In the past decade, the study of NK cells was transformed by the discovery of three ways these “innate” immune cells display adaptive immune behavior, including the ability to form long-lasting, Ag-specific memories of a wide variety of immunogens. In this review, we examine these types of NK cell memory, highlighting their unique features and underlying similarities. We explore those similarities in depth, focusing on the role that Ly49 receptors play in various types of NK cell memory. From this Ly49 dependency, we will build a model by which we understand the three types of NK cell memory as aspects of what is ultimately the same adaptive immune process, rather than separate facets of NK cell biology. We hope that a defined model for NK cell memory will empower collaboration between researchers of these three fields to further our understanding of this surprising and clinically promising immune response.

Natural killer cells are a key component of the innate immune response. These cells employ a broad array of activating and inhibitory receptors to patrol the body and detect subtle changes on a potential target cell that might give away a nascent tumor or virus infection. As described in the missing-self hypothesis, NK cells use inhibitory receptors, such as mouse Ly49 or human killer cell Ig-like receptor (KIR) family receptors, to detect levels of proteins associated with health, such as class I MHC (MHC-I) molecules (1, 2). Cells with normal MHC-I expression engage these inhibitory receptors and prevent NK cell killing, whereas cells that have lost MHC-I (typically indicative of cancer or a virus infection) cannot inhibit the NK cell. In this case, the NK cell is then empowered to both kill the offending cell before a disease can take hold and to release signaling molecules to orchestrate an entire immune response.

Although this missing-self innate immune response makes NK cells a potent first line of defense against tumors and viruses, NK cells were always believed to be just that: cells that prevented or delayed infection until a more powerful adaptive immune response could come online. It was a great surprise, then, when evidence was presented that showed NK cells displaying adaptive, T cell–like behavior. The first reports showed that NK cells could develop Ag-specific memories in the absence of T and B cells (which until then were believed to be the sole Ag-specific cells in the immune system) (3). Other reports soon emerged showing that NK cells could expand in response to CMV, then contract and form a memory pool that was more protective on subsequent exposure (4), or that NK cells could remember a previous activation state and respond more rapidly upon future activations in general (5). Research into these adaptive NK cell responses soon crystallized into a model that supported three distinct types of NK cell memory (6). For the purposes of this review, they will be called “adaptive hepatic memory,” “CMV-reactive memory,” and “cytokine-induced memory-like” NK cell responses (Fig. 1).

FIGURE 1.

Overview of the three types of NK cell memory.

FIGURE 1.

Overview of the three types of NK cell memory.

Close modal

The years since then have allowed more discoveries along each of these three branches of NK cell memory. Taking advantage of these new findings, this review will revisit the three types of NK cell memory, highlighting their unique hallmarks, protective capacities, and many similarities. In light of these more recent discoveries, we will then reexamine the three-branched model of NK cell memory to evaluate whether it is more appropriate to consider the three types of NK cell memory as different facets of the same phenomenon or whether they are truly distinct forms of adaptive immunity.

Adaptive hepatic memory.

The original discovery of adaptive immune responses mediated by NK cells came from the von Andrian laboratory, which reported Ag-specific contact hypersensitivity responses against haptens (small chemical Ags) in mice completely lacking T cells and B cells, which until that moment were understood to be the only adaptive immune cells (3). These memory NK cells were found uniquely in the liver (3, 7), which was attributed to their dependence on the chemokine receptor CXCR6 (8). Mice lacking CXCR6 displayed no signs of adaptive NK cell memory. Surprisingly, however, blocking CXCR6 or neutralizing its ligand CXCL16 during the recall response in vitro enhanced the NK cell memory response, leading to the development of a model in which CXCR6 expression causes memory NK cell homing to the liver but suppresses NK cell functions in favor of longevity (8). This model has more recently been solidified by the Tian laboratory, which has demonstrated that IL-7Rα–expressing group 1 innate lymphoid cells (which contain NK cells as well as other, NK-like innate lymphocytes) initially traffic to the skin-draining lymph node following hapten sensitization in a CXCR3-dependent manner (9). These cells gain their memory potential in the lymph node, upregulating memory-associated surface receptors including Ly49 receptors and CXCR6. This CXCR6 upregulation along with an increased CD49a expression ultimately leads to their liver residence and long-term maintenance through liver-secreted IL-7 (9). Somewhat paradoxically, studies in primates have not found this liver-specific homing, instead identifying NK cells in the liver or spleen that display signs of memory (10, 11). Why different species have different homing sites for memory NK cells is still unknown, but given the many differences between rodent and primate NK cells, it is perhaps not surprising.

The von Andrian laboratory later expanded these findings to include conventional protein-based Ags, showing that various inactivated viruses or virus-like particles could provoke Ag-specific allergic responses or even mediate Ag-specific protection against lethal infections (8). In particular, they found that HIV Ags could provoke Ag-specific NK responses in mice, making it unlikely that these responses were the result of some germline-encoded receptor for common mouse pathogens. Other groups have subsequently found evidence to support this diversity in options for NK cell Ag specificity, finding specific NK cell responses to influenza virus (12), HSV-2 (13), amoxicillin (14), Salmonella (15), vaccinia (16), and Ebola virus (17). Unlike other forms of NK cell memory, these responses are all Ag specific: for example, NK cells with enhanced influenza reactivity do not show memory to respiratory syncytial virus (12), and HSV-primed cells have no memory against B16F10 melanoma (13). Although the initial findings were performed in mice, these later results provide evidence for NK cell memory in macaques (10), zebrafish (18), and humans (11, 12, 19).

Unfortunately, how NK cells specifically recognize Ags for which there is likely no germline-encoded receptor is still unclear. T and B cell populations are able to generate their broad reactivity by random recombination of the T cell or BCR genes using the RAG proteins (20). However, this form of NK cell memory was specifically discovered in Rag-deficient animals, meaning that these cells must generate diversity in another manner. Moreover, to our knowledge, no publication to date has described a recombining receptor within a population of NK cells. Discovering how these NK cells generate their Ag specificity will likely uncover an entirely novel process within adaptive immunity, which is what led our group to explore the possibility that Ly49 receptors are somehow involved in NK cell Ag specificity (21).

The original paper describing adaptive hepatic NK cell memory found that it is performed by NK cells expressing Ly49C and/or Ly49I (Ly49C/I) (3). Conventionally, Ly49C/I play two roles in NK cell biology. As MHC-I receptors, they actively participate in the NK cell missing-self response, inhibiting NK cell killing against a target with normal levels of MHC-I expression as described above (1, 22). As self-reactive receptors, they also participate in a process known as NK cell education or licensing, by which the NK cell learns what a “normal” level of MHC-I is (2326). Essentially, by demonstrating an ability to be inhibited through these Ly49 receptors, the NK cell is licensed to perform its full range of cytotoxic effector functions.

Recently, however, another feature of Ly49C/I has been brought to light. Work from the Kane laboratory has shown that Ly49C in particular is sensitive to the peptide presented on the MHC-I it is binding (27). We hypothesized that this peptide sensitivity might be involved in the Ag specificity of the NK cell memory response and have recently published findings that show adaptive NK cell memory is not merely associated with Ly49C/I expression but is actively dependent on their expression and activity (21). We show that Ly49C/I must be functional, MHC-I must be presenting Ag to elicit NK cell memory responses, and the Ag-specific response is uniquely sensitive to the Ly49-interacting residues of the presented Ag (Fig. 2). This Ly49 sensitivity of the Ag-specific response indicates that Ly49C/I are likely directly involved in NK cell Ag specificity. Admittedly, we do not preclude the possibility of an as-yet-undiscovered receptor’s involvement in NK cell Ag specificity; indeed, Ly49C/I expression itself is necessary but insufficient to define a memory NK cell, because many nonmemory NK cells express Ly49C/I. It is therefore still possible that an Ag-contacting receptor, which confers MHC-I–restricted Ag specificity to a memory NK cell, will be discovered that acts alongside Ly49C/I. However, we have posited a model in which this undiscovered receptor is not necessary to understand NK cell adaptive responses (21). We are hopeful that this partial explanation for the memory NK cell’s Ag specificity will soon lead to a more complete picture of adaptive NK cell responses.

