2.1 Mice

Wild-type (WT) C57BL/6J (B6) and BALB/c mice were provided by Vital River Laboratory Animal Technology Co., Ltd. Sting1^fl (C57BL/6J-Sting1^{em1(flox)Smoc}) mice, LysM^Cre (B6.129P2-Lyz2^tm1(cre)Ifo/J) mice and Lysm^CreSting1^fl B6 mice were generated in collaboration with Shanghai Model Organisms Center, Inc. All mice in this study were male, 8–10 weeks old, weighing 23–26 g, and housed in a specific pathogen-free condition at Huazhong University of Science and Technology.

2.2 Orthotopic tracheal transplantation

The transplantation model was established following a published protocol from our laboratory.^{6, 19} The details are available in the Supporting Information.

2.3 Drug treatment

All dosages and schedules were based on published protocols. Clodronate liposomes or control liposomes (200 µL; YEASEN, Cat# 40337ES) were injected via the tail vein on days 1, 7, 14, and 21 after tracheal transplantation.²⁰ The cGAS inhibitor RU.521 (5 mg/kg every 3 days; MedChemExpress, Cat# HY-114180) was administered intraperitoneally (i.p.) for 21 days post-transplantation.¹⁸ RO8191 (30 mg/kg/day; MedChemExpress, Cat# HY-W063968), an agonist of interferon-α/β receptor (IFNAR), was administered orally for 14 days post-transplantation.²¹ Recombinant mouse IFN-α2 (rIFN-α2, 400 U/g; R&D Systems, Cat# 10149-IF) and IFN-β (rIFN-β, 500–2000 U/day; Sigma, Cat# IF011) were injected i.p. on days 1, 3, 7, 14, and 21 post-transplant.^{22, 23} The IFNAR antagonist antibody (anti-IFNAR1, 250 µg; Selleck, A2460) was administered i.p. on days 1, 3, 7, 14, and 21.²⁴ CTLA4-Ig (200 µg; BioXCell, Cat# BE0099) was injected i.p. on days 0 and 2.¹⁰ The STING inhibitor C-176 (750 nmol every 2 days; MedChemExpress, Cat# HY-112906) was administered i.p. for 21 days.²⁵

2.4 Histology assessment

Histological evaluation methods followed our previous work.^{19, 26} The details are described in the Supporting Information.

2.5 RNA sequencing, single-cell RNA sequencing and bioinformatics analysis

RNA sequencing sampling was performed following the protocol described in Eur Respir J. 2021;57(3):2000344 (PMID: 33033147).²⁷ Bioinformatics analysis was conducted based on our previously published methods.²⁸ Detailed descriptions can be found in the Supporting Information.

2.6 Statistics

All statistical data are provided in Table S3. p-Values were calculated using GraphPad (version 10.0.3). Data are presented as mean ± standard deviation (SD) or mean ± standard error of the Mean (SEM). Statistical analysis between two independent groups was conducted using an unpaired Student's t-test. For comparisons among multiple groups, one- or two-way ANOVA was employed, followed by a least significant difference t post hoc test. When data did not meet the assumptions for parametric tests, the Mann‒Whitney U-test was utilised. A p-value of <.05 was considered statistically significant.

3.1 cGAS/STING pathway was activated in allograft macrophages

A murine orthotopic tracheal transplantation model was developed to replicate OB pathogenesis. Based previous studies and our team's earlier findings, allograft rejection peaked at day 7 post-transplant, making it the optimal time point for assessing immune infiltration.^{6, 7, 29} Additionally, tracheal allograft fibrosis stabilised by day 28, marking it as the ideal time for evaluating stenosis and fibrosis.²⁹

