As discussed above, the binding of PD-L1 expressed by tumor cell surface to PD-1 expressed by activated NK cells potentially inhibits NK cell-dependent immune surveillance and mediates antitumor immune responses. However, studies on the mechanism by which PD-1 inhibits NK cell response against tumors and the possibility of inducing NK cell responses via PD-1/PD-L1 blockade are still scarce. In one study, CD16- or ILCs-activated NK cells showed upregulated expression of PD-1, as well as CD69, CD107a, IFN-γ, and granzyme B; however, the expression of these molecules was downregulated and that of CD16 surface density was significantly downregulated after the binding of PD-1 to PD-L1 (33). PD-1 blockade increased NK cell activation and cytotoxicity, but only in patients with HNCs showing high PD-L1 expression (33). Another study used anti-PD-1/PD-L1 antibodies to block the interaction between PD-1and PD-L1 and found that this strategy could help reverse PD-1 inhibition and the dysfunctional state of PD-1⁺ NK cells. This blockade leads to a significant increase in NK cell cytotoxicity and cytokine production, inhibition of tumor growth in vivo, and obvious improvement in NK cell-based antitumor response (32). Similarly, a previous study found that blocking PD-1/PD-L1 signaling on the surface of NK cells significantly enhanced IFN-γ production, CD107a expression, and cell degranulation, and inhibited NK cell apoptosis in vitro. More importantly, PD-1 blocking antibody was found to significantly inhibit the growth of xenograft tumors, and NK depletion completely abolished this inhibition of tumor growth (29). However, inhibition of tumor growth could be completely abolished by NK depletion. Furthermore, PD-1 may exert its inhibitory effect on NK cells by interfering with AKT activation, and PD-1/PD-L1 blockade activates NK cells by enhancing the PI3K/AKT signaling pathway in them (29). In addition, anti-PD-1/PD-L1 antibodies can enhance the secretion of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) by inhibiting PD-1 in NK cells, effectively improving the antitumor activity of these cells (62). IFN-β has been shown to activate NK cells and induce cytotoxicity against tumor cells by upregulating NK cell-surface membrane-bound and soluble TRAIL expression, which leads to subsequent activation of the TRAIL receptor signaling pathway and apoptosis in nasopharyngeal carcinoma (NPC) cells (62). Subsequently, blocking the PD-1/PD-L1 checkpoint in the presence of IFN-β further increased the killing activity of NK cells against NPC cells, and this suggests that blocking PD-1 in activated NK cells may increase the secretion of soluble TRAIL and contribute to the killing of TRAIL NPC cells (62). This study revealed a new mechanism by which IFN-β and anti-PD-1 antibodies enhance the antitumor effect of NK cells and provide ideas for the therapeutic strategy of the combination of IFN-β and anti-PD-1.

Recently, researchers have found that repetitive irradiation can increase PD-L1 levels in non-small cell lung carcinoma (NSCLC) cells while reducing NKG2D ligand levels, which may reduce the sensitivity of lung tumor cells to the cytotoxic effect of NK cells and the possibility that tumor cells escape immune responses (63). Mechanistic studies revealed that IL-6-MEK/ERK signaling contributes most significantly to the upregulation of PD-L1 or downregulation of NKG2D ligand in radioresistant cells (63), while in another similar study, IL‐6‐JAK/STAT3 signaling was shown to contribute significantly to this process (64). Subsequently, researchers examined PD‐1 levels in NK cells and found that PD‐1 expression could not be detected in primary NK cells, but this expression increased when NK cells were exposed to tumor cells; in other words, there was a PD‐L1/PD‐1 interaction between tumor cells and NK cells, which inhibited the activity of NK cells (63). When neutralizing antibodies against PD-L1 were added to radioresistant cell/NK cell co-cultures, this resistance was reduced and the susceptibility of tumor cells to NK cell cytotoxicity increased, presumably because PD‐L1/PD‐1 interaction was blocked (63). In addition, the added PD-L1 antibodies effectively reversed NK cell activity by releasing PD-1 from the PD-L1–PD-1 complex; these antibodies also effectively reversed the expression of NKG2D ligand on tumor cells, which further enhanced the killing ability of NK cells against tumor cells (63). In summary, the PD-1/PD-L1 axis is an inhibitory/regulatory signal for the interaction between tumor cells and NK cells, and blocking of PD-1/PD-L1 interaction may be an effective antitumor immunotherapeutic strategy that is based on the reversal of NK cell dysfunction.

