Versatility in NK cell memory

The ability of certain cells of the immune system to respond more robustly upon re-exposure to the same infection constitutes the basis for vaccination. These same cells that protect the host from repeat pathogen encounter make up the adaptive arm of the immune system. For decades, T and B cells have been known to provide immune memory after vaccination; however, several recent studies have demonstrated that natural killer (NK) cells also possess the adaptive immune feature of immunological memory.

The adaptive immune system is estimated to have arisen in lower vertebrates nearly 500 million years ago¹. As a result of recombination-activating gene (RAG)-mediated recombination of genes encoding antigen recognition receptors, each individual T cell and B cell is endowed with the ability to specifically detect a specific antigen. Collectively, this army of lymphocyte clones can identify an infinite number of self and foreign antigens. Through well-characterized selection processes in the thymus and bone marrow, T and B cells, respectively, undergo an “education” to eliminate self-reactive clones, and generate an army of specialized lymphocytes that can recognize many pathogens and pathogen components.

NK cells were first identified as a population of cytotoxic cells in the periphery that kill target cells without prior sensitization of the host, resulting in the term “natural killing”². Later, NK cells were found to preferentially kill tumor cells that lacked MHC class I, leading to the “missing-self hypothesis”³. Because they were shown to respond more rapidly than T and B cells and were thought to be short-lived, NK cells were classified as innate immune cells. However, NK cells are lymphocytes that develop from a common lymphoid progenitor stem cell shared by T and B cells⁴, and in recent years, have been shown to undergo several processes that are hallmarks of adaptive immunity including developmental education^{5, 6}, clonal-like expansion⁷, memory cell generation^8–10, and recall responses^8–10. Recently, several groups have independently demonstrated that previously activated NK cells can produce long-lived progeny able to mediate robust secondary responses in vitro and in vivo^11–13. During mouse cytomegalovirus (MCMV) infection, mouse NK cells possessing the activating Ly49H receptor specific for the viral m157 glycoprotein can expand up to 1000-fold following adoptive transfer into Ly49H-deficient recipients, resulting in a long-lived population of MCMV-specific memory NK cells in lymphoid and non-lymphoid tissues able to respond robustly against subsequent MCMV challenge¹³. These findings suggest that antigen-specificity might be necessary for the generation of long-lived NK cell progeny, analogous to mechanisms observed for T cell memory, although in some circumstances cytokine-induced activation might be sufficient.

Unlike T and B cells, NK cells do not use the somatic rearrangement of receptor genes to generate their antigen receptors; therefore, the diversity of NK receptors that can be expressed by NK cells is limited to the genes present in their genome. Thus, NK cells can sense only a finite number of cognate ligands. Because many of the identified ligands of NK cell activating receptors are derived from self (in some cases upregulated by cellular stress) or certain viruses (including herpesvirus family members and influenza)¹⁴, it is speculated that the fixed number of germline-encoded NK cell receptors evolved in response to certain virus families that have threatened mammalian species¹⁵. A new study by Paust and colleagues confirms that NK cells can mediate recall responses months after initial priming, but also astonishingly reports that memory NK cells can possess a specificity against a broad spectrum of foreign antigens from chemical haptens to viruses of different families, including HIV-1, which doesn’t infect mice¹⁶. Furthermore, they indicate that the chemokine receptor CXCR6 plays an important role in the ability of memory NK cells to mediate contact hypersensitivity reactions and provide protective immunity against viral challenge¹⁶.

An earlier study from these investigators reported that only NK cells resident in the liver of mice deficient in T and B cells could mediate hapten-specific delayed-type hypersensitivity (DTH) responses¹². Surprisingly, adoptively transferred splenic NK cells could not confer upon recipient mice the ability to manifest the DTH response. Within the hepatic NK cell population, the hapten-specific memory was found in a Thy-1 and Ly49C/I-expressing subset. Because inhibitory Ly49C/I receptor expression on developing NK cells in C57BL/6 mice “licenses” these cells for proper peripheral function^17–20, NK cell education might play a role in the generation of these hapten-specific memory responses. Because of the specificity observed in NK cells against different haptens, it can be postulated that the secondary responses are mediated via activating receptors, using a mechanism similar to NK cell memory following MCMV infection^{10, 13}. These findings raise the question of what selective pressures might have endowed NK cells with mechanisms for the recognition of chemical haptens.

