STAT5 Loss Awakens the Dark Force in Natural Killer Cells

Within the tumor microenvironment the interaction of tumor cells and immune cells critically shapes the progression of malignant disease (1). There is now increasing evidence that certain types of immune cells have the potential not only to efficiently control, but also to promote tumor growth. For instance, M1 macrophages in tumors can directly kill tumor cells, whereas M2 macrophages support tumor angiogenesis and tumor development. So far, natural killer (NK) cells that efficiently lyse tumor cells and produce inflammatory cytokines such as IFNγ and TNFα are mainly associated with potent antitumor activity. Individuals with high NK cell activity have a lower risk to develop cancer (2). NK cell infiltration in the tumor is frequently correlated with improved prognosis for patients with cancer, and in many mouse tumor models depletion of NK cells leads to accelerated tumor growth (1). These studies imply an important role of NK cells in tumor immunosurveillance and support clinical application of NK cell–based therapies in the clinic for cancer treatment. More recently, additional immunoregulatory functions have been attributed to NK cells. In infectious disease, NK cells were shown to kill activated T cells and to produce IL10 and TGFβ, resulting in the downregulation of immune responses. Accordingly, a recent study revealed that high numbers of tumor-infiltrating NK cells did not correlate with favorable prognosis and even might promote tumor progression (3). The mechanisms underlying NK cell–mediated tumor progression are poorly understood. Moreover, the molecular switches determining NK cell–mediated antitumor or protumor activity have not been elucidated. Strategies to counteract the potential tumor-promoting activity of NK cells could improve current NK cell–based antitumor therapies.

The JAK–STAT pathway is activated in many tumor cells and promotes tumor progression. JAK–STAT inhibitors have been extensively explored for cancer treatment, and more than 20 JAK–STAT inhibitors are being tested in clinical trials (5). The JAK–STAT pathway is also critical for lymphocyte activation, for instance by the cytokines IL2, IL12, IL15, and IL18 that mediate NK cell activity and/or development. Accordingly, NK cells from patients with myeloproliferative neoplasm treated with a JAK inhibitor show impaired functional activity (5). A previous study by the Sexl lab (6) showed severely reduced NK cell numbers in mice with STAT5 deficiency in NKp46⁺ cells, demonstrating an essential role of STAT5 in NK-cell development and survival. NK cells from Stat5^Δ/ΔNcr1-iCre^Tg mice expressed low levels of the antiapoptotic gene Bcl2. To rescue mature NK cells in Stat5^Δ/ΔNcr1-iCre^Tg mice for functional study upon STAT5 deficiency, Gotthardt and colleagues (7) crossed Stat5^Δ/ΔNcr1-iCre^Tg mice with Vav-Bcl2 transgenic mice. Overexpression of Bcl2 rescued NK cell numbers in Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice. The developmental phenotype, however, was not completely restored because NK cells from these mice contained higher percentages of CD27⁻CD11b⁻ immature NK cells accompanied by reduced expression of the transcription factor genes Id2, Tbet, and Eomes compared with control mice. NK cells from Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice showed severe functional defects in vitro, including impaired proliferation and decreased cytotoxicity and IFNγ production toward tumor targets. These in vitro results predict that in vivo NK cells in Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice would also be unsuccessful in controlling growth of NK cell–sensitive tumors in mouse tumor models.

The study by Gotthardt and colleagues (7), however, revealed that Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice not only failed to control tumor growth of RMA-S lymphoma cells and a v-abl transformed tumor, but, intriguingly, tumor growth in these mice was even significantly increased compared with Stat5^Δ/ΔNcr1-iCre^Tg mice, implying an unexpected tumor-promoting function of STAT5-deficient NK cells (Fig. 1). In Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice, tumors reached the endpoint within 9 days, whereas only small tumors were detectable in control mice. Depletion of NK cells further confirmed that the tumor-promoting effect in Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice was mediated by NK cells.

To investigate if STAT5-deficient NK cells promote tumor progression via angiogenesis, Gotthardt and colleagues (7) analyzed levels of proangiogenic factors and found highly enhanced VEGF production by STAT5-deficient NK cells. NK cells from Stat5^Δ/ΔNcr1-iCre^Tg-Vav-Bcl2 mice show a complex phenotype characterized by multiple developmental and functional defects. Thus, in order to elucidate the role of VEGF production in NKp46⁺ cells, Gotthardt and colleagues established Vegfa^Δ/ΔNcr1-iCre^Tg mice, in which VEGFA is deficient specifically in NKp46⁺ NK cells. Indeed, in different tumor models (v-abl⁺ tumor, RMA-S, and A-MuLV–induced leukemia), a pronounced reduction of tumor burden and prolonged survival in those mice compared with control mice were observed consistent with a proangiogenic role of NK cells. Fewer CD31⁺ blood vessels in tumors of Vegfa^Δ/ΔNcr1-iCre^Tg mice compared with controls were detected. In this study, a tumor-promoting role of VEGF-producing NK cells was revealed in lymphoma models, whereas metastases of B16 melanoma were similar in VEGF-deficient or VEGF-sufficient mice, as well as in STAT5-deficient or BCL2-rescued mice. It will be important to further determine whether growth of other solid tumors such as colon cancer, melanoma, or lung carcinomas is affected by VEGF produced by NKp46⁺ cells. Because many different cell types can produce VEGF in the tumor beds, including myeloid cell subsets or tumors by themselves, it is likely that in different tumor models different relative contributions of the VEGF producers will be observed. Thus, VEGF production by NKp46⁺ cells might play a nonredundant role in some but not all tumor models.