Anti-PD-1/L1 mAbs may act on NK cells through other non-PD-1 dependent pathways in addition to the PD-1/PD-L1 pathway. Upon encountering and being activated by NK-susceptible tumor cells, NK cells not only secrete cytokines and cytolytic granules, but also upregulate the expression of PD-L1 on their own surface through the PI3K/AKT/NF-κB pathway (8). This population of PD-L1⁺ NK cells is essential for the antitumor activity of PD-L1 mAbs. Upon application of atezolizumab, an mAb medication against PD-L1, PD-L1⁺ NK cells express significantly increased levels of granzyme B, IFN-γ, and CD107a, which results in a significant increase in antitumor activity and ultimately a significant decrease in tumor burden in mice (8). Furthermore, it was confirmed that anti-PD-L1 mAbs directly activate PD-L1⁺ NK cells through the p38/NF-κB pathway in PD-L1⁻ tumors (8). Upregulation of PD-L1 by NK cells, as well as the direct effect of atezolizumab on NK cells, leads to NF-κB activation, which creates a positive feedback loop in the presence of excessive immunotherapeutic agents to consistently induce PD-L1 expression and further activate NK cells (8). In the loop, the binding of anti-PD-L1 mAbs to PD-L1 upregulates PD-L1 expression on the surface of NK cells, thus, increasing the number of binding sites for the drug; this, in turn, leads to sustained activation of p38, which further transmits strong activation signals to NK cells to maintain their cytotoxic and cytokine secretion characteristics (8). The study suggested that anti-PD-L1 mAb therapy has a unique therapeutic effect against PD-L1-negative tumors (8). Based on the fact that PD-L1⁺ NK cells act through a PD-1-independent pathway, the discovery of new antitumor mechanisms provides insights into the activation of NK cells and a potential explanation for the response to anti-PD-L1 mAb therapy in some patients who lack PD-L1 expression.

Most PD-1/PD-L1 inhibitors (including nivolumab, pembrolizumab, etc.) have human IgG4 which has low fragment crystallizable (Fc) effector activity. These would be expected to have low ADCC. Atezolizumab is aglycosylated and would be expected to have no ADCC. These all prevent potent ADCC against non-tumor cells expressing PD-L1. However, avelumab, a fully human IgG1 anti-PD-L1 mAb containing a wild-type Fc that induces ADCC (65), has shown toxicity and efficacy similar to those of Fc-modified anti-PD-1/PD-L1 mAbs in several phase I and II clinical trials (66). FcγRIIIA expressed on NK cells can activate NK cell-induced cytotoxicity by recognizing the Fc portion of tumor-bound antibodies and releasing cytotoxic factors and cytokines that recruit and activate other immune cells with specific antitumor activity in the presence of tumor antigen-targeting antibodies (16). Investigators have demonstrated that avelumab triggers and enhances NK cell-mediated ADCC against TNBC cells expressing a certain level of PD-L1, which results in a significant increase in tumor cell lysis. Park et al. (59) also demonstrated that when NK cells were co-cultured with wild-type Fc anti-PD-L1 mAbs, which can induce ADCC effects on NK cells, the cytotoxicity against PD-L1-positive tumor cell lines was significantly enhanced. Therefore, some anti-PD-L1 mAbs may act as both tumor antigen-targeting antibodies and anti-PD-L1 inhibitors, which leads to the activation of NK cell and CD8⁺ T cell and improvement in therapeutic efficacy.

