Natural killer cells have been known and actively studied for more than four decades. They were first described as large granular lymphocytes cytotoxic for various tumor cells without prior stimulation (1, 2). In addition to their cytolytic activity against neoplastic and virus-infected cells, NK cells also display immunomodulatory functions by their ability to release cytokines, like interferon-γ (IFNγ) and tumor necrosis factor-α (TNFα), and chemokines. NK cells represent 5–15% of blood lymphocytes. They are present in the bone marrow, liver, uterus, spleen, lungs, in mucosa-associated lymphoid tissues, thymus, and secondary lymphoid tissues (SLT) and are recruited in inflamed sites. In SLT, NK cells provide an early source of IFNγ and interact with dendritic cells to promote T helper cell type 1 responses (3).

Natural killer cells are now grouped in the system of innate lymphoid cells (4). These populations, mostly tissue resident and characterized by their capacity to produce high amounts of cytokines, constitute innate homologs of T helper cell (CD4) and cytotoxic T cell (CD8) subsets. ILCs are implicated in tissue homeostasis and autoimmune diseases. Their distribution and capacity to produce cytokines suggest that they may also be involved in the development or evolution of cancer. NK cells are considered as cytotoxic counterparts of ILC1, both depending on the T-bet transcription factor for their development.

Human NK cells, defined as CD45⁺/CD3⁻/CD56⁺ cells (5), are classically subdivided in two subsets based on the relative membrane expression of CD56 and CD16, the low-affinity receptor for the Fc portion of IgG (FcγRIIIA): CD56^dim NK cells that express high levels of CD16 mediate antibody dependent cell cytotoxicity (ADCC), whereas CD56^bright NK cells express no or low levels of CD16. These two subsets are present in different proportions in the different tissues. CD56^dim NK cells represent 90% of blood and splenic NK cells, while CD56^bright NK cells predominate over CD56^dim in the SLT [lymph nodes (LN) and tonsils] representing up to 90% of NK cells and also constitute the major NK subset in tissues. It is accepted that CD56^bright NK cells are less mature than CD56^dim NK cells and display an immunoregulatory function, secreting high amounts of IFNγ and TNFα. CD56^dim NK cells represent mature NK cells with a high cytotoxic activity (6).

The activation of NK cells is tightly regulated by a balance between activating and inhibitory signals delivered through engagement of numerous activating and inhibitory receptors with ligands on the target cell. Natural cytotoxicity receptors (NCRs), such as NKp46 and NKp30, are expressed by resting NK cells while NKp44 is induced after activation by cytokines, such as IL-2 and IL-15 (7, 8). The NCRs are implicated in the lysis of various tumor cells (9). The activating NK group 2 member D (NKG2D) receptor is expressed by most circulating NK cells and binds the stress-induced MHC-class I polypeptide-related sequence (MIC)-A/B molecules and UL16-binding proteins 1–6 (ULBP1–6) (10). DNAX accessory molecule-1 (DNAM-1) binds Nectin family molecules CD155 and CD112.

Natural killer cell activation is efficiently controlled by specific inhibitory NK receptors binding human leukocyte antigen of class I (HLA-class I) molecules. The C-type lectin CD94/NKG2A receptor binds HLA-E molecules (11) sensing the global HLA-class I molecules on the target while killer Ig-like receptors (KIRs) bind classical HLA-class I molecules, including HLA-C, HLA-Bw4, and some HLA-A alleles.

A link between NK cell function and cancer development was reported in a Japanese 11-year follow-up study including 3,625 patients in which cancer incidence was negatively correlated with blood NK-mediated cytotoxicity (12). Authors further showed that individuals with particular NKG2D haplotypes, HNK1/HNK1 haplotype (correlated with high NK activity) had a decreased risk of cancer compared to those with an LNK1/LNK1 haplotype (correlated with low NK activity) (13).

Additional results including ours showed the impact of NCR transcripts in the evolution of melanoma, lung cancers, and gastrointestinal stromal tumors (GIST) patients (14–16). High NKp46 correlated with better survival in metastatic melanoma patients and particular profiles of NKp30 isoforms was associated with better outcome and response to treatment in GIST patients.

The cancer immunoediting process (17) resumes cancer progression in three phases. In the elimination phase, immune cells and among them NK cells eradicate developing tumor cells. During the equilibrium phase, the immune system may select tumor variants with less immunogenicity gradually leading to the tumor escape phase and tumor progression. It is considered that most tumors at diagnosis are in the phase of immune escape associated with functionally altered tumor infiltrating NK cells (18). Tumor immunoediting selecting variants with decreased expression of stress-induced ligands provide tumor escape to NK cell-mediated lysis through activating receptors NKG2D or NKp46 (19, 20).

The challenge is thus to overcome tumor immunosuppression and restore NK cell activities. To this aim, understanding the mechanisms that lead to NK cell defects in tumor is required.

Numerous studies showed that severe quantitative and qualitative alterations of NK cells are associated with different hematological malignancies, particularly in myeloid disorders. In chronic myelogenous leukemia patients, low numbers of NK cells are associated with defects in their proliferation, and weak NK cell cytolytic functions in comparison with healthy donor blood NK cells (21). Furthermore, profound alterations in the activating receptors profile have also been reported including downregulation of NKp30 and NKp46 as well as DNAM-1, 2B4, and NKG2C on NK cells from acute myeloid leukemia (AML) patients. Decreased NKp30 and NKp46 expression was correlated with reduced NK cell killing and poor leukemia prognosis (22–25). Recently, Khaznadar et al. analyzed by cell imaging the lytic NK immunological synapse following interaction with AML cells and showed defective lytic granule polarization in NK cell-AML conjugates leading to impaired NK cell cytotoxic function (26).

