1. NK Cell Function Related to Metastasis
NK cells were first found for their capability to eliminate tumor cells without prior stimulation by antigen-presenting cells
[63,64][1][2]. Studies have reported that patients with a higher cancer incidence had inadequate peripheral NK cell cytotoxicity responses in various types of cancer
[65,66,67,68,69][3][4][5][6][7]. Additionally, the patients who suffered NK cell dysfunction had an increased rate of malignancies and metastases
[70,71,72,73,74,75,76,77][8][9][10][11][12][13][14][15]. Using an experimental metastasis assay,
wresearche
rs and others found that antibodies mediated depletion of NK function or the use of an NK cell-deficient host increased metastasis, suggesting that NK cells play an essential role in antimetastasis (
[78][16],
our the unpublished data). An intravital imaging system directly observed NK cells attack disseminated tumor cells leading to cancer cell death in mouse models
[78][16]. These observations indicated that NK cells act as a killer in tumor progression, especially cancer cells spreading from their original site to the bloodstream
[29,62][17][18]. Clinically, the patients who suffered various cancers with low amounts of peripheral or infiltrating NK cells at tumor sites have higher numbers of metastatic lesions.
[74,79,80,81,82,83,84][12][19][20][21][22][23][24]. In contrast, the patients who suffered an increased metastatic cancer risk with high levels of NK cell activating receptors (NKAR) have good prognoses
[77,85,86,87,88,89][15][25][26][27][28][29]. Moreover, a high level of IFNγ production by circulating NK cells and the presence of NKp30 are associated with a positive prediction for long-term survival in patients with breast cancer, gastrointestinal stromal tumor, or melanoma under treatments
[69,79,80,90,91,92][7][19][20][30][31][32]. More recently, a study has shown that NK cells could sustain breast cancer dormancy for controlling breast cancer liver metastasis
[93][33]. These facts suggest the notion that NK cells mediate antimetastatic effects in clinical. The notion has also been strongly linked to cancer treatment and preventing tumor metastasis. For example, after successfully removing the tumor by surgery, many patients still developed distant metastasis later; one of the important reasons is that low NK cell cytotoxicity and less IFNγ secretion link impaired NK cell function directly to increased postoperative metastases
[94,95,96][34][35][36]. Thus, the functional restoration of NK cells by agents such as arginine prevents metastases
[97,98][37][38].
2. How NK Cells Kill Metastases
It is well appreciated that the amount of positive and negative signals transmitted by NK cell activating or inhibitory receptors determines the fate of the targets
[99][39]. Normal cells typically express low amounts of activating ligands and have higher amounts of MHC-I molecules that interact with inhibitory receptors and transduce more negative signals; therefore, NK cells will not eliminate normal somatic cells. In contrast, in malignant tumor cells, the expression of activating ligands has been increased, and the positive signals will overcome the negative signals, thus ensuring that aberrant cells are destroyed by NK cells
[100,101][40][41]. The deficiency of NK cell activating receptor NCR1 in a GEM (genetically engineered mouse) model was shown to promote tumor growth
[101,102][41][42]. Interestingly, when comparing with B16 melanoma growing subcutaneously in either NCR1 heterozygous (het) or homozygous (KO) mice, the tumor size in the two groups had no significant difference. Still, there was a substantial difference in the rate of metastasis between the two groups: NCR1 KO mice appeared to have significantly higher metastatic lesions than NCR1 het mice
