Natural killer T cell‐mediated antitumor immune responses and their clinical applications

Characteristics of α‐GalCer and its antitumor effects through NKT cell activation

Primarily, α‐GalCer is the general designation for glycosphingolipids that contain a galactose carbohydrate attached by an α‐linkage to a ceramide lipid that has acyl and sphingosine chains of variable lengths. KRN7000 ((2S, 3S, 4R)‐1‐O‐(α‐D‐galactopyranosyl)‐N‐hexacosanoyl‐2‐amino‐1, 3, 4‐octadecanetriol) is a synthetic α‐GalCer that has been most frequently used in experimental studies and is usually referred to as α‐GalCer (Fig. 1). The lead compound for α‐GalCer was originally derived from marine sponge, and KRN7000 was found by screening the potential of various synthetic glycolipids to stimulate NKT cell proliferation.^{( 2 )} It has been demonstrated that a change of length of the lipid portion of α‐GalCer enables a different pattern of cytokine production from that of α‐GalCer to be induced. OCH is a short form of α‐GalCer by truncations in the acyl chain (by two carbons) and the sphingosine chain (by nine carbons), which induces preferential production of IL‐4 over IFN‐γ (Fig. 1).^{( 10 )}α‐C‐GalCer is a carbon glycoside analog of α‐GalCer, which induces preferential production of IFN‐γ over IL‐4 (Fig. 1).^{( 11 )}

It has been believed that the lipid portion of α‐GalCer interacts with the hydrophobic antigen‐binding groove of CD1d, and the carbohydrate portion is accessible for interaction with the NKT cell invariant TCR. This was directly proved by recent several studies analyzing the crystal structure of CD1d binding α‐GalCer.^{( 12} , ^{13 )} The crystal structures showed a tightly fit lipid in the CD1d binding groove, with the sphingosine chain bound in the C′ pocket and the longer acyl chain anchored in the A′ pocket. The studies also presented the CD1d structure without a lipid, which had a more open conformation of the binding groove, suggesting a dual conformation of CD1d in which the ‘open’ conformation is more able to load lipids. These structures provided clues as to how the CD1d molecule loads glycolipids, as well as data to guide the design of new therapeutic agents.

α‐GalCer presented by mouse or human CD1d strongly stimulates mouse or human NKT cells and mediates their direct or indirect antitumor responses (Fig. 2).^{( 2} , ^{14 )} The α‐GalCer/CD1d complex activates NKT cells mediating perforin‐dependent direct cytotoxicity and also upregulating CD40L on NKT cells, leading to the activation of DC to produce IL‐12.^{( 15} , ^{16 )} IL‐12 from DC activates NKT cells to produce IFN‐γ which in turn stimulates NK cells and CD8 CTLs mediating antitumor cytotoxicity,^{( 8} , ¹⁷ , ^{18 )} as in the case of microbial infection.^{( 19 )} IFN‐γ secreted by NKT and NK cells is required for the NKT cell‐mediated antitumor responses, as IFN‐γ‐deficient mice could not induce antitumor responses.^{( 17 )}α‐GalCer‐activated NKT cells also induce the maturation of DC, which contributes to the upregulation of Th1 responses.^{( 20} , ^{21 )} Regarding the molecular mechanism for the killing of tumor cells, it has been reported that IFN‐γ‐mediated TRAIL induction on NK cells plays a critical role in the antimetastatic effects of IL‐12 and α‐GalCer.^{( 22 )} It has been further indicated that perforin‐mediated killing by NKT cells is also involved in the cytotoxicity against tumor cells.^{( 6 )}

Cancer immunosurveillance by NKT cells

It has also been demonstrated that, in the absence of α‐GalCer, NKT cells contribute to the elimination of cancer cells or inhibiting expansion of cancer cell growth. In fact, experiments using Jα281‐deficient mice indicated a critical role for NKT cells in protection from spontaneous tumors initiated by a chemical carcinogen, MCA.^{( 23 )} Further evaluation of several MCA‐induced tumors, in conjunction with NKT cell‐transfer experiments, revealed that, in addition to NK cells, CD8⁺ T cells, perforin, IFN‐γ, and CD1d expression on antigen‐presenting cells, NKT cells play an important role in protection against tumor growth.^{( 24 )}