To explore mechanisms underlying allograft rejection, RNA sequencing was performed on syngrafts and allografts harvested 7 days post-transplant (Figure S1A). Differentially expressed genes analysis identified 1005 upregulated and 1257 downregulated genes in allografts compared to syngrafts (Figure S1B). Consistent with previous findings,^{8, 30, 31} pathways known to promote allograft rejection, such as the Toll-like receptor signalling pathway, were upregulated in allografts (Figure S1C). Notably, significant activation was observed in pathways related to DNA damage, particularly the cytosolic DNA sensing pathway and the RIG-I-like receptor (RLR) signalling pathway, which are closely associated with innate immunity (Figure S1D,E). Key genes in the cytosolic DNA sensing pathway (Cgas, Sting, Tbk1, Irf7, Ifna2 and Ifnb1) and RLR pathway (Ddx58 and Mavs) (Figures S1F and S2B) were markedly upregulated. This coordinated activation pattern implicated a possible involvement of the cGAS/STING signalling in allograft rejection. To further confirm the expression localisation of this pathway, we analysed a single-cell dataset (GSE160760) of tracheal grafts published by Di Campli et al.²⁷ Six major cell types were identified based on cell-specific marker genes, with macrophages emerging as the predominant population in allografts (Figure S1G,H). Subsequent analysis of Sting expression across multiple cell types revealed the highest expression levels in macrophages (Figure S1I). Furthermore, Sting expression was significantly upregulated in allograft macrophages compared to their syngraft counterparts (Figure S1J).

These findings were further supported by immunofluorescence staining, which revealed increased levels of cytosolic DNA (Figure S2A), as well as elevated expression of cGAS and STING (Figure 1A,B) in macrophages infiltrating the allografts. Additionally, Figure S2C‒E shows a pronounced increase in RIG-I and MAVS protein expression in allograft-infiltrating macrophages compared to syngrafts, while MDA5 levels remained statistically unchanged. Furthermore, there was an increased infiltration of p-TBK1⁺ and TBK1⁺ macrophages in the allografts (Figure 1C,D). Elevated levels of IFN-α and IFN-β were also observed in allografts (Figure 1E,F). Correspondingly, mRNA levels of downstream cytokines, including IFN-α, IFN-β, interleukin-1β (IL-1β) and CXCL10, were significantly upregulated (Figure 1G). Taken together, these data demonstrated the activation of the cGAS/STING pathway and the RIG-I/MAVS axis in macrophages infiltrating the allografts, suggesting their potential involvement in allograft rejection.

3.2 Inhibition of the cGAS/STING pathway ameliorated OB

Next, recipient mice were injected with clodronate liposomes to investigate the role of macrophage depletion in the development of OB (Figure S3A). As shown in Figure S3B, mice treated with clodronate liposomes exhibited a significantly reduced luminal occlusion rate in tracheal allografts. Moreover, a marked reduction in fibrosis was observed within the allografts of clodronate liposome-treated mice (Figure S3C). Subsequently, the cGAS inhibitor RU.521 and the STING inhibitor C-176 were administered to elucidate the involvement of cGAS/STING signalling in OB. Pharmacological inhibition of cGAS with RU.521 alleviated tracheal allograft stenosis and fibrosis (Figure 2A–C). Similarly, treatment with C-176 also ameliorated OB (Figure 6J–L). To determine whether the cGAS/STING pathway in macrophages contributed to OB, macrophage-specific Sting1 knockout (Lysm^CreSting1^fl) mice were generated as recipients (Figures 2D and S4A‒C). Compared to Sting1^fl control mice, Lysm^CreSting1^fl mice exhibited significantly reduced tracheal allograft stenosis and fibrosis (Figure 2E,F). Unfortunately, similar to studies on other pathways,³ blocking the cGAS/STING signalling failed to protect the airway's ciliated columnar epithelial cells (0/5), ultimately leading to epithelial shedding in the allografts (Figure 2E). In summary, these findings revealed the pathogenic effect of the cGAS/STING pathway in OB progression.

