Dynamic STING repression orchestrates immune cell development and function

Abstract

STING is an essential component of the innate immune system, yet homeostatic STING expression patterns and regulation are unknown. Using Sting1^IRES-EGFP reporter and conditional Sting1 transgenic mice, we found that regulation of STING expression is critical for immune cell development and functionality. STING expression was repressed in neutrophils and forced STING expression or signaling drove systemic inflammatory disease. During T lymphocyte development, STING expression was restricted at the double-positive stage via epigenetic silencing by DNA methyltransferase 1. Forced STING expression or signaling impaired T lymphocyte development independent of type I IFN and promoted lineage commitment to innate-like γδ T cells over adaptive αβ T cells. In the tumor microenvironment, CD8⁺ T lymphocytes repressed STING expression, correlating with features of T cell exhaustion in syngeneic mouse tumors and human colorectal cancer. Our data demonstrates the necessity of controlled, rather than ubiquitous STING expression, uncovering a previously unappreciated dimension of STING pathobiology. 

One-sentence summary:

Repression of STING expression in neutrophils and T lymphocytes is required for immune homeostasis and tumor response.

INTRODUCTION

Stimulator of interferon genes (STING) signaling is an essential component of the innate immune system involved in response to pathogenic infections and a wide breadth of human diseases. The mechanisms regulating the initiation and termination of STING signaling have been well-studied in myeloid cells in which STING primarily drives type I interferon (IFN-I) signaling. In contrast, STING in lymphoid cells predominately drives IFN-independent signaling activities, such as suppressing proliferation, endoplasmic reticulum (ER) stress, and cell death (1–7). The physiological importance of STING in T lymphocytes remains poorly understood. Additionally, STING expression patterns and functions in other immune cells remain largely unclear.

As a critical innate immune pathway, STING is thought to be widely expressed in immune cells and tissues. Chronic STING signaling has been implicated in disease pathologies associated with many immune and non-immune tissues. However, in most cases, the origin of STING signaling driving pathology is ill-defined and varies among diseases. For example, loss of the ubiquitously expressed cytosolic DNase, TREX1, results in self-DNA-mediated activation of cGAS-STING signaling (8–10). Both mice lacking Trex1 globally and specifically in dendritic cells develop systemic multi-organ inflammation, suggesting that STING signaling in a small subset of innate immune cells can be sufficient to drive systemic inflammatory pathology (11). In contrast, STING gain-of-function mutant mouse models (Sting1^N153S/+ and Sting1^V154M/+) develop interstitial lung disease, intestinal inflammation, and lymphopenia (12–18). These pathologies are drastically different from those reported in Trex1^−/− mice despite the common etiology of chronic cGAS-STING signaling. In fact, the major tissue pathologies observed in STING gain-of-function mice are instead caused by STING signaling originating in endothelial cells and T lymphocytes (16, 17, 19–21). Further, STING signaling has been implicated in many neurological diseases, where STING expression is increased in the brain but the role of STING expression and signaling remain unclear (22–27).

STING expression is regulated at both mRNA and protein levels in various disease settings (25, 28–34). For example, many cancer cells epigenetically silence STING transcription to evade innate immune signaling and anti-tumor response (29, 34–41). Intestinal epithelial cells do not express detectable STING protein at homeostasis, but STING expression increases drastically during intestinal infection or inflammation (30). Here, we established a Sting1^IRES-EGFP reporter mouse to systematically characterize STING expression spatially in tissues and temporally during immune cell development. Using this model, we found that STING expression is highly dynamic between immune cell populations and tissues. STING expression is silenced in neutrophils and restoration of STING signaling drove systemic autoinflammatory pathology. Additionally, STING expression was temporally restricted during thymopoiesis at the double-positive (DP) stage and forced STING expression and signaling impaired T lymphocyte development. Finally, STING expression was gradually repressed during tumor growth and this repression correlated with CD8 T lymphocyte exhaustion in mice and human cancer patients. Our data highlights the physiological importance of controlled, rather than ubiquitous STING expression.

RESULTS

STING-reporter mice reveal distinct patterns of STING expression

We generated a STING-reporter mouse (Sting1^IRES-EGFP) by inserting an internal ribosomal entry site (IRES)-EGFP cassette after the stop codon of the endogenous Sting1 gene locus using CRISPR/Cas9 (fig. S1, A and B). This strategy allows for temporal and spatial tracking of STING expression in mouse tissues and cells via measurement of cytoplasmic EGFP expression and fluorescence. We chose an IRES-EGFP expression strategy as opposed to a STING-EGFP fusion system because the latter is known to interfere with STING activity. In an HEK293T-based IFN-luc reporter assay, STING-EGFP fusion proteins induced a much lower level of luciferase activity compared to an untagged STING protein alone (fig. S1C).

We first examined STING expression patterns in tissues and compared the EGFP signal with that of endogenous STING antibody staining. We also confirmed the specificity of STING staining using Sting1^−/− mice (Fig. 1A). STING is expressed in lymphoid organs such as the spleen as well as pulmonary epithelium (12, 42). Endogenous STING and EGFP reporter expression patterns were similar in these tissues, validating our reporter strategy (Fig. 1A). We previously reported endogenous STING expression in Purkinje cells in the cerebellum, which we also validated here with the Sting1^IRES-EGFP reporter mouse (Fig. 1A) (27). In addition, STING expression was high in the epithelium of the fallopian tubes (Fig. 1A). The function of STING in the female reproductive tract is unclear, although we and others have previously observed breeding difficulties and unfavorable pregnancy outcomes in mouse models harboring a Sting1 gain-of-function mutant (13, 15, 18). Further, STING expression was undetectable in multiple tissues such as liver and kidney by endogenous STING staining or EGFP staining in Sting1^IRES-EGFP reporter mice (Fig. 1A). In human tissues, STING expression was also high in the tonsil, lung, and fallopian tube (fig. S1D). Together, these findings suggest that STING expression is not ubiquitous. Instead, STING expression is largely restricted to organs of the immune system and selective cell types within non-immune organs such as the lung, female reproductive tract, and Purkinje cells in the brain.

Figure 1: Landscape of STING expression in tissues and immune cells.

(A) Immunohistochemistry staining of mouse tissues from Sting1^IRES-EGFP, wild-type, and Sting1^−/− mice stained with anti-EGFP (brown) and anti-STING1 (green) antibodies. (B to D) Flow cytometry performed on single cell suspensions of whole blood (B), spleen (C), and lymph nodes (D) from 7–8-week-old Sting1^IRES-EGFP mice (n=3 biological replicates). The graphs on the left represent the percentage of each population that is EGFP-positive. The graphs on the right represent the mean fluorescence intensity (MFI) for each EGFP positive population. Data shown as mean ± SD. Each population was identified via cell surface markers and then analyzed for percentage of EGFP-positivity and MFI of EGFP. (E) Flow cytometry performed on single cell suspensions of whole blood obtained from healthy human donors. (n=3 biological replicates). The graphs on the left represent the percentage of each population that is STING-positive. The graphs on the right represent the MFI of STING for each population. Data shown as mean ± SD. See also fig. S1 and S2.

Next, we examined STING expression in immune cells at homeostasis by measuring the percentage of EGFP⁺ cells in each population as well as the mean fluorescence intensity (MFI) of EGFP via flow cytometry (Fig. 1, B to D and fig. S1, E and F). STING expression was nearly ubiquitous in circulating immune cell populations and across secondary lymphoid organs such as the spleen and lymph node with a few exceptions (Fig. 1, B to D). First, STING expression was surprisingly low to undetectable in neutrophils (CD11b⁺Ly6G⁺; Fig. 1B). Second, the MFI of EGFP in B lymphocytes (CD19⁺), dendritic cells (CD11b⁺CD11c⁺), and macrophages (CD11b⁺F4/80⁺) was higher in the lymph node than in the spleen (Fig. 1, C and D). These observations suggest that STING expression changes dynamically as immune cells traffic across lymphoid organs. Third, STING expression was generally higher in lymphoid cells than in myeloid cells. T lymphocytes (both CD4⁺ and CD8⁺) expressed the highest level of STING both in terms of percentage of EGFP⁺ cells and EGFP MFI (Fig. 1, B to D). This observation was interesting given that the biology of STING signaling in T lymphocytes is poorly understood. Additionally, we previously showed that STING functionalities in T lymphocytes are largely IFN-independent and are distinct from its IFN-dependent activities in macrophages (2).