It will be of great importance to determine the factor(s) and mechanism(s) by which NKT cells are activated in tumor‐bearing hosts in the absence of exogeneous stimulation such as α‐GalCer. In the case of microbial infection, NKT cells seem to be activated by IL‐12 produced by toll‐like receptor‐stimulated DC in the presence of a basal level of a CD1d/TCR‐mediated signal.^{( 19 )} Similarly, NKT cells might mediate antitumor responses without direct recognition of tumor‐derived antigens. As antitumor responses sometimes associate with inflammations that deliver cytokines such as IL‐12, NKT cells can be activated by IL‐12 produced during the tumor‐mediated inflammatory processes (Fig. 3a). Thus, NKT cells might be able to respond to any type of tumor cell irrelevant of the tumor specificity. However, it is still possible that NKT cells might directly recognize glycolipid components from tumor cells (Fig. 3b). In fact, some glycolipid fractions from the tumor cell membrane are known to be presented by CD1d and to be recognized by NKT cells.^{( 25 )} Consistent with this notion, an endogenous glycolipid ligand for NKT cells has been identified as an isoGb3 (Fig. 1)^{( 26 )} that can stimulate NKT cells in a way similar to α‐GalCer. Thus, it is possible that NKT cells induce antitumor responses by recognition of the cancer‐related glycolipid ligand, which might be similar to isoGb3.

Recently, Crowe et al. reported that distinct subsets of NKT cells have different antitumor capacities.^{( 27 )} This is the case not only for the endogenous antitumor immunosurveillance but also for the α‐GalCer‐induced antitumor effect, as mentioned in the former section. They compared NKT cells from liver, thymus, and spleen for their ability to mediate rejection of the MCA‐induced sarcoma cell line MCA‐1 in vivo, and found that this was observed only by liver‐derived NKT cells. Furthermore, when CD4⁺ and CD4⁻ liver‐derived NKT cells were given separately, MCA‐1 rejection was mediated primarily by the CD4⁻ fraction. They also found that liver‐derived NKT cells are superior in rejecting B16F10 melanoma metastasis when NKT cells were stimulated with α‐GalCer rather than other organ‐derived NKT cells. The impaired ability of thymus‐derived NKT cells was due to their production of IL‐4, because tumor immunity was clearly enhanced after transfer of IL‐4‐deficient thymus‐derived NKT cells. Therefore, it seems important to consider the differential potential of distinct subsets of NKT cells to promote more effective antitumor immune therapy.^{( 28 )}

Regulation of NKT cell‐mediated antitumor responses

Several recent papers indicated some cellular components which can suppress NKT cell antitumor effects. One is naturally occurring Treg. It was reported that Treg can suppress the proliferation, cytokine production, and cytotoxic activity of NKT cells.^{( 29 )} Furthermore, Nishikawa et al. found that coimmunization with DNA encoding autoantigens (i.e. Dana J‐like 2) defined by a serological expression cloning method, SEREX, leads to an immunosuppressive status mediated by Treg in which the number of NKT cells is significantly reduced.^{( 30 )} Although the precise mechanisms of interaction between Treg and NKT cells have not been revealed yet, the result suggests that the SEREX‐defined autoantigens regulate immunological homeostasis through Treg that inhibit NKT cell‐mediated antitumor immunity.

Not only experimental but also clinical studies have shown a repressed NKT cell function in cancer‐bearing status. It has been shown that the number of human Vα24 NKT cells is markedly decreased in peripheral blood from advanced cancer patients (for example, in lung cancer patients).^{( 31 )} In prostate cancer patients, the ex vivo expansion of α‐GalCer‐activated Vα24 NKT cells and the amount of IFN‐γ produced by these cells is significantly decreased.^{( 32 )} Similarly, we showed that Vα24 NKT cells from patients with different types of solid cancers failed to proliferate even with α‐GalCer stimulation, and produced lower amounts of cytokines than those from healthy individuals.^{( 33 )} These studies indicate that Vα24 NKT cells in cancer patients have some numerical and functional defects, although it has not been elucidated whether Treg or autoantigens, mentioned above, are involved in the clinical cases.

In addition to studies of cancer patients with solid tumors, Fujii et al. have indicated that patients with myelodysplastic syndromes have a severe functional deficiency in Vα24 NKT cells, but not NK cells or CD4⁺ or CD8⁺ T cells.^{( 34 )} Freshly isolated Vα24 NKT cells from patients with progressive multiple myeloma have a marked deficiency in their α‐GalCer‐dependent IFN‐γ production. This deficiency is not present in non‐progressive multiple myeloma or premalignant gammopathy patients, although Vα24 NKT cells are all detectable by flow cytometry in different stages of multiple myeloma patients (premalignant gammopathy, non‐progressive multiple myeloma, or progressive multiple myeloma).^{( 35 )} Based on these findings, it is hypothesized that presentation of tumor‐derived ligands by myeloma cells might cause Vα24 NKT dysfunction in vivo. Taken together with these findings, it seems required to restore the function of Vα24 NKT cells in some cancer patients before treating them with an NKT cell‐mediated immune therapy. In this regard, we have shown that granulocyte colony stimulating factor can partly restore the repressed NKT cell function derived from cancer patients,^{( 33 )} which might be a hint as to how to improve NKT cell immune therapy.

