, treatment, and basic research [1]. It is a single-stranded, segmented, enveloped RNA virus that belongs to the Orthomyxoviridae family. The mature virus is composed of 7 types of internal proteins: nucleoprotein, 3 polymerase proteins (PA, PB1, and PB2), 2 matrix proteins (M1 and M2), a small amount of NS2 or nuclear export protein, and 2 surface glycoproteins, HA and NA, which are important targets for neutralization antibody recognition and therapeutic drug treatments [2, 3]. In addition, the NS1 protein is translated in the nucleus during active infection, and it is a powerful weapon for the virus to interfere with early host cell responses [4].

Seasonal influenza causes ∼36,000 deaths annually in the United States alone [5]. The rapid and emergent global spread of the novel swine origin influenza virus A/H1N1 prompted the World Health Organization to raise the pandemic alert to the highest level, officially declaring that this novel, triple-reassortant virus had caused a pandemic [6]. Since its outbreak in Mexico in April 2009, many thousands of patients with influenza-like illness and associated hospitalizations have been reported, and increasing cases of deaths were announced as a result of this novel virus infection [7]. Additionally, there are still concerns about the potential for new pandemics arising from avian influenza virus strains such as those of the H5N1 subtype [8].

Prevention of seasonal influenza relies heavily on vaccines that must be formulated annually. Unfortunately, vaccinations might not provide satisfying or any protection if vaccine strains selected based on prediction do not match the circulating strains that predominate later on [9]. In addition, for the newly emerged influenza virus, it will take more than 1 week for antigen-specific B and T cell responses to form and be effective, even if there is no interference of original antigenic sin, a phenomenon observed in influenza and other infections [10]. Compared with the rapid virus replication that peaks within a few days after infection, acquired immune responses to new influenza viruses are relatively delayed. Anti-influenza innate immunity might be critical for the initial control of viral replication.

Innate immune cells, including mast cells, basophiles, macrophages, neutrophils, DCs, NK cells, NKT cells, and γδ T cells, are vital components of the innate immune system. Among those cells, macrophages, neutrophils, and DCs have been shown to clear influenza virus via phagocytosis or promotion of adaptive responses [11–19]. NK cell response to influenza infection has also been studied, and its critical involvement in the early control of influenza replication and the potential mechanisms of NK anti-influenza effector functions have been investigated extensively [20–32]. However, recent work from us and others has found that the influenza virus has developed a series of strategies to evade NK cell-mediated innate control of infection. In this review, we will focus on the potential NK evasion mechanisms used by influenza virus.

NK CELL ANTI-INFLUENZA ACTIVATION RECEPTORS

NK cells are traditionally viewed as crucial effector cells of innate immunity for defending against various infections caused by bacteria, parasites, and viruses. Like other lymphocytes, NK cells are widely distributed throughout the body and can be found in lymphoid and nonlymphoid tissues. Natural cytotoxicity and cytokine/chemokine production are 2 primary effector functions of NK cells. Both of these functions are regulated precisely through NK cell activation and inhibitory receptors such as NKG2D activation receptor and Ly49 inhibitory receptor [33, 34]. The interaction with the corresponding ligands ultimately results in signaling transductions into the nucleus that converge into the decision of whether to execute NK functions.

NCRs are another set of important NK cell triggering receptors, which include NKp46 or NCR1, NKp44 or NCR2, and NKp30 or NCR3 [35]. These receptors are Ig-like transmembrane glycoproteins, and their transmembrane regions contain positively charged amino acids allowing association with the ITAM-bearing polypeptides CD3ζ and FcRγ for NKp46 and NKp30 or KARAP/DAP12 for NKp44 [35]. Among these receptors, expression of NKp30 or NKp44 is only observed in humans and nonhuman primates [36, 37], but NKp46 is functionally expressed in several mammalian species, including human [37], macaques [36], cow [38], mouse, and rat [39, 40]. NCRs are known to be involved in the recognition of tumor targets, but the natural ligands recognized by the NCRs remain elusive with the exception of NKp30 [41, 42]. In addition to its interaction with tumor ligands, NCRs can recognize viral proteins. NKp46 has been shown to be able to interact with the HA of influenza and the HA-NA of Paramyxovirus, including Sendai virus and Newcastle disease virus [20, 43]. Besides NKp46, NKp44 can also functionally interact with HA of different influenza subtypes [24, 29, 44], although the in vivo relevance of NKp44 and HA interaction during influenza infection has not been addressed. Unlike NKp46 and NKp44, NKp30 cannot bind to influenza HA [44], but it can interact with pp65 of human CMV [45] and is crucial for NK cell-mediated clearance of other viruses [46, 47].

