Putting the natural killer cell in its place


Dr Clair Gardiner, School of Biochemistry and Immunology, Trinity College Dublin, Dublin 2, Ireland.
Email: clair.gardiner@tcd.ie
Senior author: Dr Clair Gardiner


Natural killer (NK) cells were originally described as ‘null’ lymphocytes, but we have increasing evidence of their role in recognizing pathogen, and our knowledge of NK cell receptors continues to expand exponentially. Human NK cells have many receptors for human leucoctye antigen (HLA) class I. These killer immunoglobulin-like receptors (KIRs) and CD94/NKG2 receptors can signal in both positive and negative ways to regulate NK cell functions. The inhibitory receptors are the best characterized, but even in these cases much of their functional biology remains elusive. In this review, some recent advances in terms of the three-immunoglobulin (3Ig)-domain KIRs are discussed. Natural cytotoxicity receptors (NCRs) are among the activatory receptors found on NK cells. While pathogen ligands for these receptors have been described, endogenous ligands remain elusive. NCRs and NKG2D, a receptor for stress-induced antigens, appear to play complementary functional roles in terms of NK cell activation. More recently described on NK cells are the Toll-like receptors. In particular, these receptors of the innate immune system allow NK cells to directly sense pathogen, and their ligation on accessory cells indirectly activates NK cells through cytokine production. It is becoming clear that none of these receptor systems functions in isolation and that it is the sum of the signals (which will reflect the pathogenic situation), in addition to the cytokine milieu, that will direct NK cell activation. The resulting cytotoxicity, cytokine production and direct cell–cell regulatory interactions with other cells of the immune system, for example dendritic cells, ultimately determine the role of the NK cell in the context of an overall immune response.


Natural killer (NK) cells were first identified in the 1970s by their ability to kill cancer cells and virally infected cells without prior sensitization.1–3 As such, they were considered ‘non-specific’ lymphocytes. Furthermore, as they did not express cognate receptors generated by recombination,4,5 T and B lymphocytes triumphed and dominated the field of immunology for many years as their complex receptor systems and biology were elucidated. However, the discovery of NK cell receptor systems complex enough to satisfy even the most avid T-cell receptor zealot has refocused energy on these cells. Indeed, we no longer think of NK cells as ‘non-specific’, but rather as having very defined, and sometimes very subtle, specificities.

Cytotoxicity is perhaps the best characterized effector function of NK cells, and targets include tumour cells,6 virally infected cells,7,8 cells infected with intracellular bacterial pathogens9,10 and, as more recently reported, immature dendritic cells (iDCs).11–13 NK cells also produce a range of cytokines, including haematopoietic factors such as interleukin (IL)-3 and granulocyte–macrophage colony-stimulating factor (GM-CSF), tumour necrosis factor (TNF)-α and regulatory cytokines such as transforming growth factor (TGF)-β and interferon (IFN)-γ.14 In both viral and bacterial models of infection, IFN-γ production by NK cells has been shown to be a key event in successful resolution of infection15–17. As a general rule, IL-12 produced very early in infection is responsible for driving NK cells to produce IFN-γ.17,18 Thus, accessory cells are important in the NK cell response to pathogen, and NK cell recognition of pathogen is indirect. However, it has become apparent that direct recognition of pathogen by NK cells can also occur with corresponding activation of effector functions (see below). In terms of cytotoxicity, recognition of appropriate targets is the fundamental first step in the process, and most recent progress has been made in this area. NK cells have conventional activatory receptors but, more unusually, they also express constitutively active inhibitory receptors. It appears that the net sum of inhibition and activation signalling ultimately controls NK cell function. NK cell receptors broadly fall into two categories based on their ability (or inability) to recognize human leucocyte antigen (HLA) class I. In this review, we present a synopsis of new information on HLA class I receptors but also describe more recently characterized receptors, including natural cytotoxicity receptors (NCRs) and Toll-like receptors (TLRs), and address how these receptors function in concert to regulate NK cell function.