Repression of antitumor immune response by type II NKT cells

It has been shown that type II NKT cells act as regulatory cells that suppress CD8 CTL‐mediated antitumor responses. Terabe et al. reported that, using a model of tumor recurrence, CD8 CTL‐mediated tumor immunosurveillance of a transformed fibrosarcoma is suppressed by IL‐13 produced by CD1d‐restricted T cells with an activated IL‐4Rα‐STAT6 signaling pathway.^{( 36 )} They subsequently showed that TGF‐β produced by CD11b⁺Gr‐1⁺ cells, which is dependent on IL‐13 from CD1d‐restricted T cells, is responsible for this negative regulation.^{( 37 )} In their report, they showed that blocking of TGF‐β or depletion of Gr‐1⁺ cells in vivo prevented tumor recurrence. Recently, they compared the negative regulation of effects observed in Jα281‐deficient mice (lacking type I NKT cells) with those in CD1d‐deficient mice (lacking both type I and type II NKT cells), and identified type II NKT cells as the regulatory cells involved in the suppression, because CD1d‐deficient mice but not Jα281‐deficient mice failed to suppress antitumor responses.^{( 38 )} They have also reported that, in CD1d‐deficient mice, rejection of K7M2 mouse osteosarcoma was significantly accelerated, suggesting that CD1d‐restricted NKT cells naturally repress the rejection of the osteosarcoma.^{( 39 )} However the authors did not identify the NKT cells as type I or type II. In contrast with previous observations, the repression of this tumor rejection mediated by CD1d‐restricted NKT cells was not dependent on IL‐4Rα‐STAT6 signaling, including IL‐13, or on TGF‐β. Therefore, CD1d‐restricted NKT cells might regulate antitumor immune responses through several different pathways. Furthermore, it is conceivable that type I and II NKT cells might interact mutually to repress the other's function and form a negative feedback loop in the antitumor immune responses.^{( 28 )}

Anti‐tumor effects by activated human Vα24 NKT cells

Although the frequency of invariant Vα24 NKT cells is very low in humans, Vα24 NKT cells can be expanded by the stimulation of PBMC with α‐GalCer in an in vitro culture.^{( 14} , ⁴⁰ , ^{41 )} The in vitro expanded human Vα24 NKT cell lines show substantial cytotoxicity against tumor target cells.^{( 14} , ³⁵ , ⁴² , ^{43 )} Concerning the direct cytotoxic mechanism in Vα24 NKT cells, van der Vliet et al. have shown that freshly isolated and expanded Vα24 NKT cells express granzyme B but not FasL.^{( 41 )} However, FasL expression is inducible in CD4⁺ human NKT cells after stimulation with PMA/ionomycin.^{( 44 )} In contrast, the CD4⁻ subset of human NKT cells selectively produces TNF‐α and express perforin after activation.^{( 44 )} We have also demonstrated that Vα24 NKT cells possess cytoplasmic perforin and kill U937 cells mainly through a perforin‐mediated pathway.^{( 14 )} Moreover, human NKT cells can express TRAIL, thus inducing apoptosis in TRAIL‐sensitive leukemic cells.^{( 7 )} Therefore, it is likely that the various activities of human NKT cells in tumor rejection result from the use of a distinct cytotoxic machinery adjusted to the microenvironment and the sensitivity of target cells. Moreover, it is highly conceivable that human Vα24 NKT cells might contribute to activating other immune cells such as NK cells or DC (Fig. 1).

Clinical trials in cancer patients with α‐GalCer‐activated Vα24 NKT cells

Clinical studies on NKT cell therapy using α‐GalCer or α‐GalCer‐pulsed DC have been initiated. A phase I study of direct intravenous injection of α‐GalCer was initially carried out in the Netherlands.^{( 45 )} In this study, no dose‐limiting toxicity was observed, and the injection of α‐GalCer over a wide range of doses was well tolerated in cancer patients. However, an increase in serum cytokine levels (TNF‐α and GM‐CSF) was observed only in patients with high Vα24 NKT cell frequency, and no clinical antitumor effect was recorded.

