Head and neck squamous cell carcinoma (HNSCC) is one of the most common cancer with 878348 new cases and 444347 deaths in 2020 worldwide (1). HNSCC includes carcinoma originating from oral cavity, nasopharynx, hypopharynx, oropharynx and larynx, is closely associated to alcohol abuse and tobacco smoke and it may be associated with infection with oncogenic strains of human papillomavirus (HPV), such as HPV-16 and HPV-18. HPV positive HNSCC have a better prognosis than HPV-negative (2, 3). Treatment depends on the stage and the site of the disease, primary treatment options are surgical resection followed by chemotherapy and radiotherapy. The monoclonal antibody Cetuximab directed against epidermal growth factor receptor (EGFR) is used alone or in combination with chemotherapy, for the treatment of patients with recurrent or metastatic disease (4). Recently, the immune checkpoint inhibitors Nivolumab and Pembrolizumab have also been approved for the treatment of cisplatin-refractory recurrent or metastatic HNSCC (5, 6). Most patients have an advanced stage disease at diagnosis, because of the lack of important symptoms or reduced symptom awareness (7) and a local recurrence and lymphnode metastasis are frequently observed.

Unfortunately, five year survival rate is quite low, (56%-62%) and it has modestly improved over the years (8–10). A better survival is associated with the presence of tumor immune infiltrate (11–13), as shown by patients with a tumor microenvironment enriched in tumor infiltrating lymphocytes (TIL). Interestingly, most HPV positive HNSCC tumors are highly infiltrated by lymphocytes and have the best survival rates (13, 14), suggesting that the number of TIL plays an important role in the control of tumor growth. A recent study has shown that tumor immune microenvironment (TIME) in HNSCC is associated with immune checkpoint therapy response and better prognosis (15). Furthermore, the presence of defective NK genes related with NK cell cytotoxicity pathway and phenotypes, were associated with higher risk of developing cancer (15). Increasing data have shown that NK cells are not merely innate effector cells, but they are fundamental in the recruitment and activation of adaptive immunity, through the secretion of IFNγ which induces activation of T helper cells (15, 16). Moreover, in recent years different studies supported an increasing potential of NK cells in the control of different types of cancers, including HNSCC, where a favorable prognosis has been associated with NK cell infiltration (17–21).

However, a number of immune escape mechanisms are exploited by tumors to avoid NK cell recognition, thus a better knowledge of peripheral and intratumoral NK cell profile in patients with HNSCC is needed to identify possible therapeutic targets.

The activity of NK cells is finely tuned by several membrane receptors able to switch the NK cell function towards activation or inhibition after ligand recognition and interaction (22). The low affinity FcγRIIIA (CD16) is one of the most potent activating receptors of NK cells that mediates antibody-dependent cell-mediated cytotoxicity (ADCC) (23). It is associated with ITAM-containing

CD3 ζ-chain and its activation allows NK cells to recognize and kill antibody-coated target cells via ADCC and to release immunoregulatory cytokines.

Spontaneous NK cell lytic activity is mediated by several activating receptor-ligand axes, including the natural killer group 2, member D (NKG2D) activating receptor, and its ligands. NKG2D is a dimeric, type II membrane protein constitutively expressed by NK cells in humans, as well as by almost all CD8+ αβ T cells and γδ T cells (24). NKG2D receptors recognize several molecules, such as MHC class I chain-related molecules (MIC)A/B and the cytomegalovirus UL-16 protein (ULBP1-6), the expression of which is induced on the cell surface as a result of cellular stress conditions (25). Therefore, NKG2D plays a pivotal role in immune surveillance and antitumor immune responses (26–28). NKp30 is a natural cytotoxicity receptor constitutively expressed by NK cells whose engagement by its ligand transduces a strong activation signal to the cell (29). Siglec-7 is an inhibitory NK receptor whose reduced frequency is associated with decreased cell function in physiologic (30) and pathologic conditions (31–34).

PD-1 and NKG2A are checkpoint molecules that have been shown to harness NK and T cell immune responses in cancer (35). NKG2A forms a heterodimer with CD94. The NKG2A/CD94 complex binds to the non-classical MHC I molecule HLA-E in humans and transduces inhibitory signals which suppress NK and CD8+ T cell activity (36). Interestingly, NK cells are endowed by innate and adaptive properties, being able of long-term ‘memory-like’ immune responses (37, 38). Individual repertoire of NK cells may be shaped by environmental exposure, such as CMV infection, which is able to lead to a stable expansion of adaptive NK cell population characterized by the expression of NKG2C activating receptor (39–44). Sequence variations in CMV UL40-encoded peptides control the activation and expansion of adaptive NK cells (40–42), and the peptide variant VMAPRTLFL induces increased proliferation of NKG2C population compared with other CMV peptides (40, 41). Such expanded NK cells are identified by the lack of FcϵR1γ adaptor protein and display an altered receptor expression pattern characterized by reduced expression of NKp46, NKp30, CD16, Siglec-7 and NKG2A compared to conventional, FcεR1γ positive NK cells. Importantly, they have increased ability to perform ADCC and to produce IFN gamma (44–48).