The ability of NK cells to produce VEGF has been described before (8, 9). In patients with non–small cell lung carcinoma, tumor-infiltrating NK cells could produce increased VEGFA and promoted angiogenesis ex vivo (8). Moreover, NK cells from decidua were also shown to produce VEGF and to support angiogenesis (9). Gotthardt and colleagues (7), however, now reveal a novel mechanism leading to VEGF production in NK cells. ChIP assays demonstrate binding of STAT5 to the Vegfa promoter. The Vegfa mRNA particularly increased in Stat5b knockout but not in Stat5a knockout NK cells, indicating STAT5B as the main regulator of Vegfa transcription. The mechanisms for how STAT5 suppresses Vegfa transcription and whether STAT5B needs interaction with other transcription factors to suppress Vegfa transcription have not yet been elucidated. Importantly, cytokines also modulated STAT5 activation and VEGF production by NK cells. For instance, treatment with the cytokines IL10, IL12, IL18, IL21, and IFNβ decreased STAT5 activity in NK cells accompanied by increased Vegfa mRNA levels. These results imply that also within the tumor microenvironment VEGF production by NK cells might be regulated by cytokines. Similar to mouse NK cells, human peripheral blood NK cells, in particular the CD56^brightCD16⁻ subset, produced VEGFA correlating with low expression of STAT5A, STAT5B, and BCL2. This inverse correlation between STAT5B and VEGFA expression suggests that STAT5B also suppresses VEGFA expression in human NK cells. Because VEGF-producing NK cells have been detected in human tumors, it will be informative to assess the STAT5 levels in NK cells in correlation to the cytokine milieu within the tumors. Moreover, Vegfa^Δ/ΔNcr1-iCre^Tg mice will provide valuable tools to determine the role of VEGF production by NK cells in the decidua. Finally, NKp46⁺ cells comprise not only conventional NK cells but also subsets of innate lymphoid cells (ILC) that orchestrate tissue homeostasis and immune responses within tissues, and it will be important to determine whether ILCs also contribute to VEGF production within tissues.

The study by Gotthardt and colleagues (7) implies that strategies of activating the STAT5 pathway, such as application of IL2 and IL15, might not only enhance the cytolytic activity and cytokine production by NK cells, but also help to counteract VEGF production by NK cells during cancer therapy. Thus, the data presented in this study support current NK-based therapeutic strategies, including injections of IL2 or IL15 into patients with cancer or targeted therapies, including cytokines such as trispecific killer engagers (10). The cytokines IL12, IL18, IL21, and IFNβ decreased STAT5 activity and induced VEGF expression in NK cells; however, these cytokines are also known to be potent activators of NK cell cytotoxicity or cytokine production. Thus, the protumor or antitumor effect of NK cells induced by those cytokines needs further investigation.

Considerable numbers of JAK–STAT inhibitors are currently being tested in clinical trials for treatment of different types of malignancies (4). In the study by Gotthardt and colleagues (7), in vitro treatment of murine or human NK cells with ruxolitinib, the first JAK inhibitor approved by the FDA (4), which targets JAK1, JAK2, and JAK3 upstream of STATs, greatly enhanced expression of Vegfa. Most importantly, larger tumors were observed in control mice bearing v-abl⁺ tumors after injection of ruxolitinib compared with Vegfa^Δ/ΔNcr1-iCre^Tg mice. These results imply that JAK–STAT inhibitors might induce NK cell–mediated tumor progression in certain patients by NK cell–derived VEGFA. Thus, the combination of JAK–STAT inhibitors with antiangiogenic therapy should be considered. Meanwhile, considerable effort is focusing on developing novel STAT inhibitors with a more selective targeting region in a specific JAK or STAT (4). Clinical application of such selective JAK–STAT inhibitors could potentially destroy tumors without awakening the Dark Force of immune cells promoting tumor growth.

No potential conflicts of interest were disclosed.

The authors thank Margareta, Ana, Julia, Lea, and all members of the Cerwenka lab for discussions, and in particular for contributing to the title of the article.

The Cerwenka lab is funded by grants from German Cancer Aid, Deutsche Krebshilfe (110442 and 111455), and the German Research Foundation (DFG; RTG2099) to A. Cerwenka.