We have previously shown that NK cells are defective in cancer patients 24,34–36. Similarly, Ovarian cancer patients have significant deficiency in the numbers of PBMCs, in particular NK cells. Both the numbers and function of NK cells are compromised in these patients. Due to this deficiency any arising PDCSCs that have lost or downmodulated MHC-class I and/or II will not be eliminated since primary activated NK cells are well known to target aggressive CSCs tumors 19,37, and thus, the patient will be predisposed for recurrences. Indeed, our global analysis of genes which may potentially contribute to disease recurrences points to a decrease in MHC-class I and MHC-class II genes (Fig. 2). Moreover, recurrent ovarian tumors are shown to demonstrate increased cell cycle markers, decreased differentiation and a decrease in immune cell infiltration (Fig. 2A). Therefore, due to severe decrease in NK cells in patients such aggressive tumors will not be controlled, and thus will give rise to recurrent tumors. Therefore, cell therapy using sNK cells should be able to control disease progression, because they are not only equipped to lyse PDCSCs significantly, but they are also capable of targeting differentiated tumors with the high expression of MHC-class I, a promising and unique attribute of sNK cells which were not observed in primary activated NK cells, since increase in MHC-class I expression on target cells is known to inhibit NK cell function through the inhibitory receptors.
There are many advantages to using off-the-shelf sNK cells for tumor immunotherapy. First, unlike CAR-T therapy, they will be readily available to use in therapy and one does not have to wait to generate them from the patient cells. Second, there have not been any reports of NK cell therapy causing graft vs. host disease (GVHD) and/or cytokine storm syndrome 38,39. These important advantages make sNK cells an ideal cell to use in tumor immunotherapy.
We have previously shown that primary activated NK cells were responsible to lyse and differentiate PDCSCs, whereas well-differentiated tumors were not or minimally lysed by primary NK cells 14,28. By using four specific biomarkers of CD44, MHC-class I, CD54 and PD-L1 we were able to distinguish between PDCSCs and well differentiated tumors in a number of tumor models including pancreatic, oral, ovarian and hepatic tumors, and study their interaction with the primary activated NK cells and compare to sNK cell function 13,14,19,24,36,40,41 (manuscript in prep.). There was a great correlation between the differentiation state of ovarian tumors and their susceptibility to primary NK cell-mediated cytotoxicity 19. In addition, using primary activated NK cells against a number of ovarian tumors at different states of differentiation we could distinguish between PDCSCs and well differentiated tumors 19. Thus, based on susceptibility to primary NK cell-mediated cytotoxicity and the four biomarkers, mentioned above on a number of ovarian tumor cells 19, we selected two ovarian tumor lines of OVCAR8 with PDCSC phenotype and OVCAR4 for well differentiated phenotype to further understand their dynamic differences, and susceptibility to primary activated NK cells and sNK cells. Our results indicated that similar to other tumor models, primary NK cells were able to lyse OVCAR8 PDCSCs and much less OVCAR4 or NK-supernatant differentiated OVCAR4, whereas sNK cells not only had much higher levels of cytotoxicity than primary NK cells against OVCAR8 PDCSCs, but they were also able to target OVCAR4 or NK-supernatant differentiated OVCAR4 much more than their primary activated NK cells (Figs. 4 and 6). Similarly, sNK cells were able to target either patient derived ovarian tumors or PDXs at much higher levels than primary activated NK cells. The highest fold increases in sNK compared to primary activated NK cells were seen against either patient derived tumor cells or patient derived PDXs.
In this paper we present the significant attributes of sNK cells armed with ability to kill ovarian tumors irrespective of the levels of MHC-class I expression, or the stage of differentiation of the tumors. These attributes of sNK cells are very different from the conventional primary activated NK cells which lack ability to target differentiated tumors since they express increased levels of MHC-class I, delivering inhibitory signal to primary NK cells, and therefore cause suppression of NK cell function against tumors with increased MHC-class I expression 42. Since sNK cells are laboratory generated, at the moment it is not clear whether these cells are laboratory created, or such subpopulation of NK cells exist either in the blood or tissues of humans. Indeed, we have not observed such cell populations either in patients or obtained with other activators of NK cells in the absence of osteoclasts as feeder cells. Other NK cell activation methods tested, such as the use of tumor cells as feeder cells, or treatment with the combination of multiple cytokines or the use of engineered tumors to express 4IBBL and IL-21 do not increase cytotoxicity of primary activated NK cells against differentiated tumors 43. Considering that osteoclasts are usually found in the bone microenvironment, it is possible that these cells are not in the periphery and thus we may not have such subpopulation of NK cells in periphery. Alternatively, there may be such subpopulation of NK cells in patients which have not yet been identified or characterized. In addition, in the bone marrow microenvironment the majority of NK cells are in immature state, therefore may not have the capability to give rise to sNK cells until they mature and exist to the periphery. sNK cells, therefore, have characteristic of classical primary NK cells since they can highly target cells which have lost MHC-class I, but also have acquired characteristics of T cells which can target tumors with high MHC-class I expression levels.
In the sNK engineering process, before exposure to osteoclasts, the NK cells are treated with IL-2 and anti-CD16 mAbs in which NK cells cease to kill tumors but they are programmed to secrete cytokines, a concept coined as split anergy 24,29,44. Indeed, these are the cells that give rise to sNK cells after culture with osteoclasts. In addition, IL-2 + anti-CD16 mAbs activated NK cells also express higher levels of 4IBB and 4IBBL, therefore it is possible that in the presence of osteoclasts in which higher expressions of NKG2DL and cytokines such as IL-15, IL-12, IL-18 etc are present, NK cells greatly expand due to the increase in survival, activation and secretion of key cytokines such as IFN-γ and TNF-α which are known to expand NK cells 26,45. Future work should focus on identifying endogenous sNK-like cells in humans if such cells exist.