We further analyzed STING expression in immune cell populations in human peripheral blood from three healthy donors (Fig. 1E and fig. S2A). Only a small percentage of neutrophils stained positive with an anti-STING antibody, consistent with our observations in mice. Human peripheral B lymphocytes also expressed low levels of STING (both percentage and MFI), which is distinct from our observations in mice. This observation was interesting because a gain-of-function mutation in STING (HsSTING-V155M) that causes STING-associated vasculopathy with onset in infancy (SAVI) only impacts T lymphocytes in humans while the equivalent mutation in mice (MmSting-V154M) impacts both B and T lymphocytes (12, 13).

STING expression and signaling are detrimental for mature neutrophil survival

Given the importance of STING signaling in innate immunity, we were intrigued by the extremely low level of STING expression in circulating neutrophils. Neutrophils are the most abundant innate immune cells in circulation with a very short half-life. They respond to infection by phagocytosing microbes and releasing neutrophil extracellular traps (NETs) to immobilize targets. To determine whether the lack of STING expression in neutrophils is physiologically important, we generated mice with “forced” STING expression and signaling in neutrophils by crossing a neutrophil-specific cre-driver (Mrp8-cre) to LoxP-STOP-LoxP Sting1-N153S mice (LSL-Sting1^N153S/+Mrp8-cre) (16). This gain-of-function mutant (N153S) activates STING signaling independent of upstream cGAS expression, which is low in neutrophils (12, 16, 43, 44). LSL-Sting1^N153S/+Mrp8-cre mice exhibited reduced body weight and overall survival compared to littermate cre⁻ controls (Fig. 2, A and B). These mice also developed hair loss with variable penetrance (fig. S3A). Analysis of neutrophils in peripheral blood by flow cytometry revealed a reduction of mature circulating neutrophils (CD11b⁺Ly6G^hi) and a compensatory increase in immature neutrophils (CD11b⁺Ly6G^lo) in LSL-Sting1^N153S/+Mrp8-cre mice compared to littermate cre⁻ controls (Fig. 2D). This phenotype (decreased mature and increased immature neutrophils in the blood) is commonly observed during inflammation as mature neutrophils migrate into peripheral tissues and immature neutrophils are prematurely released from the bone marrow (45, 46).

Figure 2: Elevated STING expression and signaling in neutrophils leads to systemic inflammation.

(A) Representative image (left) of 5-week-old LSL-Sting1^N153S/+ (cre-negative control) and LSL-Sting1^N153S/+Mrp8-cre littermates and quantification of body weight (right) of 6-week-old LSL-Sting1^N153S/+ (n=10) and LSL-Sting1^N153S/+Mrp8-cre (n=6) littermates. Data are shown as mean ± SD and are the pool of three independent experiments. Unpaired t test. ****P<0.0001. (B) Survival curve of LSL-Sting1^N153S/+ (black, n=13) and LSL-Sting1^N153S/+Mrp8-cre (red, n=12) littermates. Log-rank test. **P<0.01. (C) Representative image (left) of 6-week-old LSL-Sting1^N153S/+ and LSL-Sting1^N153S/+Mrp8-cre littermates and quantification of normalized splenic weight (right) of 6-week-old LSL-Sting1^N153S/+ (n=10) and LSL-Sting1^N153S/+Mrp8-cre (n=6) littermates. Data are shown as mean ± SD and are the pool of three independent experiments. Unpaired t test. ****P<0.0001. (D) Representative dot plot, gating, and quantification of mature (Ly6G^hi, M) and immature (Ly6G^lo, Imm) neutrophils in 6-week-old LSL-Sting1^N153S/+ (n=10) and LSL-Sting1^N153S/+Mrp8-cre (n=6) littermates. Data are shown as mean ± SD and are the pool of three independent experiments. Unpaired t test. **P<0.01, ***P<0.001, ****P<0.0001 (E and F) Multiplex quantification of serum cytokines and chemokines from indicated genotypes (top). (E) A heatmap summarizing all data (n=5 mice per genotype). (F) selected cytokines. Data shown as mean ± SD. Unpaired t test. *P<0.05. **P<0.01, ****P<0.0001, ns, not significant. (G) Representative image of spleens taken from 6-week-old LSL-Sting1^WT/+ and LSL-Sting1^WT/+Mrp8-cre littermates. (H) Confocal fluorescent microscopy images showing STING colocalization with the ER. Bone-marrow-derived neutrophils were isolated from LSL-Sting1^WT/+Mrp8-cre mice, and then stained with α-STING antibody (green), α-PDI antibody (an ER marker, red), and DAPI (nucleus marker, blue). See also fig. S3 and S4.

Indeed, LSL-Sting1^N153S/+Mrp8-cre mice exhibited a systemic autoinflammatory phenotype manifesting as massive splenomegaly, thymic atrophy, and widespread lymphadenopathy (Fig. 2C and fig. S3B). Histological examination of the spleen revealed complete disruption of white pulp architecture with the presence of neutrophil debris (fig. S3C). Serum cytokine analysis showed that LSL-Sting1^N153S/+Mrp8-cre mice had increased serum levels of cytokines and chemokines such as eotaxin, G-CSF, KC (CXCL1), MIP1β (CCL4), and RANTES (CCL5) (Fig. 2, E and F). However, we detected little to no IFNβ protein in the serum by ELISA (fig. S3D). Flow cytometry analysis of T lymphocyte populations revealed that LSL-Sting1^N153S/+Mrp8-cre mice had a lower percentage of CD4⁺ and CD8⁺ lymphocytes in the spleen and increased expression of CD95 (Fas), suggesting a higher level of activation (fig. S3, E and F). We also observed decreased naïve and effector memory CD8⁺ T lymphocytes with a corresponding increase in central memory CD8⁺ T lymphocytes in the spleen (fig. S3E). Together, these data suggest that STING expression and signaling are detrimental to mature neutrophils, and inadvertent STING signaling in neutrophils alone is sufficient to induce strong systemic inflammation and pathology.

As a control for the Mrp8-cre, we generated LSL-Sting1^N153S/+UBC^ERT2-cre mice (whole body inducible knock-in). After one tamoxifen injection, LSL-Sting1^N153S/+UBC^ERT2-cre mice became severely inflamed, moribund, and died as soon as 15h post-injection with a septicemic phenotype (fig. S3H). Analysis of circulating lymphocytes revealed an increase in the percentage of neutrophils, as well as increased mature and decreased immature neutrophils, consistent with an immediate short-lived inflammatory response (fig. S3G). We also observed a decreased percentage of B lymphocytes and no changes in CD4⁺ or CD8⁺ T lymphocytes within 15h of tamoxifen injection (fig. S3I). The acuity of the phenotype precluded the development of tissue abnormalities as seen in the LSL-Sting1^N153S/+Mrp8-cre mice.

Next, we generated LSL-Sting1^WT/+Mrp8-cre mice, in which wildtype Sting1 is expressed in neutrophils (fig. S4A). In this model, STING signaling would require additional endogenous cues for activation (e.g. DNA-cGAS-cGAMP-STING). These mice developed splenomegaly around 11 weeks of age, suggesting that increased STING expression in neutrophils alone can also drive pathologic inflammation (Fig. 2G). STING protein colocalized with the ER marker, PDI, suggesting that STING is properly localized when expressed in neutrophils in these mice (Fig. 2H).

We next examined endogenous STING function in neutrophils ex vivo. We isolated bone marrow-derived neutrophils from C57BL/6J mice (CD11b⁺Ly6G⁺, 98% purity) and stimulated them with the cell-permeable STING agonist diABZI (fig. S3, J and K). STING agonist treatment induced the expression of Ifnb1, Cxcl10, Ifit1, Il6 in neutrophils, suggesting that the low level of STING in these cells is capable of immune signaling (fig. S3K). Treatment with an IFNAR1-blocking antibody (α-IFNAR1) did not affect STING-mediated induction of immune gene expression, suggesting that the induction is cell-intrinsic (fig. S3L). However, STING activation in C57BL/6J neutrophils did not induce NETosis or oxidative burst as measured by ROS production (fig. S3, M and N). As a positive control, phorbol myristate acetate (PMA) treatment induced robust NET and ROS production in C57BL/6J neutrophils (fig. S3, M and N). Treatment with STING agonist, diABZI, also did not affect neutrophil phagocytosis of IgG-FITC beads (fig. S3O). Taken together, our data suggest that neutrophils repress STING expression to avoid inflammatory signaling at homeostasis, and that the primary function of STING signaling in neutrophils is likely cytokine and chemokine secretion to induce inflammation.

