Perspective: Natural Killer Cells: Bridging Innate and Adaptive Immunity?

New findings indicate that natural killer cells can respond to target cells in a peptide-specific manner and could be involved in the memory response to specific antigens, suggesting that their role in antiviral immunity should be reassessed

By Galit Alter and Marcus Altfeld*

All organisms, ranging from the simplest unicellular ones to humans, are subject to relentless attack by pathogens(1) and so have consequently evolved the means to defend themselves by dedicating a large number of genes to ensuring their survival(1,2). In higher animals, starting with the jawed fish, these defense mechanisms have resulted in the development of two elaborate but critical arms of the immune system, innate and adaptive immunity.

Innate immunity is the most ancient and can be found in relatively simple organisms as diverse as plants andDrosophila sp(1). The innate response is induced rapidly following infection and relies on the activation of various cell subsets via specific cellular receptors(3) or pattern-recognition molecules(4) to alert these cells of the invading agents' presence. Once activated, innate immune effectors, including monocytes, macrophages, neutrophils, and natural killer (NK) cells(5) are responsible for containing and clearing the infection and secreting a number of immunomodulatory cytokines that drive the adaptive immune response(6). This adaptive response, geared towards providing antigen-specificity and immunological memory, complements but does not replace the activity of the innate immune response(2).

Of particular interest when considering HIV-1 infection are NK cells, which represent a subset of innate effector cells that are critical for the control and clearance of a number of viral infections(6,7). NK-cell deficiencies in humans and NK-cell depletions in mice can result in recurrent viral infections and death(7-9). Human NK cells can be subdivided into at least three separate subpopulations(10-12). The first subgroup includes the CD3negCD56brightCD16neg NK cells, or "immunomodulatory" NK cells, that express little perforin but secrete large quantities of inflammatory and antiviral cytokines and chemokines. The second group is the CD3negCD56dimCD16pos, or "cytolytic" NK cells, which express high quantities of perforin and are able to mediate both direct cellular cytotoxicity and antibody-dependent cellular cytotoxicity. Recently, a third group of NK cells was described that is preferentially expanded in the context of viremic HIV-1 infection, the CD3negCD56negCD16pos ("anergic") NK cells(10,12). How do NK cells recognize their target cells?

Unlike B and T cells, NK cells do not express unique clonally distributed receptors for specific antigens, rather they express many different promiscuous stimulatory and inhibitory receptors that can be divided into at least four classes(13): the killer immunoglobulin-like receptors (KIRs), the C-type lectin receptors, the natural cytotoxicity receptors (NCRs), and the toll-like receptors (TLRs) (Figure 1). Fourteen KIRs have been described to date. These receptors are activating or inhibitory in nature and their predominant ligand comprises HLA-class I molecules that are expressed on all nucleated cells(3). The C-type lectin receptors (NKG2A-E in humans, Ly49 in mice) monitor the expression of HLA-E, HLA-G, and non-classical MHC-class I-homologs (MIC A/B)(13). Three NCRs have been identified to date (NKp30, NKp33, NKp36) but their ligands remain largely undefined(13). TLRs are pathogen-associated, pattern-recognition receptors involved in the non-specific recognition of infection(14). While the NCRs are exclusively expressed on NK cells, KIRs, NKG2A, and TLRs are found on other cells of the immune system.