FIGURE 2.

Evidence of the extent of Ag specificity of adaptive hepatic memory. NK cell memory response to hapten, peptide, or protein Ag challenge in mice is dependent on Ly49C/I and on prior sensitization with the same or highly similar Ag. Lack of memory response (indicated by blue line) when mice are challenged with a distinct Ag (blue) versus same or similar Ag (red). All the above memory responses were observed in Rag1−/− mice, * indicates memory phenotype observed in CD8α−/− mice.

FIGURE 2.

Evidence of the extent of Ag specificity of adaptive hepatic memory. NK cell memory response to hapten, peptide, or protein Ag challenge in mice is dependent on Ly49C/I and on prior sensitization with the same or highly similar Ag. Lack of memory response (indicated by blue line) when mice are challenged with a distinct Ag (blue) versus same or similar Ag (red). All the above memory responses were observed in Rag1−/− mice, * indicates memory phenotype observed in CD8α−/− mice.

Close modal

Beyond this Ly49C/I dependency, there are several other key features of adaptive hepatic NK cell responses. The original discovery of adaptive NK cell memory cells also described them as expressing Thy1, NKG2D, CD18, and L-selectin, as well as requiring P/E-selectin for their function (3). Later work expanded this list to include IFN-γ (15, 28), the aryl hydrocarbon receptor (AhR) (29), IL-12 (28), and type-I IFNR (IFNAR) (28) as required for NK cell memory. In humans, NK cells with apparent memory activity can be found in the pleural fluid of tuberculosis patients. These cells are CD56+ CD16 CD45RO+ CXCR3+ NK cells that respond rapidly to IL-12 by increasing cytotoxicity, CD69, CD25, and NKG2D (19). Of particular note is the requirement for IL-12, which bears a striking resemblance to the cytokine-induced memory-like cells discussed below and the lack of CD16, in contrast to the CMV-reactive memory cells. A more recent paper from the Paust laboratory has identified human NK cells in humanized mouse livers that display adaptive NK cell memory and express CXCR6, Eomes, and CD69 (11). This same study also identified a similar NK cell subset that was recruited to sites of vesicular stomatitis virus challenge in human volunteers and displayed marked inflammation-induced cytotoxicity increases, reminiscent of the above IL-12–induced cytotoxic response.

Finally, if Ag specificity is the hallmark of this form of NK cell memory, it raises the question of how Ag is presented to these cells. A publication from the Hornung laboratory took advantage of a unique, NK cell–stimulating hapten to demonstrate that the APC for adaptive NK cell memory is the macrophage, using an inflammasome-dependent mechanism (30). Although this might be a unique property of the hapten used in their study, macrophages are classic examples of good APCs. Moreover, the inflammasome is a source of potent NK cell–stimulating cytokines like IL-18, making this a plausible model for Ag presentation to memory NK cells. Furthermore, in SIV-infected primates, a robust NK cell infiltration 1 d postinfection correlates with a strong inflammasome gene signature (31), and mice lacking inflammatory CCR2+ monocytes were reported to show no NK cell–mediated benefit from a Candida albicans vaccine (32), supporting the hypothesis that macrophages, or perhaps all monocyte-derived APCs, and the inflammasome play a central role in adaptive NK cell memory responses.

Our own research into the role of Ly49 receptors in adaptive NK cell memory has also revealed a feature of how Ags are presented to memory NK cells. As mentioned above, we found that Ly49 receptors were responding to peptides presented on MHC-I molecules during an NK cell memory response, and perhaps unsurprisingly, memory NK cells did not respond to Ags presented on MHC class II molecules. However, we were also able to show that whole-protein sensitization with chicken OVA was able to provoke Ag-specific ear swelling responses when challenged with the OVA-derived SIINFEKL peptide (a hallmark of adaptive memory) and even mediate robust anticancer responses when challenged with OVA-expressing tumors (21). That a whole protein, administered exogenously, was capable of sensitizing NK cells to respond to MHC-I–presented peptides implies that the APC for NK cell memory is capable of processing and cross-presenting exogenous peptides using MHC-I (Fig. 3). Dendritic cells (DCs), especially CD8α+ DCs, are well-known cross-presenters and have a wealth of literature describing their close interactions with NK cells. Although these interactions have historically been understood as providing cytokine signals to NK cells, it is possible that there has been an as-yet-undetected Ag presentation event contained within these same DC:NK interactions. This is especially pertinent given that IL-12, IL-15, and IL-18 are produced by DCs when interacting with NK cells (33, 34) and are the key cytokines provoking memory-like responses in NK cells, as discussed below.

FIGURE 3.

Summary of recent findings about adaptive NK cell memory and open questions that remain.

FIGURE 3.

Summary of recent findings about adaptive NK cell memory and open questions that remain.

Close modal

CMV-reactive memory cells.

Around the same time as the identification of memory NK cells in mouse livers, evidence was accumulating that CMV infection could shape the NK cell repertoire (35, 36). This ultimately was recognized as another form of NK cell memory, based on the relevant NK cells undergoing a T cell–like expansion and memory pool formation in response to murine CMV (MCMV) (4). Upon challenge, the NK cell population in question would rapidly expand and clear the virus infection, then undergo a contraction phase that resulted in a small pool of memory NK cells. Just like their T cell counterparts, these memory NK cells were more reactive upon a subsequent exposure and would again expand to protect the host. Unlike the hepatic NK cell memory, however, the formation of these MCMV-reactive NK memory cells required what is now a well-defined interaction between the NK cell activating receptor, Ly49H, and the m157 protein of MCMV (37). Although the NK cell memory response to MCMV infection in mice mirrors the Ag dependence of hepatic memory, it is characteristically distinct from hepatic memory, in that the recall responses generated through Ly49H are restricted to m157 Ag recognition unlike the adaptable capacity hepatic memory NK cells exhibit to a variety of Ags. Later experiments showed that NK cells expressing Ly49D, the other B6 mouse activating receptor, could also undergo expansion, contraction, and memory pool formation upon exposure to their target Ag, the H-2d MHC-I molecules, provided that a virus infection was ongoing at the time (38).

The MCMV-mediated NK cell memory phenomenon might have been dismissed as a mouse-specific oddity, part of the ongoing and well-documented arms race between CMV and NK cells, except that a similar pattern was noticed in humans. Around 40–70% of the population is infected with human CMV (HCMV) (39). HCMV+ individuals, like the MCMV-infected mice, tend to have a noticeably expanded population of NK cells; in this case, these cells are marked by their NKG2C expression (35). Further study of these cells has revealed a signature surface phenotype: they tend to be CD57+ NKG2A NKG2C+ NK cells (40). They are also frequently CD16+ and lack the FcεRγ1 adaptor protein and the downstream mediators Syk and Eat-2 (40, 41). This CD16+ FcεRγ1 phenotype (known as the g NK cell phenotype) is significant in a study of NK cell memory, because it is associated with extremely potent, mature NK cells that are very responsive to CD16 stimulation and less responsive to other cytokine or activating signals, recapitulating the observed Ag specificity of a memory cell (42). Indeed, the g phenotype might be more important to CMV-reactive memory NK cells than NKG2C, because an expanded NK cell memory population exists in HCMV+ humans that lack NKG2C expression. In both NKG2C-null individuals, as well as in those with NKG2C, effective CMV-reactive memory is correlated with CD2 activity synergizing with CD16 to provide protection (43).