When the effect of sNK was determined in a long-term assay using eSight, important differences between primary activated NK cells and those of sNK cells were observed. In all different assessments sNK cells had a profound ability to lyse and prevent tumor growth (Figs. 5 and 6, and supplemental videos). Particularly, when sNK cells were cultured in the presence of well differentiated oral OSCCs and ovarian OVCAR4 tumors, the ability of sNK cells to lyse such tumors was superior than primary activated NK cells (45, and Figs. 5 and 6, and S3, and supplemental videos). Indeed, sNK cells lysed these tumors and was responsible in growth suppression of these tumors in a long-term assessment. However, primary activated NK cells even though initially had some suppressive activity, were not able to maintain suppression of tumors, and the tumor growth started picking up after initial decrease (Figs. 5 and 6, and S3, and supplemental videos). Unlike IL-2 activated primary NK cells, addition of sNK cells to the tumor cells induced a rapid lysis of tumor cells, and subsequent aggregation of the dead tumors at the end of the incubation period (supplemental videos).
Our previous findings demonstrated that sNK cells persist in the tumor microenvironment, and were able to resist suppression by tumor cells, unlike IL-2 activated primary NK cells 19,40. In addition, in our recent single cell RNAseq analysis, sNK cells demonstrated decreased gene expression of NKG2A and CD94, potentially providing the rationale for the decreased inhibition through the inhibitory MHC-class I ligands. Therefore, sNK cells were greatly equipped to lyse the PDCSCs as well as well differentiated tumors at a much higher rate than primary activated NK cells, and the extent of differences in killing well-differentiated tumors was much higher than IL-2 activated NK cells, 40 (Fig. 4, and supplemental videos).
We have shown previously and, in this paper, that sNK cells secrete higher levels of IFN-γ and TNF-α and the combination of these two cytokines drive differentiation of tumor cells rendering them more resistant to primary activated NK cells, but not to sNK cells 13 (Fig. 3). Addition of antibodies to both IFN-γ and TNF-α completely abrogated NK supernatant mediated differentiation of tumor cells, and tumors remained susceptible to primary NK cell mediated cytotoxicity 19,46. Similarly, differentiation of many ovarian tumors with the exception of OVCAR8 either by NK supernatants or recombinant IFN-γ and TNF-α rendered them resistant to primary NK cell mediated cytotoxicity but not to sNK cells (Fig. 4). Indeed, when cytotoxicity was assessed by eSight at much longer time (i.e, 40 hours) primary NK cells were able to initially lyse some NK-supernatant differentiated OVCAR4 tumors, but the suppression of growth did not last since the tumor continued to grow in the presence of activated primary NK cells, whereas sNK cells completely eliminated these tumors and suppression continued until the assay was terminated (Fig. 6). These differences were not seen in OVCAR8 since primary IL-2 activated NK cells eliminated these tumors albeit sNK cells were still more potent when compared to primary activated NK cells in killing of tumor cells (Fig. 6). This is due to the resistance of sNK cells to tumor mediated cell death since sNK cells were found to exhibit the least amount or no cell death due to increased anti-apoptotic genes, whereas almost half of the primary activated NK cells had undergone cell death in the presence of the tumors in our preliminary studies. In accordance, more viable tumors were seen in the presence of IL-2 activated NK cells both in non-NK-supernatant differentiated OVCAR4, as well as NK-supernatant differentiated tumors when compared to sNK cells, whereas more OVCAR8 tumors were eliminated by primary IL-2 activated NK cells, albeit sNK cells still mediated the highest lysis of OVCAR8 tumor cells. Indeed, in all tumor models tested sNK cells had much higher ability to lyse tumor cells than the primary IL-2 activated NK cells. sNK cells exhibited higher binding avidity as determined by z-Movi when compared to IL-2 activated primary NK cells indicating their increased potential to form conjugates with the tumor cells. Increased in avidity of immune cells has been correlated with increase effectiveness of NK and T cells in lysing tumors in vivo47.
There is significant correlation between the ability of NK cells to lyse aggressive patient derived tumors and the MHC-class I expression. The lower the levels of MHC-class I on ovarian tumors, the better targets they are for NK cells (Fig. 7). These tumors will not be targeted by CD8 + T cells, since they do not express adequate levels of MHC-class I and/or MHC-class II. In addition, they are highly resistant to chemotherapeutic and radiotherapeutic strategies 13,28. Differentiation of such aggressive tumors with NK cell supernatants, are capable of inducing increased tumor susceptibility to chemotherapeutic and radiotherapeutic strategies 28,45. Moreover, increase in MHC-class I will activate CD8 + T cells, thus a combination of sNK cells with anti-PD1 antibody treatment should exhibit synergistic effect in augmentation of immune activation as previously reported in hu-BLT mice 13. Interestingly, OVCAR8 PDCSCs have lost ability to upregulate MHC-class I when differentiated by NK supernatants. This is opposite to what was found with a number of other tumors, in which differentiation by NK supernatants, in general, was able to increase MHC-class I 28,46.
Based on the data obtained in our studies, in recurrent ovarian tumors, a subpopulation of tumors with loss of MHC-class I may survive, potentially due to defective nature of NK cells in cancer patients, and the use of sNK cells in immunotherapy should provide effective strategy in targeting of such tumors to prevent recurrences.