STING expression is temporally restricted during lymphocyte development

STING activation in mature T and B lymphocytes leads to reduced proliferation and even cell death (1, 3–7, 47, 48). However, very little is known about STING function during T and B lymphocyte development. We next examined STING expression at each stage of T lymphocyte development in the thymus using Sting1^IRES-EGFP mice (Fig. 3A and fig. S5, A and B). STING expression was dynamic during T lymphocyte development (Fig. 3, B to E). STING expression was initially high in double-negative (DN: CD4⁻CD8⁻) thymocytes and decreased to a very low level in the double-positive stages (DP: CD4⁺CD8⁺). STING expression then returned to high levels in single-positive thymocytes (SP: CD4⁺ or CD8⁺) (Fig. 3, B and C). We confirmed this pattern of expression using an antibody for endogenous STING protein via flow cytometry in C57BL/6J mice (Fig. 3, D and E). Spatially, this can be seen by high STING expression in the medulla (post-selection SP thymocytes) relative to diffuse staining of the cortex where most cells are DP (CD4⁺CD8⁺) (Fig. 3F). As a negative control, EGFP expression did not change during T lymphocyte development in the thymus of β-actin-EGFP reporter mice (fig. S5C). We also analyzed a publicly available single-cell RNAseq data set for STING expression in a thymus taken from a C57BL/6J mouse and observed a similar “high-low-high” pattern in DN-DP-SP thymocytes, respectively (Fig. 3G) (49). To examine whether DP TCRβ^−/lo/int thymocytes are functionally deficient in STING signaling, we isolated DP TCRβ^−/lo/int thymocytes from C57BL/6J mice and stimulated with diABZI to induce cell death. As a positive and negative control, we isolated splenic T lymphocytes from C57BL/6J and Sting1^−/− mice, respectively. diABZI treatment induced robust T lymphocyte death in wildtype but not Sting1^−/− T lymphocytes (Fig. 3H). Importantly, diABZI treatment induced less cell death in DP TCRβ^−/lo/int thymocytes compared to wildtype splenic T lymphocytes, suggesting that reduced STING expression in DP TCRβ^−/lo/int thymocytes protects them from STING-mediated cell death (Fig. 3H).

Figure 3: STING expression is temporally restricted during T lymphocyte development.

(A) Diagram showing the developmental stages of thymocytes and cell surface markers of each stage in the thymus. DN, double negative (CD4⁻CD8⁻). ISP, immature single positive. DP, double positive (CD4⁺CD8⁺). SP, single positive (CD4⁺ or CD8⁺), SM, semi-mature (CD4⁺ or CD8⁺), M, mature (CD4⁺ or CD8⁺). (B and C) Flow cytometry analysis of EGFP expression in developing thymocyte populations in 5–6-week-old Sting1^IRES-EGFP mice (n=3 biological replicates). (B) bar graph of percentage of EGFP-positivity at each stage. (C) bar graph of EGFP mean fluorescence intensity (MFI) of each stage. Data shown as mean ± SD. Each population was identified via cell surface markers and then analyzed for percentage of EGFP-positivity and MFI of EGFP. (D and E) Flow cytometry analysis of STING expression using α-STING antibody staining of thymus from 5–6-week-old C57BL/6J mice (n=5 biological replicates). (D) bar graph of percentage of STING-positivity at each stage. (E) bar graph of α-STING MFI of each stage. Data are shown as mean ± SD. (F) Representative IHC staining of STING1 expression (left) and H&E staining (right) of thymus taken from 8-week-old C57BL/6J mice. The thymic medulla is outlined in a dashed line. (G) Sting1 expression from publicly available scRNA sequencing data obtained from thymus taken from 11–13-week-old female and male C57BL/6J mice accessed via the Immunological Genome Project Single Cell Profiling dataset. Data visualized using the Broad Institute Single Cell Portal. (H) Flow cytometry analysis of STING agonist diABZI-induced T cell death. C57BL/6J splenic T cells, Sting1^−/− splenic T cells, and C57BL/6J DP TCRβ^−/lo/int T cells (sorted from thymus) were stimulated with vehicle or diABZI (1μM) for 16 h, stained with Zombie dye for cell death and then analyzed by flow cytometry. (n=4 biological replicates) Data shown as mean ± SD. Two-way ANOVA with post-hoc Šidák’s multiple comparison test. **P<0.01, ****P<0.0001. See also fig. S5 and S6.

STING expression during B lymphocyte development also followed a “high-low-high” pattern (fig. S6, A to F). Pre-pro-B lymphocytes expressed relatively high levels of STING which decreased as the cell approaches the small pre-B stage (fig. S6, A to D). STING expression then gradually increased to that of a mature B lymphocyte as it leaves the bone marrow and enters the periphery (fig. S6, A to D). Overall, these data reveal that STING expression is temporally restricted during lymphocyte development.

STING expression and signaling impair lymphocyte development

We next tested whether temporal restriction of STING expression during T lymphocyte development is physiologically necessary. We crossed LSL-Sting1^N153S/+ mice to a proximal Lck-cre driver to force STING expression and signaling during T lymphocyte development from the DN2 stage. LSL-Sting1^N153S/+Lck-cre mice had reduced thymus size and overall cellularity, indicating reduced thymopoiesis compared to cre⁻ littermate controls (Fig. 4, A and B). Supporting reduced thymopoiesis, these mice displayed decreased output of CD4⁺ and CD8⁺ T lymphocytes to the spleen (Fig. 4C). Detailed analysis of thymocyte populations demonstrated an increase of thymocytes at early stages (DN1-ISP) and a decrease in later stages (DP-SP4 and SP8), suggesting a developmental block that occurs at the transition from DN to DP thymocytes (Fig. 4, D and E). This block corresponded to the stage at which endogenous STING expression was downregulated from high in the DN populations to low in the DP populations (Fig. 3B). This block was not due to cre-mediated toxicity as we did not observe this phenotype in Lck-cre/+ mice when compared to cre⁻ littermates (fig. S7, A to C).

Figure 4: STING expression and signaling impairs T lymphocyte development.

(A) Quantification of thymocyte numbers obtained from thymus taken from 5-week-old LSL-Sting1^N153S/+ (n=6) and LSL-Sting1^N153S/+Lck-cre (n=6) littermates. Data shown as mean ± SD. Unpaired t test. ***P<0.001. (B) Representative image of thymus in LSL-Sting1^N153S/+ and LSL-Sting1^N153S/+Lck-cre littermates. (C) Absolute splenocyte count (left) and flow cytometry analysis of total splenic CD4⁺ (middle) and CD8⁺ (right) T lymphocyte counts from 7-week-old LSL-Sting1^N153S/+(n=6) and LSL-Sting1^N153S/+Lck-cre (n=6) littermates. Data shown as mean ± SD. Unpaired t test. *P<0.05. (D and E) Flow cytometry analysis of relative (D) and absolute (E) counts of thymocytes at each stage of development in single cell suspensions of thymus taken from 5-week-old LSL-Sting1^N153S/+ (n=6) and LSL-Sting1^N153S/+Lck-cre (n=6) littermates. Data shown as mean ± SD. Unpaired t test. *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001. (F) Flow cytometry analysis of thymocytes at each stage of development in single cell suspensions of thymus taken from 4-week-old LSL-Sting1^N153S/+Lck-cre mice treated with α-IFNAR antibody (n=5) or isotype IgG antibody (n=4) (50 μg every other day). Unpaired t test. **P<0.01. (G) Absolute splenocyte count (left) and flow cytometry analysis of total splenic CD4⁺ (middle) and CD8⁺ (right) T lymphocyte counts from 7-week-old LSL-Sting1^WT/+(n=8) and LSL-Sting1^WT/+Lck-cre (n=6) littermates. Data shown as mean ± SD. Unpaired t test. **P<0.01. (H) Quantification of thymocyte numbers obtained from thymus taken from 5–6 week-old LSL-Sting1^WT/+ (n=11) and LSL-Sting1^WT/+Lck-cre (n=17) littermates. Data shown as mean ± SD. Unpaired t test. *P<0.05 (I and J) Flow cytometry analysis of relative (I) and absolute (J) counts of thymocytes at each stage of development in single cell suspensions of thymus taken from 5–6 week-old LSL-Sting1^WT/+ (n=11) and LSL-Sting1^N153S/+Lck-cre (n=17) littermates. Data shown as mean ± SD. Unpaired t test. *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001. DN, double negative (CD4⁻CD8⁻). ISP, immature single positive. DP, double positive (CD4⁺CD8⁺). SP, single positive (CD4⁺ or CD8⁺), SM, semi-mature (CD4⁺ or CD8⁺), M, mature (CD4⁺ or CD8⁺). See also fig. S7 and S8.