If the hallmark of adaptive hepatic memory is its Ag specificity, the hallmark of CMV-reactive memory is this notable expansion and contraction that the relevant NK cell pool undergoes, whether it expresses Ly49H (4), NKG2C (35), or CD16 (44) as its activating receptor. Further research has uncovered key features of these phases. Expansion requires the transcription factor Zbtb32 (45) and is at least somewhat dependent on the cytokines IL-12 (46), IL-18 (47), and IL-33 (48), all of which are induced following CMV infection, as well as the intracellular isoform of osteopontin (49). IFNAR is also required for an efficient memory expansion: although IFNAR‒/‒ NK cells expand normally upon CMV challenge, they are quickly killed off and do not lead to a memory pool (50). Contraction is also a regulated process that is necessary for an effective NK cell memory response. Mice lacking the proapoptotic factor Bim have too little contraction, resulting in an uncharacteristically large but ultimately nonprotective Ly49H+ pool of NK cells following MCMV challenge (51). Conversely, mice lacking mitophagy display too much memory contraction and similarly fail to develop protective memory (52).

Ultimately, the successful formation of a CMV-reactive memory pool is correlated with protection from a number of diseases. Leukemia outcomes are better following a bone marrow transplant if there is a CMV-reactive memory NK cell expansion (53), and HIV-exposed seronegative individuals have expanded CMV-reactive memory NK cells with very high IFN-γ expression (54). Moreover, the CMV-reactive memory NK cells are resistant to myeloid-derived suppressor cell–imposed inhibition (55), making them promising agents in attempts at cancer immune therapies.

Cytokine-induced memory-like NK cells.

Around the same time that CMV-reactive cells were discovered, work from the Yokoyama group reported yet another memory phenotype in NK cells. They found that NK cells previously activated with IL-12 and IL-18 were much more responsive to IL-15 than naive cells in terms of proliferation and IFN-γ expression (5). This was later recapitulated in vivo (56) and expanded to include many different stimuli, including cross-linking Ly49H or NKR-P1C (5), coculturing with the YAC-1 cancer cell line (57), or LPS-induced inflammation (58), all of which could provoke a stronger response from preactivated cells than from naive. This lack of Ag specificity gave rise to the term “memory-like,” because the classical definition of immunological memory includes both enhanced recall responses and Ag specificity, but these cells were clearly carrying memories of a previous activation state. The enhanced responsiveness of cytokine-activated cells persisted for 3–4 wk in most reports and was stable after two or three generations of NK cell division in an adoptive transfer model, so could not be explained as simply residual activation from the cytokine stimulation (5, 56). Keeping with this generational stability, one study found that the cytokine-induced memory-like NK cells had distinct demethylation across immune promoters; this pattern was more similar to a CD8 T cell’s promoter landscape than that of a naive NK cell (59).

In contrast to adaptive hepatic memory, cytokine memory’s hallmark is a noted lack of Ag specificity. It appears to result from several combinations of cytokine pretreatment, with both IL-12 + IL-15 and IL-12 + IL-18 resulting in enhanced responsiveness to various stimuli (5, 56, 60, 61). It also can arise from vaccination directly, as NK cells from influenza-vaccinated individuals were more responsive to IL-12 + IL-15 or IL-12 + IL-18 (60). Similarly, a rabies vaccine was reported to enhance NK cell IFN-γ in an IL-12 + IL-18–dependent manner, although this also required IL-2 from CD4 T cells (62). Along these lines, human cytokine-induced memory-like cells were observed to increase their expression of CD25 and were much more IL-2 sensitive than naive NK cells (63). The observed roles for cytokine-mediated formation of memory in NK cells is akin to the essential memory cell programming roles that cytokines IL-12 and IFN-α/β play in determining whether naive CD8 T cells form a protective memory population following Ag exposure (64).

Beyond a lack of specificity, cytokine-induced memory-like responses are fairly short-lived. Although it is stable across several generations of NK cells in adoptive transfers, the cytokine memory state begins to fade by 21 d and, under some stimuli, cannot be detected after 4 wk (57, 61). Compare this to reports of adaptive NK cell memory present after 4 mo in mice (8), a year in macaques (10), or 9 mo (41) to several decades (11) in humans. Cytokine-induced memory-like responses are also similar to a recently observed phenomenon among innate monocytes called cellular training, whereby a monocyte activated through any stimulus displays an enhanced responsiveness to any future stimulus (65). In monocytes, this is attributed to histone marks allowing faster transcription of inflammatory genes (32, 66), similar to the demethylation patterns observed in memory versus naive NK cells. It may be that the cytokine-induced memory-like state is another example of cellular training.

Even in this brief overview, it is clear that the three types of NK cell memory are phenotypically distinct (Table I). Adaptive hepatic memory is characterized by its Ag specificity, whereas cytokine-induced memory-like NK cells are noticeably nonspecific, and CMV-reactive NK cells are Ag dependent, but are only as specific as the battery of activating receptors they express: that is, Ly49H+ cells will react to m157, Ly49D+ cells will react to H-2d, and double-positive cells will readily react to both (38). However, there are also some marked similarities between the three types, enough that considering the three as facets of the same response can be a useful exercise in broadening our understanding of NK cell memory.

Table I.
Key characteristics of NK cell memory responses types
Adaptive Hepatic Memory ResponseCMV-Reactive Memory ResponseCytokine-Induced Memory-like Response
Cells involved Liver-homing NK cell or ILC1 Conventional NK cells Conventional NK cells (possibly liver-homing NK cells or ILC1) 
Ag specificity Ag specific and Ag adaptable Ag dependent Nonspecific (trained immunity) 
Cytokine requirement IL-12, IL-18 IL-12, IL-18, IL-33 IL-12 + IL-15, IL-12 + IL-18 
Ag tested Haptens, peptides, whole protein, viruses (Ebola, HIV1, HSV-2, MCMV (m157), HCMV LPS, influenza and rabies vaccine 
influenza A/B, vesicular stomatitis virus), 
S. Typhimurium, tumors 
Key receptors involved Ly49C/I, CXCR6, IL7Ra, CD49a Ly49H, CD16, NKG2C (human) N/A 
References (3, 8, 9, 12–17, 21, 28, 29) (4, 35, 37, 40, 44, 46–48) (5, 56–61) 
Adaptive Hepatic Memory ResponseCMV-Reactive Memory ResponseCytokine-Induced Memory-like Response
Cells involved Liver-homing NK cell or ILC1 Conventional NK cells Conventional NK cells (possibly liver-homing NK cells or ILC1) 
Ag specificity Ag specific and Ag adaptable Ag dependent Nonspecific (trained immunity) 
Cytokine requirement IL-12, IL-18 IL-12, IL-18, IL-33 IL-12 + IL-15, IL-12 + IL-18 
Ag tested Haptens, peptides, whole protein, viruses (Ebola, HIV1, HSV-2, MCMV (m157), HCMV LPS, influenza and rabies vaccine 
influenza A/B, vesicular stomatitis virus), 
S. Typhimurium, tumors 
Key receptors involved Ly49C/I, CXCR6, IL7Ra, CD49a Ly49H, CD16, NKG2C (human) N/A 
References (3, 8, 9, 12–17, 21, 28, 29) (4, 35, 37, 40, 44, 46–48) (5, 56–61) 

N/A, not applicable.

