Because the use of immune checkpoint inhibitors (monoclonal antibodies against CTLA4, PD-1, or its ligand, PD-L1) has become more commonplace in cancer therapy, much research has been spurred to gain a more thorough understanding of how the immune system recognizes and mounts a defense against tumors. In a study published in the Proceedings of the National Academy of the Sciences (2017;114:1637-1642), Zhijian Chen, PhD, Professor in the Department of Molecular Biology and Investigator in the Howard Hughes Medical Institute, UT Southwestern Medical Center, Dallas, presented the results from a preclinical study that evaluated a murine anti-PD-L1 antibody in wild-type mice and in those lacking the ability to produce the cytosolic DNA sensor cyclic guanosine monophosphate (GMP)-adenosine monophosphate (AMP) synthase (cGAS).
The recognition of DNA in the cytosol as being pathogen-associated provides a useful mechanism for detecting a number of microbes that require DNA for their life cycle or contain DNA. Additionally, when the presence of human DNA, which is normally confined to the nucleus or the mitochondria, is detected in the cytosol of immune or nonimmune cells, a strong innate immune response can be triggered. This can result in the production of type I interferons (IFNs) and other pro-inflammatory cytokines.
One major sensor for this cytosolic DNA is the enzyme cGAS, which selectively binds to DNA of any sequence and catalyzes the formation of cyclic guanosine monophosphate-adenosine monophosphate (cGAMP). Once bound to DNA, cGAS adopts an activated conformation that then permits it to catalyze the formation of cGAMP from guanosine triphosphate (GTP) and adenosine triphosphate (ATP). This cyclic dinucleotide is created via the formation of two phosphoester linkages: one between the 5'-phosphate of GMP and the 3'-hydroxyl of AMP; the other between the 5'-phosphate of AMP and the 2'-hydroxyl of GMP.
Once formed, cGAMP then serves as a second messenger, binding to and activating STING, an adaptor protein that is localized on the endoplasmic reticulum. This cGAMP-bound protein then activates protein kinases TBK1 (TANK-binding kinase 1) and IKK (IΚB kinase), which then results in the activation of interferon regulatory factor 3 (IRF3) and nuclear factor kappa-light-chain-enhancer of B cells (NF-ΚB). These transcription factors then enter the nucleus and jointly induce immune-related and inflammatory gene products, such as tumor necrosis factor alpha (TNFα) and type I IFNs.
Because cGAS can recognize both foreign and native DNA (i.e., from the individual), activation of this enzyme has been associated with autoimmune disease states arising from the accumulation of cytoplasmic DNA in mice deficient in the DNA degrading enzymes DNAse, DNAse II, or Trex1.
Another potential source of activation for cGAS-STING pathway is the presence of tumor DNA, which may enter the cytoplasm when tumor cells are taken up by phagocytes (e.g., dendritic cells).
“In recent studies, STING-deficient tumor-bearing mice have been shown to be less responsive to treatment with radiation or immunotherapies based on the blockage of PD-1, PD-L1, CTLA4, or CD47 immunosuppressive pathways,” Chen explained. “Although, there are other studies that suggest STING activation may contribute to tumor growth and metastasis via induction of an immunosuppressive tumor microenvironment; consequently, STING's role in tumor immunity appears to be rather complex and poorly understood.”
Although immunotherapy has become a successful treatment for many cancer patients, the majority do not show a robust response to these antibodies. The effectiveness of immune checkpoint inhibitor therapy is dependent upon the patient's immune system.
“The most relevant aspects of this antitumor immunity are the recognition of tumor antigens and the generation of tumor-specific cytotoxic T cells,” Chen clarified. “To gain a better understanding of the cGAS-STING pathway and the role it plays in tumor immunity, we utilized the particularly aggressive B16F10 melanoma cell line in mouse xenograft studies.”
In Vivo Mouse Xenograft Models
Aggressive B16F10 melanoma tumor cells were injected subcutaneously into wild-type (WT), cGAS-deficient (cGAS-/-) and STING-deficient (Sting golden ticket, Stinggt/gt) mice. The mice were then split into control and treatment groups. The treatment mice were dosed with a murine PD-L1 antibody via IP injection. Then, tumor volumes and survival were recorded for both groups.
“There was little difference observed in the tumor growth between the three groups of mice in the control group, which received no antibody” Chen noted. “In the mice that were treated with the PD-L1 antibody, only the WT mice showed decreased tumor volumes and longer survival; the cGAS-/- and the Stinggt/gt mice clearly had impaired responses to the antibody.” Interestingly, similar PD-L1 expression levels were noted for tumor and dendritic cells harvested from all three groups of the treated mice. “This suggests that the impaired tumor immunity observed in the cGAS-/- and the Stinggt/gt mice was not the result of decreased PD-L1 expression levels,” Chen explained.
These data would seem to indicate that both cGAS and STING are required for PD-L1 antibody tumor response in this mouse melanoma model. “Both the cGAS-/- mice, which express STING but have impaired ability to generate cGAMP, and the Stinggt/gt mice, which have the ability to generate cGAMP but lack the STING adaptor protein, have impaired response to PD-L1 antibody therapy,” Chen stated.
Tumor-Infiltrating T Cells Analyses
Investigation of how the cGAS-STING pathway might alter the antitumor efficacy of PD-L1 antibody therapy was accomplished by inoculating WT, cGAS-/-, and Stinggt/gt mice with B16 melanoma cells that reliably expressed chicken ovalbumin (B16-Ova). At 7 and 10 days post inoculation, mice were treated with the murine PD-L1 antibody; then on day 14, the tumors were harvested. At harvest, the leukocytes were stained with a variety of antibodies, including ones that recognized ovalbumin, CD8 (tumor specific CD8 T cells), CD 45 (leukocytes), CD3 (T cells), CD4 & CD25 (regulatory T cells), and CD69 (activated T cells).