We previously showed that STING activities in T lymphocytes are largely IFN-independent (2). To assess the role of IFN-I signaling in the STING-mediated block of T lymphocyte development, we treated LSL-Sting1^N153S/+Lck-cre mice with an α-IFNAR1 blocking antibody or IgG isotype control. α-IFNAR1 blocking antibody treatment eliminated cell surface IFNAR1 expression on cells taken from the blood and spleen and reduced cell surface IFNAR1 expression on thymocytes (fig. S7D). However, most T lymphocyte populations in the thymus of LSL-Sting1^N153S/+Lck-cre mice remained unchanged (except for a decrease in DN4) after 3 weeks of α-IFNAR1 antibody treatment (Fig. 4F). These data suggest the block in T lymphocyte development mediated by forced STING expression occurred independently of IFN-I signaling.

We also generated LSL-Sting1^WT/+Lck-cre mice to only force STING expression but not signaling. These mice require further endogenous cues for STING activation and signaling. Forced STING expression alone in LSL-Sting1^WT/+Lck-cre mice caused a similar developmental block at the transition from DN to DP thymocytes, resulting in decreased output of CD8⁺ T lymphocytes (Fig. 4, G to J). The overall phenotype was less severe when compared to LSL-Sting1^N153S/+Lck-cre mice, which could be due to the relatively low level of cGAS expression in the thymus (fig. S7E) (49, 50). Of note, STING expression does not seem to be required for T lymphocyte development as Sting1^−/− mice did not display a developmental block in vivo or in fetal thymic organ culture (fig. S7, F to I). Together, these data suggest that transient restriction of STING expression and signaling is important to ensure effective T lymphocyte development and output.

To evaluate the role of STING signaling during B lymphocyte development, we generated LSL-Sting1^N153S/+Cd79a-cre mice to force STING signaling from the earliest stage of B lymphocyte development in the bone marrow. We observed a near complete block in B lymphocyte development from the pre-pro B to the pro-B stage (fig. S8, A to E). This block nearly eliminated the peripheral population of B-1 and B-2 lymphocytes in the spleen (fig. S8, F to I). These data suggest that restricting STING expression and signaling is also critical for proper B lymphocyte development in mice.

STING signaling promotes survival of γδ T lymphocytes over αβ T lymphocytes

Our data thus far raises two questions: how STING signaling impairs T lymphocyte development and what regulates STING expression in the thymus. To investigate the role of STING signaling in T lymphocyte development, we performed RNA-seq of sorted DP (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes from Sting1^N153S/+Lck-cre and cre⁻ littermate control mice. At this stage of T lymphocyte development, STING expression was normally low to none (Fig. 3B) and forced STING expression overcame this restriction to drive signaling. Differential gene expression analysis identified 2,010 upregulated and 1,931 downregulated genes in Sting1^N153S/+Lck-cre mice (Fig. 5A). Ingenuity Pathway Analysis (IPA) of upregulated genes demonstrated enrichment for pathways related to inflammatory signaling including Interferon Signaling, IL-8 Signaling and Pathogen Induced Cytokine Storm Signaling (Fig. 5B). These findings are consistent with known STING signaling activities and confirmed that the transgene was active in thymocytes (Fig. 5, C to E). Top downregulated genes were dominated by T cell receptor (TCR) alpha (Tra) genes (Fig. 5F). Upon further examination of other TCR genes, TCR beta (Trb) genes were unchanged (Fig. 5F). However, the expression of TCR gamma (Trg) and delta (Trd) genes were increased 2 to 10-fold in the thymus of Sting1^N153S/+Lck-cre mice compared to cre⁻ littermate controls (Fig. 5F).

Figure 5: STING signaling favors γδ T lymphocytes survival over αβ T lymphocytes.

(A) Volcano plot showing all differentially expressed genes (DEGs) in LSL-Sting1^N153S/+Lck-cre (n=4 biological replicates) versus LSL-Sting1^N153S/+ (n=4 biological replicates) sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes taken from 5-week-old mice. Horizontal dashed line, P=0.05 (B) IPA of 2-fold or greater differentially upregulated genes in LSL-Sting1^N153S/+Lck-cre sorted thymocytes. Top pathways were ranked with −log₁₀(p-value). (C) Fold change of Sting1 expression in LSL-Sting1^N153S/+Lck-cre (n=4) versus LSL-Sting1^N153S/+ (n=4) sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes. Data shown as mean ± SD. (D) Fold change of selected interferon-stimulated genes (ISGs) in the Interferon Signaling pathway from LSL-Sting1^N153S/+Lck-cre (n=4) versus LSL-Sting1^N153S/+ (n=4) sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes. Data shown as mean ± SD. (E) Fold change of top upregulated genes in the IL-8 Signaling pathway from LSL-Sting1^N153S/+Lck-cre (n=4) versus LSL-Sting1^N153S/+ (n=4) sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes. Data shown as mean ± SD. (F) Fold change of T cell receptor (TCR) family genes in LSL-Sting1^N153S/+Lck-cre (n=4) versus LSL-Sting1^N153S/+ (n=4) sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes. (G and H) Flow cytometry analysis of γδ T lymphocytes from the thymus (G) and spleen (H) of 8-week-old LSL-Sting1^N153S/+ (n=18) and LSL-Sting1^N153S/+Lck-cre (n=15) littermates. Data shown as mean ± SD. Unpaired t test. *P<0.05, ***P<0.001. ns, non-significant.

This phenomenon of decreased expression of TCRα genes and corresponding increased expression of TCRγ and TCRδ genes suggests that forced STING expression may skew T lymphocyte lineage commitment. To confirm this, we measured γδ T lymphocytes in Sting1^N153S/+Lck-cre mice via flow cytometry. We observed an approximately 4-fold increase in the percentage and a 2-fold increase in absolute number of γδ T lymphocytes in the thymus of Sting1^N153S/+Lck-cre mice (Fig. 5G). We also analyzed γδ T lymphocytes in the spleen and observed an approximately 2-fold increase in the percentage in of γδ T lymphocytes in Sting1^N153S/+Lck-cre mice (Fig. 5H). Together, these data suggest that STING signaling skews T lymphocyte lineage commitment from adaptive αβ to more innate-like γδ T lymphocytes.

STING expression is epigenetically regulated in immune cells in vivo

We next investigated how STING expression is regulated during T lymphocyte development. In cancer cells, STING expression is often epigenetically silenced, and restoration of STING expression inhibits cancer cell proliferation (29, 34–41, 51, 52). We analyzed chromatin accessibility of the Sting1 transcriptional start site (TSS) using a publicly available ATAC-seq dataset of wild-type mouse thymus (49). Accessibility of the Sting1 TSS demonstrated a similar high-low-high pattern with the lowest chromatin accessibility at DP stages which corresponds with the lowest STING expression at this stage (fig. S9A) and suggests that STING expression is epigenetically silenced in DP T lymphocytes. We next asked whether STING expression can be restored by DNA methyltransferase inhibitors. Treatment of HEK293T cells, which do not express STING, with Decitabine (DAC, 5-aza-deoxycytidine, DNMT1/3 inhibitor) or GSK3685032 (selective DNMT1 inhibitor) for 48h restored STING expression in a dose-dependent manner (fig. S9B).

Next, we treated C57BL/6J and Sting1^−/− mice with DAC in vivo by I.P. injection every other day for 7 days. On day 7, STING protein levels in the thymus were increased in DAC-treated mice (Fig. 6A). Following treatment of Sting1^IRES-EGFP mice with DAC, EGFP MFI was increased in mature CD3⁺, CD4⁺, and CD8⁺ T lymphocytes in the spleen (Fig. 6B). Additionally, DAC treatment increased STING expression in myeloid and B lymphocyte populations in the blood, as measured by percentage of EGFP⁺ cells or EGFP MFI (Fig. 6, C and D). Finally, we treated Sting1^IRES-EGFP mice with a selective DNMT1 inhibitor, GSK3685032, that can be administered for an extended period without overt toxicity (subcutaneous twice daily for 14 days) (53, 54). GSK3685032 treatment increased STING expression (EGFP) in DP TCRβ^−/lo/int thymocytes (Fig. 6E). GSK3685032 treatment also increased the percentage of γδ T lymphocytes in the thymus and the blood (Fig. 6F) and increased STING expression (EGFP) in γδ T lymphocytes in the thymus and spleen (Fig. 6, G to I). These data suggest that STING expression is epigenetically regulated, by DNMT1 and likely others, in immune cells in vivo.