HLA class I specific inhibitory receptors

NK cells are programmed to kill and require inhibitory signals from normal autologous cells to prevent unwanted cell death. The best characterized inhibitory signals are those transduced by HLA-specific receptors, including members of the killer immunoglobulin-like receptor (KIR) family of receptors and the CD94/NKG2A heterodimer that belongs to a lectin-like family of receptors.19 Under normal circumstances, HLA class I antigen expression maintains NK cell tolerance. However, in certain pathological conditions, for example viral infections or cancer, HLA class I expression is commonly altered, thereby removing normal inhibition. The resulting NK cell activation, including cytotoxicity and cytokine production, is a fundamental component of the early immune response. Encoded on chromosome 19, KIRs comprise a family of 15 receptors that include both inhibitory and activatory members. Inhibitory receptors are characterized by immunotyrosine-based inhibitory motifs (ITIMs) in their cytoplasmic tails, while activatory receptors recruit positive signalling adapter molecules. Expression is restricted to NK cells and a small subset of T cells (for a detailed description, see References20 and 21). These receptors bind to subgroups of HLA class I molecules (see Table 1). Ligation of inhibitory KIRs by their HLA ligands on healthy cells results in the inhibition of NK cell activation and protection of the HLA+ cell from NK cell lysis. A detailed discussion of all KIRs is beyond the scope of this review. Current information on three-immunoglobulin (3Ig)-domain KIRs, including KIR3DL1, a focus of study in our laboratory, is discussed.

Table 1.  Natural killer (NK) cell receptors
  1. KIR, killer immunoglobulin-like receptor; LILR, leukocyte immunoglobulin-like receptor; DNAM, DNAX accessory molecule 1; NTB–A, NK-, T- and B-cell antigen; LFA, lymphocyte function-associated antigen; TLR, Toll-like receptor; HLA, human leucocyte antigen; MIC, MHC class I chain related; HA, haemagglutinin; IgG, immunoglobulin G; PVR, poliovirus receptor; ICAM, intercellular adhesion molecule; PAMP, pathogen-associated molecular pattern; ds, double-stranded.

Immunoglobulin-like receptors
KIR2DL1,HLA-C, C2Inhibitory72, 73
KIR2DS1HLA-C, C2Activatory74
KIR2DL2/3HLA-C, C1Inhibitory72, 73
KIR2DS2HLA-C, C1Activatory74
KIR3DL1HLA-B Bw4Inhibitory22
KIR3DS1HLA-B Bw4?Activatory? 
KIR3DL2HLA-A A3, A11Inhibitory29
LILR family (ILT)HLASome inhibitory, some activatorySee 76
Lectin-like receptors
Cytotoxicity receptors
NKp44Unknown cellular ligand, viral HAActivatory50
NKp46Unknown cellular ligand, viral HAActivatory78
DNAM-1PVR (CD155), nectin-2 (CD112)Activatory,80
NKp80UnknownActivatory, coreceptor81
CD59UnknownActivatory, coreceptor82
NTB-ANTB-AActivatory, coreceptor83
2B4 (CD244)CD48Activatory, coreceptor84, 85
CD2LFA-2Activatory, coreceptor86
TLRPAMPs, e.g. dsRNAActivatory41–43, 58


KIR3DL1 is a well-characterized receptor that recognizes a subgroup of HLA-B molecules expressing the serological Bw4 epitope, which is defined by positions 77–83 on the α1 helix of the heavy chain.22 The Bw4 epitope is present on approximately one-third of all HLA-B alleles and is defined by the presence of leucine at position 82 and arginine at position 83, while other residues show variation among alleles. A 70-kDa protein, KIR3DL1 contains three extracellular immunoglobulin (Ig) domains (3D) and a long (L) intracellular domain that contains two ITIMs. Ligation of KIR3DL1 by its cellular ligand results in phosphorylation of these ITIMs and the recruitment of phosphatases, including SHP-1 and SHP-2, that inhibit activation pathways.23,24 KIR3DL1 is present in approximately 95% of the population and, like most KIRs, is highly polymorphic, with approximately 35 known alleles (P. Norman and P. Parham, Stanford University, Stanford, CA, personal communication). Polymorphism has been shown to affect antibody binding25 and may have functional consequences. Indeed, we have preliminary evidence that different alleles vary in their ability to recognize an identical Bw4 target antigen (manuscript in preparation, G. O'Connor and C. Gardiner).

