What’s in a name? Memory NK cells for cancer immunotherapy

Abstract

The discovery that natural killer (NK) cells can retain features of “memory” from previous stimulation and pathogen exposure was a landmark advance highlighting one of many ways in which NK cells of the innate immune system resemble T cells of the adaptive immune system. This ability to “remember” prior stimulation to bring about enhanced protection of the host sparked significant excitement regarding potential therapeutic applications. Yet, how closely the features of naïve and memory NK cells recapitulate those of T cells remains unclear. Nonetheless, despite unresolved questions about the immunobiology of naïve and memory NK cells, the application of memory NK cells to the clinic for cancer and other indications has gathered steam to meet the unmet need for novel immunotherapies. Recent work from Arellano-Ballestero et al highlights this evolving field and the current state of the art with memory NK cells. Application of these cells to the clinic is progressing with promising results, but important questions remain about the essential molecular, phenotypic, and functional characteristics that define a memory NK cell.

Commentary

In Shakespeare’s classic Romeo and Juliet, Juliet proclaims: “What’s in a name? That which we call a rose by any other name would smell as sweet”. Though this line is commonly invoked to state that the names of things do not affect their true essence, the brilliance of Shakespeare is that this adage simultaneously captures the concept that names are not the true essence of things, but also that names carry significance in how things are viewed and judged. Within the field of immunology, this effect of names is relevant to how “memory” natural killer (NK) cells are viewed and applied clinically.

From their initial description and discovery in the 1960s and 1970s, NK cells have classically been described as non-T, non-B lymphoid cells of the innate immune system capable of spontaneous killing of target cells in an major histocompability complex (MHC)-independent manner. They are characterized as short-lived and continuously replenishing throughout life in marked contrast to T cells. In 2006, O’Leary and colleagues published one of several paradigm-shifting papers that identified a subset of NK cells that could mediate an adaptive, hapten-specific response with the capacity for recall responses after apparent resolution of the initial exposure.¹ A subsequent landmark paper by Sun and colleagues in 2009² demonstrated in a mouse model of cytomegalovirus (MCMV) that NK cells bearing the virus-specific Ly49H receptor proliferated by several orders of magnitude in the acute setting and could also be recovered from various organs for several months following a contraction phase. Importantly, they also observed that these MCMV “primed” NK cells could be transferred and re-expanded with protective immunity after MCMV challenge in naïve mice. Other studies have reported that prior exposure to certain cytokine combinations resulted in increased and accelerated NK cell responses to subsequent cytokine exposure in a non-antigen-specific manner.^{3 4} These observations of antigen-specific recall or accelerated responses suggested that “memory”, “memory-like”, or “adaptive” NK cells could exist and could offer a major therapeutic impact, especially in cancer therapy.

Following those and other seminal publications, the field of NK cell memory has evolved with studies ultimately identifying and characterizing memory-like features of NK cells in mice,^{2 3} humans,⁴ and non-human primates.⁵ These memory-like features of prolonged survival and immunologic recall have made them an appealing target for cancer immunotherapy, though these observations are based on viral infection models and not necessarily tumor rechallenge showing improved responses of a memory-like NK population. However, results from clinical trials do suggest that the memory-like NK population may mediate improved antitumor effects and demonstrate prolonged persistence in myeloid leukemias,6,9 though no clinical trial has definitively shown that receipt of memory-like NK cells leads to better outcomes compared with conventional NK cells. The prospect of a memory NK cell population being used to overcome the deficiencies of prior NK cell trials showing limited clinical efficacy is appealing given the relatively short lifespan of NK cells. However, what is currently described as “memory” may, or may not, align with what was called “memory” in those initial descriptions. By using “memory” in the name, however, there are certain qualities that may be inferred as analogous to classical T cell memory without rigorously recapitulating those cardinal features. Hence, more characterization of these properties and cells is needed.

The recent JITC publication from Arellano-Ballestero and colleagues¹⁰ adds to our understanding of memory NK cells through sophisticated and extensive characterization of memory-like NK cells using distinct cell culture methods. Using a combination of transcriptomics, proteomics, and metabolomics, the investigators present similarities and differences in these memory-like NK cells as compared with resting NK cells. Critical to this study are the methods for generating memory-like NK cells. Here, NK cells were exposed to either (1) low-dose Interleukin (IL)-15, (2) INB16 (a unique subclone of malignant cells from the CTV-1 leukemia cell line) or, (3) a combination of IL-12, IL-15, and IL-18. After a 16-hour culture, NK cells from these different conditions were collected and analyzed. What underlies this analysis is the assumption that these diverse activation strategies generate a homogenous memory-like NK cell population that reflects the previously described abilities of memory NK cells, namely persistence and recall.

