It is well known that chronic stimulation of the TCR on T cells leads to activation and exhaustion, but the effect of chronic stimulation on NK cells is not as well understood. To investigate this, Merino et al. induced chronic stimulation in CD3-CD56dimCD57+NKG2C+ NK cells, deemed “adaptive” NK cells, which have a genetic profile resembling CD8+ effector memory T cells. Their results were recently published in The Journal of Clinical Investigation.
To begin, Merino et al. isolated NK cells from human cytomegalovirus (HCMV) seropositive individuals whose NK cells would have already been primed in vivo through NKG2C/HLA-E interactions. In a simple in vitro study, adaptive NK cells were cultured with IL-15 and with one of three plate-bound antibodies: a control antibody, an antibody that specifically targets the NK cell activating receptor NKG2C, or an antibody that targets both NKG2C and its inhibitory counterpart, NKG2A. Chronic stimulation with either experimental antibody induced expansion of NK cells expressing NKG2C and/or the maturation marker CD57. Interestingly, cells cultured in the presence of the NKG2A/C antibody showed enriched NKG2C expression relative to cells treated with the NKG2C antibody. Further, IL-15 stimulation was found to induce NKG2A expression and suppress NKG2C expression on adaptive NK cells over time. Together, this suggested that overstimulation of NKG2C may lead to downmodulation or internalization of the receptor, while NKG2A stimulation may counter overactivation – a homeostatic mechanism to limit exhaustion.
When naive CD8+ T cells are chronically stimulated, they tend to downregulate CD45RA expression in exchange for CD45RO expression. Looking for a similar isoform switch in NK cells, Merino et al. found that NK cells freshly isolated from HCMV seropositive donors uniformly showed high CD45RA expression and low CD45RO expression. Exposure to control antibody induced an isoform switch in some cells, while exposure to anti-NKG2C or anti-NKG2C/A induced isoform switching from CD45RA to CD45RO in the majority of cells. Within NKG2C+ cells, isoform switching was even more prominent and was most significant in the group treated with anti-NKG2C. Compared to CD45RA+ NK cells, CD45RO+ cells showed enhanced proliferative capacity.
Having determined that chronic stimulation enhances NK cell activation, Merino et al. investigated whether it also induced NK cell exhaustion. They found that chronic stimulation with either anti-NKG2C or anti-NKG2A/C led to significant upregulation of LAG-3 and upregulation of PD-1 on a portion of LAG-3+ cells. This suggested that LAG-3 may be an early indicator of exhaustion, while PD-1 may arise later on more severely exhausted cells. Exposure to anti-NKG2C induced the highest portion of NKG2C+ NK cells expressing both checkpoint receptors.
Next, Merino et al. investigated the functionality of adaptive NK cells both before and after chronic antigen stimulation. In response to K562 myeloid leukemia cells, freshly isolated NKG2C+ NK cells produced more IFNγ than NKG2C- NK cells. After culture with IL-15 and control antibody, anti-NKG2C, or anti-NKG2A/C, NKG2C+ NK cells expressing LAG-3 produced less IFNγ against K562 targets than NKG2C+ NK cells that were LAG-3-. Higher dosing of IL-15 also induced higher expression of LAG-3, PD-1, and CD45RO on NKG2C+ NK cells, indicating that cytokine stimulation contributes to effects of chronic stimulation in adaptive NK cells.
NK cells from HCMV seronegative donors did not show the same levels of expansion, upregulation of maturation markers or inhibitory molecules, or change in effector function upon chronic stimulation with IL-15 and anti-NGK2C or anti-NKG2A/C, indicating the importance of priming through NKG2C. In NK cells from HCMV seropositive donors, chronic stimulation of other activating receptors, including NKp30 and NKG2D, did not induce relative expansion of NKG2C+ NK cells, but did enhance CD45 isoform switching and checkpoint receptor expression on NKG2C+ NK cells, particularly on those cultured with anti-NKG2C. Together, these results indicate that chronic stimulation induces activation and exhaustion specifically through NKG2C on adaptive NK cells.
Moving beyond plate-bound agonist antibodies, Merino et al. cultured adaptive NK cells in the presence of human umbilical vein endothelial cells infected with a version of HCMV that expresses GFP. Although the frequency of NKG2C+ NK cells was similar in infected and mock-infected cultures, NKG2C+ NK cells from infected cultures showed more prominent isoform switching as well as upregulated LAG-3 and PD-1.