Figure 6: STING expression is epigenetically regulated in immune cells.

(A) Western blot analysis of STING and DNMT1 protein expression levels in thymus taken from 7-week-old C57BL/6J and Sting1^−/− mice at day 7 after 3 intraperitoneal (I.P) injections of 0.4 mg/kg Decitabine (DAC) or vehicle control given on days 1, 3, and 5. (n=3) (B) Flow cytometry analysis of EGFP MFI in CD3⁺, CD4⁺, and CD8⁺ circulating T cells from 4-week-old Sting1^IRES-EGFP mice at day 15 after 6 intraperitoneal injections of 0.4 mg/kg Decitabine (DAC) (n=6) or vehicle control (n=9) given on days 1, 3, 5, 8, 10, 12. Data are shown as mean ± SD. Unpaired t test. ****P<0.0001. (C and D) Flow cytometry analysis of EGFP expression (% EGFP-positivity and EGFP MFI) and representative EGFP histograms in circulating CD11b⁺ myeloid (C), and CD19⁺ B cells (D) in whole blood taken from Sting1^IRES-EGFP mice at day 5 after 2 intraperitoneal injections of 0.4 mg/kg Decitabine (DAC) (n=4) or vehicle control (n=3) on days 1 and 3. Data are shown as mean ± SD. Unpaired t test. **P<0.01. Each population was identified via cell surface markers and then analyzed for percentage of EGFP-positivity and MFI of EGFP. (E) Flow cytometry analysis of EGFP expression (% EGFP-positivity and EGFP mean fluorescence intensity, MFI) in double positive, DP (CD4⁺ CD8⁺) TCRβ^−/lo (left-most panels) and double positive (CD4⁺ CD8⁺) TCRβ^int (right-most panels) in 4-week-old Sting1^IRES-EGFP mice at day 15 after 14 days of twice daily subcutaneous (S.C.) injection of 45 mg/kg GSK3685032 (n=7) or vehicle control (n=7). Data shown as mean ± SD. Unpaired t test. ****P<0.0001. (F) Flow cytometry analysis of γδ T lymphocytes in the thymus, spleen and blood of Sting1^IRES-EGFP mice after GSK3685032 (n=10) or vehicle control (n=7) treatment as in E. Data shown as mean ± SD. Unpaired t test. **P<0.001, ****P<0.0001. ns, non-significant. (G to I) Flow cytometry analysis of EGFP expression in γδ T lymphocytes in the thymus (G), spleen (H) and blood (I) of Sting1^IRES-EGFP mice after GSK3685032 or vehicle control treatment as in F. Data shown as mean ± SD. Unpaired t test. **P<0.001, ****P<0.0001. ns, non-significant. See also fig. S9.

STING expression is bimodal in mature lymphocytes

T lymphocytes are heterogenous, even within a defined subset, in their ability to proliferate in response to antigens and effector functions. To determine whether STING expression is also heterogenous within mature T lymphocyte subsets, we measured STING expression in in CD4⁺ and CD8⁺ naïve, effector, central memory, effector memory, follicular helper, and stem cell memory T cells from Sting1^IRES-EGFP reporter mice (fig. S10, A to E). Naïve T lymphocytes (CD4⁺ and CD8⁺) showed uniformly high STING expression whereas effector T lymphocytes consisted of both a STING⁺ and STING⁻ population in a distinctive bimodal distribution (fig. S10, A and B). Approximately 50% of CD4⁺ and 70% of CD8⁺ effector T cells were STING⁻, respectively. A similar distribution of STING expression was observed in effector memory T cells. In contrast, γδ T cells in the thymus, spleen, and blood all uniformly expressed high levels of STING, with the highest expression (by MFI) in the blood (fig. S10, F and G). In mature B lymphocytes, STING expression was also bimodal in B-2 cells overall and in more specific subsets such as marginal zone and follicular B lymphocytes (fig. S11, A to E). These data reveal that STING expression is heterogenous in mature lymphocytes. Additionally, effector T lymphocytes consist of a population of STING⁺ and STING⁻ cells. Given the antiproliferative activity of STING in T lymphocytes, this may reflect a mechanism to control the proliferative capacity and activation of effector T lymphocytes.

T lymphocytes restrict STING expression in the tumor microenvironment

Next, we investigated the dynamics of STING expression in T lymphocytes during a tumor challenge. We injected MC38 cells subcutaneously into Sting1^IRES-EGFP mice. MC38 tumors are immunologically “hot” tumors that attract large numbers of tumor-infiltrating T lymphocytes (TIL). After allowing the tumor to grow for 20 days, we isolated T lymphocytes from the tumor and peripheral immune tissues including the spleen, draining and non-draining lymph nodes, and peripheral blood (Fig. 7A and fig. S12). While STING expression remained high in circulating and peripheral T lymphocytes, a substantial fraction of CD4⁺ and CD8⁺ TILs lost STING expression (Fig. 7A). We observed a continued decline of STING expression in CD8⁺ TILs as the tumor grew (Fig. 7B). In some cases, 80% of CD8⁺ TILs become STING⁻ by day 25 when tumor size approached 2000 mm³ (Fig. 7B). Using an additional murine lung cancer model (KP67–1 cells (55)), we injected KP67–1 cells subcutaneously into Sting1^IRES-EGFP mice. Both CD4⁺ and CD8⁺ TILs had decreased STING expression compared to T cells in non-tumor tissues (Fig. 7C).

Figure 7: T lymphocytes repress STING expression in the tumor microenvironment.

(A) Flow cytometry analysis of EGFP expression (% EGFP-positivity) in CD4⁺ (left) and CD8⁺ (right) T lymphocytes in indicated tissues from 9-week-old Sting1^IRES-EGFP mice (n=6) at day 20 after subcutaneous injection of 5 × 10⁵ MC38 tumor cells. Data shown as mean ± SD and are representative of 3 independent experiments. One-way ANOVA with post-hoc Tukey’s multiple comparisons test. ****P<0.0001. (B) Flow cytometry analysis of EGFP expression (% EGFP-positivity) in CD4⁺ (left) and CD8⁺ (right) T lymphocytes in MC38 tumors from 8–10-week-old Sting1^IRES-EGFP mice (n=5–6 per time point) at day 15, 20, and 25 after subcutaneous injection of 5 × 10⁵ MC38 tumor cells. Data are shown as mean ± SD. One-way ANOVA with post-hoc Tukey’s multiple comparisons test. ns, non-significant, *P<0.05, and **P<0.01. (C) Flow cytometry analysis as in (B) but using KP67–1 tumor-bearing mice. (n=8) One-way ANOVA with post-hoc Tukey’s multiple comparisons test. ****P<0.0001. (D) Diagram showing adoptive transfer via intravenous injection of 5 × 10⁶ sorted EGFP⁺ splenic T cells from Sting1^IRES-EGFP mice to Rag1^−/− mice 7 days prior to subcutaneous injection of 5 × 10⁵ MC38 tumor cells. (E) Flow cytometry analysis of EGFP expression (% EGFP-positivity) in adoptively transferred CD4⁺ (left) and CD8⁺ (right) T lymphocytes in indicated tissues from 6-week-old Rag1^−/− mice (n=5) at day 23 after subcutaneous injection of 5 × 10⁵ MC38 tumor cells. Data shown as mean ± SD. Unpaired t test. ****P<0.0001. (F to H) Flow cytometry analysis of T cell exhaustion markers (indicated on the top). Tumors are from 8–10-week-old Sting1^IRES-EGFP mice (n=6) at day 20 after subcutaneous injection of 5 × 10⁵ MC-38 cells. Data shown as mean ± SD. Paired t test. *P<0.05, **P<0.01, ***P<0.001. Each population was identified via cell surface markers and then analyzed for percentage of EGFP-positivity. (I) Violin plot (left) of Sting1 expression in effector (T_EFF) and exhausted (T_EXH) T lymphocytes and dot plot (right) of population-defining markers for T_EFF and T_EXH cells. Unpaired t test. See also fig. S12.

