A non-interferon dependent role of STING signaling in pulmonary hypertension

Abstract

Background:

Patients with constitutive activation of DNA sensing pathway through stimulator of interferon genes (STING), such as those with STING-Associated Vasculopathy with onset in Infancy (SAVI), develop pulmonary hypertension (PH). However, the role of STING-signaling in general PH patients is heretofore undescribed. Here, we seek to investigate the role of STING in PH development.

Methods:

STING expression in patient lung samples was examined. PH was induced in global STING deficient mice and global type I Inteferon receptor 1 deficient mice using bleomycin or chronic hypoxia exposure. PH development was evaluated by right ventricular systolic pressure and Fulton index, with additional histological and flow cytometric analysis. VEGF expression on murine immune cells were quantified and evaluated with multiplex and flow cytometry. Human myeloid-derived cells were differentiated from PBMC and treated with either STING agonist or STING antagonist for evaluation of VEGF secretion.

Results:

Global STING deficiency protects mice from PH development, and STING-associated PH appears independent of type I Interferon (IFN) signaling. Furthermore, a role for STING-VEGF signaling pathway in PH development was demonstrated, with altered VEGF secretion in murine pulmonary infiltrated myeloid cells in a STING-dependent manner. In addition, pharmacological manipulation of STING in human myeloid-derived cells supports in vivo findings. Finally, a potential role of STING-VEGF mediated apoptosis in disease development and progression was illustrated, providing a roadmap toward potential therapeutic applications.

Conclusions:

Overall, these data provide concrete evidence of STING involvement in PH, establishing biological plausibility for STING-related therapies in PH treatment.

Graphical Abstract

Introduction

Pulmonary hypertension (PH) is a devastating disease, affecting approximately 15-50 million people worldwide¹. World Health Organization (WHO) Group 3 PH – PH secondary to chronic lung disease and/or hypoxia – is the second leading cause of PH in developed countries, with rising incidence globally. Unfortunately, patients with Group 3 PH have the highest mortality rate among all cohorts of pulmonary vascular disease^2-4. Group 3 PH patients also utilize the health care system up to four times more than patients without complicating PH, contributing to significant health care cost burden⁵. Moreover, there is a significant therapeutic gap for afflicted patients, with only one PH-specific drug indicated for use in these patients⁶ and no known disease-modifying therapies. More research is therefore needed to understand disease pathogenesis.

Previous studies have demonstrated a critical role for immune cells in PH progression^7,8. Enhanced discernment of the various immune cell subtypes in PH represents a still novel area of research into disease-modifying therapy. In patients with genetic predisposition to constitutive interferon production and corresponding immune cell activation, a family of illnesses referred to as “interferonopathies”, there is a proclivity to PH development^9-13. Such disease states provide a novel link between systemic inflammation and pulmonary vascular remodeling. The canonical interferonopathy that predisposes to development of pulmonary vasculopathy, termed STING-associated Vasculopathy onset in Infancy (SAVI), involves activating mutations in the upstream regulator of type I interferon (IFN) signaling, Stimulator of Interferon Genes (STING)^12,13. STING is a cellular gatekeeper against various cytosolic DNA-containing pathogens¹⁴, detecting and responding to – evolutionarily – viral DNA. Upon activation of STING, a series of downstream transcription factors, including Interferon regulator factor 3 (IRF3), are subsequently activated. Phosphorylated IRF3 enters the nucleus and promotes transcription of type I IFN genes and interferon-response genes^14,15, resulting in immune cell activation. Notably, a recent study of STING pathway in COVID-19 revealed a pathogenic role for STING related to interferon signaling, contributing to tissue damage and disease progression¹⁶. Likewise, STING promotes inflammation in a variety of disparate disease models^17-20, playing an important role in anticipatory immune homeostasis²¹, contributing in particular to parenchymal lung disease and systemic vasculitis²². Similarly, STING activation in smooth muscle cells was recently reported to result in cell death, contributing to aortic aneurysm²³, while STING agonist has been shown to promote apoptosis of tumor endothelial cells in different cancer models²⁴. Both studies highlight the relevance of STING in regulation of cells in the vasculature. To date, there is no data on STING involvement in pulmonary vascular disease, although the role of Type I IFN in PH has been addressed by different groups, albeit with contradicting results^25,26. PH has also been associated with DNA damage due to drugs and toxins²⁷, with an associated increase in level of circulating DNA due to predisposing disease-associated mutations^28-30, making it plausible that downstream detection of cell-free DNA, through STING, may contribute to disease.

Herein, we report on the relevance of STING in patients with Group 3 PH. We then describe that global STING deficient (STING^−/−) mice are protected against PH in two complementary disease models, with phenotype developing independent of changes in interferon signaling. Finally, we show an evolving role for STING-VEGF signaling in protection against PH, using PH mouse models and human primary cells. Together, our data support a role for STING in PH development, laying the groundwork for STING-based therapies that can be utilized in PH patients³¹.

