Since their discovery,1–4 a large number of studies have demonstrated natural killer (NK) cell-mediated killing of many different types of tumor cell lines in vitro and, in experimental models, in vivo.5, 6 More recently, studies have also shown that NK cells are involved in rejection responses against experimentally induced and spontaneously developing tumors in mice.7–9 Indirect evidence for NK cell targeting of human tumors has come from allogeneic stem cell transplantation (SCT) against human cancer, in particular haploidentical SCT against acute myeloid leukemias (AML),7, 8 and from observations of NK cell infiltration in some solid tumors.9 A correlation between increased cancer risk and low NK cell-mediated cytotoxicity has been observed in one study.10 More direct evidence for NK cell targeting of human tumors has emerged from recent studies involving adoptive transfer of NK cells to cancer patients,11 where some responses have been detected. These findings have opened up the possibility for future NK cell-based immunotherapy against some malignancies.12–14
To better understand requirements for NK cell recognition of human tumors in the affected patient, and prerequisites for successful human NK cell-based adoptive immunotherapy, studies have more recently been initiated to assess the ability of human NK cells to target freshly isolated human tumor cells. Here, we provide a short review of studies that have attempted such an approach. We have deliberately decided not to review the large area of research on the expression of ligands for NK cell receptors on human tumor cells, an important field of research but outside the scope of this review. Several well-written reviews exist on this topic.15, 16 Nor have we reviewed the emerging literature on modulation of the receptor repertoire on NK cells within the tumor environment. Prior to the discussion on NK cell recognition of human tumors, we briefly summarize the molecular basis for NK cell recognition of human target cells.
Molecular basis for NK cell recognition of human target cells
NK cell recognition of target cells is a tightly regulated process involving the interaction of specific ligands on target cells with NK cell receptors and subsequent integration of signals derived from such receptors in the responding NK cell.17, 18 The earliest insights into the molecular specificity of NK cells emerges from observations that NK cell cytotoxicity was triggered by tumor cells lacking expression of certain (or all) self-major histocompatibility complex (MHC) Class I molecules, a phenomenon referred to as “missing-self” recognition.19, 20 These observations led to the identification of specific NK cell inhibitory receptors that recognize MHC Class I molecules.21–23 However, NK cells also need stimulation via specific receptors recognizing ligands expressed by the target cells.24–26 More recently, we have understood that the latter recognition events confer one additional degree of specificity in the tumor cell killing by NK cells.
Receptors mediating NK cell activation
NK cells have been known for a long time to express FcRγIIIR (CD16), a receptor that binds the Fc-portion of IgG, involved in antibody-dependent cellular cytotoxicity. More recently, several other NK cell activation receptors have been discovered and characterized,27 all of which may contribute to “natural cytoxicity” by NK cells. One important group of human NK cell activation receptors is the so-called natural cytotoxicity receptors (NCRs), NKp30, NKp46 and NKp44. Two of these, NKp30 and NKp46, are constitutively expressed on all peripheral blood NK cells, whereas NKp44 is induced and expressed by IL-2-activated NK cells.28 The role of these receptors in tumor cell killing have been demonstrated by receptor masking with anti-NCR antibodies inhibiting NK cell-mediated lysis of many different tumor cell lines.29–32 Indirect evidence for NCR ligand expression on several tumor types is provided by the use of soluble NCR fusion proteins.33 However, despite considerable effort to identify cellular ligands for the NCRs, only one candidate ligand has been described so far, i.e., the human leukocyte antigen (HLA) B-associated transcript 3.34 Other important activation receptors are NKG2D and DNAM-1. Of these, NKG2D is particularly well characterized.35 Human NKG2D recognizes the stress inducible molecules MHC Class I-related chain (MIC)A and MICB as well as the UL16-binding proteins. DNAM-1 recognizes the poliovirus receptor (PVR) and Nectin-2. In addition to these receptors, many other activation receptors, including 2B4 (CD244), NTBA, NKp80, CD2, CD11a/CD18 and CD59, have also been characterized.27 Several of these may have important coactivating or costimulatory functions in NK cell activation and tumor cell recognition.27, 36 The relative importance of different NK cell activation receptors and their ligands in recognizing human tumors is only partially known.