[102,103][42][43]. This indicates that the absence of an activating receptor dramatically affects metastasis development but has no evident effect on the primary tumor, implying that the level of NK cell activating signaling plays a vital role in controlling tumor metastasis. Subsequently, it was found that elevated secretion of IFNγ by NK cells enhanced the level of FN1 in the tumor, resulting in architectural alteration and less aggressive metastasis. In contrast, deficient NCR1 abolished the release of IFNγ
[103][43]. Recent studies have also reported that NK cells are necessary for the selective destruction of circulating single breast tumor cells
[104,105][44][45]. Extensive studies also support that NK cells control metastasis by activating receptor signaling. For example, NK cells efficiently eliminate metastatic melanoma cells when overexpressing ligands for NKp44, NKp46, and DNAM-1
[74][12]. In contrast, GEM mice deficient in DNAM-1
[106[46][47],
107], Tlr3 (regulating NK cell responses to cytokines)
[108][48], Il2rg (ablating NKp46+ NK cells)
[109][49], or T-bet (regulating the differentiation of NK cells)
[110,111][50][51] are susceptible to metastatic colonization. Interestingly, the metastatic potential could be prevented by transplanting bulk NK cells
[110,111,112,113,114,115,116][50][51][52][53][54][55][56]. Moreover, treating NK cells with activating cytokine IL-15 can restore protection from metastasis in Tbx21-deficient mice
[111][51]. Deletion of a negative regulator Clbl also promotes NK cell-dependent antimetastatic effects in mice
[115][55]. Furthermore, deficient endogenous IL-15 inhibitor of Cish in NK cells (Cish
−/−NK cells) leads to NK cell hyperactivation; transplanting the Cish
−/−NK cells into mice could robustly abrogate the metastatic phenotype of highly metastatic B16F10 melanoma cells
[117,118][57][58]. Decrease in DNAM-1 limits NK cytotoxicity and blocks IFNγ production, while the overexpression of DNAM-1 ligand in tumor cells causes a decrease in NKG2D ligands, indicating that NK cell-induced killing is initiated by DNAM-1 or NKG2D signaling pathway
[74,117,119][12][57][59]. Moreover, NK cell depletion with antibodies of NK1.1 or asiago-GM markedly enhances the metastasis
[116,120,121,122][56][60][61][62]. Mice with NK cell activating factors, including IFNγ, perforin 1 (PRF1), or TRAIL deficiency by gene knockout or antibody-caused neutralization, were more susceptible to metastatic incidences following challenges with tumor cell inoculation or with carcinogen-induced tumors
[123,124,125][63][64][65]. NK cells also produce and release IFNγ and TNFα to function on macrophages and dendritic cells for enhancing the immune response
[126][66]. Interestingly, using an image tracker system, a recent study functionally visualized that NK cells directly contacted metastatic tumor cells rapidly, leading to its ERK activation and metastatic tumor cell apoptosis
[78][16]. They then confirmed metastatic tumor cell death related to DNAM-1 activation in NK cells
[78][16]. Whereas a more recent study has shown that activation of inhibitory receptor NKG2A/HLA-E signaling promoted the distant metastasis of PDAC; blocking this pathway provokes NK cells and inhibits PDAC metastasis
[127][67]. Whether the NK cells destroy metastatic cancer depends on the amounts of signals from activating and inhibitory receptors in the NK cells. Activating receptors interact with molecules on the surface of cancer cells and ‘turn on’ activation of the NK cell. Inhibitory receptors on NK cells check the signals from its ligands on metastatic cancer cells to block the ability of NK cell-mediated killing. Metastatic tumor cells often lose NK cell inhibitory receptor ligands of MHC-I, which reduce the inhibitory signal and leave them vulnerable to NK cell killing. Activating signals through activating receptors in NK cells promote NK proliferation and stimulate the secretion of cytotoxic granules to release perforin and granzymes, leading to metastatic tumor lysis.