Methods

Availability of Data

The data, analytical methods, and study materials that support the findings of this study are available from the corresponding author upon resonable request.

Please see the Major Resources Table in the Supplemental Material for details of material used.

Animal Studies

Global STING deficient (STING^−/− or MPYS^−/−) and global IFNAR1 deficient (IFNAR1^−/−) mice were purchased from Jackson Laboratory and then crossed with in-house C75BL/6J background for maintenance. All transgenic mice generated in this study were C57BL/6J background. In-house C57BL/6J mice were used as controls for global knock-out mice. Sex-matched 8-10 weeks mice were used (n=2-4/mice/sex/experimental group) after confirmation of appropriate genotype. No sex-specific effect was observed across experiments; therefore, results reported were inclusive of sex. Animal experiments and maintenance were approved by the Institutional Animal Care and Use Committee of University of Florida (IACUC; Protocol 08702). Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny, Browne, Cuthill, Emerson, & Altman, 2010) and with recommendations made by the Arteriosclerosis, Thrombosis, and Vascular Biology Journal. Of note, experimental results for STING^−/− mice were pooled from 3 distinct experiments (n=5-8/experiment), in which the lung tissues were collected for different purposes, including flow cytometry, scRNAseq, histology, and cytokine analysis, all detailed below.

Models of pulmonary hypertension

Pulmonary hypertension (PH) was induced in mice with bleomycin (Millipore Sigma #9041934) injection or chronic hypoxia exposure.

Bleomycin:

Mice received intraperitoneal (i.p.) injections of bleomycin at 0.018U/g twice a week for 4 weeks. The weight of animals was monitored throughout injection period. A 20% body weight loss resulted in temporary termination of bleomycin treatment. Injection was resumed when the animal regained at least 10% of loss weight. Euthanasia and data collection were performed 5 days after final injection, on day 33 of bleomycin protocol.

Chronic hypoxia:

Mice undergoing chronic hypoxia exposure were placed in a normobaric ventilated chamber, in which the level of O₂ is controlled through the flow of N₂ (O₂ monitor/controller and chamber, Coy Laboratory). O₂ and CO₂ concentration was monitored continuously, such that their concentrations remained at 10% and 0.1%, respectively. Exposure to normal air was limited to only during water and food changes. All mice were sacrificed for analysis after 4 weeks (28 days) of exposure.

Finally, 10-week-old C75BL/6J and STING^−/− mice were injected subcutaneously with SU5416 (Sigma Aldrich #S8442) suspended in a mixture of 0.5% carboxymethylcellulose sodium (Sigma Aldrich C4888), 0.9% sodium chloride (Sigma Aldrich #7647145), 0.4% polysorbate 80 (Sigma Aldrich #9005656), and 0.9% benzyl alcohol (Sigma Aldrich #100516) in deionized water. Mice received injections once a week at 20mg/kg body weight per dose while being exposed to hypoxia for 4 weeks. Control mice were injected with the same volume of vehicle. After 4 weeks of exposure to hypoxia, mice were returned to normoxia for 1 week before euthanasia.

Primer sequences in mice

See Table 1 for primer sequences in mice.

Table 1:

Primer sequences in mice

Gene	Primer	Sequence (5’ → 3’)
TMEM173 (STING)	Forward	TTTTCATCTGCCTTCCAGGT
	Reverse	GCGCACACACACTAAAAACTG
IFNAR1	Forward	ACTCAGGTTCGCTCCATCAG
	Reverse	CTTTTAACCACTTCGCCTCGT

Pulmonary hemodynamic assessment

Mice were put under deep anesthesia with intraperitoneal (i.p.) injection of 25% Avertin (2,2,2-tribromoethanol, Fisher Scientific^™ #AC421432500) in PBS at 16mg/kg dose. A 1.4-French-pressure-volume microtip catheter (Millar Instruments, SPR-839) was inserted through a right internal jugular incision and threaded down into the right ventricle. The catheter was connected to a signal processor (PowerLab and ADInstruments), and the invasive right ventricular systolic pressure (RVSP; mmHg) was recorded digitally and displayed with Chart5. After stable measurements of a minimum of 5 minutes, animals were euthanized with the removal of hearts and lungs for subsequent analysis. The right ventricular was separated from the heart after removal of the atria, to which the weights of both right ventricle (RV) and left ventricle plus septum (LV+S) were obtained. Right ventricular hypertrophy was then calculated using RV/LV+S (%) ratio (Fulton Index). RVSP measurement success rate was 90% across experiments, as typical for the complexity of the measure. Tissues from non-RVSP mice were all processed for flow cytometry and histology, detailed below.

