Research Interests
Cytotoxic cells, including natural killer (NK) cells and cytotoxic T cells, recognize and kill infected and transformed cells. Our laboratory investigates mechanisms of recognition of cancer cells and infected cells by natural killer cells and T cells. Furthermore, we investigate mechanisms by which tumors and infectious agents naturally activate or inhibit productive responses by cytotoxic cells, or desensitize the cells by inducing anergy. Our aim is to harness these findings to improve existing immunotherapies for cancer and design new ones. We have employed our understanding of NK recognition, activation, and inhibition to establish therapy combinations that super-activate NK cells, while at the same time preventing desensitization and inhibition of the cells. Our approaches have promise for activating T cell responses as well. These combinations show promise as new therapies for cancers that are refractory to existing immunotherapies.
Current Projects
Natural killer cell activation. We are investigating mechanisms by which tumors activate NK cell cytotoxicity against tumors. Though activated NK cells potently kill many tumors in vitro, they display relatively little or no activity when isolated from normal naïve animals or patients. Introduction of an exogenous tumor can nevertheless activate potent anti-tumor responses by NK cells. We are investigating the role of the cGAS-STING pathway in this process. This pathway responds to cytosolic dsDNA and induces production of type 1 interferons and other immune activators. Our results demonstrate that tumor cells activate the cGAS-STING pathway, leading to initiation of potent NK cell responses. Furthermore, activation of the CGAS-STING pathway with STING agonists induces powerful anti-tumor responses by NK cells. This approach is being evaluated in combination with other NK-targeted therapies in preclinical testing, and shows considerable promise.
Tumor cell recognition by NK cells. Our group has investigated the role of the NKG2D receptor in tumor recognition by NK cells. NKG2D recognizes up to 8 different, though related, cell surface ligands called NKG2D ligands. NKG2D ligands are poorly expressed by normal cells, but are frequently upregulated in tumor cells and in certain infected cells. Engagement of a ligand on an unhealthy cell causes the NK cell to kill it and secrete cytokines. We have identified numerous stress pathways that induce different NKG2D ligands on cells, including (i) the DNA damage response; (ii) the integrated stress response; (iii) hyper-proliferation; (iv) the PI-3 kinase pathway (in collaboration with the Coscoy lab), which is activated in virus-infected cells and cancer cells. We generated knockout mice lacking NKG2D and demonstrated a defect in tumor immunosurveillance, providing evidence for innate surveillance of cancer.
NK cell inhibition by KIR/Ly49 and checkpoint receptors. NK cells express inhibitory receptors (Ly49 in mice, KIR in human) that recognize class I MHC molecules and function to inhibit the lysis of cells that express class I molecules and function to inhibit the lysis of cells that express class I molecules normally, and allow the destruction of those that do not. Each NK cell stably expresses 3-4 of a panel of 10 possible inhibitory receptors, which bind to partially overlapping sets of polymorphic MHC I molecules. The selection of inhibitory receptors expressed by a given NK cell is determined randomly. We are dissecting the molecular mechanisms that select a random set of genes for expression and prevent other genes from being expressed. The mechanisms are central to how NK cells function, and are likely also important for understanding a variety of developmental decisions in immune cells and other lineages.
NK cells also express other inhibitory receptors. We have found that a fraction of NK cells within tumors frequently upregulate inhibitory checkpoint receptors, including PD-1, Lag3, and others. The PD-1+ NK cells are the most intrinsically active NK cells, but are inhibited if they engage tumors that express the PD-L1 ligand for NKG2D. We found that blockade of PD-1 or PD-L1 unleashes NK-mediated tumor regression, just as it unleashes anti-tumor T cell responses. Hence "checkpoint blockade" is relevant for enhancing NK responses against tumors, just as it induces T cell responses against tumors.
NK cell tolerance and desensitization to MHC I-deficient cells. Although NK cells are considered components of the innate immune-response, we discovered that they have the potential to attack self cells (e.g. self cells that lack MHC I expression). This potential autoreactivity must be limited by mechanisms that render NK cells self-tolerant. The self-tolerance mechanism (also called NK cell "licensing") represents a tuning mechanism that sets the triggering threshold of individual NK cells so as to maximize reactivity against unhealthy cells while preventing reactivity against self-cells, and works by desensitizing autoreactive NK cells. Hence, when mature NK cells from normal animals are exposed to an environment of MHC I-deficient cells, they are rapidly desensitized and rendered unresponsive to those cells. We have found that desensitization is regulated by cytokines, such that exposure to NK-activating IL-12/IL-18 or IL-2 family cytokines can reverse NK cell desensitization in certain contexts. Hence, NK cells infiltrating MHC I-deficient tumors are eventually desensitized and rendered ineffective, but cytokine administration can partially reverse desensitization, restoring anti-tumor responses and extending survival of tumor-bearing animals.
NK cell desensitization to normal cells expressing NKG2D ligands: a mechanism that restrains NK cell activity. Interestingly, when NK cells are exposed to cells expressing activating ligands, desensitization can also occur. We have discovered a fascinating example that occurs naturally in normal mice and to a greater extent in tumors. Though NKG2D ligands are thought to be expressed only by unhealthy cells. we found that endothelial cells within the lymph nodes of normal animals express the RAE-1 NKG2D ligands. When NK cells enter lymoh nodes, the NK cells encounter the RAE-1 molecules, resulting in desensitization of the NK cells. Interestingly, RAE-1 is strongly upregulated on endothelial cells within tumors, resulting in even greater desensitization of the NK cells that penetrate into a tumor. Hence, NK cells in RAE-1-deficient mice, or in NKG2D-knockout mice, exhibit higher functional activity than those in WT mice, especially in the context of tumors. Such mice therefore show greater NK-mediated rejection of tumors that are NK-sensitive but lack NKG2D ligands. Interestingly, blockade of NKG2D in vivo, with a shed, high affinity ligand that is naturally produced in tumor-bearing mice, also leads to greater rejection of such tumors. These findings suggest that shedding of high affinity NKG2D ligands may lead to blockade of the specific NKG2D ligands expressed by endothelial cells may represent a novel approach for immunotherapy of cancer.
Immunotherapy of tumors by targeting NK cells. The scope of T cell-targeted therapies may be limited because some tumors lose expression of MHC I molecules, while others do not express immunogenic peptide epitopes. Yet many or most tumors are sensitive to NK-mediated killing. Hence, therapies that target NK cells may represent a broad and important mode of cancer immunotherapy. We are testing the immunotherapeutic efficacy of combinations of agents that enhance activation of cytotoxic cells, prevent inhibition of the cells, and block desensitization of NK cells. The agents are being tested in preclinical cancer models, including tumor-transplant models and genetically engineered models of spontaneous cancer. As agents that activate NK cells and T cells we are employing STING agonists; as agents that prevent inhibition, we are employing checkpoint receptor antibodies such anti-PD-1; as agents that block or reverse desensitization we are testing cytokines and antibodies that block desensitizing NKG2D ligands. Preliminary results show that combination therapies targeting these independent mechanisms are more effective than monotherapy and promising for further development.
In recent studies we are collaborating with colleagues at clinical centers to evaluate translating our results in mouse models to human cancer. Using primary tumors from pediatric and adult patients we are testing the generality of our findings and mice and evaluating the potential of these treatments using cell culture models.