NK CELL-MEDIATED CONTROL OF INFLUENZA INFECTION

It has been observed that shortly after influenza virus infection, NK cells exhibit increased killing activity, suggesting a potential role of NK cells in the innate defense against influenza [22]. In vivo NK cell depletion studies using antiasialo GM1 or NK1.1 antibodies confirmed that NK cells can contribute to the control of influenza infection [21, 48, 49]. Recently, NKp46 or NCR1 has been identified as a novel mouse NK cell marker [50]. Compared with NK1.1 or DX5, which has been used frequently for defining mouse NK cells in the past, NKp46 is expressed almost exclusively on NK cells in all mouse strains tested [50]. As aforementioned, NKp46 can recognize influenza HA protein, and this interaction is mediated through the sialic acid contained on the NKp46 molecule itself [23, 29]. Interaction of NKp46 with viral HA can lead to enhanced cytotoxicity against infected targets and cytokine production by activated NK cells [20]. These biological effects of the NKp46 and HA interaction can be blocked specifically by anti-NKp46 antibody or anti-HA serum treatment [20]. To test further the physiological significance of the NKp46 and HA interaction, mice with nonfunctional NCR1 (Ncr1^gfp/gfp) were generated [23]. When challenged with influenza PR8 virus, a much higher mortality was observed in the NCR1-deficient mice, clearly indicating a crucial role of NKp46 in NK cell-mediated control of influenza infection.

Recent analyses have found that 3 branched O-glycan sequences on the NKp46 glycosylation site, which carries Thr225, are important for the recognition of influenza H1 subtypes [31]. However, because of the structural variations among HA subtypes, the property and even the consequence of the interaction of NKp46 with different HA subtypes can vary. This possibility has been confirmed by the most recent observation that the interaction site for avian H5 protein used by NKp46 is not the same as that used for H1 protein recognition [32]. More interestingly, unlike its interaction with H1 protein, NKp46 recognition of H5 protein could not initiate direct killing of infected targets unless additional activation signals were present. Nevertheless, when mice with NCR1 deficiency were challenged with H1 or H5 virus, high virus titer for H1 and H5 viruses was detected, and little or no virus was found in WT control mice [32], further suggesting NKp46-mediated cytotoxicity contributes to NK cell anti-influenza responses. In summary, NK cells can fight against influenza virus infection, primarily through NKp46-mediated recognition, and the interaction of NKp46 with HA is probably more complicated than we thought previously.

EVASION OF NK ACTIVITY DURING INFLUENZA INFECTION

Although NK cells have the demonstrated ability to control influenza virus infection, influenza virus seems able to replicate efficiently and release new virions into the respiratory tract rapidly. Why is it that NK cells cannot control influenza infection completely? One possibility is that initial propagation of influenza virus happens so quickly that it leads to a robust production of virus that overwhelms the limited number of surrounding NK cells. Indeed, during the first few days after infection, the absolute number of NK cells in the lung is low, and robust infiltration of NK cells together with other immune cells does not occur until Day 5 after infection [51–53]. It is known that influenza virus can produce new viral proteins within a few hours after infection, and the viral titer in the lung peaks within 3 days [54]. Thus, compared with the rapid viral synthesis and assembly, the recruitment of NK cells into the lung is relatively slow, providing sufficient time for influenza virus to complete its life cycle. It is also possible that influenza virus has developed strategies to counteract NK cell recognition. Below are detailed discussions of these potential evasion mechanisms that have been documented in the literature.