Flow cytometry

Mouse left lung was cut into small pieces and digested for 1h with 10mL of DNAse I (10mg/mL, Sigma Aldrich #10104159001) and type I collagenase (100mg/mL, Sigma Aldrich #11088882001) in PBS 10% FBS (Thermo Fisher #A4766801) at 37°C at 200rpm on a shaker. The digested homogenate was filtered through a 70μM cell strainer and red blood cells were lysed using ammonium chloride lysis buffer (KD Medical #50-1019080). Single-cell suspensions obtained after using a 70μM cell strainer were stained with fluorochrome-conjugated surface antibodies for 30 minutes on ice, Fixable Viability Dye (eBioscience #65-0865-14) for 30 minutes on ice, fixed, permeabilized, and then incubated with intracellular markers for 30 minutes on ice. Data was acquired using BD FACSymphony A3 Cytometer (BD Biosciences) with 5 lasers and analyzed with FlowJo version 10 software.

FACS antibodies

A comprehensive list of FACS antibodies used is presented in Table 2. Working concentration is shown in Supplemental Material.

Table 2:

FACS antibodies for flow cytometry used in human and mice

Target	Supplier
Mouse
CD11b ef450	Affymetrix
CD11b BV786	Biolegend
CD11c PE-Cy7	Affymetrix
Ly6C BV711	BioLegend
Ly6G BV605	BioLegend
PD-L1BV785	BioLegend
FVD APC-Cy7	eBioscience
αSMA AF405	Novus Biology
CD31 BB700	Thermofisher
VEGFa FITC	Novus Biology
CD3 PE-Cy7	Biolegend
CD4 BV605	Biolegend
CD8 BV785	Biolegend
CD62L APC	Biolegend
CD25 FITC	Biolegend
PD-1 PE-CF594	BD Biosciences
Foxp3 PE	Biolegend
Human
CD33 APC	BD Biosciences
HLA-DR BV785	Biolegend
CD11b Pacific Blue	BD Biosciences
CD15 PE-Cy7	Biolegend
CD14 BV605	BD Biosciences
FVD APC-Cy7	eBioscience
STING PE-CF594	BD Biosciences
VEGFa AF488	Novus Biology
CD4 PE	BD Biosciences
CD8 FITC	BD Biosciences

cDNA library construction and single-cell RNA-seq

Mouse whole lungs were perfused with 10mL PBS to remove blood cells. Lung tissues were then cut into small pieces and incubated in 3mL RPMI 1640 with 10% FBS (Gibco #A4766801), 5% Liberase (Sigma Aldrich #05401127001), and 1000U/mL DNAse I (Sigma Aldrich #9003989) for 1h at 100rpm and 37°C. Tissues were triturated with 3mL syringes and 18G needles until complete dissociation. Cells are filtered through 70μm cell strainer, washed 3 times with DPBS 2% FBS and 1M EDTA (KD Medical #RGF-3130) to remove debris. The remaining red blood cells were lysed with ammonium chloride lysis buffer (KD Medical #50-1019080). Single cells were captured in the 10X Genomic Chromium Single Cell 3’ Solution, and RNAseq libraries were prepared following manufacturer’s protocol. The libraries were subjected to high-throughput sequencing on an Illumina NovaSeq 6000 platform, targeting 6000-8000 cells per sample with a sequencing depth of 20 million reads of 150-bp paired-end reads.

Process and quality control of the single-cell RNA-seq data

The raw sequencing reads were aligned with mouse genome mm10 provided on CellRanger website by 10X Genomics. The mapped reads were then used for unique molecular identifier (UMI) counting, following the standard CellRanger pipeline for quality control as recommended by the manufacturer (10X Genomics). In short, cells with UMI counts lower than 500 or a feature count less than 200 were excluded. In addition, cells with greater than 30% RNA content made up of the most common genes were excluded, as they accounted for empty droplets with free-floating RNA. Cells with greater than 30% RNA content mapped to the mitochondrial genome were also eliminated as they indicated poor quality. Lastly, cells with abnormal high counts were discarded. Subsequently, the filtered single cells were imported into R package “Seurat” (version 4.0) for clustering of data and calculating differential gene expression, following standard pipeline per manufacture instruction (10x Genomics). FGSEA package with gene sets from Molecular Signature Database, enrichPlot, and clusterProfile was used for generating differential expression of type I IFN and VEGF-related genes. Markers for different populations of cells (CD11b⁺, CD11c⁺, stromal, endothelial, and myeloid) were obtained from previously described work^32-36.

scRNAseq data can be assessed on GEO database. Access link provided in Major Resources Table. R codes used for scRNA-seq analysis can be found on GitHub depository “A non-inteferon dependent role of STING in pulmonary hypertension”. Access link provided in Major Resources Table

Mouse histological staining

Mouse lung right lower lobe, upon harvest, was fixed in formalin overnight. Fixed tissues were paraffin-embedded, cut with a Leica RM2235 Microtome, and stained for Masson Trichrome and α-smooth muscle actin to assess inflammation and identify muscularized pulmonary vessels³⁷.