The magnitude of the ADCC effect of NK cells evoked by anti-PD-L1 mAbs may be modulated by other factors. First, polymorphisms in FcγRIIIA may affect NK cell-mediated interindividual variability in the ADCC effect (67). Although this association was identified in studies involving different tumor antigen-targeting mAbs of the IgG1 isotype (rituximab, trastuzumab, and cetuximab), it may not be generalizable to different settings (68). Preliminary in vitro results showed that avelumab-mediated ADCC was more effective in patients with NK cells expressing the high-affinity CD16 valine (V) allele than in those with NK cells expressing the low-affinity phenylalanine (F) allele and F/F genotype (65); however, this difference in effectiveness needs to be verified in clinical studies. Second, the release of some cytokines, such as IL-2 and IL-15, as well as subsequently stimulated IFN-γ, can stimulate the enhancement of avelumab-triggered cytokine production and degranulation in NK cells while increasing lytic activity against tumor cells (67). In addition, in a previous study, investigators enhanced the ADCC of avelumab against many types of cancer cells through epigenetic priming of NK cells and tumors (69). This evidence suggests that therapeutic strategies that block PD-1/PD-L1 while inducing the ADCC action of NK cells may enhance existing immunotherapeutic efficacy. Considering that the magnitude of NK cell-based ADCC effects induced by anti-PD-L1 mAbs can be regulated by some factors such as immunomodulators (IL-15 or IL-2), combining drugs targeting these factors (67) may improve the effectiveness of immunotherapeutic strategy.

Bromodomain proteins, such as BRD1 and BRD4, play a role in the development of immune and hematological cells and in the regulation of tumor inflammation (70–72). In a previous study on high-grade serous ovarian cancer (HGSC), BRD1 expression was found to be low in tumor cells and high in immune cells, which is associated with significant downregulation of T cell- and NK cell-surface activity markers (GZMA, GZMB, IFNG, and NKG7) and upregulation of the naïve T cell marker TCF7 (73). This study performed immune function and single-cell RNA-seq transcriptional profiling of novel HGSC organoid/immune cell co-cultures treated with unique bispecific anti-PD-1/PD-L1 antibodies versus monospecific anti-PD-1 or anti-PD-L1 antibodies (control) (73). It revealed that bispecific antibodies uniquely induced a transition from inert to more active and cytotoxic NK cell phenotypes and a transition from naïve to more active and cytotoxic progenitor-exhausted phenotypes of CD8⁺ T cells after treatment. It was further found that these superior cell state changes were driven in part by the downregulation of BRD1 expression induced by bispecific antibodies in immune cells (73). The inhibitory effect of the small molecule inhibitor, BAY-299, on BRD1 partially leads to increased NK cell maturation, activation, and tumor cell killing via alteration of the chromatin pathway of key immune transcription factors (e.g., GATA3, TBX21, and TBXT), induces similar state transitions in immune cells in vitro and in vivo and demonstrates the in vivo efficacy of bispecific antibodies (73). Therefore, changes in the activity and cytotoxic status of NK cells and T cells may be the key to driving the induction of effective antitumor immune responses using bispecific antibodies, by partially eliminating some TME-driven dysfunction through epigenetic changes (BRD1 downregulation or inhibition).