Receptors mediating NK cell inhibition
Activation of NK cells is under the control of inhibitory receptors.37 Inhibitory receptors bind classical and/or nonclassical MHC Class I molecules.17, 18 These molecules are normally expressed on most healthy cells in the body but are often lost upon malignant transformation or during tumor evolution. In humans, KIR and CD94/NKG2A play major roles as HLA-Class I-specific inhibitory NK cell receptors. KIRs recognize groups of HLA-A, -B and -C alleles, whereas CD94/NKG2A, B receptors recognize HLA-E molecules.17, 18 In contrast to the case for most activation receptors and inhibitory CD94/NKG2A, B receptors, individuals differ in the number and type of inherited KIR genes, and specific KIR gene products are expressed on distinct subsets of NK cells. Thus, many NK cells express only a few of many possible inhibitory KIR. Most functionally mature NK cells, however, express at least one inhibitory receptor (i.e., KIR and/or CD94/NKG2A), that is specific for a self-MHC Class I ligand. The clonal distribution of KIRs results in a system allowing NK cells to detect cells lacking expression of single MHC Class I alleles.38 This diversity is potentially beneficial in settings of SCT and adoptive NK cell-based immunotherapy against cancer.12, 39 Normally, NK inhibition dominates over activation. In some situations, however, the activation signals may override the inhibitory signals mediated by MHC Class I molecules, as has been demonstrated, e.g., for NKG2D-mediated triggering of some MHC Class I expressing tumor cell lines.40, 41
NK cell targeting of freshly explanted human tumor cells ex vivo
Studies of interactions between human NK cells and freshly isolated human tumor cells are of importance to unravel the susceptibility of different tumor types to NK cell-mediated lysis. Even though studies of tumor cell lines have contributed to our current knowledge of NK cell specificity and function, such studies have limitations in certain respects. Many differences between primary tumor cells and tumor cell lines are associated with altered proliferation rates and disrupted tissue organization.42, 43 These and other insights have more recently prompted studies of interactions between freshly isolated human tumor and NK cells.
Isolation of fresh tumor cells for usage as targets for NK cells
Isolation of fresh human tumor cells to be used as targets for NK cells is not always a straightforward process. Several obstacles may limit the possibilities of obtaining tumor material adequate for experimental studies. Human tumor targets of hematological origin can often be isolated directly from bone marrow aspirates or from peripheral blood. Fresh human tumor cells derived from solid malignancies are normally not as easily obtainable. Here, tumor material must be derived from fine needle aspirations, tissue biopsies or from surgically removed tumor samples. Certain solid tumors metastasizing to the bone marrow and/or peripheral blood may be isolated from these locations, however, in this case the frequency of tumor cells may be too low to allow further processing. In some cases, solid malignancies give rise to effusions, including pleural and peritoneal effusions, which may be relatively more available sources of tumor cells.
Several additional obstacles are encountered when processing fresh tumor samples. Separation of tumor cells from solid tumor masses often requires enzyme digestion or manual cutting and filtration through strainers, and/or other tumor cell separation procedures such as, e.g., magnetic bead separation. Processing of fresh tumor material, including enzymatic digestion of solid tumor tissue, carries the risk of introducing changes in the ligand expression and possibly other properties of the fresh tumor cells. Moreover, the number of tumor cells obtained may be low, e.g., due to the nature of the sample (small tumor biopsy or fine needle aspirate). Another problem that may be encountered is low viability of the collected material due to necrosis within the tumor tissue. Even though viable fresh tumor samples are obtained and successfully prepared in single cell suspensions, the material may still contain a mix of normal (nontransformed) and tumor cells. Thus, difficulties in processing tumor material may limit the available methods for studying NK cell recognition of freshly isolated tumor cells.