To better understand how chronic stimulation induces changes in adaptive NK cells, Merino et al. performed whole genome DNA methylation arrays on NK cells from HCMV seropositive donors cultured with IL-15 and either anti-NKG2A/C or anti-NKG2C. Following stimulation with either antibody, adaptive NK cells showed a gene signature of hypomethylation, including hypomethylation of genes like LAG3, PDCD1, and TIGIT, which are known to be associated with exhaustion in CD8+ T cells. This showed that NK cell exhaustion in response to chronic stimulation was epigenetically regulated.
Overall, Merino et al. show that chronic stimulation of adaptive NK cells drives expansion and activation, but also checkpoint inhibitor expression and reduced cytotoxic function against a tumor target. Meanwhile, inhibitory NKG2A may help limit excessive activation and terminal exhaustion. Much like chronic stimulation-induced exhaustion in CD8+ T cells, induced exhaustion in NK cells was epigenetically regulated. Because upregulation of checkpoint inhibitors correlated with limited cytotoxic function, exhausted adaptive NK cells are likely affected along with T cells by checkpoint blockade therapies like anti-PD-1 and anti-LAG-3.
by Lauren Hitchings
This week, Aimee Merino, first author on the paper and Frank Cichocki, lead author, took the time to answer our questions.
What prompted you to tackle this research question?
A.M: Adaptive NK cells are longer-lived than canonical NK cells and share many features with memory T cells. These adaptive NK cells are only found in individuals who have been infected with cytomegalovirus (CMV) but since CMV seroprevalence is estimated at 60 to 80% of the population, this is relevant to the majority of cancer patients. We know that cancer provokes exhaustion in T cells with well described metabolic, epigenetic, and transcriptional changes that allow for immune escape and tumor growth. We were interested in determining if adaptive NK cells might undergo similar changes under inflammatory conditions. We found that adaptive NK cells did become exhausted under conditions with elevated cytokines and engagement of their activating receptors. This impacted their ability to secrete cytokines and led to checkpoint receptor expression.
Immunotherapy has been the greatest advancement in cancer care of the past decade and has significantly expanded the treatment options available for many patients. Currently available immunotherapies predominantly target checkpoint receptors relevant to T cell exhaustion. It is our hope that a greater understanding of NK cell exhaustion will expand the immunotherapy targets that we have available to engage the innate arm of the immune system.
F.C: We were initially motivated to test whether we could expand adaptive NK cells from HCMV seropositive individuals using a simple culture system with anti-NKG2C antibodies. Along the way, we found that we could indeed expand adaptive NK cells, but that chronic stimulation also induced epigenetic reprogramming and functional exhaustion.
What was the most surprising finding of this study for you?
A.M: We were surprised to find that adaptive NK cells virtually never express the checkpoint receptor PD-1 until after expression of LAG-3. By examining different time points, we consistently found that LAG3 was expressed within days of exposure to inflammatory cytokines and activating receptor engagement. PD-1 was virtually never expressed on NK cells that did not already express LAG-3, indicating that LAG-3 is the earliest checkpoint receptor expressed as adaptive NK move towards an exhausted phenotype. This is important because it has been shown in T cells that expression of more checkpoint receptors indicates a lower likelihood of response to immunotherapy, further highlighting the importance of treatment options targeted to rejuvenation of NK cell function.
F.C: The most surprising finding to me was how similar the epigenetic reprogramming of exhausted adaptive NK cells is to that described in the literature for exhausted memory CD8+ T cells. In my opinion, this provides further support for the idea that adaptive NK cells possess memory characteristics.
What was the coolest thing you’ve learned (about) recently outside of the lab?
A.M: One interesting area outside of our laboratory is the recent focus on the microbiome and its role in cancer responses. It has been fascinating to see the many publications describing how the gut microbiome impacts everything from the metabolism of cancer drugs to responses to stem cell transplant. We look forward, as that field matures, to gaining a better understanding of how the microbiome interacts with and influences the immune system throughout the body.
F.C: The coolest thing that I’ve been learning about outside of the lab is how to do basic computer programming using LINUX.