In addition to monitoring levels of HLA expression (frequently altered in viral and transformation events), KIR3DL1 is sensitive to the peptide presented by the HLA molecule.26,27 Sensitivity to peptide presented by HLA class I has also been found for KIR2DL1 and KIR3DL228,29 (see also below). Unlike T-cell receptors, which display very fine peptide specificity, KIR3DL1 shows a broader specificity, with several unrelated peptides capable of preventing KIR3DL1 recognition, thus leading to KIR3DL1+ NK cell activation. This allows the possibility of NK cell detection of viral infection or transformation as a result of a perturbation in the inhibitory signal to KIR upon presentation of viral or tumour-derived peptide. This possibility was nicely illustrated by the discovery that retrovirally transduced T cells became a target for autologous NK cell lysis as a result of the presentation of a resistance gene-derived peptide that prevented KIR3DL1 recognition.30

KIR3DS1, an allele of KIR3DL1, is highly homologous to KIR3DL1 extracellularly but resembles activatory receptors cytoplasmically. A genetic study has found that KIR3DS1, when expressed in combination with HLA-Bw4 molecules with isoleucine at position 80 (Bw4-80Ile), is associated with delayed progression to acquired immune deficiency syndrome (AIDS).31 One attractive hypothesis is that KIR3DS1+ NK cells recognize and kill HLA Bw4-80Ile+ human immunodeficiency virus (HIV)-infected targets, allowing control of infection. Despite wide interest in this receptor, there is a lack of information on the expression and function of this important allele. Lack of antibody reagents against specific KIR molecules has hampered detailed analysis, and indeed KIR3DS1 cell surface expression has not been definitively established. However, we have recently generated some experimental evidence to suggest that KIR3DS1 is indeed present on the cell surface (manuscript in preparation, G. O'Connor and C. Gardiner). The ligand specificity of KIR3DS1 and its activatory nature are also inferred from its similarity to KIR3DL1 extracellularly and activatory receptors intracellularly, but remain to be proved by formal demonstration.

KIR3DL2, which is structurally similar to KIR3DL1, is one of the three framework KIR genes (along with KIR3DL3 and KIR2DL4) present on all haplotypes. The ligand of KIR3DL2 has been an issue of dispute, with difficulty experienced in repeating initial observations that it bound to HLA-A3 and HLA-A11.32,33 This discrepancy has been recently resolved by the finding that the interaction between KIR3DL2 and these HLA-A alleles is highly peptide dependent.29 Indeed, of a panel of eight peptides tested in tetramer binding assays, only one, an Epstein Barr virus (EBV)-derived peptide, allowed recognition. In support of this finding, initial experiments that observed protection of HLA-A3 or HLA-A11 positive cells from lysis by KIR3DL2+ NK cells used EBV-transformed B-cell lines as target. The functional significance of this finding is still unclear, as the existence of an inhibitory pathogen-derived peptide-specific receptor is counterintuitive.


In parallel to the KIRs, NK cells express CD94 and NKG2 heterodimer receptors, members of the C-type lectin-like family, which recognize the non-classical class I HLA-E molecule.34 HLA-E binds and presents the leader peptide of most HLA-A, -B and -C molecules as well as the non-classical HLA-G molecule.35 Therefore, while KIRs are sensitive to individual HLA allele changes, they are complemented functionally by CD94/NKG2A, which responds to changes in global HLA expression. This allows NK cells to respond to both subtle and major alterations in HLA expression caused by different pathological situations.

CD94 is expressed as a disulphide-linked heterodimer with members of the NKG2 family, including NKG2A, which transduces inhibitory signals, and NKG2C, which transduces activatory signals. As with KIR molecules, the inhibitory receptor has a higher binding affinity and inhibition dominates.36 As CD94/NKG2 compliments KIR functionally, its expression and activity must be taken into account in any experiments involving human NK cells. The converse also holds in that KIR expression must also be considered when investigating CD94/NKG2 in functional experiments. Herein lies a major limitation in terms of performing experimental work. In order to dissect out receptor specificities definitively, NK cells must be cloned, characterized and screened for receptor expression and then used in functional experiments. This is not a trivial task and is both time-consuming and technically demanding. Given the clonal distribution of NK cell receptors, the efficiency of isolating clones expressing a relevant repertoire of receptors is low.