Methods for studying NK cell-mediated recognition of fresh human tumor cells
Cytotoxicity assays performed in vitro have been classified in three categories.44 The first category is indirect and based on the measurement of effector cell activity. The other two categories are direct and based on either release of substances or uptake/retention of substances detected in target cells upon killing.
Indirect measurements of NK cell-mediated killing of fresh tumor targets can be acquired from assessing degranulation and cytokine release by the effectors upon coincubation with the tumor cells. CD107a expression on the cell surfaces of effector cells can be analyzed by flow cytometry and is now frequently used as a surrogate marker for degranulation by lymphocytes.45, 46 The upregulation of CD107a on NK cells correlates with granule exocytosis and the release of perforin and granzyme B.46 Thus, the frequency of CD107a positive NK cells, upon stimulation with target cells, often correlates to the killing of the corresponding targets.47, 48 This type of indirect measurements of NK cell killing may allow identification of subsets of NK cells with the most prominent tumor targeting capacity. This can be achieved by combining the functional read-out with high resolution phenotyping of distinct NK cell populations expressing combinations of receptors or other surface molecules of interest in multi-parameter flow cytometry.49 Identification and isolation of the tumor reactive NK cell subset could be used, e.g., in settings of adoptive immunotherapy to select the best donor for a given patient, which theoretically can improve outcome of the therapy.50 Moreover, indirect measurement of NK cell responses using multiparameter flow cytometry allows the simultaneous assessment of degranulation and cytokine production and studies of multiple aspects of NK cell functionality.47, 49
However, there is still a need for a more direct measurement of target cell killing. NK cell-mediated lysis of tumor targets is detectable by several methods. The standard method for assessing NK cell-mediated killing is the 51chromium (51Cr) release assay. This method is based on release of a radioactive isotope upon lysis of target cells preloaded with 51Cr.51, 52 This method requires few cells and is, therefore, attractive. Unfortunately, it has some drawbacks. First of all, the method is based on work with a radioactive substance. Moreover, and perhaps more importantly, many primary tumor cells are difficult to label with radioactive isotopes and display high spontaneous leakage of 51Cr. Several other methods for evaluating killing of the target cell have been established over the last years, in particular protocols in which NK cell-mediated lysis of target cells can be detected by flow cytometry. Non-radioactive flow cytometry-based methods can be used to identify cells killed by NK cell-induced apoptosis.53 Several pathways of apoptosis can be studied, e.g., by measuring granzyme B activity (granzyme/perforin pathway) or caspase-8 activity (FAS-FASL or TRAIL pathways) inside the target cell. It is also possible to analyze the activity of effector-caspases, reflecting the overall apoptosis activity inside the target cell. Flow cytometry-based detection of apoptosis through cleavage of caspase substrates represents 1 example.48, 51 The target cells are coincubated with NK cells following a short incubation with a caspase substrate that diffuses freely over the cell membrane. Once the substrate is specifically cleaved by activated caspases in a target cell undergoing apoptosis, a quencher is removed leaving a fluorescent and polarized product trapped in the cell. Other types of flow cytometry-based assays involve techniques where dying target cells are detected by staining with fluorescent reagents, including 7-aminoactinomycin D (7-AAD) or propidium iodide, which bind to DNA in apoptotic cells54 (Fig. 1). Importantly, the use of flow cytometry-based assays for measuring target cell killing may also permit assessment of NK cell specificity by discriminating tumor cells from normal cells as shown in Figure 1.
In many of the studies of NK cell interactions with fresh human tumor cells, a combination of detection of effector activity and subsequent detection of induced killing (apoptosis or lysis) has been used.