The authors of prior reports have presented persuasive analogies between “adaptive” NK responses and those of well-described naïve T-cell responses leading to the subsequent generation of a small population of memory T cells.² Showing that a memory-like NK population could be induced in mouse³ and human⁴ NK cells, investigators have used a similar cytokine cocktail to activate the NK cells, then rested these cells for up to 3 weeks and detected still enhanced responses in this population. However, in the current study, the resting phase was omitted, and it remains unclear if these different methods of memory-like NK cell generation reflect the characteristics of memory NK cells, or rather unique activation states through distinct receptor/ligand interactions (ie, IL2R, NKG2D, KIRs, exposure to tumor targets) without true memory formation.

Should memory NK cells follow a developmental paradigm and exhibit properties similar to T cells, it becomes paramount to understand the biological niche of these cells. This may be in the context of viral resistance, particularly with chronic or latent viruses such as CMV. In the prior work investigating MCMV, those mice without the memory-like NK cell population were more likely to succumb to MCMV infection.² Complementary observations have also been made in humans with studies showing preferential expansion of unique NKG2C+NK cells in primary CMV infection¹¹ and following CMV reactivation in CMV seropositive patients receiving a hematopoietic stem cell transplant from CMV-seropositive donors.¹² These observations in humans lend support for a subset of NK cells that exhibit memory-like characteristics but are aimed at controlling acute and recurrent CMV infections. Thus, it is still unclear how to delineate a true memory population compared with NK cells being continuously activated by low-level viral exposure in these situations.

The not insignificant differences between T and NK cell biology complicate the narrative. There is a clear need for long-lived memory T cells throughout the life of the host as naïve T cells decrease and the thymus involutes with age. In contrast, bone marrow-derived NK cells appear to be replenished for the life of the host with a fairly rapid turnover rate (although there remains disagreement on the lifespan of human NK cells in vivo), and there is conflicting evidence on the changes in quantity and quality of circulating NK cells as humans age. The source of the NK cells also matters as investigation into NK cells within different tissues (vs more commonly used peripheral blood) has provided critical insights into the changes related to age, sex, and infectious exposures.¹³ Tissue-resident NK cells may be the area where we find the most rationale and evidence for longer-lived NK cells with “memory” characteristics, particularly in response to latent or chronic viral infections such as hepatitis viruses and Epstein-Barr virus (EBV).

Arellano-Ballestero and colleagues conclude their paper showing immune phenotyping from three patients with hematologic malignancies treated with a novel, replication-incompetent drug product hypothesized to induce prolonged tumor killing similar to the tumor-primed memory-like NK cells. Through serial blood draws from patients on trial, the investigators show similar phenotypic changes to those they observed in vitro—specifically, increased expression of granzyme B, CD57, and CD2. In the two patients with longer follow-up, these changes are identified in NK cells 29–73 days after the initiation of treatment, suggesting a prolonged effect on the circulating NK cells. A confounding variable not investigated is whether the drug therapy is still present within the circulation, since if it were present, this could potentially represent prolonged stimulation as compared with the development of a true memory-like NK cell.

Overall, evidence from diverse preclinical experimentation and clinical studies indicate that memory-like NK cells mediate increased cytotoxicity, have higher Interferon (IFN)-γ production, and have longer-term persistence in vivo with enhanced antitumor effects. Yet, key questions remain regarding what truly distinguishes a memory NK cell from other activated or stimulated NK cell subsets, particularly over longer periods of time and different degrees of stimulation. The clinical application of NK cells with memory-like properties holds promise to realize the long-desired breakthrough potential of NK cell-based approaches for cancer immunotherapy. This therapy holds great potential and early phase clinical trials are ongoing in hematologic and solid malignancies. Whether these therapies improve outcomes through induction of “memory” NK cells, or some other mechanism may be less important since our primary objective is more effective and durable cell therapy products for our patients. With that objective in mind, Arellano-Ballestero and colleagues seem to be on the right track regardless of what these cells are named.