Masson Trichrome Staining (MTC) and semi-quantitative inflammation scoring: lung inflammation was evaluated on trichrome-stained lung sections using a 0 to 4 scale, with a score of 0, normal lung architecture; 1, increased thickness of some (≤50%) of interalveolar septa; 2, thickening of >50% of interalveolar septa without formation of fibrotic foci; 3, thickening of the interalveolar septa with formation of isolated fibrotic foci; and 4, formation of multiple fibrotic foci with total or subtotal distortion of parenchymal architecture. Blinded evaluation was performed on 10 randomized sequential, nonoverlapping fields (magnification 10x) of lung parenchyma from each specimen. The mean score for the 10 fields represented the score for each specimen.

α-Smooth muscle actin (αSMA) staining and muscularized vessel count: formalin-fixed lung section was stained for rabbit polyclonal αSMA (Abcam #ab5694; diluted 1:1000 in in PBS 10% BSA, not reused; blocking reagent goat serum (Abcam #ab7481), followed by goat anti-rabbit IgG HPR (Abcam #ab205718). The protein of interest is visualized in brown, using DAB Substrate kit (Abcam #ab64238).

Stained lung specimens were then randomized and blindly assessed for pulmonary vessel counts on 10 sequential, nonoverlapping fields (magnification 10x) of lung parenchyma from each specimen. Muscularized pulmonary vessel was visualized as either partially or completely circumferential. Vessel was considered small if ≤50μM, medium if 50-150μM, or large if >150μM. Numbers of complete or partial, small, medium, and large, were quantified in the report of muscularized vessels.

Image processing and acquisition

Images of representative tissues from histological staining were taken with Keyence BZ-X microscope at 10x, 20x, 40x, and 100x magnification. Image processing was performed using BZ-X-Analyzer software (Keyence).

Patient samples

Paraffin-embedded lung tissues of interstitial lung disease (ILD) and chronic obstructive lung disease (COPD) patients were obtained from the Lung Tissue Research Consortium (LTRC), approved by the Institutional Review Board of the University of Florida (IRB201501114). Subjects were individuals undergoing lung surgery for nodules or masses, having a biopsy for diagnosis of possible ILD, or undergoing therapeutic surgery for established lung disease (lung volume reduction surgery or lung transplant). ILD and COPD were diagnosed by pulmonary function tests and pathological examinations. Exclusion criteria included age less than 21 years and diagnosis of cystic fibrosis and pulmonary hypertension. Details about LTRC can be found at https://biolincc.nhlbi.nih.gov/studies/ltrc/.

In addition, paraffin-embedded lung tissues from the lower lobe of PH patients were obtained from the University of Florida Shands Hospital as part of a local registry and biorepository³⁸, which captures all patients referred for PH evaluation. Subjects were included if they were clinically phenotyped as pulmonary fibrosis (IPF defined as ATS/ERS 2011 guidelines³⁹) with no PH or pulmonary fibrosis with PH, assessed by standard right heart catheterization hemodynamic parameters and echocardiographic criteria. Exclusion criteria included history of chronic thromboembolic pulmonary hypertension, left heart failure, obstructive sleep apnea, and history of congenital heart disease.

Tissues were cut with a Leica RM2235 Microtome and stained for rabbit polyclonal STING/TMEM173 antibody (Novus Biologicals #NBP2-2468310; diluted 1:100 in PBS 10% BSA, not reused), followed by goat anti-rabbit IgG H&L (Abcam #ab6721). The protein of interest is visualized in brown, using DAB Substrate kit (Abcam #ab64238).

Cytokine/Chemokine analysis

32 analytes (cytokines, chemokines, growth factors) in mouse whole lung protein were quantified using EMD Millipore’s MILLIPLEX Magnetic Bead Panel (Millipore Sigma #MCYTOMAG-70K), including Eotaxin, G-CSF, GM-CSF, IFNγ, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12 (p40), IL-12 (p70), IL-13, IL-15, IL-17, IP-10, KC, LIF, LIX, MCP-1, M-CSF, MIG, MIP-1α, MIP-1β, MIP-2, RANTES, TNFα, and VEGF.

Type I IFN ProQuantum assay

Serum collected from mice was analyzed for type I IFNα and IFNβ using Mouse IFN ProQuantum Immunoassay Kit (Invitrogen #A46736 and #A47435), following manufacturer’s protocols.

Human PBMC-differentiated myeloid-derived cells

Samples of 10mL of blood were obtained from healthy individuals participated in the clinical study “Role of hypoxia-induced factor in pulmonary vascular remodeling caused by parenchymal lung disease” (IRB201400744) at University of Florida Shands Hospital. Peripheral blood mononuclear cells (PBMCs) were isolated with Lymphoprep (StemCell #07851), following the manufacturer's protocol.