In contrast to clinical protocols using the soluble form of α‐GalCer, we have found that α‐GalCer pulsed on DC produces enhanced antitumor effects in mice.^{( 46} , ^{47 )}α‐GalCer‐pulsed DC effectively activated NKT cells in vivo, resulting in the inhibition of tumor metastasis.^{( 47 )} Moreover, a complete inhibition of B16 melanoma metastasis in the liver was obtained even by a delayed treatment with α‐GalCer‐pulsed DC. That is, when α‐GalCer‐pulsed DC were transferred into syngeneic mice 7 days after tumor cell injection (which allows the formation of multiple, if small, metastatic nodules), the liver metastasis was significantly diminished.^{( 47 )} Based on the in vivo effects of α‐GalCer‐pulsed DC in the therapeutic setting, we started to consider a clinical trial using α‐GalCer‐pulsed DC in cancer patients.

We have previously demonstrated the α‐GalCer‐induced expansion of NKT cells and the efficient inhibition of tumor growth in vivo in murine lung cancer models,^{( 48 )} therefore we chose lung cancer as the target for NKT cell therapy. We undertook a phase I clinical study to evaluate the safety and efficacy of α‐GalCer‐pulsed DC in patients with advanced lung cancer.^{( 49 )} In this study, patients with advanced non‐small cell lung cancer or recurrent lung cancer received intravenous injections of α‐GalCer‐pulsed DC (level 1: 5 × 10⁷/m²; level 2: 2.5 × 10⁸/m²; and level 3: 1 × 10⁹/m²) to test the safety, feasibility, and clinical response. Eleven patients were enrolled in this study, and no severe adverse events were observed in any patient. After the first and second injection of α‐GalCer‐pulsed DC, a dramatic increase in peripheral blood Vα24 NKT cells was observed in one case and significant responses were seen in two cases receiving the level 3 dose. Although no patient was found to meet the criteria for partial or complete responses, two cases in the level 3 group remained unchanged for more than two years with good quality of life. This initial study indicates that α‐GalCer‐pulsed DC therapy was well tolerated and could be safely done even in patients with advanced disease.

In this clinical trial, whole PBMC cultured with IL‐2 and GM‐CSF were used as the ‘DC’ loading α‐GalCer. In a further study, interestingly, it was demonstrated that IL‐2/G‐CSF‐treated PBMC were superior to moDC developed with IL‐4 and GM‐CSF in their ability to expand Vα24 NKT cells and to induce their IFN‐γ production.^{( 50 )} CD11c⁺ cells in the IL‐2/G‐CSF‐cultured PBMC showed a mature phenotype without further stimulation and exerted potent stimulatory activity on Vα24 NKT cells to enable them to produce IFN‐γ preferentially at an extent equivalent to mature moDC induced by stimulation with lipopolysaccharide (LPS) or a cytokine cocktail. Surprisingly, cocultivation with CD11c⁻ cells in the IL‐2/G‐CSF‐cultured PBMC induced maturation of moDC. This was shown to be due to TNF‐α produced by CD11c⁻CD3⁺ T cells in the IL‐2/G‐CSF‐cultured PBMC. Thus, the maturation of DC induced by CD11c⁻ T cells through TNF‐α production appeared, in this study, to result in the efficient expansion and activation of Vα24 NKT cells to produce IFN‐γ preferentially.^{( 50 )}

We have now extended the clinical trial with α‐GalCer‐pulsed DC (at 1 × 10⁹/m²) as a phase IIa study to test immunological responses, safety, and clinical responses. We are planning to complete this study of 20 patients with primary lung cancer by the end of 2006. So far, we have observed no severe adverse events. We have essentially confirmed the observation in the previous phase I study regarding the immunological and clinical responses.^{( 49 )}

Nieda et al. have also reported the results of a clinical trial with α‐GalCer‐pulsed DC in 12 patients with metastatic malignancies.^{( 51 )} In this study, it was shown that activated Vα24 NKT cells induce subsequent activation in T cells and NK cells, and lead to an increase in serum IFN‐γ. Further, it was found that serum tumor markers were significantly decreased in two patients with adenocarcinoma, indicating antitumor effects of α‐GalCer‐pulsed DC.

Conclusions

We have summarized and discussed characteristics and the mechanisms of NKT cell‐mediated antitumor immune responses. Recent studies revealed that some cellular components can repress NKT cell function in cancer, and that a certain subset of NKT cells plays a role in repressing antitumor immune responses. Therefore, we must pay attention to the functions of several other immune components surrounding the CD1d/NKT cell system when considering NKT cell‐mediated immune therapy against cancer. Although clearly in its infancy, the use of α‐GalCer‐pulsed DC therapy to bolster antitumor immunity might provide a potent new tool in a rationally conceived arsenal designed to modulate the immune response against cancer.