ESCAPE OF NK RECOGNITION BY MUTATION OF VIRAL HA

Although the interaction of NKp46 with influenza HA is an important control strategy for NK cells to recognize and kill influenza-infected targets, it is well known that influenza virus is able to change viral HA and NA rapidly and frequently to avoid antibody neutralization [55]. Is it possible that mutation of HA would have an effect on the NKp46-mediated signaling? A study from Owen et al. [56] provides a good illustration. It was found that compared with old H3N2 influenza isolates (1969–1996), recent H3N2 isolates (1999–2003) displayed lower SA binding affinity because of the modified glycosylation on HA protein. Further, when target cells were infected with these recent isolates for NK cell sensitization, NK killing was weaker than that against infection with old isolates.

It is an established concept that NK cell natural recognition and killing are regulated by signals from NK inhibitory receptors that interact with MHC class I on the targets. However, the difference in the killing of targets infected with recent and old H3N2 isolates was not a result of differential expression of the MHC class I molecule on infected targets, as the expression of MHC class I was not altered after infection with new or old virus isolates, which confirmed other published results [57]. Analysis of HA protein sequences of the recent H3N2 viruses identified 2 new glycosylation sites that are absent in the old strains, and genetic deletion of 1 of the 2 glycosylation sites resulted in similar HA surface expression compared with the parent virus but improved NK cell killing, indicating that the glycosylation modification on HA alone can alter NK cell recognition of influenza-infected targets. Thus, frequent HA mutation is an important strategy for influenza virus to avoid not only antibody neutralization but also NK cell recognition. It remains critical to determine whether this observation applies to other HA serotypes.

REGULATION OF NK RECOGNITION BY THE LEVEL OF VIRAL HA

NKp46 on NK cells and HA on influenza virus-infected targets need to interact with each other to initiate NK cell-mediated recognition and elimination of the infected targets. As influenza infection or vaccination does not seem to alter the expression of NKp46 on NK cells [58, 59], although increased levels of NKp46 can enhance its binding with HA [29], the consequence of the interaction of NKp46 with HA is likely to be influenced by the level of viral HA expressed on the infected target cell surface or on the surface of the free virions encountered. Indeed, recent in vitro studies have demonstrated that the amount of HA input affects the binding of NKp46 with HA [29]. This is also true for the binding of NKp44 to HA [24]. In a range of HA concentrations from 0.075 to 0.3 ug/ml, NKp46 binding to HA increased in a linear manner [29]. NKp44 binding to HA also increased stepwise, as the coated HA amount went up from 200 ng to 400 ng, 800 ng, and 1600 ng [24]. Unfortunately, these studies have not shown whether increased binding of the NCR correlates with enhanced NK functions.

After infection, the synthesized influenza HA protein is predominately delivered into the apical surface of the infected lung epithelial cells and concentrated in the membrane lipid raft to mediate virion assembly and fusion with the cell membrane for virus release through the respiratory tract into the air [60, 61]. Replication of influenza virus usually peaks within 2–3 days after infection and declines dramatically after Day 5 [54]. Thus, it is possible that NK cells are exposed to sufficient amounts of HAs and consequently activated for cytolytic function shortly after infection. This possibility was suggested by studies in the early 1980s [22, 62], which found that the cytotoxicity of lung NK cells against YAC-1 tumor target cells increased within 48–72 h after PR8 virus infection; meanwhile, spleen NK cell cytotoxicity was not triggered. After Day 6 of influenza infection, NK cell cytotoxicity became undetectable [62]. The peaked NK cytotoxicity, between Days 2 and 3 after influenza infection, was confirmed later by Liu et al. [28]. Thus, as influenza virus replication slows down, and the surface HA on the epithelium or HA on the viruses becomes less abundant or unavailable, the latter infiltrated NK cells might not be triggered adequately to perform their cytolytic function.