Importantly, the intimate relationship between immune pressure and leukemogenesis has been suggested in two recent studies. Stringaris et al. described an immunoediting process induced by AML blasts that limits NK cell control of leukemia. They showed that abnormal NKG2A expression and TNFα production predict a poor response to chemotherapy in AML patients (27). Conversely, Khaznadar et al. showed that NK cell defects in AML patients at diagnosis could be associated with a specific transcriptional program in AML blasts and with patient’s outcome including relapse occurrence (28).

Furthermore, the beneficial role in the graft-versus-leukemia (GvL) of allogeneic NK cells for leukemic patients receiving allogeneic hematopoietic stem cell transplantation (HSCT) is well documented (29). Several studies showed that NK cells have a potent GvL effect in both KIR/HLA-class I-mismatched and -matched donor–recipient combinations after allogenic HSCT in AML patients (30–32). Moreover, rapid NK recovery after HSCT is also associated with a greater GvL effect and improved outcome in AML patients (33).

In situ detection of NK cells infiltrating various human tumors/tissues was carried out, leading sometimes to divergent results due to the disparity of NK cell markers used (CD57, CD56, NKp46, double CD3/CD56 staining). However, several reports showed that NK cells can infiltrate clear-cell renal cell carcinoma (34), melanoma (35), non-small cell lung cancer (NSCLC) (36), breast cancer (BC) (37), GIST (38), and colorectal carcinoma (CRC) (39) although NK cells were mainly localized at the tumor’s periphery. In several tumors, infiltrations by NK cells were reported to have a prognostic value. Increased overall survival was associated with a high NK cell infiltrate within the tumor or tumor stroma in lung adenocarcinoma (40), metastatic renal carcinoma (41), and lung metastasis of renal cancer (42). Elevated number of NK cells was associated with reduced risk of cancer progression in prostate cancer (43), with a reduced risk of death in squamous cell lung cancer (44), and a better prognosis in gastric carcinoma (45) and CRC (46). In addition, the number of NKp46⁺ NK cells was found inversely correlated with metastasis occurrence in patients with GIST (47). Furthermore, a positive association between a high numbers of tumor infiltrating CD56⁺ NK cells with a regression of melanocytic lesions was observed (48).

In most tumor types studied, ex vivo tumor-infiltrating NK cells displayed severe phenotypic and functional alterations compared to blood NK cells and more interestingly compared to NK cells present in adjacent normal tissues. Those alterations affected the expression of activating receptors including NKp30, CD16, DNAM-1, and ILT2 on NK cells from patients with non-invasive and invasive BC (49) or NSCLC (36). A concomitant-increased expression of the inhibitory molecule NKG2A was also observed in BC (49). This deficient phenotype was associated with impaired functions including decreased cytotoxicity against tumor cells (36, 49) and reduced IFNγ production (36). Recently, Carrega et al. reported that lung and BC tissues were highly enriched in CD56^brightperforin^low NK cell subset compared to matched normal tissues (37). It is of note that comparison between NK cells from tumor and normal adjacent tissue is required for better understanding of the effect of the tumor environment on their activation.

Interestingly, our team recently identified in tumor draining LN from melanoma and BC patients, the presence of a CD56^brightCD16⁺ NK-cell subset that displays higher expression of activating receptors, perforin molecules, and performs ADCC (50). We found that different NK receptors regulate the two LN-NK cell subsets in melanoma and BC (personal communication) and that NK-infiltrating LN recapitulate the alterations reported in the primary tumors. The presence of CD16⁺ NK cells in certain tumors (51) and metastatic LN emphasizes the interest for ADCC function of such NK cells.

The advent of targeted therapies that counteract a vital cellular process within the tumor cell greatly improved cancer treatment strategies. Thus, mitogen-activated protein kinase (MAPK) inhibitors that control the mutation-driven oncogenic pathway present in most cancers are new efficient players in the arsenal of therapies for cancer patients. In addition, monoclonal antibodies (mAbs) that recognize tumor-associated antigens have been established as one of the most successful therapeutic strategies for both hematologic malignancies and solid tumors. These mAbs may activate antibody-dependent cell-mediated cytotoxicity involving NK cells.

Combining targeted therapies and methods to stimulate patient’s immune players is actively evaluated and represents a promising and natural evolution in cancer treatment as this could ally immediate efficiency, specificity, and long-term antitumor efficacy.

It is of note that targeted therapies also display off-target effects, connecting oncogenesis to immunosurveillance. We discuss below the interest of NK cell-based therapies in the context of such tumor-targeted therapies (Figure 1).

Restoring NK cell functions in addition to administration of tumor-specific therapies with kinase inhibitors or tumor-specific mAbs may benefit patients. It would increase the control of residual tumor cells, enhance mAbs efficiency, and promote the adaptive immune response necessary for long-lasting protective immunity. In that context, cytokines, blockade of inhibitory NK receptors (KIRs, NKG2A), or transfer of alloreactive NK cells are promising NK-based therapies.

The study protocol was approved by an ethic committee “Ile de France” (CPP: 2834), and the Declaration of Helsinki protocols were followed.

MM, ND, and AC wrote the manuscript; MM did the figure; AF, PG, and AT read and corrected the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.