Isolated PBMCs were diluted to 5x10⁵ cell/mL and cultured in RPMI 1640 (Thermofisher #12633-012) 10% FBS (Gibco #A4766801), 1% Glutamine (Thermofisher #35050-061), 1% antibiotic antimycotic solution (Sigma Aldrich #A5955), and 10ng/mL hGM-CSF (Preprotech #300-03). Media was changed on day 4. On day 7, cells were collected for downstream assay and flow cytometry using the following markers: CD33/CD11b/HLA-DR/CD14/CD15/FVD.

Full list of FACS antibodies used for validation is listed in Table 2.

Human T cell suppression assay

PBMCs were obtained as described above. T cells were collected using EasySep Human T Cell Isolation Kit (StemCell #17951), following manufacturer instructions. Isolated T cells were resuspended in 1mL PBS and labeled with 1uL Cell Trace Violet (Thermofisher #C34557) for 15 minutes at 37°C in the dark. Labeled T cells were resuspended in culture media (RPMI 1640 (Thermofisher #12633-012) 10% FBS (Gibco #A4766801), 1% Glutamine (Thermofisher #35050-061), and 1% antibiotic antimycotic solution (Sigma #A5955)) at 5x10⁵ cells/mL.

These T cells were then seeded in 96 well-plate, stimulated with Dynabeads Human T activator CD3/CD28 beads (Gibco #11161D), and cocultured with PBMC-differentiated myeloid-derived cells at 1:1, 1:2, and 1:4 (T:MDSC) ratio at 37°C with 5% CO₂. On day 5, cells were collected for flow cytometry using the following markers: CD4/CD8/FVD. Percent suppression is calculated as {[1-(proliferation with MDSC/proliferation without MDSC]x100}.

Full list of FACS antibodies used is listed in Table 2.

Coculture of human PBMC-differentiated myeloid-derived cells and human pulmonary arterial endothelial cells (PAECs).

Human PBMC-differentiated myeloid-derived cells were obtained as described above and resuspended to 1x10⁶ cells/mL in RPMI 1640 (Thermofisher #12633-012) 10% FBS (Gibco #A4766801), 1% Glutamine (Thermofisher #35050-061), and 1% antibiotic antimycotic solution (Sigma #A5955). Cells were cultured with either vehicle (DMSO), 150ng/mL diABZI (STING agonist, Invivogen #tlrl-diabzi), or 4ug/mL H151 (STING antagonist, Invivogen #inh-h151) for 12 hours.

Human PAECs were purchased from AATC (#PCS-100-022) and cultured per manufacturer instructions for 24h prior to coculture assay in complete PAEC culture media (Vascular Cell Basal Medium (AATC #PCS-100-040) + EC Growth Kit BBE (AATC #PCS-100-030)).

Human PBMC-differentiated myeloid-derived cells, after culture with STING agonist and antagonist, were washed and cocultured with PAEC at 1:1 ratio in complete PAEC culture media for 12h. Cells were then collected for flow cytometry using the following markers: CD33/CD11b/HLA-DR/CD14/CD15/FVD/VEGF.

Full list of FACS antibodies used is listed in Table 2.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 9.0 software. Quantitative data is presented as mean±SEM. Each data point on bar graphs represents an individual mouse. Box violin shows mean, mode, and interquartile range of data set. Data from each graph was collected across 1-3 individual experiment(s). Power analysis revealed 80 to >90% power to detect a change in the mean with n=4. Therefore, all data sets were performed with n=4-13. Data was pooled for biological replications performed in individual experiments for statistical analysis. Mean difference across groups was performed using unpaired two-tailed Welch t-test (unequal variance assumption). A p-value of <0.05 was considered statistically significant.

No multiplicity adjustments or post hoc test was performed.

Results

Patients with Group 3 PH may have differential STING expression within the lung.

To investigate STING relevance to Group 3 PH, we examined STING expression grossly in patients with PH and associated chronic parenchymal pulmonary disease of idiopathic pulmonary fibrosis (IPF). To this end, lung sections from IPF patients with and without PH in a previously reported cohort⁴⁰ were examined. A qualitative presence of STING expression in the lung sections of patients with IPF and IPF+PH compared to healthy individuals was found (Fig. 1 and Fig. 1S). As a disease control, we also examined sections from patients with chronic obstructive pulmonary disease (COPD) with and without PH, which also demonstrated differences in pulmonary STING expression (Fig. S2). From these qualitative images, we assumed the potential relevance of STING to PH development, emboldening further mechanistic preclinical investigation into these hypothesis-generating data.

Figure 1: Patients with idiopathic pulmonary fibrosis (IPF), with or without PH, may have differential expression in STING expression within the lung.

Representative images of immunohistochemical (IHC) staining for STING (brown, arrowheads indicates pulmonary vessels) in formalin-fixed lung sections of healthy individuals and patients with IPF-PH at 20x magnification. Scale bar = 50μm. Each image represents an individual donor (n=5/group)

Global STING deficient (STING^−/−) mice are protected against PH.