Effect of HIV-1 infection on NK cells

A large number of studies have assessed the impact of HIV-1 infection on the NK cell compartment. During acute HIV-1 infection, NK cells are significantly expanded(10) and activated. In particular, the cytolytic CD3negCD56dimCD16pos NK cells are preferentially expanded, express high levels of KIR, and exhibit strong responses to MHC-devoid target cells at this early stage of the infection when the adaptive arm of the immune response is just developing. Following the emergence of virus-specific T-cell responses, the number of cytolytic CD3negCD56dimCD16pos NK cells decreases, and is reduced in chronic viremic HIV-1 infection compared to non-infected individuals(10). In parallel to this loss of functional CD56pos NK cells, a population of anergic CD3negCD56negCD16pos NK cells is expanded in chronic infection(10,12). These changes in the NK-cell compartment result in an overall stable number of NK cells that respond vigorously to MHC-devoid target cells in chronic viremic infection, but also in a significant reduction in NK-cell cytolysis due to the replacement of functional NK cells by CD56neg anergic NK cells. This accumulation of anergic NK cells in chronic HIV-1 infection may also contribute to the overall immunodeficiency in progressive infection, making the host more susceptible to opportunistic infections and tumors. In contrast to HIV-1 infection, there is a marked preservation of functionally-active NK cells, as well as T and B cells, in the less-pathogenic HIV-2 infection(15) and in SIV-infected sooty mangabeys(16)..

The precise mechanisms that lead to these differences in the functional impairment of different leukocyte subsets between HIV-1 and HIV-2 infection (as well as SIV-infected macaques and sooty mangabeys) are still largely unknown. However, data presented by Mark Feinberg at the recent Keystone meeting have begun to address the role of immune activation as the underlying difference between SIV-susceptible macaques in contrast to sooty mangabeys, which serve as the natural host of this infection. Feinberg presented provocative data demonstrating a marked reduction in the innate response to become activated by SIV via TLR in sooty mangabeys, including a reduction in IFN-a production by plasmacytoid dendritic cells potentially resulting in reduced NK-cell proliferation.

In contrast, the innate response was strongly activated in SIV-infected macaques. Given the critical role of the initial innate immune response in initiating the adaptive immune response, it is likely that subdued innate immune response, both in acute as well as in chronic infection, may result in the generation of moderate NK- and T-cell responses and consequently less general immune activation in infected sooty mangabeys. These studies again emphasize the crucial interplay between the innate and adaptive immune responses in AIDS-virus infections and the potential consequences of manipulating the innate response for HIV-1 pathogenesis.

NK cells suppress HIV-1 replication

Recent observations from epidemiological studies suggest that NK cells may play a significant role in the control of HIV-1 replication and disease progression. A strong association between the expression of a particular activating KIR, KIR3DS1, and its ligand, HLA-B Bw4 with an isoleucine at position 80 (Bw4 80I), and slower HIV-1 disease progression has been noted(17). One potential interpretation of these data is that KIR3DS1 may play a direct role in sensing changes in the expression of its ligand on infected cells in subjects with HIV-1 infection.

More recent work from Carrington's group, presented at the last American Association of Immunologists meeting in Boston(18), has also begun to shed light on a potential protective role of the expression levels of KIR3DL1, an inhibitory receptor binding to HLA-B Bw4. KIR3DL1 is a highly polymorphic allele and different subtypes lead to different expression levels of this receptor on the surface of NK cells. Interestingly, the presence of two high-expressing KIR3DL1 alleles in the presence of HLA-B57, an HLA-B Bw4 allele already associated with slower disease progression, renders the protective effect of HLA-B57 even greater. Given the significant effect of HLA-B alleles on HIV-1 disease outcome it is quite intriguing that KIRs expressed on NK cells, and not only the TCRs of CD8+ T cells, interact with HLA-B molecules and that the expression level of KIR3DL1 further modulates the effect of HLA-B alleles on HIV-1 disease progression, suggesting a potential role for NK cells in controlling HIV-1 replication.

Further evidence for NK-cell mediated immune pressure in HIV-1 infection comes from studies assessing the changes in the surface expression of HLA class I molecules on HIV-1-infected T cells. Nef has been demonstrated to selectively downmodulate expression of HLA class I A and B molecules on the surface of HIV-1-infected cells(19), which subvert the virus-specific CD8+ T-cell response that recognizes viral epitopes presented by these HLA class I molecules. However this downmodulation may render these infected cells more susceptible to KIR-mediated recognition and destruction, since modulation of the levels of HLA class I molecules on the surface of either malignant or infected cells serves as an important trigger of NK cell activity.