Although the distinctions in Ag specificity of memory responses mediated by hepatic memory NK cells and MCMV-reactive NK memory cells could reflect intrinsic differences in expression, binding, and signaling characteristics of the NK cell receptors involved, the resultant memory formation mediated by each receptor shares a common cytokine requirement. Adaptive hepatic memory fails in mice lacking IL-12 (28), and there is mounting evidence that the inflammasome, responsible for the release of active IL-18, is central to inducing NK cell memory responses (30, 31). Similarly, mice lacking either the IL-12 (46) or IL-18R (47) fail to form a CMV-reactive memory pool on MCMV challenge, although IL-18 was not required for other, non-CMV–induced NK cell expansion. Additionally, the osteopontin requirement for these cells appears to be mediated through the IL-15 cytokine (49). Finally, IL-12 and IL-18, alongside a survival cytokine like IL-2 or IL-15, are the core cytokines needed to generate cytokine-induced memory-like NK cells (5). In this way, a common cytokine milieu appears to be critical in determining whether Ag encountered by NK cells leads to a protective memory population.

The accumulated evidence from the study of NK cell memory suggests that these responses are only differentiated on their method of generating Ag specificity, one of which is still unknown. Different phenotypes of NK cell memory are based only on the different receptors used to generate Ag specificity. This may also dictate the extent of Ag specificity or “Ag adaptability,” such as to either a broad array of Ags, as seen with Ly49C/I (Fig. 2), or a limited Ag type, as by Ly49H. The cytokine requirements shared across the different forms of NK cell memory can be understood as priming signals for NK cells, which only support the formation of a long-lived NK cell memory pool that acquires Ag specificity. This parallels the very same cytokine requirements observed for Ag-specific T cells to undergo memory formation. Viewed in this light, it is possible to understand a single type of NK cell memory, which relies on the “prememory” or “trainable” state from cytokine induction that is guided by the Ag-sp. act. of one or more NK cell receptors.

Additionally, evidence is emerging that suggests there may be more similarities between the Ag specificity displayed by adaptive hepatic NK cells and CMV-reactive NK cells than originally appreciated. In humans, alterations in HLA-E peptide presentation from HCMV infection has been associated with changes in the resulting NK cell memory response (67, 68). Exploring the relevance of this finding in the context of CMV-reactive NK cell memory goes beyond the scope of this review; an article doing just that has recently been published by Rölle and colleagues (69). For our purposes, it is enough to note that in both adaptive hepatic and CMV-reactive forms of NK cell memory, we now have evidence that a C-type lectin complex (Ly49 and NKG2C/CD94, respectively) expressed by the memory NK cell has a direct involvement in recognizing MHC-I–presented peptide changes and driving the resulting memory response. Again, this is not to say that these two lectin:MHC interactions are identical; in particular, NKG2C is an activating receptor, whereas the Ly49 receptors that drive adaptive hepatic NK cell memory are inhibitory. Instead, we take this as further support that the different NK cell memory phenotypes are more akin to variations on a common theme than to truly distinct cellular phenomena.

Given the broad array of activating and inhibitory receptors NK cells possess, having several strategies for acquiring Ag specificity is not surprising and may not indicate fundamentally different responses if a different method of specificity is used. One question this model does not address is the variability in how specific an NK cell memory response is. Adaptive hepatic responses are very specific, only responding to challenge Ags that match the sensitization Ag. Conversely, there are mixed reports with CMV-reactive memory. In some cases, it too is reported as very specific: mice with robust CMV-reactive memory pools were observed to have blunted responses to influenza virus or Listeria (70), and a recent report indicates that CMV-reactive NK cells undergo avidity selection during expansion, which could further increase their specificity (71). However, in other cases, there appears to be no conflict in cross-reactivity. As mentioned above, Ly49H+ memory NK cells could respond well to H-2d provided they also expressed Ly49D (38), NKG2C+ memory NK cell expansion correlates with improved leukemia survival (53), and an HIV-exposed seronegative status correlates with potent CMV-reactive memory (54).

One potential solution to this disparity in responses is the correlation of the g CD16-dependent NK cell memory cells with CMV memory status. Of note, the study that found reduced cross-reactivity of Ly49H+ memory cells to influenza virus or Listeria did so in a very controlled, adoptive transfer model, where no Ab response to either influenza virus or Listeria could have occurred. Conversely, the studies that found broadly enhanced protection from the CMV-reactive memory cells did so by observing complex immune situations, which likely had some involvement from Ab responses. Much like how Ly49H+Ly49D+ NK cells could efficiently act through both receptors, CMV-reactive cells that also express CD16 would be broadly active against any target for which there was an Ab response because a variety of Abs all signal through CD16. In support of this, CMV-reactive memory cells were reported to be effective against influenza virus infection, but only in the presence of influenza virus-reactive Abs (44).

Put another way, all forms of NK cell memory are only able to react to stimuli for which they have a receptor. In the case of CMV-reactive memory, these cells tend to coexpress CD16 and so will be broadly active memory cells, recognizing anything that an Ab can recognize. In the case of adaptive hepatic NK cells, these cells are reported to be less likely to express CD16, explaining their more rigid Ag recognition. Granted, this tendency for CMV-reactive memory cells to coexpress CD16 and for adaptive hepatic memory cells to avoid it may indicate that these are two fundamentally distinct forms of NK cell memory. Instead, however, there may be a single NK cell memory precursor, which stochastically either expresses CD16 or not and which will then develop into either a broadly-reactive CD16+ NK cell memory cell or a narrowly focused CD16 cell (Fig. 4). Hopefully, a deeper understanding of how hepatic memory cell specificity is generated will answer this question.

FIGURE 4.

A unified model of NK cell memory, in which the cytokine-induced memory-like state induced in cells with heterogeneous activating receptor expression can lead to diverse NK cell memory phenotypes from the same underlying process. The cells that mediate adaptive hepatic memory may originate from liver-homing type I innate lymphoid cells or NK cells.

FIGURE 4.

A unified model of NK cell memory, in which the cytokine-induced memory-like state induced in cells with heterogeneous activating receptor expression can lead to diverse NK cell memory phenotypes from the same underlying process. The cells that mediate adaptive hepatic memory may originate from liver-homing type I innate lymphoid cells or NK cells.

Close modal

The study of NK cell memory is still in its infancy, and so there are few examples of its applicability in clinical medicine. Already, however, a clinical trial in a small cohort of acute myeloid leukemia patients has indicated that transplantation with cytokine-induced memory-like NK cells shows clinical promise; donor memory NK cells were associated with an improved antileukemia response ex vivo and ultimately an overall response rate of 55% (72). Similar results in a preclinical xenograft model of ovarian cancer suggests that this benefit may be extensible into solid tumors as well (73). Additionally, as mentioned above, preliminary studies have shown that bone marrow transplants from donors with a pool of CMV-reactive memory NK cells result in improved antileukemia outcomes (53). Together, these findings suggest that NK cell memory could be clinically relevant for improving cancer outcomes even without specific Ag targeting. Indeed, advances in oncolytic virus therapy have focused on engineering viruses that infect tumor cells and simultaneously deliver immune-modulating cytokines and chemokines (7476). An oncolytic virus that delivers IL-12 and IL-18 after infecting a cancer cell could establish the ideal environment for naturally developing adaptive NK cell memory to the cancer in question without having to uncover and design relevant peptide targets against that cancer.