Given that STING appears to be expressed in the lungs of patients with chronic parenchymal lung disease and the aforementioned pulmonary vascular manifestations of constitutive STING activation, we hypothesized that deletion of STING would protect against PH development. Conceptually, this hypothesis was supported by evidence of STING signaling in pathologic systemic vascular remodeling²³. To that end, we induced PH in sex-matched 8–10-week-old global STING deficient (STING^−/−) mice using either bleomycin (Blm) or chronic hypoxia (Hx) as standard models for PH, as previously described^41,42. STING's deletion was confirmed with genotyping and immunoblot analysis of whole lung protein (Fig. S4A). Importantly, no spontaneous cardiopulmonary phenotype was found in STING^−/− mice prior to PH induction with either bleomycin or chronic hypoxia. In stark contrast, bleomycin- and chronic hypoxia-induced PH control mice were found to have elevated invasive RVSP, while STING^−/− mice displayed significantly lower pressures of less than 25mmHg (Fig. 2, A and C) (control Blm: 27.4mmHg; STING^−/− Blm: 23.9mmHg; control Hx: 30.7mmHg; STING^−/− Hx: 24.5mmHg). Fulton Index (the ratio of mass of RV over mass of left ventricular and septum (LV+S)) was decreased in STING^−/− mice exposed to chronic hypoxia (Fig. 2B). No such change was observed in STING^−/− mice treated with bleomycin, an expected finding given previously described lack of right heart remodeling in the bleomycin-induced PH model^43,44. Given that bleomycin induces PH concurrent with development of fibrosis, we performed Masson Trichrome (MTC) staining on control and bleomycin treated groups to assess the degree of semi-quantitative fibrotic changes. Correlating with physiological data, inflammation/fibrosis assessment through staining showed an anticipated decrease in mature collagen deposition and inflammatory changes in the lung of STING^−/− mice treated with bleomycin compared to control (Fig. 2, D and E). In addition, immunohistochemical (IHC) staining for α-smooth muscle actin (αSMA) revealed a decrease in total muscularized pulmonary vessels, primarily in the small vessels, in both bleomycin-treated and chronic hypoxic STING^−/− mice (Fig. 2, F and G). As mentioned previously, no sex-specific effect was observed in experimental results.

Figure 2: Global STING deficient (STING^−/−) mice are protected against elevation of pulmonary pressure secondary to bleomycin or chronic hypoxia.

(A) RVSP measurement of C57BL/6J (control) and global STING deficient (STING^−/−) mice subjected to normoxia (Nx), bleomycin (Blm) injection or chronic hypoxia (Hx) exposure. (B) Fulton Index (RV/LV+S) of mice from dedicated experimental groups. (C) RVSP curve representation of control and STING^−/− mice treated with bleomycin or exposed to chronic hypoxia. (D) Quantification of inflammation scoring from MTC-stained lung sections of mice undergoing bleomycin treatment or chronic hypoxia exposure. (E) Representative images of MTC-stained formalin-fixed lung sections of control and STING^−/− mice subjected to normoxia, bleomycin, or chronic hypoxia. Scale bar = 200μm at 10x magnification. (F) Representative images of IHC staining for αSMA (brown, arrowheads) in formalin-fixed lung sections of mice from designated experimental groups. Scale bar = 200μm at 10x magnification. (G) Quantification of muscularized pulmonary vessels (small, medium, large) from αSMA IHC staining of lung sections of mice from all experimental groups. Each dot represents an individual mouse (n=8-13/group). Data was collected across 1-3 experiment(s). Column represents mean ± SEM. Box violin shows mean, mode, and interquartile range of data set. Statistical significance was determined using unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant. P values from (H) were calculated from total vessels (small+medium+large) of mice from experimental groups.

Pulmonary vascular disease pathogenesis invokes an important role for myeloid cells, specifically in prior reports, the immunoregulatory role of myeloid-derived suppressor cells (MDSCs)^37,40-42. While the context and classification of MDSCs continue to evolve, they are commonly described as morphologically immature myeloid cells that are similar to either monocytes (Mo-MDSC, CD11b⁺Ly6C^hiLy6G⁻) or neutrophils (PMN-MDSC, CD11b⁺Ly6C^lo Ly6G⁺), capable of suppressing adaptive immune cell responses, primarily that of T cells⁴⁵. As STING is an important regulator of innate and adaptive immunity, we hypothesized significant changes in the make-up of pulmonary infiltrated myeloid cells in STING^−/− mice, correlating with RVSP changes. Indeed, under chronic hypoxia exposure, there was a decrease in the number of pulmonary infiltrated CD11b⁺ cells in STING^−/− mice compared to controls (Fig. 3, A and D). We observed a similar trend of decreased pulmonary infiltrated subpopulations of myeloid cells in PH-induced STING^−/− mice, including the CD11b⁺Ly6C^hiLy6G⁻ and CD11b⁺Ly6C^loLy6G⁺ cell sub-populations (Fig. 3, B, C and E). Such difference was not as profound in STING^−/− mice subjected to bleomycin treatment, potentially related to the complexity of the immunologic response to systemic bleomycin (Fig. 3, A - E). The gating strategy for myeloid cells and their subpopulations is shown in Fig. S3.