3.3 Sting1 deficiency in macrophages limited alloreactive T-cell responses

To comprehensively evaluate the influence of the cGAS/STING pathway in macrophages on allograft rejection, immune cell infiltration was assessed by immunofluorescence staining. Sting1 deficiency in macrophages resulted in reduced infiltration of neutrophils and macrophages in the allografts (Figure 3A,B). Lysm^CreSting1^fl mice also exhibited decreased CD4⁺ and CD8⁺ T cells in tracheal allografts (Figure 3C,D). Given that T-cell-mediated adaptive immune responses are key determinants of allograft rejection, we characterised the molecular and functional phenotypes of splenic T cells on day 7 post-transplant (Figures 3E and S5A). Flow cytometry revealed lower proportions of cells expressing the antigen-experienced marker CD44 or the proliferation marker Ki-67 in both CD8⁺ and CD4⁺ T cells in Lysm^CreSting1^fl mice (Figure 3E). Moreover, Sting1 ablation in macrophages reduced the proportion of cells expressing effector molecules (IFN-γ, perforin and granzyme B [GZMB]) in splenic CD8⁺ T cells. Interestingly, while the proportion of cells expressing IFN-γ or IL-17A in CD4⁺ T cells remained unchanged, there was a notable increase in FOXP3-expressing CD4⁺ T cells (Tregs) in Lysm^CreSting1^fl mice. These data suggested that Sting1 deletion in macrophages limited alloantigen-reactive T-cell responses, thereby contributing to the attenuation of transplant rejection.

3.4 Sting ablation inhibited macrophage IFN-I production and antigen presentation

To clarify how the cGAS/STING pathway in macrophages contributed to allograft rejection and OB exacerbation, allografts from Sting1^fl mice and Lysm^CreSting1^fl mice were harvested for RNA sequencing (Figure 4A). GSEA revealed that pathways related to classical cytokines, such as the transforming growth factor-beta (TGF-β) signalling pathway, were inhibited by macrophage-specific Sting knockout (Figure S6A). Pathways involved in the IFN response, such as the DDX58/IFIH1-mediated induction of IFN-α/β pathway, were also downregulated in Lysm^CreSting1^fl mice (Figure 4B). Heatmap in Figure S7A shows that Sting1 ablation led to reduced mRNA expression of RIG-I, MDA5 and MAVS. Immunofluorescence staining further revealed decreased expression levels of RIG-I and MAVS in macrophages within allografts from Lysm^CreSting1^fl mice, whereas MDA5 expression showed no statistically significant difference (Figure S7B‒D). These findings suggested that macrophage-specific Sting1 deletion suppressed activation of the RIG-I/MAVS pathway in allografts. Notably, the petal diagram in Figure 4C shows that Ifna2 and Ifnb1 were both enriched and downregulated in these IFN-related pathways. Experimental results further confirmed that Sting1 ablation in macrophages reduced the infiltration of IFN-α and IFN-β in allografts at both the protein and mRNA levels (Figure 4D‒F). Plasma levels of IFN-α and IFN-β were also lower in Lysm^CreSting1^fl mice (Figure 4G). Moreover, macrophage-specific Sting1 knockout reduced the proportion of cells producing IFN-α and IFN-β within macrophages (Figures 4H,I and S8A). Overall, Sting1 deficiency inhibited IFN-I production in macrophages.

Macrophages not only secrete cytokines that directly damage allografts, but also act as antigen-presenting cells to promote alloreactive T-cell responses.^{32, 33} GSEA revealed that the MHC class II antigen presentation pathway, reflecting antigen-presenting function, was downregulated in allografts from Lysm^CreSting1^fl mice (Figure S6B). As shown in Figure S6C, mRNA levels of the pro-inflammatory marker NOS2 and co-stimulatory molecules CD80 and CD86 were also reduced. Flow cytometry further demonstrated that Sting1 knockout decreased the expression of NOS2 (Figure S6D), MHC class II (Figure S6E) and co-stimulatory molecules (CD80 and CD86) in macrophages (Figure S6F,G), indicating a diminished antigen-presenting capacity. Collectively, these findings suggested that Sting1 deletion reduced IFN-I production and impaired the antigen-presenting function of macrophages.