Although, NK cytotoxicity parallels influenza replication, overexposure to viral particles or HA can lead to impaired NK cell function. Ali et al. [63] reported that preincubation with a high concentration of HA (from 1 ug/ml to 10 ug/ml) inhibited NK cell killing significantly. Interestingly, this inhibition also seemed to be HA dose-dependent. At 10 ug/ml, NK cytotoxicity against target cells was blocked almost completely [63]. Treatment with IFN-α, a strong NK cell stimulator, could not reverse the inhibitory effect, suggesting that a high concentration of HA could mediate strong inhibitory signals against NK cell activation. The HA dose-dependent inhibition of NK cell cytotoxicity has also been reported recently by Mao et al. [64], who demonstrated further the mechanism of the influenza HA or virion-mediated negative effect on NK cell cytotoxicity. Mao et al. [64] observed that after HA was added to NK cell culture, it was internalized into NK cells via the sialic acids, resulting in down-regulated expression of the CD3 ζ chain, which is coupled with NKp46 for transmission of NKp46-mediated signaling. The decreased expression of the ζ chain impaired Syk and ERK phosphorylation and finally caused the reduced, redirected cytotoxicity of NK cells induced by the anti-NKp46 antibody [64]. A similar observation was found by Du et al. [65] in their recent study, which showed that NK cell activation by pseudotyped particles expressing influenza HA and NA is viral dose-dependent, yet increasing the viral HA titer attenuated NK cell activation and effector functions further.

Taken together, these studies show that the anti-influenza responses mediated by NK cells are regulated by influenza HA protein. Although NK cytotoxicity reaches its maximum level at the peak of HA expression or virion production, this cytotoxicity can be a compromised response, as NK cells at this time-point can be overwhelmingly exposed to a high level of HAs, which are rapidly generated and accumulated within 2–3 days after infection. In addition to influenza HA, NK cell functional suppression mediated by other viral proteins has been reported, for example, HIV Nef [66] and Tat [67] and hepatitis C virus E2 [68], indicating a universal strategy used by viruses to escape NK cell recognition.

ESCAPE OF NK RECOGNITION BY REORGANIZING MHC I

NK cell function is balanced by signals from activating and inhibitory receptors [33, 69]. NK cell-activating receptors recognize their ligands, which are usually not present in the normal self cells but rather, in targets under infection or tumorigenesis. This recognition and interaction trigger signals leading to NK cell cytotoxicity and cytokine generation. NK cells also express inhibitory receptors that interact with specific MHC class I molecules to initiate signals that inhibit NK activation. The absence or reduction of MHC I expression can release the suppression and thus, trigger NK cell activation. This is known as “miss-self” recognition, originally proposed by Kärre et al. [70]. Several virus infections can down-regulate MHC I, but surprisingly, in vitro data show that influenza virus infections do not alter the MHC I expression on the infected targets [71, 72]. However, after influenza virus infection, the MHC I proteins can be redistributed into cell-surface lipid raft microdomains so that their interaction with inhibitory receptors is enhanced, making NK cell killing against influenza-infected targets less effective [71]. Thus, influenza virus can inhibit NK cell function through relocalizing MHC class I molecules without affecting their expression level.

ESCAPE OF NK RECOGNITION BY INFECTING NK CELLS

One of the viral immune-evasion tactics is to infect and kill immune cells, including NK cells. A few viruses, for example, HIV, herpes simplex virus, and Ectromelia virus, have been demonstrated to possess this ability to infect NK cells [73–75]. Recently, we and others [76, 77] have found that influenza virus can also inhibit NK cell functions by infecting NK cells. Mao et al. [77] shows influenza virus A/Hong Kong/54/98 (H1N1) can enter human NK cells in vitro through clathrin- and caveolin-dependent endocytosis. Although the virus replicates inefficiently, the abortive infection induced marked NK cell apoptosis, resulting in reduced NK cytotoxicity [77]. Clinical investigations have found that the number of peripheral NK cells in influenza-infected patients was reduced or not even detectable in severely infected patients [58, 78]. Although these studies did not provide detailed explanations for their observations, it is possible that the reduction or absence of NK cells is associated with direct infection of NK cells but not simply a result of their migration into respiratory tissues, as in the infected lung, NK cells could not be detected [78].