The loss of STING expression in CD8⁺ TILs could be due to cell-intrinsic transcriptional silencing of STING or selective recruitment of STING⁻ T lymphocytes into the tumor. To determine whether STING expression in T lymphocytes is transcriptionally silenced cell-intrinsically, we adoptively transferred sorted 100% STING⁺ T lymphocytes into tumor-bearing Rag1^−/− mice (Fig. 7D). After tumor growth, we again observed a substantial STING⁻ intratumoral population of T lymphocytes despite the expression in circulating cells remaining uniformly high (Fig. 7E). These data suggest that STING expression is transcriptionally silenced in TILs, likely after they have encountered tumor antigen and begin to proliferate. Hyperproliferation and chronic exposure to tumor antigen leads to T lymphocyte exhaustion. We stained for exhaustion markers, PD-1, LAG-3, and TIGIT. A higher proportion of STING⁻ TILs were exhausted (PD-1⁺LAG-3⁺, PD-1⁺TIGIT⁺) when compared to their STING⁺ counterparts (Fig. 7, F and G). PD-1⁺TIM3⁺ TILs correlated with STING positivity, likely due to TIM3 being IFN-inducible (Fig. 7H) (56). To further corroborate these findings, we analyzed scRNAseq data from a CAR-T model, where murine CD19-targeting CAR-T lymphocytes were injected into wild-type mice bearing CD19-expressing B16 melanoma tumors (57). In this model, Sting1 expression was lower in exhausted T lymphocytes compared to effector T lymphocytes (Fig. 7I).

Finally, we analyzed STING expression in a cohort of colorectal cancer patients. A total of 57 patients diagnosed with colorectal cancer were analyzed (table S1). We obtained fresh cancer and adjacent non-cancerous tissues (para-cancer) for flow cytometry analysis with anti-STING and T cell markers (Fig. 8A and fig. S13). STING expression was lower in intratumoral CD8⁺ T cells compared to CD8⁺ T cells in adjacent non-cancerous tissues, similar to our findings in mice (Fig. 8B). As expected, CD8⁺ T cells with an exhausted phenotype (CD8⁺PD-1⁺LAG3⁺) were higher in cancer compared to non-cancerous tissues (Fig. 8C). In both tissues, the majority of CD8⁺PD-1⁺LAG3⁺ T cells were STING⁻, similar to our findings in mice (Fig. 8, D and E). Together, our data suggest that T lymphocytes restrict STING expression in the tumor microenvironment and loss of STING expression strongly correlates with expression of markers associated with T cell exhaustion.

Figure 8. STING expression analysis in a cohort of human colorectal cancer patients.

(A) Diagram showing experimental workflow. (B) Flow cytometry analysis of STING expression in CD8⁺ T cells in cancer or adjacent non-cancer tissue (Para-cancer). Left, representative FACS plots. Right, graphical summary of data (n=57). Paired t test. ****P<0.0001. (C) Flow cytometry analysis of T cell exhaustion (PD-1⁺LAG-3⁺) in cancer or adjacent non-cancer tissue (Para-cancer). Left, representative flow cytometry plots. Right, graphical summary of data (n=57). Paired t test. ****P<0.0001. (D) Pie chart showing percentages of CD8⁺PD-1⁺LAG-3⁺ T cells that are STING positive or negative in cancer or para-cancer tissues. Paired t test. *P<0.05. See also fig. S13.

DISCUSSION

Previous studies have established the cellular and molecular mechanisms of STING activation and downstream signaling outcomes, although a comprehensive analysis of STING expression in tissues and immune cells is lacking. We established a Sting1^IRES-EGFP reporter mouse to define STING expression in tissues and cells in vivo. Using this mouse model, we uncovered multiple interesting aspects of STING biology encompassing neutrophil homeostasis, adaptive lymphocyte development, and T lymphocyte lineage commitment and response to tumor. We used a LoxP-STOP-LoxP transgenic approach to express either gain-of-function Sting1-N153S or wild-type Sting1 at the ROSA26 locus under control of the CAG promoter to avoid chromatin regulatory mechanisms at the endogenous Sting1 locus and to thereby “force” STING expression or signaling. We observed similar phenotypes for both alleles, although Sting1-N153S phenotypes were generally stronger than Sting1, which rely on endogenous cues (e.g., DNA-cGAS-cGAMP) that could be limiting in certain cell types.

One major finding of our work is that both mouse and human neutrophils expressed very low levels of STING (approximately 10% or less were STING⁺). C57BL/6J mouse neutrophils still responded to STING agonist stimulation ex vivo and activated immune-related gene expression, but STING activation did not affect neutrophil functions such as NETosis, ROS production, or phagocytosis. Forced STING expression in neutrophils with the Sting1 allele or the gain-of-function Sting-N153S allele led to systemic inflammation due to overproduction of cytokines and chemokines. Thus, our data suggest that STING signaling is detrimental for neutrophils at homeostasis. Neutrophil cGAS-STING signaling also induced inflammation in disease settings such as pulmonary and myocardial ischemia (MI) (58, 59). In the MI model, cGas^−/− or Sting^−/− did not affect neutrophil functions such as ROS production, maturation, activation, or degranulation. Rather, cGAS-STING is required for an immune transcriptional program that regulates neutrophil production and differentiation. Therefore, it is possible that STING expression may be induced in neutrophils in disease settings to promote inflammation.

Another interesting finding is the multifaceted nature of STING function in T lymphocytes. STING expression was temporally restricted during T lymphocyte development in the thymus at the DP stage, and forced STING expression with the Sting1 allele or the Sting-N153S allele from early progenitors impaired T lymphocyte development independent of IFN-I signaling. Mechanistically, STING expression was epigenetically repressed in DP TCRβ^−/lo/int thymocytes by DNMT1 and possibly other regulators. However, the physiological reason for STING repression in DP TCRβ^−/lo/int thymocytes warrants further investigation. DP TCRβ^−/lo/int thymocytes were resistant to STING-mediated cell death in vitro. Most DP T cells will undergo cell death in the context of positive and negative selection in the thymus, a process that could release cellular DNA capable of activating cGAS-STING signaling and thereby skew selection. The transition from DN to DP also coincides with first TCRβ and then TCRα gene rearrangement. This process is known to produce excision DNA products that could also activate cGAS-STING signaling, however we did not directly test whether sensing of these DNA products is responsible for the T lymphocyte developmental block that we observe. Additionally, while the STING-mediated block in T lymphocyte development was independent of IFNAR signaling (IFN-I), we did not test the role of IFN-II or IFN-II signaling in this context. Further investigation is needed to elucidate the physiological requirement for STING repression in DP TCRβ^−/lo/int thymocytes.

Forced STING signaling favored T lymphocyte lineage commitment to innate-like γδ T cells over adaptive αβ T cells. This is interesting because we have limited knowledge of the factors driving lineage commitment of αβ and γδ T lymphocytes in the thymus. The first round of TCR gene rearrangement (TCRβ, γ, and δ) occurs at the DN2/3 stages. The rearrangement of TCRα genes occurs subsequently during the transition to double-positive (DP). The strength of the TCR signal is believed to dictate αβ verses γδ linage commitment, with a weak signal promoting αβ fate and a strong signal promoting γδ fate (60–62). Most T lymphocytes become αβ T cells while only 1–4% of T cells in the thymus and secondary lymphoid organs are γδ T cells (63). We and others previously showed that STING activation in mature T lymphocytes promotes calcium release from the ER that synergizes with TCR signaling (6, 64, 65). Therefore, it is likely that STING signaling at the DN stages increases TCR signaling strength to favor γδ T cell fate. γδ T lymphocytes can provide an early source of IFNγ in response to infection, tissue damage, and tumor establishment (66–69). Therefore, skewing towards innate-like γδ T cells could prime stronger adaptive immune responses.

In addition, our work suggests that tumor-infiltrating T lymphocytes progressively repress STING expression as a tumor grows. One possibility is that T lymphocytes repress STING expression upon recognition of tumor antigen, but this remains to be tested using a defined tumor or model antigen. We observed a bimodal distribution of STING expression in T lymphocytes at homeostasis and nearly 50% of effector T lymphocytes expressed low levels of STING. Repressing STING expression in tumor-infiltrating T lymphocytes may be a protective mechanism for T lymphocytes to circumvent STING-mediated cell death. Consistent with this notion, we have previously shown that Sting1^−/− T lymphocytes control tumors better than wildtype T lymphocytes (2). We also observed a strong correlation between repressed STING expression and signs of CD8⁺ T cell exhaustion in both mouse syngeneic tumor models and in a cohort of human colorectal cancer patients. The role of STING during the process of T lymphocyte exhaustion requires further investigation.