Studies on the killing of fresh tumor targets by NK cells
Direct evidence for NK cell-mediated killing of freshly isolated human tumors reported so far in the literature is based on a limited number of studies, some rather old55–59 and some more recent (Table I).48, 60–67 Although, in some cases hampered by technical difficulties, including monitoring the specific lysis of fresh tumor cells within heterogeneous patient-derived cell populations, recent reports have provided evidence for NK cell targeting of human AML, multiple myelomas, neuroblastomas, gastric cancers, renal cell carcinomas, colon carcinomas and ovarian carcinomas. Below, we briefly discuss the results from some of the more recently published studies.
Table I. Studies on NK Cell Recognition of Freshly Isolated Human Tumor Cells1
NK cell source
NK cell preparation
NK cell stimulation
The table is not intended to provide a complete survey of all studies on NK cell-mediated recognition of fresh human tumor targets. A particular focus has been on more recently published studies.
Acute lymphatic leukemia
Expanded for 10–12 days on feeders and activated with IL-2, IL-12 and IL-15
NK cell-mediated lysis of primary acute lymphatic leukemia blasts has been observed with autologous NK cells expanded in vitro.60 One recently published study focusing on NK cell killing of primary AML blasts used NK cell lines with single KIR specificities for HLA Class I allotypes. This study nicely demonstrates NK cell recognition of freshly isolated primary AML blasts and indicates a beneficial role for KIR ligand mismatching in tumor cell killing.61 Tumor cell killing was predominantly observed in monoblastic cells expressing NKG2D ligands, whereas myeloblastic cells lacking corresponding ligands were resistant to lysis. Induction of cell surface NKG2D ligands by treatment with the histone deacetylase inhibitor, valporic acid, rendered cells more sensitive to NK cell-mediated lysis.61 This study pointed out the possibility of using alloreactive HLA Class I-mismatched NK cells in combination with pharmacologic induction of NKG2D in clinical evaluations as a novel approach to immunotherapy for AML.
NK cell-mediated killing of freshly isolated multiple myeloma cells has also been demonstrated using either allogeneic or autologous NK cells.62–65 However, the interaction governing NK cell-mediated killing varies among the studies. One study, using allogeneic NK cells, revealed a prominent role for the DNAM-1 receptor, as demonstrated by antibody masking of activating NK cell receptors.63 Another study using antibody blockade of autologous NK cells indicated that several activating receptors might contribute to lysis of multiple myeloma cells in this setting.65 The recognition of patient-derived multiple myeloma by autologous NK cells, as demonstrated with either IL-2 stimulated or long term expanded autologous NK cells,64, 65 led to speculation that this tumor might be targeted by immunotherapeutic strategies involving autologous NK cells. It should be noted that multiple myeloma frequently display reduced levels of HLA Class I on the cell surface which may explain the effectiveness of autologous NK cell preparation in this setting. Indeed, NK cell killing correlated inversely with the level of HLA Class I on the myeloma cells.64
Freshly isolated neuroblastoma cells, obtained from bone marrow samples from patients with metastasizing diseases, represent one solid tumor characterized with respect to NK cell susceptibility.66 Killing of freshly isolated neuroblastoma cells was shown to involve NKp30 and NKp46. A significant heterogeneity in susceptibility to lysis was found among neuroblastomas derived from different patients. Interestingly, susceptibility to lysis directly correlated with the surface expression of the DNAM-1 ligand PVR. The PVR-expressing neuroblastoma cells were efficiently killed by NK cells, and monoclonal antibody (mAb) masking of either DNAM-1 on the NK cells or PVR on the tumor cells resulted in strong inhibition of tumor cell lysis. These findings indicate that assessment of PVR levels on cell surfaces may represent a criterion for predicting the susceptibility of neuroblastomas to NK cell-mediated killing.