Only WT mice displayed reduced tumor volumes and increased leukocyte infiltration into the tumor mass. Among the cell types that were found at elevated numbers in these WT mice tumors were ovalbumin-specific CD8 T cells, CD69-positive (i.e., activated) CD8 and CD4 T cells, and regulatory CD4 T cells. When asked to explain these results, Chen commented, “These data would seem to suggest that both cGAS and STING are necessary for the generation and infiltration of leukocytes, including antigen-specific activated T cells, in tumors.”
Intramuscular cGAMP Injection
Since cGAS-/- mice seem to have impaired response to PD-L1 antibody therapy, it would appear that the product of this enzyme, cGAMP, may enhance antitumor immunity. Previous studies in mice by the Chen group have shown that intramuscular injection of cGAMP boosts antibody production as well as activation of CD8 and CD4 T cells.
To assess its antitumor activity, cGAMP was injected into the legs of WT mice distant from the flanks where tumor inoculation with B16 tumor cells had occurred. Differing doses of cGAMP were injected at days 4, 8, and 12 in these mice.
“Treatment with 10 μg of cGAMP was as effective as 200 μg of the murine PD-L1 antibody. Additionally, the combination of cGAMP and PD-L1 antibody was even more effective than either therapy alone,” Chen noted. “Titration studies showed that cGAMP enhanced the antitumor effects of the PD-L1 antibody in a dose-dependent manner.” No weight loss or other side effects were noted in those mice dosed with cGAMP alone or in combination with the PD-L1 antibody.
“We were able to show that injection of cGAMP at a site distant from the tumor can have a profound enhancement of the antitumor effect of the PD-L1 antibody. These results were somewhat surprising, as cGAMP, which has two highly polar phosphodiester moieties that could hinder its ability to enter cells, was able to display potent in vivo antitumor activity without the use of any transfection reagent or other delivery vehicle,” Chen observed.
Dendritic Cell Activation
Some possible means for cGAMP cell entry into dendritic cells (DCs) are phagocytosis, or perhaps pinocytosis. To assess the feasibility for this mechanism, bone marrow-derived DCs (BMDCs) were co-cultured with either granulocyte macrophage colony-stimulating factor (stimulates conventional DC differentiation) or Flt3 ligand (stimulates differentiation of plasmacytoid DCs). These DCs and those isolated directly from spleens were then treated with different concentrations of cGAMP without the use of a transfection reagent. This treatment resulted in the expression of the T cell co-stimulatory ligand CD86 as well as the production of IFNβ. “These results showed that DCs can directly take up extracellular cGAMP, which may explain the in vivo efficacy noted with intramuscular dosing of tumor-bearing mice,” Chen explained.
Tumor-Associated Antigen Cross-Presentation
Further study of the mechanism by which antitumor immunity is enhanced via cGAMP was next undertaken. In these experiments, BMDCs were harvested from WT, cGAS-/-, and Stinggt/gt mice and then subsequently incubated with irradiated B16-Ova cells and varying concentrations of cGAMP. Purified CD11c+ DCs were then incubated with CD8 T cells harvested from transgenic mice that expressed the T-cell receptor that targets the ovalbumin peptide. CD8 T-cell activation was gauged by the expression of the CD69 activation marker.
“Our results showed that cGAMP strongly stimulated activation in CD8 T cells obtained from WT and cGAS-/- mice, but not in those from Stinggt/gt mice,” Chen noted. “These results showed that cGAMP stimulates the cross presentation of tumor antigens to CD8 T cells in a STING dependent manner.”
When asked to summarize the results, Chen said, “With the B16 melanoma xenograft study, we were able to show that the antitumor efficacy for the murine PD-L1 was only noted in the WT mice, not those deficient in cGAS or STING. Additionally, the antibody treatment resulted in a large increase in tumor-infiltrating leukocytes, including tumor antigen-specific CD8 T cells.
“One possible explanation for this observation is that in taking up dead tumor cells, the DCs have their cytoplasm exposed to the tumor DNA, which then activates the cGAS-STING pathway, leading to increased T-cell priming and cytokine production for recruitment of leukocytes to the tumor site.”
Chen then offered an alternative explanation. “It is also possible that the PD-L1 antibody acts directly upon the dendritic cells. An earlier study showed that the blocking of PD-L1 expression in human myeloid DCs enhanced antitumor T-cell activation, which suggested that the PD-L1 blockade not only allowed killing of tumor cells that expressed PD-L1, but also promoted T-cell priming by DCs. Our data would seem to indicate that the cGAS pathway is required for PD-L1 blockade-mediated T-cell priming.
“The results for cGAMP intramuscular injection were especially encouraging, as no transfection reagent or other delivery vehicle was required to elicit this potent in vivo antitumor response,” he continued.
When asked about future directions for this research, Chen stated, “Clearly, additional research is needed to determine in which cells the cGAS pathway plays a role in activating antitumor responses, as well as the mechanism by which tumor DNA is delivered to the cytosol to initiate the cGAS-STING pathway. In our study, we utilized the PD-L1 antibody; however, further studies would need to be done to evaluate the role of cGAS in other immune checkpoint antibodies (e.g., PD-1, CTLA4, and CD47).”
When discussing the cell line used, Chen offered the following observations: “We utilized a very aggressive melanoma cell line for our mouse models; however, in the future, we should evaluate the role of cGAS using other models, including endogenous tumors in genetically engineered mice and transplanted tumors in syngeneic mice.”
Richard Simoneaux is a contributing writer.