Lastly, our findings suggest the potential for severe lymphotoxicity of STING agonists as therapeutics. Multiple STING agonists are being developed for cancer immunotherapy, although early results demonstrate limited efficacy. In these clinical trials, even localized intratumoral injection of STING agonists nearly immediately entered systemic circulation (70). Our data suggest that STING agonist exposure to the thymus, particularly STING^hi DN T lymphocytes could impair thymopoiesis and T lymphocyte development. Also, STING activation in mature T lymphocytes leads to cell death, representing an additional mechanism to limit an efficacious anti-tumor response.

In summary, our findings reveal complex dynamics of STING expression that orchestrate both innate and adaptive immune cell function. We focused on immune cells in this study, but STING expression could also be dynamic in other non-immune cell types and during disease onset. Our data highlights the utility of the Sting1^IRES-EGFP reporter mouse as a tool for the community to further understand STING pathobiology in mouse models of human disease.

MATERIALS AND METHODS

Study design

The objective of this study was to elucidate patterns of STING protein expression and understand the physiological importance of regulated STING protein expression from immune cell development to maturity. We generated a STING reporter mouse and characterized STING expression in mouse and human peripheral immune cell populations using flow cytometry and immunohistochemistry. We then used transgenic conditional knock-in mice to restore STING expression and signaling in neutrophils. We evaluated systemic inflammatory disease and the effect of STING on neutrophil functionality using flow cytometry, cytokine analysis, histopathology, and in vitro functional assays of neutrophils. To investigate the role of STING during immune cell development, we characterized STING expression during T and B lymphocyte development using flow cytometry. We then determined whether the regulation of STING expression was required for T and B lymphocyte development by forcing STING expression and signaling during development using transgenic conditional knock-in mice. Using RNAseq, we evaluated the functional outcome of forced STING signaling during T lymphocyte development. We used pharmacological inhibition of DNMT1 in mice to evaluate the role for epigenetic regulation of STING expression. Finally, we used murine tumor models and a human patient cohort to evaluate the dynamics of STING expression in T lymphocytes during tumor growth.

Sample sizes were chosen based on previous studies and littermate availability in the case of live experimental animals. Tests of normality were used to determine data distribution, or the data was assumed to be normally distributed in all statistical tests used for data analysis. Details regarding biological and technical replicates are as indicated in figure captions. Mice used in in vivo experiments were sex- and age-matched as available. Experimental mice were cohoused and littermate controls. Randomization of treatment was applied to in vivo experiments and researchers were blinded to genotype and treatment status during tissue collection, data acquisition, and data analysis.

Mice

C57BL/6J, Rag1^−/−, MRP8-Cre-IRES/EGFP, Lck-Cre, β-actin-EGFP, CD79a-Cre and UBC^ERT2-cre mice were purchased from the Jackson Laboratory (#000664, #002216, #021614, #003802, #006567, #020505, and #007001). Sting1^−/− mice were obtained from Glen Barber (Univ. of Miami). LSL-Sting1^N153S/+ mice were obtained from Jonathan Miner (University of Pennsylvania). Sting1^IRES-EGFP and LSL-Sting1 mice were generated by Cyagen Biosciences Inc. Sting1^IRES-EGFP mice were generated by coinjection of gRNA to murine Sting1, donor vector containing “IRES-EGFP” cassette, and cas9 mRNA into fertilized mouse eggs. LSL-Sting1 mice were generated by coinjection of gRNA to the mouse ROSA26 gene locus, donor vector containing “CAG-promoter-loxP-PGK-Neo-6*SV40 pA-loxP-Kozak-murine Sting1 CDS-IRES-EGFP-rBG pA” cassette and cas9 mRNA into fertilized mouse eggs. F0 founder animals were identified by PCR followed by sequencing analysis and bred to wildtype (C57BL/6J) to generate F1 animals. Tamoxifen-mediated cre expression was induced with I.P injection of 75 mg/kg Tamoxifen (Sigma-Aldrich, T5648) in corn oil at a stock concentration of 20 mg/mL. Both male and female mice were used in the study. All mice were housed in standard pathogen free barrier facilities under 12h light-dark cycles. Animal work was approved by the Institutional Animal Care and Use Committee at University of Texas Southwestern Medical Center.

Histopathology, immunohistochemistry, and confocal immunofluorescence microscopy

Mouse tissues collected for histopathology and immunohistochemistry staining were taken from C57BL/6J, Sting1^−/− or Sting1^IRES-EGFP mice after whole body perfusion with PBS by cardiac puncture. Spleen, thymus, fallopian tube, and kidney were directly fixed for 24 hours at 4°C in 4% paraformaldehyde (PFA) and then switched to PBS. Brains and livers were prepared by whole body perfusion of PBS and fixation with 4% PFA via cardiac puncture. Once fixed, brains and livers were removed from the mouse to a 50 mL conical containing 4% PFA for 24–48 hours at 4°C. Lungs were flushed of red blood cells by injection of 3–5 mL PBS into the right cardiac ventricle, followed by intratracheal inflation with 1 mL 4% PFA. Lungs were transferred to a 50 mL conical containing 4% PFA for 1 hour at RT and then switched to PBS at 4°C. Paraffin embedding, sectioning, immunofluorescence, and hematoxylin/eosin (H&E) staining of C57BL/6J and Sting1^−/− mouse tissues were performed at UT Southwestern Tissue Management Shared Resource Core. Briefly, paraffin sections were deparaffinized in xylene and serial ethanol followed by H&E staining or antigen epitope retrieval prior to incubation with primary and secondary antibodies. Conventional IHC staining of EGFP from Sting1^IRES-EGFP mouse tissues was performed by HistoWiz (Brooklyn, New York). Primary antibodies used include anti-STING (19851–1-AP, Proteintech, 1:500), anti-EGFP (ab183734 1:100 or ab6556 1:300, Abcam), anti-Neutrophil Elastase (ab68672, Abcam, final 5 μg/mL), and anti-Human/Mouse Myeloperoxidase/MPO (AF3667, Biotechne, final 10 μg/mL). Secondary antibodies used for immunofluorescence include Donkey anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor^™ Plus 488 (A-32790, ThermoFisher, 1:200), Cy^™3 AffiniPure^™ Donkey Anti-Goat IgG (H+L) (705–165-147, Jackson Immunoresearch, final 7.5 μg/mL) and Alexa Fluor^® 647 AffiniPure^™ Donkey Anti-Rabbit IgG (H+L) (711–605-152, Jackson Immunoresearch, final 7.5 μg/mL). Images were captured using a Zeiss Axioscan Z1 slide scanner (fluorescent) or Hamamatsu Nanozoomer 2.0HT for brightfield.

Flow cytometry staining

The cell surface antibodies and corresponding isotype controls used for flow cytometry are listed in table S2. Dead cells were excluded using Zombie Aqua^™ Fixable Viability Kit (BioLegend, 423102) followed by blocking of Fc receptors using Mouse TruStain FcX (101320, BioLegend) Human TruStain FcX (422302, BioLegend) prior to staining with cell surface antibodies. Cells were stained for cell surface markers in FACS buffer (2% FBS in PBS, 200 mM EDTA) with indicated antibodies for 30 minutes at 4°C and washed with FACS buffer. For intracellular staining (anti-STING, anti-CD68 only), cells were first stained for surface markers, then fixed and permeabilized in Cytofix/Cytoperm^™ solution at 4°C for 20 minutes, washed in Perm/Wash buffer and stained with anti-STING (Proteintech, 19851–1-AP, 1:5000) for 1h at 4°C. After washing, cells were stained with donkey anti-Rabbit IgG AlexaFluor^™ Plus 488 (A-32790, ThermoFisher, 1:1000) at 4°C for 1h. Data was collected on Beckman-Coulter CytoFLEX and analyzed using FlowJo software (version 10.8.1). To ensure an unbiased assessment, all subpopulations were identified before analysis of EGFP percentage and mean fluorescence intensity (MFI).

Neutrophil isolation

Neutrophils were isolated from mouse bone marrow with density gradient centrifugation as described previously (71). Briefly, bone marrow cells were flushed out of femurs and tibias and red blood cells were lysed. Bone marrow cells were overlayed on Histopaque-based density gradient and centrifuged for 30 mins at 872 × g at room temperature without brake. Neutrophils were collected and washed. The purity of isolated neutrophils (CD11b⁺Ly6G⁺) was confirmed with flow cytometry.