3. Immunosuppression of NK in Tumor Metastasis
The effectiveness of NK cell killing of metastatic tumors depends not only on the immensity of the NK cell response, such as the activation of NK cells, but also on the capacity of cancer cells to evade destruction. Tumor cells can develop a wide range of strategies to elude detection and destruction by NK cells, continue to grow at distant sites, and then form metastases
[61][68]. The process by which tumor cells develop the intrinsic properties to avoid recognition and elimination by NK cells is referred to as “escape”
[61][68]. The efficiency of escape determines the success of disseminated tumor cells in giving rise to distant metastases. Most strategies tumor cells use to evade NK cell killing are described here. First, tumor cells can increase ligands of NK inhibitory receptors such as CD111
[128][69], PD-L1
[129][70], HLA-G
[130][71], galectin-9
[131][72], HMGB-1
[132[73][74][75],
133,134], and CEACAM-1
[135][76], therefore the activation of inhibitory receptor signal pathways
[127][67]. The expression of nonclassical HLA-G is reported in various metastatic cancers, which ties up the inhibitory receptor of LIR-1 to transduce the inhibitory signals to NK cells and inhibit the NK cell proliferation and gene expression
[130][71]. In fact, cancer cells frequently lose their MHC-1 to escape T cell attack but are vulnerable to NK cell killing, and the metastatic tumor has too many tricks to manage the MHC-1 level for refraining both T cell and NK cell elimination
[67][5], such as genetic and epigenetic modification, and transcriptional and translational regulation
[136,137,138][77][78][79]. Second, metastatic cancer cells decrease the ligands of NK cells activating receptors (NKAR)
[139,140][80][81]. For example, metastatic tumor cells often shed the NKG2D ligand of MICA and MICB proteins
[141][82] through proteolytic proteins such as ADAM10, ADAM17, and MMP14
[142[83][84],
143], thereby producing soluble variants of ligands, acting as molecular decoys for blocking NK cell activation
[144,145][85][86]. Indeed, in many different cohorts of cancers, patients with advanced stage tumors often have a high level of soluble NKAR ligands
[146,147][87][88]. Loss of PVR and nectin 2, a ligand for the NK cell activating receptor DNAM-1, abolished NK cell-mediated destruction of metastatic melanoma B16F10
[78][16]; the consequence of deletion of DNAM-1 ligand caused due to failure to activation of ERK signal pathway in NK cells
[78][16]. The downregulation of a death receptor FAS in metastatic tumors is another way to turn off NK cells for killing the metastatic cells through the FAS/FASL pathway
[148][89]. Tumor cells can also decrease immunostimulatory factors within the TME. Tumor cells can secrete IL10, CXCL8, and TGFB1, which directly reduce NK cell cytotoxic functions
[149,150,151][90][91][92] and/or recruit other immune cells such as Treg cells
[152][93], MDSCs
[153][94], CD11b + Ly6G + neutrophils
[122[62][95],
154], and DC
[155][96], thus inhibiting NK cell functions indirectly. Analysis of the secreted protein profiling of progressive cancers showed little amounts of NK cell-stimulating molecules such as IFN
[156][97] and IL-15
[157][98]. Tumor cells can rewrite their metabolic programming and release the metabolites that affect the TME to interfere with the antimetastatic functions of NK cells. For example, overexpression of the ectonucleoside triphosphate diphosphohydrolase 1 (ENTPD1) in metastatic tumor cells hydrolyze extracellular ATP into AMP to convert it to adenosine in hypoxia condition caused by HIF-1
[158,159][99][100]. Adenosine unleashes powerful immunosuppressive effects on NK cells via adenosine A2a receptor (ADORA2A) signaling
[159,160,161,162][100][101][102][103]. Hypoxia in TME favors metastatic tumor cells for avoiding NK cell elimination by releasing TGFB1 and miRNAs through exosome to target NKG2D
[163][104] and trigger cell autophagy to block tumor cells to GZMB-mediated lysis
[164][105]. In contrast, hyperoxia promotes the destruction of metastases through immune responses of CD8+ CTLs and NK cells
[165][106]. Inhibition of ENTPD1 by small molecular inhibitor polyoxometalate-1 or Entpd1 deletion blocked the metastasis of melanoma and colon cancer cells
[166][107]. In addition, lactate produced by tumors creates a favorable niche for metastasis and modulates the TME preventing NK cell activation
[167,168][108][109]. Furthermore, tumor-derived stromal inflammation has a condition-dependent effect on controlling metastasis by NK cells
[169,170][110][111]. Recent studies showed other proteins such as PAEP, pp12, and pp14 upregulated in metastatic tumor cells to hamper NK cell function
[171,172][112][113]. Last, some reports have shown that metastases exhibited higher stemness features than primary tumors, suggesting that metastatic tumor cells may obtain immune escape abilities from stem cells or dormant cells
[173,174][114][115] that could shield their proliferation and hide in distant sites for a long time.