Activatory receptors and their ligands

Our appreciation of and interest in activatory receptors on NK cells has ebbed and flowed over the years as initial discoveries failed to account for all the experimental evidence and to demonstrate how NK cells function. Indeed, it has only been since the discovery of the importance of inhibitory receptors in NK cell biology that activatory receptors have again become a hot research topic. NK cells express many activatory receptors including CD2, CD16, lymphocyte function-associated antigen (LFA)-137 and, as more recently described, the activatory KIRs,19,21 NKG2C,38 NCRs,39 NKG2D40 and TLRs.41–43

HLA-specific receptors

Among the KIR family are members that do not contain an inhibitory signalling motif and couple to signalling molecules that transduce activatory signals.21,44 Extracellularly, many of the activatory KIRs are highly homologous to inhibitory members of the KIR family (2SD1 and 2DL1) and bind to the same HLA ligands (Table 1). The activatory KIRs bind with lower affinity than their inhibitory counterparts,36,45 and non-major histocompatibility complex (MHC) ligands have been identified for some KIRs. KIR haplotypes vary in the number of activatory KIRs present, which ranges from one to six.19 In people with an AA haplotype (the simplest haplotype known), the only activatory KIR present in the genome is KIR2DS4, a common allele of which is not expressed.46 Thus, in these individuals, there is no activatory KIR present on NK cells, raising a question over the role played by activatory KIR. Perhaps, in these individuals, the CD94/NKG2C receptor is sufficient, which suggests redundancy in HLA-associated activatory receptor usage. This idea is supported by a detailed study of NK cell receptor usage in two individuals,47 which appeared to show a differential dependence on either KIR or the CD94/NKG2 receptor system, depending on the HLA background of the donors.

NCR and NKG2D receptors do not recognize HLA class I ligands

The activatory NCRs and NKG2D receptors have received most recent attention (for a detailed review, see reference 39). Their expression is predominantly restricted to NK cells, and they have the ability to activate NK cells in the absence of additional stimuli. NKp4648 and NKp3049 are expressed by all NK cells, while NKp4450 expression is restricted to activated NK cells. Although not an absolute phenomenon, NK cell clones can be identified as NCRbright or NCRdull based on the surface receptor densities and high levels of NCR correlated with high natural cytotoxicity against many target cells.48,49 Indeed, a complimentary relationship between NCR and NKG2D appears to exist, and NCRdull clones kill tumour cells in an NKG2D-dependent manner.51 Together, NCRs and NKG2D accounted for virtually all cytotoxicity mediated by NK cells against a wide variety of tumour target cell types. While the ligands for NKG2D are known to be the stress-induced antigens MHC class I chain related (MIC)A and MICB,40 the endogenous ligands for the NCRs remain to be identified.51 It appears, however, that NCRs may have been co-opted by NK cells to recognize pathogen-specific moieties, as both NKp46 and NKp44 recognize virus-specific haemagglutinins and facilitate NK cell lysis of virally infected cells.52,53 In contrast, NKp30 appears to have been targeted by the pp65 protein of human cytomegalovirus (HCMV), which binds NKp30 and inhibits NK cell cytotoxicity.54 In the mouse, the activatory Ly49H receptor (a KIR functional homologue involved in recognition of H-2 antigen) specifically recognizes the m157 protein of mouse cytomegalovirus (MCMV).55 The selection of NK cell receptors by the immune system for specific recognition of pathogen and their targeting by specific viral proteins highlight their importance in the immune response to these infections.

TLRs: direct and indirect pathogen recognition by NK cells

TLRs are a recently described family of innate immune receptors which recognize conserved pathogen-associated molecular patterns (PAMPs); for example, lipopolysaccharide (LPS), a component of the Gram-negative bacterial cell wall, is recognized by TLR4.56 Ten mammalian TLRs have been identified to date, and there has been a huge expansion in terms of elucidating TLR ligand recognition and signalling pathways (for comprehensive reviews, see references56 and 57). Given that NK cells play an important role in the early innate immune response to pathogen, and the fact that they lack cognate receptors for the generation of a specific adaptive immune response, it is attractive to speculate that NK cells express TLRs and that these would function in the early sensing of infection. Indeed, this has proved to be the case, and reports are beginning to emerge detailing a role for TLR2,43 TLR3,41,42,58 TLR941 and, more recently, TLR7 and TLR8 in NK cell biology.