3.5 rIFN-α2 treatment aggravated OB in Lysm^CreSting1^fl mice

To investigate the role of IFN-I in OB, recipient mice were treated with an antagonist antibody targeting IFNAR1 (anti-IFNAR1) (Figure S9A). The results demonstrated that anti-IFNAR1 treatment significantly alleviated tracheal allograft stenosis and fibrosis, implicating the involvement of IFN-I in allograft rejection (Figure S9B,C). Subsequently, to further assess whether the cGAS/STING pathway exacerbated allograft rejection via the IFN-I axis, Lysm^CreSting1^fl mice were administered RO8191 (30 mg/kg/day for 14 days), an IFNAR agonist.²¹ As expected, RO8191 treatment aggravated stenosis and fibrosis of the transplanted trachea, with the airway epithelium remaining in a state of shedding (Figure 5A,B). Although no difference in neutrophil infiltration was observed (Figure 5C), the infiltration of macrophages, CD4⁺ T cells, and CD8⁺ T cells was increased in the RO8191-treated Lysm^CreSting1^fl mice (Figure 5D‒F). Additionally, RO8191 elevated the proportion of Ki-67⁺ cells within both CD4⁺ and CD8⁺ T-cell populations (Figure 5G,H), as well as the percentage of CD8⁺ T cells expressing IFN-γ (Figure 5I). Conversely, RO8191 decreased the proportion of Tregs within the CD4⁺ T-cell population (Figure 5J). These findings suggest that IFN-I contributes to allograft rejection as a downstream effector of the cGAS/STING pathway; however, it remains unclear whether IFN-α or IFN-β is responsible.

Therefore, we supplemented the Lysm^CreSting1^fl mice with either rIFN-α2 (400 U/g/day) or rIFN-β (500, 1000 or 2000 U/day) on days 1, 3, 7, 14 and 21 post-transplant.^{22, 23} Intriguingly, rIFN-α2 exacerbated OB, whereas rIFN-β had no observable effect (Figure 5A,B). Correspondingly, rIFN-α2 increased immune cell infiltration (neutrophils, macrophages and CD4⁺/CD8⁺ T cells) within the allografts (Figure 5C‒F), elevated the percentage of Ki-67⁺ cells among both CD4⁺ and CD8⁺ T cells (Figure 5G,I), increased the proportion of IFN-γ⁺ cells among splenic CD8⁺ T cells (Figure 5H), and reduced the proportion of Tregs within splenic CD4⁺ T cells (Figure 5J). Overall, these data indicated that IFN-α2, rather than IFN-β, was a key downstream effector through which the cGAS/STING pathway promoted allograft rejection.

3.6 STING inhibition enhanced the immunosuppressive effect of CTLA4-Ig

CTLA4-Ig has been approved by the Food and Drug Administration for immunosuppressive therapy in kidney transplantation.³⁴ Previous studies have shown that CTLA4-Ig could alleviate tracheal stenosis and fibrosis.³ However, it shares a crucial limitation with macrophage-specific Sting1 knockout: neither approach can protect the airway epithelium from damage. Considering combination therapy is an effective strategy to enhance treatment efficacy, and that FTY720 combined with CTLA4-Ig have been reported to protect respiratory epithelium of tracheal allografts,³⁵ we sought to determine whether combining STING inhibition with CTLA4-Ig treatment could preserve the ciliated columnar epithelium. Thus, Sting1^fl and Lysm^CreSting1^fl mice were injected with either CTLA4-Ig or vehicle (200 µg/day) on days 0 and 2 (Figure 6A). Consistent with previous findings,³⁵ CTLA4-Ig alone reduced tracheal allograft stenosis and fibrosis but failed to preserve the ciliated columnar epithelium (Figure 6B,C). Remarkably, CTLA4-Ig treatment in Lysm^CreSting1^fl mice not only mitigated airway stenosis and fibrosis but also preserved the ciliated columnar epithelium (4/5)—an effect not observed with either treatment alone (Figure 6B,C). Moreover, Sting1 ablation enhanced the inhibitory effect of CTLA4-Ig on alloreactive T-cell responses. While there was no significant difference in memory CD8⁺ T cells between the combination group and CTLA4-Ig alone (Figure 6D), combination therapy significantly reduced the proportion of memory CD4⁺ T cells (Figure 6E), as well as the Ki-67⁺ population in both CD4⁺ and CD8⁺ T cells (Figure 6F,G). Furthermore, it led to a marked increase in Tregs among CD4⁺ T cells (Figure 6I), without significantly affecting effector CD8⁺ T cells (Figure 6H). Additionally, considering that pharmacological intervention is more clinically translatable than genetic knockout, we treated recipient mice with a combination of CTLA4-Ig and the STING inhibitor C-176 (Figure 6J), and found that this combination effectively protected the airway epithelium (4/5) (Figure 6K,L). Collectively, these results demonstrated that STING inhibition potentiated the therapeutic efficacy of CTLA4-Ig in OB.