Figure 3: STING^−/− mice exposed to chronic hypoxia display a decrease in pulmonary infiltrated inflammatory cells.

(A – C) Flow cytometric quantification of pulmonary infiltrated (A) CD11b⁺, (B) CD11b⁺Ly6C^hiLy6G⁻, and (C) CD11b⁺Ly6C^loLy6G⁺ cells of control and STING^−/− mice from different experimental groups. (D and E) Flow plot representation of (D) CD11b⁺ and (E) CD11b⁺Ly6C^loLy6G⁺ cells from designated experiment groups. Each dot represents an individual mouse (n=8-13/group). Data was collected across 1-3 independent experiment(s). Column represents mean ± SEM. Statistical significance was determined by unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant.

MDSC affects T cell activation through a variety of pathways, including induction of an exhaustive phenotype via PD-1 upregulation on effector T cells, well documented in many cancer models⁴⁶, as well as in the pulmonary fibrosis-associated PH mouse model³⁷. Thus, we hypothesized a decrease in PD-1/PD-L1 axis expression in immune cells, correlating with physiological change in RVSP and Fulton Index across experimental groups. However, we found no significant difference in PD-L1 expression on CD11b⁺ and CD11b⁺Ly6C^hiLy6G⁻ cells between control and STING^−/− mice exposed to chronic hypoxia (Fig. 4, A and B). We did note a decrease in PD-L1 expression on CD11b⁺Ly6C^loLy6G⁺ cells in chronic hypoxic STING^−/− mice (Fig. 4, C and D), corresponding to our previous preclinical and clinical reports⁴⁰. In addition, minimal differences in PD-L1 expression were observed in cells of bleomycin treated control and STING^−/− mice (Fig. 4, A - C), consistent with our data on bleomycin treated mice (Fig. 3, A - E).

Figure 4: Evidence for immune cell exhaustion is diminished in STING^−/− mice exposed to chronic hypoxia.

(A – C) Flow cytometric quantification of PD-L1 expression on pulmonary infiltrated (A) CD11b⁺, (B) CD11b⁺Ly6C^hiLy6G⁻, and (C) CD11b⁺Ly6C^loLy6G⁺ cells of control and STING^−/− mice from all experimental groups. (D) Flow plot representation of PD-L1 on CD11b⁺Ly6C^loLy6G⁺ cells from control (blue) and STING^−/− (red) mice. (E and F) Flow cytometric quantification of (E) CD4⁺ T cells and (F) mean fluorescence intensity (MFI) of PD-1 expression on CD4⁺ T cells from control and STING^−/− mice at baseline or exposed to chronic hypoxia. (G and H) Flow cytometric quantification of (G) CD8⁺ T cells and (H) MFI of CD62L expression on CD8⁺ T cells from designated experimental groups. (I) Flow chart representation of CD4⁺, CD8⁺, PD-1, and CD62L expression on respected cell group in control (blue) and STING^−/− (red) mice underwent chronic hypoxia exposure. (J – K) Flow cytometric quantification and representation of (J) regulatory T cells (CD4⁺CD25⁺Foxp3⁺) and (K) PD-1 expression on regulatory T cells from dedicated experimental groups. Each dot represents an individual mouse (n=4-10/group). Data was collected across 3 independent experiments. Column represents mean ± SEM. Statistical significance was determined using unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant.

Given described changes in myeloid cell populations in the chronic hypoxia exposed, but not bleomycin treated mice, we asked whether pulmonary T cell function is altered in chronically hypoxic STING^−/− mice. In response, we found that STING^−/− mice exposed to chronic hypoxia displayed a significantly higher number of CD4⁺ T cells compared to control, countered by significantly lower PD-1 expression (Fig. 4, E, F and I). In contrast, the mice exhibited a decrease in the absolute number of presumably infiltrative CD8⁺ T cells, accompanied by lower expression of CD62L, indicating reduced effector function of T cells and likely decreased inflammatory response (Fig. 4, G - I). Interestingly, a significantly lower number of regulatory T cells (CD4⁺CD25⁺Foxp3⁺) with decreased PD-1 expression was seen, consistent with the exhaustive immunophenotype (Fig. 4, J - L). Taken together, we conclude from these data that STING contributes to hypoxia-induced PH development, partly through the complex regulatory network of innate and adaptive immunoregulatory cells.

Deficiency of type I IFN signaling does not protect against PH development secondary to bleomycin and chronic hypoxia.