If we understand NK cell memory as a single response that arises from the combination of the correct cytokine milieu (IL-12 + IL-15 + IL-18) and the stochastic expression of the relevant receptor (Ly49H, CD16, Ly49C/I, etc.), it is possible that introducing new relevant receptors to NK cells could expand the possibilities offered by NK cell memory. Already, NK cells are being examined as targets for chimeric Ag receptor (CAR) therapy for a number of reasons, including their potential for reduced toxicity and dramatically reduced cost when compared with more traditional CAR T cell–based approaches (77, 78). Currently, CAR-NK cells are generated by expanding NK cells from cord blood or the NK-like NK-92 cell line using IL-2 or IL-15. However, if our unified model of NK cell memory is correct, it would predict that CAR-NK cells expanded in the presence of IL-12 and IL-18 would exist as memory NK cells when introduced into the patient. This is especially enticing, because memory NK cells could exhibit enhanced activity against their target on a cell-by-cell basis, independent of any enhanced activity from the clonal expansion of the memory NK cell (4, 28), meaning that these memory CAR-NK cells would be more effective cancer controllers than might be predicted based on the study of naive NK cells. Additionally, as observed with memory NK cells coexpressing Ly49H and Ly49D or CD16, memory NK cells can express a battery of activating receptors and respond to targets that are recognized by any of these receptors, making it likely that a memory CAR-NK cell can better control escape mutants than a CAR-T cell, because the mutant would have to escape recognition from the CAR itself as well as any Ab signaling through CD16 and any normal NK cell stress sensing through the natural cytotoxicity receptors.

In reviewing the three types of NK cell memory, we have highlighted their similarities and differences, and in so doing have proposed that these three types of memory may in fact be facets of the same response. Cytokine induction primes the NK cell to become a memory cell, and then Ag specificity, either through the Ly49C/I-related mechanism of hepatic memory or through a germline receptor like NKG2C, Ly49H, or CD16, commits the cell to an Ag-specific memory cell. Unfortunately, an incomplete understanding of this immune phenomenon prevents a definitive analysis of whether NK cell memory is one or several distinct processes. However, it is our hope that by focusing on the similarities between these forms of memory, the three fields can each be strengthened by each other’s findings, and together we can advance our understanding of a novel immune response that has already shown potential in cancer vaccines, HIV therapy, tuberculosis, and many other health care challenges.

We thank members of the Makrigiannis laboratory for critical reading of the manuscript.

This work was supported by a Project Grant from the Canadian Institutes for Health Research (MOP-155906) (to A.P.M.).

Abbreviations used in this article:

     
  • CAR

    chimeric Ag receptor

  •  
  • DC

    dendritic cell

  •  
  • HCMV

    human CMV

  •  
  • Ly49C/I

    Ly49C and/or Ly49I

  •  
  • MCMV

    murine CMV

  •  
  • MHC-I

    class I MHC.