In contrast to its effect on HLA-A and -B molecules Nef does not downregulate expression of HLA-C molecules on the surface of infected cells, and several studies have actually demonstrated increased expression of HLA-C and HLA-E on HIV-1-infected cells(20). HLA-C and HLA-E represent the major ligands for KIR2DL and NKG2 and provide a strong inhibitory signal to NK cells. These data suggest that HIV-1 Nef may be able to tip the scales in favor of CD8+ T-cell evasion by downregulating HLA-A and HLA-B, while increased expression of HLA-C and HLA-E may protect these infected cells from the second subset of cytolytic effectors, the NK cells. Overall, these sophisticated methods that HIV-1 has developed to evade both T-cell and NK-cell mediated recognition provide direct evidence for the strong immune selection pressure exerted by NK cells in HIV-1 infection.

Despite these relative changes in the expression of HLA class I molecules on infected cells, functional in vitro studies have shown that blocking inhibitory KIR on NK cells(20), and in particular NK-cell clones bearing fewer copies of inhibitory KIRs, can lead to lysis of HIV-1-infected cells and inhibit viral replication. How can the different receptors expressed on NK cells mediate this recognition of HIV-1-infected target cells? At least two models have been hypothesized. The first proposes that HIV-1 infection induces HLA-independent changes in cell surface molecules (in addition to the changes in HLA class I molecules described above) that are subsequently recognized by activating NK-cell receptors, resulting in lysis of infected cells. An alternative model suggests that HIV-1 infection results in a change in the repertoire of peptides presented by HLA class I molecules such that these novel peptide-HLA class I complexes no longer allow for recognition by inhibitory KIRs-or, alternatively, bind more effectively to activating KIRs-and so target cells are lysed.

TCRs expressed on CD8+ T cells bind to HLA-class I molecules in such a way that the TCR broadly covers the epitope-presenting HLA groove centered around position 5 of the epitope, whereas KIR binds in a different location (Figure 2). The crystal structure of an HLA/epitope/KIR complex (KIR2DL2 in complex with HLA-Cw3) has been resolved(21) and demonstrates that KIR interaction with the peptide-HLA class I complex is centered around amino acid position 7 and 8 of the epitope, close to the C-terminal portion of the peptide binding groove. This resolution of the structure of the HLA/epitope/KIR complex validated a series of previous studies demonstrating that KIRs can discriminate between peptides presented by HLA-B*2705 (KIR3DL1)(22), HLA-Cw*0304 (KIR2DL2)(23), HLA-Cw*0401 (KIR2DL1)(24), HLA-Cw7 (KIR2DL2)(25), and, more recently, HLA-A3/A11 (KIR3DL2)(26). These studies have confirmed the critical role of the residue in position 8 on the HLA-presented epitopes in the peptide-specific interaction with KIR, as individual amino acid changes in that position determine recognition of the HLA/peptide complex by the respective KIRs. Taken together these data demonstrate that the sequence of the peptide presented by HLA class I molecules can influence the ligation by inhibitory and activating KIRs expressed on NK cells, and thereby modulate target cell recognition and lysis. Based on these structural and experimental data on the peptide-specificity of the KIR/HLA interaction, it is tempting to speculate that the recognition of HIV-1-infected target cells by NK cells is more 'specific' than previously thought.

NK cells: a reassessment?

Recent published data challenge the dogma that NK cells are simply mediators of the innate immune response, and also suggest the involvement of NK cells in a memory response to hapten-induced contact hypersensitivity (CHS)(27,28). These studies demonstrate that hepatic Ly49C+ NK cells alone, derived from a previously sensitized mouse, were able to mount detectable CHS upon adoptive transfer into a naïve animal. These data, in the SCID mouse model in the absence of B and T cells, are very exciting but preliminary, and their relevance for human NK cell biology requires further investigation. But given this potential role of NK cells in secondary memory responses and the fact that some degree of peptide specificity, in particular around residue 8 of the epitope presented by HLA class I molecules, appears to play a significant role in NK-cell mediated recognition of target cells, it is probably time that we reevaluate the role of NK cells in antiviral immunity.