Regulative effects of NK and CD8⁺ T cells on each other have been reported in many infection models (74, 75) and antigen-independent IL-2 models (76), and PD-1/PD-L1 inhibitors may indirectly alter NK cell function by affecting CD8⁺ T cell activity. NK cell exhaustion (NCE) has been identified as a self-regulatory mechanism responsible for inducing dysfunctional phenotypes to prevent exacerbated immune responses under conditions of chronic stimulation (77), and it is a crucial mechanism involved in tumor or viral evasion of immune responses. IL-2 is a pleiotropic cytokine that activates T cells, NK cells, and dendritic cells. IL-2 binds to IL-2 receptors (IL-2Rs), including CD32 (IL-2Rγc), CD122 (IL-2Rβ), and CD25 (IL-2Rα). IL-2Rs showed different affinities for cytokines, with CD25 (IL-2Rα) having the highest IL-2 affinity. CD25 is constitutively expressed on Tregs; therefore, IL-2 treatment can lead to the expansion of these cells, thus mediating NK and CD8⁺ T cell suppression through multiple mechanisms (76, 78, 79). A recent study evaluated the role of the PD-1/PD-L1 pathway in NK cell activation, function, and depletion in a mouse model of IL-2-dependent depletion (80) and found that anti-PD-1 therapy provides an activating advantage to CD8⁺ T cells in the competition for IL-2 by promoting CD8⁺ T cell expansion, activation, and functional phenotype; consequently, the quantity of stimulatory cytokines available to NK cells becomes limited, which results in a delayed NCE, improved NK cell activation, higher proliferative capacity, and enhanced granzyme B production. This evidence suggests that the phenotypic and functional benefits observed after the chronic stimulation of NK cells by anti-PD-1 therapy are indirectly mediated by the effect of anti-PD-1 antibodies on CD8⁺ T cells, rather than by the direct effect of anti-PD-1 drugs on NK cells; this can be demonstrated by the reversal of the benefit of anti-PD-1 therapy on NK cells by the depletion of CD8⁺ T cells. Disruption of the balance between NK and CD8⁺ T cells affects the homeostatic response to each other’s stimuli, and this resource competition (i.e., IL-2) delays the onset of NCE. Thus, there is a delicate balance between CD8⁺ T cells, Tregs, and NK cells that is partly regulated by their ability to respond to cytokines. Achieving a balance between these immune cells may be important for achieving long-term efficacy of immunotherapy. Although dual regulation between CD8⁺ T cells and NK cells is caused by competition for space and resources, direct or indirect lysis, and functional inhibition, these two populations can also work together to mount a stronger response. Another study found that dual blockade of PD-1 and IL-10 enhanced cytokine secretion, NK degranulation, and killing target cell function of NK cells by restoring HIV-specific CD4⁺ T cell function, thus establishing a previously unappreciated relationship between CD4⁺ T cell injury and NK cell depletion in HIV infection (81). This important evidence also fully suggests that PD-1 blockade may enhance immunotherapy by improving collaboration between CD4⁺ T cells and NK cells.

Immunosuppressive cells such as tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs), and Tregs in the TME inhibit NK cell proliferation, infiltration, and activation by secreting immunosuppressive cytokines (TGF-β and IL-10) or by interfering with NK cell receptor expression and activation; thus, these cells facilitate tumor immune escape and promote progression and metastasis (82, 83). It can, therefore, be hypothesized that PD-1/PD-L1 blockade prevents the induction of immunosuppression and improves NK cell efficacy to increase the survival of a tumor-bearing animal. A previous study showed that anti-PD-L1 therapy has no direct effect on the cytotoxicity or cytokine secretion of PD-1-negative PM21 particle-expanded NK cells in response to PD-L1⁺ targets in vitro; however, secretion of a large quantity of IFN-γ by NK cells significantly improved antitumor efficacy, and long-lasting retention of their cytotoxic phenotype by NK cells can be observed in vivo (84). Thus, the investigators continued to explore the specific mechanism by which anti-PD-L1 therapy enhances the antitumor efficacy of NK cells. They found that PM21-NK cells are highly cytotoxic to tumor cells and secrete IFN-γ upon stimulation, which leads to the induction of PD-L1 expression in tumors and consequently to the induction and in situ proliferation of Tregs in the TME (84). Subsequently, blockade of PD-L1 mitigated the induction of Treg expansion and associated immunosuppressive responses, which in turn improved the NK cell phenotype and cytotoxic activity, as well as survival in treated animals (84). The number of CD57⁺ NK cells has increased, a population with high cytotoxic capacity and responsiveness via CD16 binding, and the presence of these cells is correlated with better outcomes in patients with squamous cell lung cancer in other reports (85, 86). The expansion of Tregs may occur in the TME with PD-L1 expression, and Tregs have been reported to suppress the survival and cytotoxic function of NK cells through several mechanisms, including TGF-β surface presentation (87–89). Thus, anti-PD-L1 therapy may improve NK cell antitumor efficacy by reducing in situ Treg generation and preventing Treg induction.