1
Kärre
,
K.
,
H.-G.
Ljunggren
,
G.
Piontek
,
R.
Kiessling
.
1986
.
Selective rejection of H-2-deficient lymphoma variants suggests alternative immune defence strategy.
Nature
319
:
675
678
.
2
Ljunggren
,
H. G.
,
K.
Kärre
.
1990
.
In search of the ‘missing self’: MHC molecules and NK cell recognition.
Immunol. Today
11
:
237
244
.
3
O’Leary
,
J. G.
,
M.
Goodarzi
,
D. L.
Drayton
,
U. H.
von Andrian
.
2006
.
T cell- and B cell-independent adaptive immunity mediated by natural killer cells.
Nat. Immunol.
7
:
507
516
.
4
Sun
,
J. C.
,
J. N.
Beilke
,
L. L.
Lanier
.
2009
.
Adaptive immune features of natural killer cells. [Published erratum appears in 2009 Nature 457: 1168.]
Nature
457
:
557
561
.
5
Cooper
,
M. A.
,
J. M.
Elliott
,
P. A.
Keyel
,
L.
Yang
,
J. A.
Carrero
,
W. M.
Yokoyama
.
2009
.
Cytokine-induced memory-like natural killer cells.
Proc. Natl. Acad. Sci. USA
106
:
1915
1919
.
6
Min-Oo
,
G.
,
Y.
Kamimura
,
D. W.
Hendricks
,
T.
Nabekura
,
L. L.
Lanier
.
2013
.
Natural killer cells: walking three paths down memory lane.
Trends Immunol.
34
:
251
258
.
7
Kawahara
,
M.
,
N.
Hasegawa
,
H.
Takaku
.
2016
.
Murine splenic natural killer cells do not develop immunological memory after re-encounter with Mycobacterium bovis BCG.
PLoS One
11
: e0152051.
8
Paust
,
S.
,
H. S.
Gill
,
B.-Z.
Wang
,
M. P.
Flynn
,
E. A.
Moseman
,
B.
Senman
,
M.
Szczepanik
,
A.
Telenti
,
P. W.
Askenase
,
R. W.
Compans
,
U. H.
von Andrian
.
2010
.
Critical role for the chemokine receptor CXCR6 in NK cell-mediated antigen-specific memory of haptens and viruses.
Nat. Immunol.
11
:
1127
1135
.
9
Wang
,
X.
,
H.
Peng
,
J.
Cong
,
X.
Wang
,
Z.
Lian
,
H.
Wei
,
R.
Sun
,
Z.
Tian
.
2018
.
Memory formation and long-term maintenance of IL-7Rα+ ILC1s via a lymph node-liver axis. [Published erratum appears in 2019 Nat. Commun. 10: 160.]
Nat. Commun.
9
:
4854
.
10
Reeves
,
R. K.
,
H.
Li
,
S.
Jost
,
E.
Blass
,
H.
Li
,
J. L.
Schafer
,
V.
Varner
,
C.
Manickam
,
L.
Eslamizar
,
M.
Altfeld
, et al
.
2015
.
Antigen-specific NK cell memory in rhesus macaques.
Nat. Immunol.
16
:
927
932
.
11
Nikzad
,
R.
,
L. S.
Angelo
,
K.
Aviles-Padilla
,
D. T.
Le
,
V. K.
Singh
,
L.
Bimler
,
M.
Vukmanovic-Stejic
,
E.
Vendrame
,
T.
Ranganath
,
L.
Simpson
, et al
.
2019
.
Human natural killer cells mediate adaptive immunity to viral antigens.
Sci. Immunol.
4
: eaat8116t.
12
Dou
,
Y.
,
B.
Fu
,
R.
Sun
,
W.
Li
,
W.
Hu
,
Z.
Tian
,
H.
Wei
.
2015
.
Influenza vaccine induces intracellular immune memory of human NK cells.
PLoS One
10
: e0121258.
13
Abdul-Careem
,
M. F.
,
A. J.
Lee
,
E. A.
Pek
,
N.
Gill
,
A. E.
Gillgrass
,
M. V.
Chew
,
S.
Reid
,
A. A.
Ashkar
.
2012
.
Genital HSV-2 infection induces short-term NK cell memory.
PLoS One
7
: e32821.
14
Chaves
,
P.
,
M. J.
Torres
,
A.
Aranda
,
S.
Lopez
,
G.
Canto
,
M.
Blanca
,
C.
Mayorga
.
2010
.
Natural killer-dendritic cell interaction in lymphocyte responses in hypersensitivity reactions to betalactams.
Allergy
65
:
1600
1608
.
15
Kupz
,
A.
,
T. A.
Scott
,
G. T.
Belz
,
D. M.
Andrews
,
M.
Greyer
,
A. M.
Lew
,
A. G.
Brooks
,
M. J.
Smyth
,
R.
Curtiss
III
,
S.
Bedoui
,
R. A.
Strugnell
.
2013
.
Contribution of Thy1+ NK cells to protective IFN-γ production during Salmonella typhimurium infections.
Proc. Natl. Acad. Sci. USA
110
:
2252
2257
.
16
Gillard
,
G. O.
,
M.
Bivas-Benita
,
A.-H.
Hovav
,
L. E.
Grandpre
,
M. W.
Panas
,
M. S.
Seaman
,
B. F.
Haynes
,
N. L.
Letvin
.
2011
.
Thy1+ NK [corrected] cells from vaccinia virus-primed mice confer protection against vaccinia virus challenge in the absence of adaptive lymphocytes. [Published erratum appears in 2011 PLoS Pathog. 7.]
PLoS Pathog.
7
: e1002141.
17
Williams
,
K. J. N.
,
X.
Qiu
,
L.
Fernando
,
S. M.
Jones
,
J. B.
Alimonti
.
2015
.
VSVΔG/EBOV GP-induced innate protection enhances natural killer cell activity to increase survival in a lethal mouse adapted Ebola virus infection.
Viral Immunol.
28
:
51
61
.
18
Hohn
,
C.
,
L.
Petrie-Hanson
.
2012
.
Rag1-/- mutant zebrafish demonstrate specific protection following bacterial re-exposure.
PLoS One
7
: e44451.
19
Fu
,
X.
,
Y.
Liu
,
L.
Li
,
Q.
Li
,
D.
Qiao
,
H.
Wang
,
S.
Lao
,
Y.
Fan
,
C.
Wu
.
2011
.
Human natural killer cells expressing the memory-associated marker CD45RO from tuberculous pleurisy respond more strongly and rapidly than CD45RO- natural killer cells following stimulation with interleukin-12.
Immunology
134
:
41
49
.
20
Oettinger
,
M. A.
,
D. G.
Schatz
,
C.
Gorka
,
D.
Baltimore
.
1990
.
RAG-1 and RAG-2, adjacent genes that synergistically activate V(D)J recombination.
Science
248
:
1517
1523
.
21
Wight
,
A.
,
A. B.
Mahmoud
,
M.
Scur
,
M. M.
Tu
,
M. M. A.
Rahim
,
S.
Sad
,
A. P.
Makrigiannis
.
2018
.
Critical role for the Ly49 family of class I MHC receptors in adaptive natural killer cell responses.
Proc. Natl. Acad. Sci. USA
115
:
11579
11584
.
22
Hanke
,
T.
,
H.
Takizawa
,
C. W.
McMahon
,
D. H.
Busch
,
E. G.
Pamer
,
J. D.
Miller
,
J. D.
Altman
,
Y.
Liu
,
D.
Cado
,
F. A.
Lemonnier
, et al
.
1999
.
Direct assessment of MHC class I binding by seven Ly49 inhibitory NK cell receptors.
Immunity
11
:
67
77
.
23
Bix
,
M.
,
N.-S.
Liao
,
M.
Zijlstra
,
J.
Loring
,
R.
Jaenisch
,
D.
Raulet
.
1991
.
Rejection of class I MHC-deficient haemopoietic cells by irradiated MHC-matched mice.
Nature
349
:
329
331
.
24
Höglund
,
P.
,
C.
Ohlén
,
E.
Carbone
,
L.
Franksson
,
H. G.
Ljunggren
,
A.
Latour
,
B.
Koller
,
K.
Kärre
.
1991
.
Recognition of beta 2-microglobulin-negative (beta 2m-) T-cell blasts by natural killer cells from normal but not from beta 2m- mice: nonresponsiveness controlled by beta 2m- bone marrow in chimeric mice.
Proc. Natl. Acad. Sci. USA
88
:
10332
10336
.
25
Sundbäck
,
J.
,
K.
Kärre
,
C. L.
Sentman
.
1996
.
Cloning of minimally divergent allelic forms of the natural killer (NK) receptor Ly-49C, differentially controlled by host genes in the MHC and NK gene complexes.
J. Immunol.
157
:
3936
3942
.
26
Yu
,
Y. Y. L.
,
T.
George
,
J. R.
Dorfman
,
J.
Roland
,
V.
Kumar
,
M.
Bennett
.
1996
.
The role of Ly49A and 5E6(Ly49C) molecules in hybrid resistance mediated by murine natural killer cells against normal T cell blasts.
Immunity
4
:
67
76
.
27
Marquez
,
E. A.
,
K. P.
Kane
.
2015
.