It is interesting that, despite the obvious importance of TLRs in viral infection, the first evidence for TLR function in NK cells came from models of parasitic and bacterial infections. Lipophosphoglycan from Leishmania major stimulated NK cells to produce the proinflammatory cytokines IFN-γ and TNF-α through cell surface ligation of TLR2.43 Subsequently, NK cells were shown to respond to the outer membrane protein A from Klebsiella pneumoniae and flagellin from Escherichia coli through TLR2 and TLR5, respectively. These TLR agonists acted directly and in synergy with proinflammatory cytokines, to induce IFN-γ production. However, the role of TLR3 in NK cells has been the best characterized to date, with complimentary, and occasionally conflicting, data emerging from different laboratories.41,42,58 Despite an early report to the contrary,59 there is consensus that NK cells express TLR3, and, as it has been known for many years that NK cells respond to polyinosinic-polycytidylic acid [poly I:C, a synthetic analogue of double-stranded RNA (dsRNA)], data soon emerged to support TLR3 as the receptor mediating this effect. While the original study reported direct activation of human NK cells by poly I:C,42 a second study reported that IL-12 was required for this to occur, i.e. there was a requirement for accessory cell-derived cytokine.41 Our findings support both a direct and an indirect role of poly I:C activation through TLR3. While highly purified NK cells can be directly activated, this effect is significantly enhanced by accessory cells, as demonstrated by cytokine blocking and cell reconstitution experiments. In fact, even very low numbers of monocytes can have potent effects on NK cell activation.60

TLR9 is activated by double-stranded unmethylated CpG motifs which are present in bacterial and, occasionally, viral genomes.57 CpG has been reported to activate human NK cells in vitro, although this was accessory cell dependent as IL-12 was required for this effect.41 When the immune response to MCMV was examined in vivo, it was surprising to find that infections were more severe in TLR9–/– than TLR3–/– mice.61,62 In fact, although TLR3–/– mice did have an impaired immune response to MCMV, IFN-γ production by NK cells was relatively normal, while it was almost completely absent in TLR9–/– animals.62 In addition, type 1 IFN production was also more profoundly affected in TLR9–/– mice.62 It has also been shown that activation of NK cells in MCMV is indirect and mediated by dendritic cells (DCs) stimulated through TLR9.61 Differences in TLR expression and function have been reported between humans and mice, and such differences appear to exist for TLR9 as, in contrast to human cells, murine NK cells do not express TLR9 or respond to TLR9 agonists in vitro.61 Further studies are required to confirm these findings.

We have recently reported that TLR7 and TLR8 are also expressed by NK cells.60 The physiological ligands for TLR7 and TLR8 are viral single-stranded RNA (ssRNA).63–65 NK cells respond to the TLR7/8 agonist R848 and demonstrate increased cytotoxicity and cytokine production. While NK cells required priming with cytokine to directly transduce signal in response to R848, they were potently activated by accessory cell-derived cytokines. In fact, IFN-γ production by NK cells in response to R848 was entirely IL-12, and therefore accessory cell dependent. The increased cytotoxicity observed was also partly attributable to the accessory cell involvement.60

Finally, in contrast to expression on other lymphocyte subsets, we also have observed that TLR4 is expressed at the surface of human NK cells (Fig. 1). In fact, almost all CD56dim NK cells expressed TLR4, while TLR4 was not expressed by the functionally distinct subset of CD56bright NK cells. However, NK cells did not respond to LPS stimulation (unpublished observations, O. Hart), which may reflect lack of required costimulatory signals such as CD14 or MD2. Therefore, although NK cells express virtually all TLRs, not all are inherently functional, and indirect activation of NK cells by TLR ligand on accessory cells is likely to play a significant role in the NK cell response to microbial pathogens.

Figure 1.

Toll-like receptor 4 (TLR4) is expressed on the surface of human natural killer (NK) cells. All CD56dim NK cells expressed TLR4 on their cell surface, in contrast to the CD56bright subset of NK cells. Cells were stained with anti-CD56 (BD Biosciences, Oxford, UK) and anti-TLR4 antibodies (E-biosciences, San Diego, CA), and analysed using a FACSCalibur flow cytometer (Becton Dickenson, Mountain View, CA).