Paust and colleagues further investigated these hepatic NK cells that mediated DTH responses by extending their studies to include viral infections¹⁶. In a series of complex adoptive transfer experiments, RAG-deficient mice lacking T and B cells were first injected with a chemical hapten (OXA or DNFB), a virus-like particle, or a UV-inactivated virus to generate an immune response (Figure 1A). Several months after initial priming, NK cells from liver, lung, and spleen were isolated from these antigen-primed RAG-deficient mice and separately injected into recipient mice lacking T, B, and NK cells. Following homologous challenge with the same hapten or virus used in priming, or heterologous challenge, DTH responses and anti-viral immunity were measured. Consistent with their previous findings, homologous hapten exposure resulted in a DTH response in the recipient mouse, as measured by inflammation and thickening of the ear; if a different hapten than the one used in priming was injected into recipients, no DTH response was measured (Figure 1A). Only NK cells isolated from the livers and lungs, but not spleen, of primed mice mediated the DTH response. Surprisingly, hepatic NK cells primed by viral particles or inactivated virus also generated virus-specific DTH responses in recipient mice following homologous virus challenge. In addition, only primed hepatic NK cells, and not primed splenic NK cells or naïve hepatic NK cells, were able to enhance survival of recipient mice challenged with influenza A or VSV.

Figure 1. Memory resides in a population of hepatic CXCR6-expressing NK cells.

(A) RAG-deficient mice, which lack T and B cells, were injected with a specific chemical hapten or virus (antigen/pathogen A). Several months later, NK cells were isolated from the liver, lungs, and spleen, and transferred into recipient mice lacking T, B, and NK cells (mice in which genes encoding Rag and the common gamma subunit of the IL-2 receptor have been ablated). DTH or anti-viral immunity were measured following challenge with the same hapten or virus (depicted as a blue “A”) or a different one (depicted as a red “B”). +, positive response; −, no response detected. (B) Hepatic NK cells were isolated from RAG-deficient CXCR6 reporter mice injected with a specific chemical hapten or virus several months earlier, and equal numbers of CXCR6⁺ (GFP⁺ shown in green) and CXCR6⁻ (GFP⁻) NK cells were transferred into recipient mice in which genes encoding Rag and the common gamma subunit of the IL-2 receptor have been ablated. DTH or anti-viral immunity were measured following challenge with the same or different hapten or virus.

Because hepatic NK cells were the predominant population demonstrating specific recall responses, the role of chemokine receptor CXCR6 was explored in both hapten- and virus-specific NK cell memory. CXCR6 (CD186) and its ligand CXCL16 (constitutively expressed on hepatic endothelium) were previously shown to be crucial for the survival of hepatic NK T cells²¹. Using GFP reporter mice, CXCR6 was found to be expressed on a higher percentage of NK cells in the liver (~50%) compared to other tissues¹⁶. The reporter mice were primed with hapten or virus and several months later CXCR6⁺ (GFP⁺) and CXCR6⁻ (GFP⁻) NK cells were transferred into recipient mice (Figure 1B). Upon homologous hapten exposure or virus challenge, DTH responses and anti-viral immunity were only detected in mice receiving previously primed CXCR6⁺ NK cells (Figure 1B). Interestingly, blocking with an antibody against CXCR6 also ablated secondary responses to hapten and virus re-exposure both in vivo and in vitro. Additional adoptive transfer experiments showed CXCR6 to be important for the survival and homeostasis of hepatic NK cells.

The study by Paust and colleagues adds to the mounting evidence that NK cells are more versatile than previously appreciated, in terms of functional ability and generation of long-lived memory, and now broad antigen-specificity. In fact, the ability of NK cells to recognize a variety of cognate ligands ranging from chemical haptens to different classes of viruses is arguably more surprising than the their ability to demonstrate immune memory. Although RAG expression has been observed in a subset of NK cells²², it is not believed to that RAG plays any role in the function and expression of Ly49 receptors in mice or the corresponding KIR family of NK receptors in humans. Thus, without the limitless antigen receptor repertoire of T and B cells, we are left to wonder what molecular mechanisms are at work in NK cells to allow them to recognize and “remember” such a broad spectrum of antigens. Nonetheless, these new findings¹⁶ – along with the previous demonstration of NK cell memory against cytomegalovirus¹³ – warrant further studies to explore whether vaccination of NK cells might provide new opportunities to immunize against pathogens that have proven difficult to control using conventional B and T cell vaccination.