Another solid tumor tested for NK cell-mediated recognition is ovarian carcinoma. This tumor type often gives rise to ascitic fluid containing tumor cells in suspension in the peritoneal cavity of affected individuals. Such cells represent a source of fresh tumor targets useful for studying NK cell susceptibility. Similar to fresh neuroblastomas, ovarian carcinoma cells are also sensitive to lysis by allogeneic NK cells.48 In this study, tumor cell killing was measured by monitoring cleavage of a caspase-6 substrate in tumor cells. This was associated with NK cell degranulation and subsequent detection of granzyme B inside the target cells. In line with typical profiles of NK cell receptor ligand expression, mAb-mediated blockade of activating receptor pathways revealed a dominant role for DNAM-1 and a complementary contribution of NKG2D-signaling in tumor cell recognition.A hallmark of this study was its emphasis on recognition of ovarian carcinoma cells by resting (non-IL-2 activated) NK cells. The results suggest that ovarian carcinomas are potential targets for adoptive NK cell-based immunotherapy.
Observations of NK cell-mediated killing of a limited number of freshly isolated human tumor cells of various histotypes, including gastric, ovarian, colon and renal cell cancers have also been made in a recently published study.67 Although the observations are based on few experiments, the results indicate an enhanced killing in the KIR-ligand mismatched setting. These studies also put forward the possibilities of using alloreactive NK cells as a basis for the design of new cell therapy based approaches against solid tumors.
Conclusion of studies addressing NK cell interaction with freshly isolated human tumor targets
We have summarized some of the available evidence for NK cell targeting of freshly explanted human tumors. Different aspects regarding the rationale for working with fresh tumor cells and associated technical difficulties have been discussed. The recognition of fresh tumor targets has in some studies been addressed on both the level of the effector and the tumor in the different studies. In such studies, flow cytometry-based methods for assessing NK cell reactivity toward tumor cells within (often) heterogeneous cell population are invaluable. Data from the studies presented in this review point toward a tumor specificity of the NK cells, sparing normal tissue.48, 65 The receptor–ligand interactions that govern tumor recognition vary between the different tumor types studied. For example, activation signals through the DNAM-1 receptor seem to play an important role in the recognition of neuroblastomas and ovarian carcinomas, whereas signaling via NKG2D is important in the recognition of AML blasts.48, 60, 61, 66 Killing of the tumor types studied seem often not mediated solely via one receptor, but often seem to rely on a complex pattern of receptor–ligand engagements as demonstrated in several of the reviewed papers. The varying ligand profile of fresh tumor cells, even within specific tumor types, represents a hallmark of many tumors and highlights the importance of moving the focus from homogeneous cell lines to complex freshly isolated, nonpropagated, patient-derived tumor targets when studies relating to human immunotherapy are addressed. In general, fresh tumor targets seem less sensitive to NK cell-mediated lysis compared with many cell lines of the same origin, as exemplified by differences between fresh neuroblastomas and established neuroblastoma cell lines.66
As discussed, allogeneic NK cells have been used in several of the studies presented here.48, 61, 62, 64, 66, 67 KIR-ligand mismatched allogeneic NK cells was shown to more efficiently kill monoblastic AML blasts when compared with matched NK cells.61 The cytotoxicity of allogeneic NK cells was also further enhanced in some of the studies by either culturing of the effector cell or by prestimulation with cytokines.48, 61, 63, 66, 67 Interestingly, autologous NK cells derived from patients suffering from multiple myeloma showed cytotoxicity against fresh autologous tumor targets after activation and/or expansion in vitro.60, 65 Thus, the source of NK cells and type of prestimulation are important factors to consider when studying the interaction of NK cells and fresh human tumor targets.
Notably, all studies published so far on the killing of fresh human tumor cells, to our knowledge, have assessed NK cell-mediated killing by using short-term assays that detect killing mainly mediated via degranulation and release of granzymes and perforin. One important factor to take into consideration when addressing NK cell susceptibility of tumors in the future is the role for death receptor pathways, including FASL and TRAIL.
In this review, we have provided evidence that NK cells can target freshly explanted human tumor cells. The studies reveal that NK cell activity is tightly regulated by complex receptor–ligand interactions mediating a fine-tuned target reactivity. Studies performed in vitro with new techniques demonstrating efficient NK cell targeting of freshly isolated human tumor cells are likely to provide new and better insights into NK cell-mediated recognition of human cancer.