Identities of P2 and P3 residues of H-2Kb-bound peptides determine mouse Ly49C recognition.
PLoS One
10
: e0131308.
28
Majewska-Szczepanik
,
M.
,
S.
Paust
,
U. H.
von Andrian
,
P. W.
Askenase
,
M.
Szczepanik
.
2013
.
Natural killer cell-mediated contact sensitivity develops rapidly and depends on interferon-α, interferon-γ and interleukin-12.
Immunology
140
:
98
110
.
29
Zhang
,
L. H.
,
J. H.
Shin
,
M. D.
Haggadone
,
J. B.
Sunwoo
.
2016
.
The aryl hydrocarbon receptor is required for the maintenance of liver-resident natural killer cells.
J. Exp. Med.
213
:
2249
2257
.
30
van den Boorn
,
J. G.
,
C.
Jakobs
,
C.
Hagen
,
M.
Renn
,
R. M.
Luiten
,
C. J. M.
Melief
,
T.
Tüting
,
N.
Garbi
,
G.
Hartmann
,
V.
Hornung
.
2016
.
Inflammasome-dependent induction of adaptive NK cell memory.
Immunity
44
:
1406
1421
.
31
Barouch
,
D. H.
,
K.
Ghneim
,
W. J.
Bosche
,
Y.
Li
,
B.
Berkemeier
,
M.
Hull
,
S.
Bhattacharyya
,
M.
Cameron
,
J.
Liu
,
K.
Smith
, et al
.
2016
.
Rapid inflammasome activation following mucosal SIV infection of rhesus monkeys.
Cell
165
:
656
667
.
32
Quintin
,
J.
,
S.
Saeed
,
J. H. A.
Martens
,
E. J.
Giamarellos-Bourboulis
,
D. C.
Ifrim
,
C.
Logie
,
L.
Jacobs
,
T.
Jansen
,
B.-J.
Kullberg
,
C.
Wijmenga
, et al
.
2012
.
Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes.
Cell Host Microbe
12
:
223
232
.
33
Koka
,
R.
,
P.
Burkett
,
M.
Chien
,
S.
Chai
,
D. L.
Boone
,
A.
Ma
.
2004
.
Cutting edge: murine dendritic cells require IL-15R alpha to prime NK cells.
J. Immunol.
173
:
3594
3598
.
34
Humann
,
J.
,
L. L.
Lenz
.
2010
.
Activation of naive NK cells in response to Listeria monocytogenes requires IL-18 and contact with infected dendritic cells.
J. Immunol.
184
:
5172
5178
.
35
Gumá
,
M.
,
A.
Angulo
,
C.
Vilches
,
N.
Gómez-Lozano
,
N.
Malats
,
M.
López-Botet
.
2004
.
Imprint of human cytomegalovirus infection on the NK cell receptor repertoire.
Blood
104
:
3664
3671
.
36
Gumá
,
M.
,
M.
Budt
,
A.
Sáez
,
T.
Brckalo
,
H.
Hengel
,
A.
Angulo
,
M.
López-Botet
.
2006
.
Expansion of CD94/NKG2C+ NK cells in response to human cytomegalovirus-infected fibroblasts.
Blood
107
:
3624
3631
.
37
Brown
,
M. G.
,
A. O.
Dokun
,
J. W.
Heusel
,
H. R.
Smith
,
D. L.
Beckman
,
E. A.
Blattenberger
,
C. E.
Dubbelde
,
L. R.
Stone
,
A. A.
Scalzo
,
W. M.
Yokoyama
.
2001
.
Vital involvement of a natural killer cell activation receptor in resistance to viral infection.
Science
292
:
934
937
.
38
Nabekura
,
T.
,
L. L.
Lanier
.
2014
.
Antigen-specific expansion and differentiation of natural killer cells by alloantigen stimulation.
J. Exp. Med.
211
:
2455
2465
.
39
Freeman
,
R. B.
 Jr
2009
.
The ‘indirect’ effects of cytomegalovirus infection.
Am. J. Transplant.
9
:
2453
2458
.
40
Muccio
,
L.
,
A.
Bertaina
,
M.
Falco
,
D.
Pende
,
R.
Meazza
,
M.
Lopez-Botet
,
L.
Moretta
,
F.
Locatelli
,
A.
Moretta
,
M.
Della Chiesa
.
2016
.
Analysis of memory-like natural killer cells in human Cytomegalovirus-infected children undergoing αβ+T and B cell-depleted hematopoietic stem cell transplantation for hematological malignancies.
Haematologica
101
:
371
381
.
41
Schlums
,
H.
,
F.
Cichocki
,
B.
Tesi
,
J.
Theorell
,
V.
Beziat
,
T. D.
Holmes
,
H.
Han
,
S. C. C.
Chiang
,
B.
Foley
,
K.
Mattsson
, et al
.
2015
.
Cytomegalovirus infection drives adaptive epigenetic diversification of NK cells with altered signaling and effector function.
Immunity
42
:
443
456
.
42
Zhang
,
T.
,
J. M.
Scott
,
I.
Hwang
,
S.
Kim
.
2013
.
Cutting edge: antibody-dependent memory-like NK cells distinguished by FcRγ deficiency.
J. Immunol.
190
:
1402
1406
.
43
Liu
,
L. L.
,
J.
Landskron
,
E. H.
Ask
,
M.
Enqvist
,
E.
Sohlberg
,
J. A.
Traherne
,
Q.
Hammer
,
J. P.
Goodridge
,
S.
Larsson
,
J.
Jayaraman
, et al
.
2016
.
Critical role of CD2 co-stimulation in adaptive natural killer cell responses revealed in NKG2C-deficient humans.
Cell Rep.
15
:
1088
1099
.
44
Lee
,
J.
,
T.
Zhang
,
I.
Hwang
,
A.
Kim
,
L.
Nitschke
,
M.
Kim
,
J. M.
Scott
,
Y.
Kamimura
,
L. L.
Lanier
,
S.
Kim
.
2015
.
Epigenetic modification and antibody-dependent expansion of memory-like NK cells in human cytomegalovirus-infected individuals.
Immunity
42
:
431
442
.
45
Beaulieu
,
A. M.
,
C. L.
Zawislak
,
T.
Nakayama
,
J. C.
Sun
.
2014
.
The transcription factor Zbtb32 controls the proliferative burst of virus-specific natural killer cells responding to infection.
Nat. Immunol.
15
:
546
553
.
46
Sun
,
J. C.
,
S.
Madera
,
N. A.
Bezman
,
J. N.
Beilke
,
M. H.
Kaplan
,
L. L.
Lanier
.
2012
.
Proinflammatory cytokine signaling required for the generation of natural killer cell memory.
J. Exp. Med.
209
:
947
954
.
47
Madera
,
S.
,
J. C.
Sun
.
2015
.
Cutting edge: stage-specific requirement of IL-18 for antiviral NK cell expansion.
J. Immunol.
194
:
1408
1412
.
48
Nabekura
,
T.
,
J.-P.
Girard
,
L. L.
Lanier
.
2015
.
IL-33 receptor ST2 amplifies the expansion of NK cells and enhances host defense during mouse cytomegalovirus infection.
J. Immunol.
194
:
5948
5952
.
49
Leavenworth
,
J. W.
,
B.
Verbinnen
,
Q.
Wang
,
E.
Shen
,
H.
Cantor
.
2015
.
Intracellular osteopontin regulates homeostasis and function of natural killer cells.
Proc. Natl. Acad. Sci. USA
112
:
494
499
.
50
Madera
,
S.
,
M.
Rapp
,
M. A.
Firth
,
J. N.
Beilke
,
L. L.
Lanier
,
J. C.
Sun
.
2016
.
Type I IFN promotes NK cell expansion during viral infection by protecting NK cells against fratricide.
J. Exp. Med.
213
:
225
233
.
51
Min-Oo
,
G.
,
N. A.
Bezman
,
S.
Madera
,
J. C.
Sun
,
L. L.
Lanier
.
2014
.
Proapoptotic Bim regulates antigen-specific NK cell contraction and the generation of the memory NK cell pool after cytomegalovirus infection.
J. Exp. Med.
211
:
1289
1296
.
52
O’Sullivan
,
T. E.
,
L. R.
Johnson
,
H. H.
Kang
,
J. C.
Sun
.
2015
.
BNIP3- and BNIP3L-mediated mitophagy promotes the generation of natural killer cell memory.
Immunity
43
:
331
342
.
53
Cichocki
,
F.
,
S.
Cooley
,
Z.
Davis
,
T. E.
DeFor
,
H.
Schlums
,
B.
Zhang
,
C. G.
Brunstein
,
B. R.
Blazar
,
J.
Wagner
,
D. J.
Diamond
, et al
.
2016
.
CD56dimCD57+NKG2C+ NK cell expansion is associated with reduced leukemia relapse after reduced intensity HCT.
Leukemia
30
:
456
463
.
54
Lima
,
J. F.
,
L. M. S.
Oliveira
,
N. Z.
Pereira
,
G. E.
Mitsunari
,
A. J. S.
Duarte
,
M. N.
Sato
.
2014
.