Recently, we also reported that mouse NK cells are susceptible to influenza virus infection [76]. We demonstrated that freshly isolated and in vitro IL-2-expanded NK cells express α-2,3 and α-2,6 sialic acids, known receptors for influenza HA recognition and initiation of infection. Viral M2 protein could be detected after influenza PR8 infection on cultured mouse NK cells. This was consistent with the data of Mao et al. [77], in which high levels of M gene expression in human NK cells after exposure to influenza virus were detected by quantitative PCR methods. Similar to human NK cells, mouse NK cells do not support productive viral propagation. In addition to the in vitro infectivity study, we showed that influenza virus can infect NK cells in vivo. Anti-M2 protein staining on lung tissue sections was strongly detected at Day 4 after infection when virus titer was at its maximum level. M2 protein staining on the infected lung tissues colocalized with NKp46, a recently identified NK cell-specific marker [50], and lysosome-associated membrane protein 1, a marker for NK cell lytic granules [79], suggesting influenza virus can enter and infect NK cells in vivo. In fact, M2 protein can be detected in lung NK cells for up to 10 days after infection, despite the decline of staining intensity after Day 4 of infection. We demonstrated further that influenza infection of mouse NK cells can also induce apoptosis and inhibit NK cell effector functions, including reduced cytotoxicity against different target tumor cells and impaired ability to generate cytokines and chemokines upon different stimulations. The ability of influenza virus to infect mouse NK cells in vivo has also been confirmed by a recent report using influenza GFP reporter virus [80]. In brief, these studies provide clear evidence that influenza virus can affect NK function by infecting NK cells to avoid innate attack.

CONCLUDING REMARKS

Influenza virus is well known for its continuous antigenic evolution through antigenic drift and shift that leads to the escape of the existing memory responses induced by natural infections or vaccinations. The most recent outbreak of the 2009 swine-like H1N1 influenza pandemic is just another illustration of how a new influenza virus emerges and threatens human beings [5]. Therefore, influenza virus remains one of the most important viral pathogens for human health and a big challenge for influenza researchers [54].

In contrast to substantial characterization and mechanistic studies of adaptive immune responses against influenza infection, the involvement of innate cellular immunity is just beginning to be appreciated. Among the innate effectors, NK cells can contribute, directly and indirectly, to controlling influenza infection. NCRs play an important role in the initiation of recognition of influenza-infected targets, but it is unlikely that they are the only NK activation receptors involved in anti-influenza responses. For example, the NKG2D receptor has also been suggested to contribute to the killing of influenza-infected targets in vitro and in vivo [26, 32]. Furthermore, other pathways in which NK cells can be activated during infections have been demonstrated, such as cross-talk between NK cells and other innate immune cells, including macrophages and DCs [81, 82]. Infection of these cells by influenza virus and the subsequent activation on NK have also been studied using coculture assays [26, 27], but the in vivo relevance of these findings has not been investigated, which represents an important direction for understanding the mechanism of NK cell activation after influenza infection.

Although NK cell activation is important for controlling influenza infection, unfortunately, the influenza virus has also developed several strategies for evading NK cell clearance, which, however, further highlights the vital role of NK cells in fighting against influenza. In addition to the influenza virus-mediated suppression of NK cell function, several other factors can affect NK cell functions during influenza infection and vaccination. These include differential genetic expression, aging, nutrition, and immune status [49, 51, 83–88]. Thus, how to reverse the immune suppression of NK cells effectively and boost NK cell activity are crucial for fully using NK cells for prevention and treatment of emerging and severe influenza infection.

AUTHORSHIP

H.G. initiated the work, searched the literature, and wrote the paper. P.K. and S.M. revised the paper. S.M. provided mentorship.