Balance of signals is context dependent

A recurring theme in NK cell biology is the integration and balancing of positive and negative signals. Activation can occur as a result of either a decrease in inhibitory signalling or an increase in the ligation of activatory receptors. The importance of the role played by different receptors is dependent on the challenge (viral infection, transformation, etc.) faced by the NK cell.

During a transformation event, a number of changes occur that can allow NK cell activation. One of these is the down-regulation by tumour cells of HLA class I molecules, which provide an inhibitory ligand for NK cells. Loss of individual HLA alleles can cause the activation of a subset of NK cells that rely on inhibitory KIR signals generated by ligation of these ligands. Stress-induced expression of heat-shock protein (hsp) 60 may also lead to activation of CD94/NKG2A+ NK cells, as a peptide derived from hsp 60 binds to HLA-E but is not recognized by CD94/NKG2A.66 Stress proteins such as MICA, MICB and UL16 binding protein (ULBPs) not generally expressed on normal cells act as ligands for the activatory receptor NKG2D.30 Ligation of NKG2D allows killing of transformed cells that express normal levels of HLA by overcoming inhibitory signals. Thus, transformed cells may be targeted for lysis by NK cells as a result of both a decrease in inhibitory signals and an increase in activatory signals.

A viral infection may lead to the activation of NK cells in a number of different ways. The expression of TLR3 by NK cells allows them to detect dsRNA, a common feature of viral replication. Cytokines produced by virally activated antigen-presenting cells (APCs), such as IL-12 and IFN-α, are potent activators of NK cell activity. NK cells may also be capable of responding to more subtle signs of viral infection. KIR-mediated inhibition of NK cells is sensitive to presented peptide, and evidence suggests that the skewing of cellular peptides that accompanies viral infection may prevent KIR recognition and render infected cells sensitive to lysis by a subset of NK cells.30 The importance of the balance of receptors is highlighted by the numerous strategies devised by viruses to subvert NK cell responses (reviewed in reference 67), which include the expression of decoy HLA-like molecules to inhibit NK cell activity and molecules such as the HCMV protein UL16, which prevents ULBP/NKG2D-mediated activation of NK cells.

NK cells direct DCs in the generation of an adaptive immune response

NK cells have long been regarded as ‘crude’ effectors of the innate immune system, but there is now mounting evidence that they are important players in setting the stage for an appropriate adaptive immune response through their interaction with DCs.11,12 DCs are crucial regulators of the adaptive T-cell response to pathogen, acting as the main APC to prime naïve T cells to mount a suitable adaptive response to pathogen. iDCs, which patrol the periphery, are specialized for antigen uptake and processing. Upon antigen capture these cells switch to an antigen-presenting mode and migrate to the lymph node for interaction with T cells. This switch is characterized by the up-regulation of costimulation markers, including HLA molecules. Evidence of interactions between these two cell types suggests that bidirectional cross-talk regulates the response of both cell types.

Activated NK cells are capable of killing iDCs, despite normal levels of HLA class I expression by these cells. Upon maturation, these cells up-regulate HLA and become resistant to NK cell lysis. Killing of iDC by NK cells is mediated almost exclusively by the activatory receptor NKp30, with little if any role for NKp44 or NKp46.13 NKp30 (and in part NKG2D) are down-regulated on the surface of NK cells in response to TGF-β1, which may provide resistance to TGF-β1-producing DCs.68 This killing of DCs that have failed to become fully activated has been proposed to form part of a ‘quality control’ system, where the quantity and quality of antigen presentation during an infection are controlled in part by NK cells.