In this study we explore immune infiltrating NK cell phenotype and function in intratumor, peritumor and peripheral blood of HNSCC patients and controls and evaluate NK cell function after CMV peptide-driven expansion of adaptive NK cells.

In this study we have examined the NK and T cell immunophenotype in peripheral blood and in tumor tissue of patients with HNSCC undergoing surgery. Other studies had previously characterized HNSCC circulating immune cells; however most were limited to assessing the distribution of the principal lymphocyte populations (20, 52–54) and only few analyzed NK cell inhibitory and activating receptors (55–57). In agreement with others, we observed reduced frequencies of the immunoregulatory CD56^bright NK cell population (58) and increased frequencies of TIM-3+ and NKG2A+ NK cells (56, 59) in the peripheral blood of patients with HNSCC compared with healthy donors. Moreover, the proportion of Siglec-7+ NK cells was significantly reduced in patients. Interestingly, a recent study showed that reduced Siglec-7 expression predicts NK cell dysfunction in hepatocellular carcinoma (60), confirming the role of this receptor in the modulation of NK cell activity, as already observed in physiological and pathological conditions (30–34).

Several recent studies pointed to the relevance of the immune infiltrate in HNSCC, showing the prognostic role of TIL specific markers, including CD56 and CD57 (20, 61, 62), CD3, CD4 and CD8 (63–68). A meta-analysis showed that M2 macrophages and CD57+ natural killer cells were the most promising predictors of survival in oral cancer patients (69).

On this basis, it has been proposed that incorporation of TIL evaluation in the TNM system could improve the prognostic performance, and help physicians in clinical decisions. NK cells are potentially very effective against tumor growth, being able to kill tumor cells when activating signals prevail on inhibitory signals. However, the tumor immunosuppressive microenvironment may inhibit NK cell function by tipping the balance toward inhibition. Our study shows that intratumoral NK cells are characterized by an inhibitory phenotypic pattern, with reduced frequencies of cells expressing the major activating receptors NKG2D, NKp30 and CD16. The importance of NKG2D in tumor control is highlighted by the fact that tumors release NKG2D ligands MICA/B and ULBP1/2/3 by proteolytic cleavage to escape NKG2D-mediated NK cell cytotoxicity (70) and that antibody targeting the site of proteolytic cleavage prevent release of MICA/B and reduce melanoma metastasis in a humanized mouse model (71). This mechanism of immune evasion has also been observed in HNSCC in an interesting study showing that increased levels of soluble NKG2D ligands are present in the plasma of HNSCC patients (72) and that depletion of these ligands from patients’ plasma restores the ability to kill target cells in vitro (73). In agreement with others (57, 74) we have observed increased frequencies of NKG2A positive intratumoral NK cells compared with matched PBMC. Different types of tumors overexpress NKG2A and its ligand HLA-E (75, 76), and blocking NKG2A is an effective strategy to enhance antitumor activity by NK and CD8 T cells (77). We observed high level of HLA-E in primary tumor cells obtained from 4 patients with HNSCC ( Supplementary Figure 5 ).

Not surprisingly, the functional activity of intratumoral NK cells was reduced, reflecting the prevalent inhibitory phenotype, which was also accompanied by an enrichment in PD-1+ TIGIT+ NK cells