Distinct natural killer cells in HIV-exposed seronegative subjects with effector cytotoxic CD56(dim) and CD56(bright) cells and memory-like CD57+NKG2C+CD56(dim) cells.
J. Acquir. Immune Defic. Syndr.
67
:
463
471
.
55
Sarhan
,
D.
,
F.
Cichocki
,
B.
Zhang
,
A.
Yingst
,
S. R.
Spellman
,
S.
Cooley
,
M. R.
Verneris
,
B. R.
Blazar
,
J. S.
Miller
.
2016
.
Adaptive NK cells with low TIGIT expression are inherently resistant to myeloid-derived suppressor cells.
Cancer Res.
76
:
5696
5706
.
56
Jin
,
F.
,
H.
Lin
,
S.
Gao
,
Z.
Hu
,
S.
Zuo
,
L.
Sun
,
C.
Jin
,
W.
Li
,
Y.
Yang
.
2016
.
The anti-tumor role of NK cells in vivo pre-activated and re-stimulated by interleukins in acute lymphoblastic leukemia.
Oncotarget
7
:
79187
79202
.
57
Keppel
,
M. P.
,
L.
Yang
,
M. A.
Cooper
.
2013
.
Murine NK cell intrinsic cytokine-induced memory-like responses are maintained following homeostatic proliferation.
J. Immunol.
190
:
4754
4762
.
58
Rasid
,
O.
,
C.
Chevalier
,
T. M. N.
Camarasa
,
C.
Fitting
,
J. M.
Cavaillon
,
M. A.
Hamon
.
2019
.
H3K4me1 supports memory-like NK cells induced by systemic inflammation.
Cell Rep.
29
:
3933
3945.e3
.
59
Wiencke
,
J. K.
,
R.
Butler
,
G.
Hsuang
,
M.
Eliot
,
S.
Kim
,
M. A.
Sepulveda
,
D.
Siegel
,
E. A.
Houseman
,
K. T.
Kelsey
.
2016
.
The DNA methylation profile of activated human natural killer cells.
Epigenetics
11
:
363
380
.
60
Goodier
,
M. R.
,
A.
Rodriguez-Galan
,
C.
Lusa
,
C. M.
Nielsen
,
A.
Darboe
,
A. L.
Moldoveanu
,
M. J.
White
,
R.
Behrens
,
E. M.
Riley
.
2016
.
Influenza vaccination generates cytokine-induced memory-like NK cells: impact of human cytomegalovirus infection.
J. Immunol.
197
:
313
325
.
61
Romee
,
R.
,
S. E.
Schneider
,
J. W.
Leong
,
J. M.
Chase
,
C. R.
Keppel
,
R. P.
Sullivan
,
M. A.
Cooper
,
T. A.
Fehniger
.
2012
.
Cytokine activation induces human memory-like NK cells.
Blood
120
:
4751
4760
.
62
Horowitz
,
A.
,
R. H.
Behrens
,
L.
Okell
,
A. R.
Fooks
,
E. M.
Riley
.
2010
.
NK cells as effectors of acquired immune responses: effector CD4+ T cell-dependent activation of NK cells following vaccination.
J. Immunol.
185
:
2808
2818
.
63
Leong
,
J. W.
,
J. M.
Chase
,
R.
Romee
,
S. E.
Schneider
,
R. P.
Sullivan
,
M. A.
Cooper
,
T. A.
Fehniger
.
2014
.
Preactivation with IL-12, IL-15, and IL-18 induces CD25 and a functional high-affinity IL-2 receptor on human cytokine-induced memory-like natural killer cells.
Biol. Blood Marrow Transplant.
20
:
463
473
.
64
Xiao
,
Z.
,
K. A.
Casey
,
S. C.
Jameson
,
J. M.
Curtsinger
,
M. F.
Mescher
.
2009
.
Programming for CD8 T cell memory development requires IL-12 or type I IFN.
J. Immunol.
182
:
2786
2794
.
65
Netea
,
M. G.
,
J.
Quintin
,
J. W. M.
van der Meer
.
2011
.
Trained immunity: a memory for innate host defense.
Cell Host Microbe
9
:
355
361
.
66
Saeed
,
S.
,
J.
Quintin
,
H. H. D.
Kerstens
,
N. A.
Rao
,
A.
Aghajanirefah
,
F.
Matarese
,
S.-C.
Cheng
,
J.
Ratter
,
K.
Berentsen
,
M. A.
van der Ent
, et al
.
2014
.
Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity.
Science
345
: 1251086.
67
Hammer
,
Q.
,
T.
Rückert
,
E. M.
Borst
,
J.
Dunst
,
A.
Haubner
,
P.
Durek
,
F.
Heinrich
,
G.
Gasparoni
,
M.
Babic
,
A.
Tomic
, et al
.
2018
.
Peptide-specific recognition of human cytomegalovirus strains controls adaptive natural killer cells.
Nat. Immunol.
19
:
453
463
.
68
Rölle
,
A.
,
M.
Meyer
,
S.
Calderazzo
,
D.
Jäger
,
F.
Momburg
.
2018
.
Distinct HLA-E peptide complexes modify antibody-driven effector functions of adaptive NK cells.
Cell Rep.
24
:
1967
1976.e4
.
69
Rölle
,
A.
,
D.
Jäger
,
F.
Momburg
.
2018
.
HLA-E peptide repertoire and dimorphism-centerpieces in the adaptive NK cell puzzle?
Front. Immunol.
9
:
2410
.
70
Min-Oo
,
G.
,
L. L.
Lanier
.
2014
.
Cytomegalovirus generates long-lived antigen-specific NK cells with diminished bystander activation to heterologous infection.
J. Exp. Med.
211
:
2669
2680
.
71
Adams
,
N. M.
,
C. D.
Geary
,
E. K.
Santosa
,
D.
Lumaquin
,
J. B.
Le Luduec
,
R.
Sottile
,
K.
van der Ploeg
,
J.
Hsu
,
B. M.
Whitlock
,
B. T.
Jackson
, et al
.
2019
.
Cytomegalovirus infection drives avidity selection of natural killer cells.
Immunity
50
:
1381
1390.e5
.
72
Romee
,
R.
,
M.
Rosario
,
M. M.
Berrien-Elliott
,
J. A.
Wagner
,
B. A.
Jewell
,
T.
Schappe
,
J. W.
Leong
,
S.
Abdel-Latif
,
S. E.
Schneider
,
S.
Willey
, et al
.
2016
.
Cytokine-induced memory-like natural killer cells exhibit enhanced responses against myeloid leukemia.
Sci. Transl. Med.
8
: 357ra123.
73
Uppendahl
,
L. D.
,
M.
Felices
,
L.
Bendzick
,
C.
Ryan
,
B.
Kodal
,
P.
Hinderlie
,
K. L. M.
Boylan
,
A. P. N.
Skubitz
,
J. S.
Miller
,
M. A.
Geller
.
2019
.
Cytokine-induced memory-like natural killer cells have enhanced function, proliferation, and in vivo expansion against ovarian cancer cells.
Gynecol. Oncol.
153
:
149
157
.
74
Patel
,
M. R.
,
B. A.
Jacobson
,
Y.
Ji
,
J.
Drees
,
S.
Tang
,
K.
Xiong
,
H.
Wang
,
J. E.
Prigge
,
A. S.
Dash
,
A. K.
Kratzke
, et al
.
2015
.
Vesicular stomatitis virus expressing interferon-β is oncolytic and promotes antitumor immune responses in a syngeneic murine model of non-small cell lung cancer.
Oncotarget
6
:
33165
33177
.
75
Li
,
J.
,
M.
O’Malley
,
J.
Urban
,
P.
Sampath
,
Z. S.
Guo
,
P.
Kalinski
,
S. H.
Thorne
,
D. L.
Bartlett
.
2011
.
Chemokine expression from oncolytic vaccinia virus enhances vaccine therapies of cancer.
Mol. Ther.
19
:
650
657
.
76
Kim
,
J. H.
,
J. Y.
Oh
,
B. H.
Park
,
D. E.
Lee
,
J. S.
Kim
,
H. E.
Park
,
M. S.
Roh
,
J. E.
Je
,
J. H.
Yoon
,
S. H.
Thorne
, et al
.
2006
.
Systemic armed oncolytic and immunologic therapy for cancer with JX-594, a targeted poxvirus expressing GM-CSF.
Mol. Ther.
14
:
361
370
.
77
Li
,
Y.
,
D. L.
Hermanson
,
B. S.
Moriarity
,
D. S.
Kaufman
.
2018
.
Human iPSC-derived natural killer cells engineered with chimeric antigen receptors enhance anti-tumor activity.
Cell Stem Cell
23
:
181
192.e5
.
78
Tang
,
X.
,
L.
Yang
,
Z.
Li
,
A. P.
Nalin
,
H.
Dai
,
T.
Xu
,
J.
Yin
,
F.
You
,
M.
Zhu
,
W.
Shen
, et al
.
2018
.
First-in-man clinical trial of CAR NK-92 cells: safety test of CD33-CAR NK-92 cells in patients with relapsed and refractory acute myeloid leukemia.
Am. J. Cancer Res.
8
:
1083
1089
.

The authors have no financial conflicts of interest.