Interaction between NK cells and DCs can also induce DC maturation, mediated by NK cell-derived cytokines. Ligation of NKp30 on NK cells by DCs promotes the secretion of TNF-α and IFN-γ, which leads to DC maturation.69 Indeed, supernatant from NKp30-triggered NK cells is sufficient to cause DC maturation and production of IL-12. Mature DCs have recently been shown to induce the activation (including cytotoxicity and IFN-γ production) and proliferation of resting NK cells. Proliferation of NK cells and production of IFN-γ appear to be confined to the CD56bright subset of NK cells.70 This subset of NK cells is characterized by a distinct chemokine receptor profile (including CD62L, CCR7 and CXCR3), perhaps allowing these cells to traffic to secondary lymphoid organs for interaction with DCs during an immune response. Different DC-derived cytokines are responsible for NK cell activation, with IL-12 promoting NK cell IFN-γ production and IL-15 inducing proliferation.71

During an immune response, NK cells are likely to encounter DCs in inflamed tissue where DC-mediated increases in NK cell cytolytic activity and IFN-γ production promote both killing of infected cells and generation of a T helper type 1 (Th1) response. In turn, NK cells can optimize the DC response by editing unsuitable iDC or promoting maturation. Thus, these two cell types can work in concert to generate a maximal adaptive response.

Back to basics

Our basic knowledge of NK cells has greatly expanded recently, and we can now explain at a molecular level many of the responses observed. Indeed, new roles for NK cells in directing both innate and adaptive immune responses have also been identified. It is somewhat ironic, perhaps, that the complexity of NK cell receptors, while initially providing a watershed for identification and molecular characterization of novel receptors, has now proved to be a rate-limiting step in terms of truly understanding how NK cells function. We have gone from having a general and global overview of the importance of NK cells in an immune response to a reductionist extreme of dissecting the minutiae of individual receptors at the molecular level. It seems an appropriate time to stand back and reflect on NK cells in the light of all this new information.

Thus, when thinking about NK cells as effector cells in the immune system, it is important to consider the full range of receptors that they express (with both HLA and non-HLA ligands), which may be differentially activated or indeed activated simultaneously to different extents, depending on the pathological environment in which they find themselves; for example, infection versus cancer, as discussed above. This should be considered when interpreting data pertinent to an individual receptor system.

It is also worth restating the obvious, that an immune system functions as a biological ‘system’ and that any system is designed to operate optimally as a whole. Therefore, while definition of events at a molecular and biochemical level is fundamental to both our understanding and our exploitation of individual events, they must be examined in the context of the global immune response. During infection, an NK cell is directly affected by the infection itself, by direct sensing of pathogen, but is also subject to indirect effects of infection through pathogen activation of other cells of the immune system, which in turn lead to downstream effects (both receptor and soluble factor mediated) on the NK cell. It is the sum of these direct and indirect interactions that will determine the extent of NK cell activation and function (Fig. 2). In addition, the NK cell itself contributes to the effector status of other cells of the immune system, for example monocytes and DCs, through both direct and indirect mechanisms. In summary, the NK cell is a complex lymphocyte that deserves increased recognition for its contribution to overall host immunity.

Figure 2.

Direct and indirect activation of human natural killer (NK) cells. NK cells receive activation signals via a variety of direct and indirect mechanisms. A number of events occur in a virally infected cell which affect NK cells. Viral protein is processed by the antigen-processing pathway and viral peptide is presented on the surface of the infected cell in the context of human leucoyte antigen (HLA) class I. The altered peptide presented may result in perturbation of an inhibitory signal through a killer immunoglobulin-like receptor (KIR) to the NK cell, resulting in its activation. KIRs are also sensitive to the loss of HLA class I from the cell surface, a mechanism used by some viruses to escape cytotoxic T lymphocyte (CTL) recognition. Type 1 interferon (IFN) is produced by virally infected cells, and this is a potent cytokine in terms of activating NK cell effector functions such as cytotoxicity. Viral antigen, for example haemagglutinin (HA), expressed at the infected cell surface can directly engage NK cell activatory receptors such as NKp46. Direct recognition of viral pathogen can also occur through sensing of viral replication products, for example double-stranded (ds) RNA, through cell surface Toll-like receptor 3 (TLR3). Viral replication products such as single-stranded (ss) and dsRNA can also indirectly activate NK cells by interaction with appropriate TLRs on accessory cells, with resultant production of NK cell stimulatory cytokines such as interleukin (IL)-12. Some similar mechanisms exist in tumour cell recognition, including down-regulation of HLA class I to escape from CTLs which can then result in NK cell activation. Different mechanisms involved include induction of stress antigens, for example MICA and MICB, which activate NK cell effector functions through NKG2D. NK cells thus receive a number of signals dependent on the immunological challenge, which can be both direct and indirect (see above), and it is the balance of these that culminates in the final NK cell response.