NETosis assay

NETosis of neutrophil was assessed by flow cytometry as described previously (72). Briefly, isolated neutrophils were stimulated with STING agonist or PMA in the presence of cell impermeable nucleic acid stain SYTOX Green or SYTOX Blue (Thermo Fisher, 10 nM) for 2 h. SYTOX Green⁺ or SYTOX Blue⁺ cells were measured by flow cytometer.

Neutrophil oxidative burst assay

Neutrophil oxidative burst was measured by flow cytometry as described previously (73). Briefly, isolated neutrophils were stimulated with STING agonist or PMA in the presence of uncharged and nonfluorescent reactive oxygen species indicator DHR123 (Thermo Fisher, 5 μM) for the indicated time. Green fluorescence of oxidized DHR was measured by flow cytometry.

Phagocytosis assay

Phagocytosis of neutrophil was assessed using latex beads coated with fluorescently labeled rabbit IgG (Phagocytosis Assay Kit (IgG FITC), Cayman, cat # 500290) per manufacturer’s instructions.

Neutrophil immunofluorescence

Isolated neutrophils were fixed and permeabilized using BD Cytofix/Cytoperm^™ Fixation/Permeabilization Kit (BD, cat# 554714). Cells were stained with primary antibodies (STING, Proteintech, cat# 19851–1-AP, 1:200; PDI, Abcam, cat# ab2792, 1:200) at 4 °C overnight. After wash with BD Perm/Wash^™ Buffer, cells were stained with Donkey anti-Rabbit IgG Alexa Fluor Plus 488 (#A-32790, ThermoFisher, 1:1000) and Donkey anti-Mouse IgG (H+L) Highly CrossAdsorbed Secondary Antibody, Alexa Fluor^™ 546 (#A10036, ThermoFisher, 1:1000) for 1 h at room temperature. Cells were mounted on a glass slide after nuclear counterstaining with DAPI and were imaged with Zeiss LSM 880 confocal microscope.

T lymphocyte cell death assay

Splenic T cells were isolated from C57BL/6J or Sting1^−/− mice using autoMACS^® Pro Separator (Pan T Cell Isolation Kit II, mouse, 130–095-130) after dissociation and ACK lysis to generate a single cell suspension as described above (see Generation of single cell suspensions of murine immune cells). Cell sorting of double positive (CD4⁺CD8⁺) TCRβ^−/lo/int was performed according to the gating strategy shown in Figure S5 using a FACS Aria sorter (BD Biosciences) at the Flow Cytometry Facility of UT Southwestern Medical Center. Isolated T lymphocytes were maintained in media supplemented with IL-7. The DP TCRβ^−/lo/int thymocytes and splenic T cells were treated with 1μM diABZI for 16h followed by cell death staining with Zombie-Aqua and analysis by flow cytometry.

Anti-IFNAR treatment

50 μg of anti-IFNAR (#I-1188; Leinco Technologies, Inc.) or equivalent amount of isotype control (#P382; Leinco Technologies, Inc.) were injected I.P. in 50μL total physiological saline every other day for 3 weeks. For the ex vivo neutrophil experiments, the cells were cultured in 100μg/mL for the duration of the stimulation.

RNA-seq and bioinformatics analysis

Cell sorting of double positive (CD4⁺CD8⁺) TCRβ^−/lo/int was performed according to the gating strategy shown in Figure S5 using a FACS Aria sorter (BD Biosciences) at the Flow Cytometry Facility of UT Southwestern Medical Center. Total RNA was isolated from sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes using the RNeasy Mini Kit (QIAGEN, 74104). RNA integrity was measured by Bioanalyzer (Agilent) by the UT Southwestern Microarray Core Facility. RNA-seq was performed by BGI Americas. Gene expression levels were quantified as Fragments Per Kilobase Million (FPKM). Differentially expressed genes (2-fold up or 2-fold down in LSL-Sting1^N153S/+Lck-cre (n=4) versus LSL-Sting1^N153S/+ (n=4) sorted double positive (CD4⁺CD8⁺) TCRβ^−/lo/int thymocytes) were analyzed using the core analysis function in Ingenuity Pathway Analysis software (QIAGEN Bioinformatics). The top pathways were grouped and plotted by bar graphs and selected pathways further presented in bar graphs showing individual genes. Data available in data file S1.

In vitro and in vivo DNMT1 inhibitor treatment

HEK293T cells were treated daily for 72h with Decitabine (Sigma, 189826), GSK3685032 (MCE, HY-139664) or vehicle control (DMSO) at indicated concentrations. In vivo drug formulations were prepared as previously described (54). GSK3685032 (45mg/kg) or vehicle (10% captisol adjusted to pH 4.5–5 with 1M acetic acid) was administered subcutaneously, twice daily at a dosing volume of 10 mL/kg. Decitabine (0.4mg/kg) or vehicle (68 mg monobasic potassium phosphate and 11.6 mg sodium hydroxide in 10 mL water) was administered by intraperitoneal injection every other day at a dosing volume of 10 mL/kg.

Tumor inoculation and processing

5 × 10⁵ MC38 were subcutaneously injected to the flank of Rag1^−/− or Sting1^IRES-EGFP mice. 2 × 10⁶ KP67–1 cells were subcutaneously injected in 50% Matrigel (CB40230, Corning^™). All protocols were approved and in compliance with UT Southwestern Institutional Animal Care and Use Committee (IACUC). Tumors were harvested at indicated time point, minced, and filtered through a 70 μm cell strainer followed by digestion in HBSS (with calcium and magnesium) containing 1 mg/mL Collagenase IV (C5138, Sigma) and 0.1 mg/mL DNase I (11284932001, Roche) for 40 minutes at 37°C with rotation. The digestion reaction was quenched with the addition of 10 mM EDTA. Red blood cells were removed by lysis (ACK lysis) and the single cell suspension was washed and filtered before counting and staining for flow cytometry.

Adoptive T lymphocyte transfer

Splenic T cells were isolated from Sting1^IRES-EGFP mice using autoMACS^® Pro Separator (Pan T Cell Isolation Kit II, mouse, 130–095-130) after dissociation and ACK lysis to generate a single cell suspension as described above (see Generation of single cell suspensions of murine immune cells). EGFP-positive T cells were sorted via fluorescence-activated cell sorting (FACS) at the UT Southwestern Medical Center Flow Cytometry Core. 5 × 10⁶ EGFP-positive T cells were intravenously injected into Rag1^−/− mice. Injected mice were analyzed for T lymphocyte reconstitution 5 days post-injection via survival bleed. Successfully reconstituted mice were subcutaneously injected with 5 × 10⁵ MC38 cells into the right flank 10 days post-adoptive transfer.

Human cancer cohort study

A total of 57 patients diagnosed with colorectal cancer were included in this study. Additional data regarding demographics and clinical status are provided in table S1. We obtained cancer and adjacent non-cancerous tissues for flow cytometry analysis. The study was approved by the Ethics Committee of Tianjin Cancer Institute and Hospital, China (protocol number EK20240319). All patients enrolled in the study provided a written informed consent. Cancer and para-cancer tissues were digested by Dnase I (Roche, 10104159001) at a concentration of 400 μg/mL and Liberase (Roche, 5401020001) at a concentration of 250 μg/mL for 1h at 37°C. Then, cells were stained with the following antibodies for flow cytometry analysis: CD45 (BD, 564105), CD3 (BD, 562280), CD4 (BD, 568675), CD8 (BD, 564526), PD-1 (BD, 564323), LAG-3 (BD, 565716), STING (BD, 564965), and Live/Dead antibody (BD, 565388).

Statistical analysis

Statistical tests are noted in figure legends. All data were shown as mean ± S.D. unless otherwise noted and all analyses were performed using Prism 10 software (Graphpad). Statistical significance was identified with *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001. Flow cytometry data were analyzed using FlowJo version 10.8.1. The number and age of individual mice analyzed per genotype or treatment is indicated in each figure legend. Schematic diagrams were created using Biorender.com.

Supplementary Material

The PDF file includes:

Materials and methods

Figs. S1 to S13

Tables S1 and S2

Data and materials availability:

Original RNA-seq data are deposited at NCBI GEO (GSE281969). Data underlying the figures is provided in tabulated form in data file S3 and uncropped immunoblots are provided in data file S2. All requests for resources and reagents should be directed to and will be fulfilled upon request by the Lead Contact, Nan Yan (nan.yan@utsouthwestern.edu). Transgenic mice generated in this work will be deposited at the Jackson Laboratories and will be made available via standard MTA. All other data needed to support the conclusions of the paper are present in the paper or the Supplementary Materials.