Antigen specificity and memory are two distinctive features attributed to the adaptive immune response. Yet it is becoming increasingly apparent that adaptive and innate immunity are not necessarily mutually exclusive. NK cells are the earliest cytolytic effector cells responding to viral infection, can secrete large amounts of cytokines and chemokines that drive the subsequent adaptive immune response, and these cells have now also been implicated to some extent in both immunological memory and peptide-specificity. This strongly suggests that NK cells are involved in both innate and adaptive immunity and in bridging these arms of the immune response. If these initial provocative findings can be generalized and reconfirmed in the context of HIV-1 infection then their implications for immunotherapy and vaccine design may be profound, as they provide an opportunity to manipulate an additional arm of the immune system. Current AIDS vaccine efforts have largely neglected to exploit the important interplay and potential interface of innate and adaptive immunity that have evolved to play synergistic roles in the fight against infection. A better understanding of the innate immune receptors and mechanisms involved in the initial recognition of HIV-1 infection will hopefully help us to better understand HIV-1 pathogenesis, and to strengthen the immunogenicity and potency of future AIDS vaccine candidates.

*Galit Alter is a research fellow leading the efforts at the Partners AIDS Research Center to elucidate the role of NK cells in the control of HIV-1 disease progression.

Marcus Altfeld is an Assistant Professor at Harvard Medical School and directs the Innate Immunity Program at the Partners AIDS Research Center at Massachusetts General Hospital.


1. F.M. Ausubel, Nat. Immunol. 6, 973 (2005)

2. M.F. Flajnik et al., Trends Immunol. 25, 640 (2004)

3. L. Moretta et al., Eur. J. Immunol. 32, 1205 (May, 2002)

4. K. Takeda et al., Annu. Rev. Immunol. 21, 335 (2003)

5. C.A. Biron et al., J. Immunol. 131, 1539 (1983)

6. C.A. Biron, Curr. Opin. Microbiol. 2, 374 (1999)

7. C.A. Biron et al., N. Engl. J. Med. 320, 1731 (1989)

8. R.M. Welsh et al., J. Immunol. 136, 1454 (1986)

9. J.F. Bukowski et al., J. Immunol. 131, 1531 (1983)

10. G. Alter et al., Blood (2005)

11. M.A. Cooper et al., Trends Immunol. 22, 633 (2001)

12. D. Mavilio et al., Proc. Natl. Acad. Sci. USA 102, 2886 (2005)

13. R. Biassoni et al., Immunol. Rev. 181, 203 (2001)

14. V. Hornung et al., J. Immunol. 168, 4531 (2002)

15. M.G. Duvall et al., J. Immunol. 176, 6973 (2006)

16. G. Silvestri et al., Immunity 18, 441 (2003)

17. M.P. Martin et al., Nat. Genet. 31, 429 (2002)

18. M. Carrington, The American Association of Immunologists Annual Meeting, Boston, MA, May 12-16, 2006

19. G.B. Cohen et al., Immunity 10, 661 (1999)

20. J.P. Ward et al., Aids 18, 1769 (2004)

21. J.C. Boyington et al., Nature 405, 537 (2000)

22. M. Peruzzi et al., J. Immunol. 157, 3350 (1996)

23. F. Zappacosta et al., Proc. Natl. Acad. Sci. USA 94, 6313 (1997)

24. S. Rajagopalan et al., J. Exp. Med. 185, 1523 (1997)

25. O. Mandelboim et al., Proc. Natl. Acad. Sci. USA 94, 4604 (1997)

26. P. Hansasuta et al., Eur. J. Immunol. 34, 1673 (2004)

27. G. O'Leary et al., Nat. Immunol. 7, 507 (2006)

28. U.H. von Andrian, Keystone Symposia, HIV-1 Pathogenesis/Vaccine Design, Keystone, CO, March 27- April 2, 2006