and a reduction in CD16, a receptor capable to bind the antibody Fc fragment, thus enabling NK cells to perform ADCC. Downregulation of CD16 is induced in different tumors and viral infections, through matrix metalloproteinases (MMPs) induced shedding (78, 79), as an escape mechanism to avoid ADCC. Among MMPs, ADAM-17 mediated cleavage has been recognized as responsible of CD16 shedding in breast cancer (78, 80). ADCC process is triggered by Cetuximab, an anti EGFR antibody, which is widely used in cancer immunotherapy, including locally advanced- or recurrent/metastatic HNSCC (4, 81). Cetuximab engagement of CD16 activates NK cells and induces IFNγ secretion (82), an important soluble factor which promotes dendritic cell maturation leading to increased tumor antigen presentation and expansion of EGFR specific T cells (83). Our data show that intratumoral NK cells have reduced ability to secrete IFNγ in the presence of cetuximab after stimulation with a potent combination of cytokines, while peripheral NK cells from patients produced IFNγ similarly to controls. Interestingly, patients with higher frequencies of TIGIT expressing NK cells produced less IFNγ in the peripheral blood and intratumor compartments. TIGIT is a co-inhibitory receptor on the immune cell surface that binds ligands CD155 and CD112 (84). Different strategies are employed to increase NK cell activity against solid tumors in a number of clinical and pre- clinical studies (85). Blocking TIGIT interaction with its ligands reduces NK cell exhaustion in vitro and in a murine model (86, 87). Moreover, several anti-TIGIT monoclonal antibodies are under investigation in phase I/II studies (88).

Similarly to NK cells, intratumoral T cells were characterized by an exhausted phenotype, with increased proportions of TIM-3, PD-1 and TIGIT expressing CD4 and CD8 T cells. Moreover, patients’ Treg cells were increased in both the circulating and intratumoral compartments, further supporting the inhibitory role of the TME.

Increased frequencies of GITR+ NK cells were found in TIL and, in agreement with others (89, 90) we also observed a significant enrichment of GITR+ CD4 and CD8 T cells. GITR is a costimulatory receptor which has a dual positive effect supporting the survival and expansion of activated T cells and at the same time deactivating the Treg population within tumors (91). For this reason, GITR represents an ideal targetable immunomodulatory receptor that could provide a powerful tool to diminish the suppressive potential of intratumoral Treg and enhance anti-tumor immunity. Of note, different clinical trials are testing GITR agonist antibodies in combination with immune checkpoint inhibitors in advanced tumors (92–95). On this basis, considering the increased frequencies of NK and T cells expressing GITR, the use of agonistic GITR combined with blocking PD-1 and/or TIGIT antibodies could be beneficial in patients with HNSCC.

Increased proportions of CXCR6 were found in intratumoral NK and T cells, probably caused by ligand-receptor chemotaxis of CXCL16/CXCR6. Interestingly, a recent study in mice showed that CXCR6 is highly expressed in intratumoral CD8+ T cells and that these cells are more immunocompetent and could enhance the effect of anti-PD-1 blockade to delay tumor progression (96). CXCR6 has been recognized as a marker for liver resident NK cells (97) and has been associated to murine CMV virus specific adaptive-memory NK cells (98, 99). Adaptive NK cells have peculiar characteristics that have been exploited in adoptive anticancer therapy, being characterized by increased ADCC activity and by resistance to myeloid-derived suppressor cells and to Treg-mediated suppression (50, 51). Interestingly, a lower leukemia relapse rate and a superior disease-free survival at 1 year has been associated with CMV reactivation and this protective effect strongly correlated with adaptive NK-cell expansion (100). We have observed an increased IFNγ secretion, but not CD107a expression, by NK cells after 12 days expansion of adaptive NK cells in CMV+ patients, confirming that these cells are specialized in IFNγ secretion after Fc binding of therapeutic antibodies and that NK cell expansion by UL40 peptide stimulation combined with IL2 is a feasible way to obtain the expansion of highly functional NK cells in CMV positive subjects.

In conclusion our data show that the HNSCC tumor microenvironment affects NK and T cells, inducing an inhibitory/exhausted phenotype which is reflected in reduced NK cell ability to produce IFNγ after cetuximab engagement. The concomitant increase of costimulatory receptor GITR in intratumoral NK and T cells suggests that agonist engagement of GITR could be an effective treatment in HNSCC patients, particularly if combined with immune checkpoint blocking antibodies anti-PD-1 and anti-TIGIT.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board and Ethical Committee of Fondazione IRCCS Policlinico San Matteo, Pavia, Italiy. Document number 20190063538. The patients/participants provided their written informed consent to participate in this study.

DM, SV, and MM provided substantial contribution to the conception and design of the study, acquisition, analysis and interpretation of data, and revised the manuscript critically for important intellectual content. SV and MM supervised the team, obtained funding and had leadership responsibility for the research activity planning and execution. BO, SM, and GP contributed to acquisition, analysis, validation and interpretation of data. GT, AL and MB were responsible for data curation and were in charge of patient care. PM performed the histopathologic analysis. All authors critically read, edited and approved the final version of the manuscript.

This work was funded by Ricerca Corrente, Ministero della Salute, Nr. 08071419 to SV.

We thank patients and donors included in the study.

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

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.