STING has been primarily described as an upstream regulator of type I interferon gene expression in response to presence of cytosolic nucleotides. Therefore, we next asked what is the pathogenic contribution of the type I IFN response to pulmonary vascular disease in our models of PH. To this end, we induced PH in sex-matched 8-10-week-old C57BL/6J and global Interferon Alpha and Beta Receptor Subunit 1 deficient (IFNAR1^−/−) mice with either bleomycin or chronic hypoxia exposure. Unexpectedly, no significant difference in RVSP and Fulton Index was found between control and IFNAR1^−/− mice in both the bleomycin and chronic hypoxia studies (Fig. 5, A, B and E). Similarly, MTC staining revealed no rescue of inflammation and fibrosis scarring in bleomycin-induced PH mice (Fig. 5, C and F). Muscularized pulmonary vessel assessment with αSMA IHC staining demonstrated no significant difference between experimental groups in both bleomycin and chronic hypoxia models (Fig. 5, D, G and H). There was, however, a significant increase in total muscularized vessels of IFNAR1^−/− mice at baseline, suggesting potential remodeling without physiologic consequence during the period of the experimental condition (Fig. 5, D and H). Interestingly, although global deletion of IFNAR1 in mice was sufficient to dampen recruitment of CD11b⁺ cells into the lungs at baseline, specifically CD11b⁺Ly6C^loLy6G⁺, but not CD11b⁺Ly6C^hiLy6G⁻ (Fig. 5, I - K), no difference was seen across exposure groups in immune cell composition (Fig. 5, I - K). These data conflict with current agreement in the field that inhibiting type I IFN signaling would be expected to decrease leukocyte recruitment in response to inflammatory stimuli^47,48. The differences seen in the collective phenotype between PH-induced STING^−/− and IFNAR1^−/− mice thus indicate a potential non-interferon dependent role of STING in PH development.

Figure 5: Deficiency of type I IFN signaling does not provide protection against PH development secondary to bleomycin or chronic hypoxia.

(A) RVSP measurement of C57BL/6J (control) and global IFNAR1 deficient (IFNAR1^−/−) mice exposed to normoxia, bleomycin injection, or chronic hypoxia exposure. (B) Fulton Index (RV/LV+S) of mice from dedicated experimental groups. (C) Representative images of MTC-stained formalin-fixed lung sections of control and IFNAR1^−/− mice exposed to either normoxia or bleomycin injection. Scale bar = 200μm at 10x magnification. (D) Representative images of IHC staining of αSMA (brown, arrowheads) in formalin-fixed lung sections of mice from designated experimental groups. Scale bar = 200μm at 10x magnification. (E) RVSP curve representation of control and IFNAR1^−/− mice exposed to chronic hypoxia. (F) Quantification of inflammation scoring from MTC-stained lung sections of mice undergoing bleomycin treatment. (G and H) Quantification of muscularized pulmonary vessels (small, medium, large, complete, and partial) from αSMA IHC staining of mice from all experimental groups. (I – K) Flow cytometric quantification of pulmonary infiltrated (I) CD11b⁺, (J) CD11b⁺Ly6C^hiLy6G⁻, and (K) CD11b⁺Ly6C^loLy6G⁺ of control and IFNAR1^−/− mice from different experimental groups. Each dot represents an individual mouse (n=8/group). Column represents mean ± SEM. Box violin shows mean, mode, and interquartile range of data set. Statistical significance was determined by unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant. P values from (H) were calculated from total vessels (small+medium+large) of mice from experimental groups.

To validate these results, we collected whole lung protein from STING^−/− mice subjected to either bleomycin or chronic hypoxia for detection of IFN-related chemokines and cytokines. Consistently, there was no significant difference in IFNγ, CXCL10, IL12 p40, CXCL9, IL2, and IL6 levels between the lungs of control and experimental mice across models (Fig. S4B). While the serum level of IFNβ was similar between WT and STING^−/− mice (Fig. S4C), IFNα level was below detection limit (data not shown). Analysis of single-cell RNA sequencing (scRNAseq) performed on STING^−/− and control mice lungs at baseline or exposed to chronic hypoxia showed little to no change in STING-related type I IFN production genes across multiple pulmonary cell populations (Fig. S4D). These data support the conclusion that STING-associated PH is not primarily regulated through type I IFN signaling.

Protection against PH in STING^−/− mice exposed to chronic hypoxia is associated with increased VEGF expression, primarily derived from CD11b⁺ and CD11c⁺ cells.

Given that STING aggravates PH independent of type I IFN signaling, we next aimed to explore an underlying contribution to disease pathogenesis. Correlative with pulmonary changes, we found significant alterations in vascular endothelial growth factor (VEGF) in whole lung cells of STING^−/− mice subjected to chronic hypoxia, quantified with multiplex (Fig. 6A). Upon further investigation using flow cytometry, we noted that the increase in VEGF level did not appear to originate from the anticipated pulmonary smooth muscle cells (Fig. 6B), or pulmonary endothelial cells (Fig. 6C). The elevation in VEGF observed in chronic hypoxic STING^−/− mice was detected to a large extent in the pulmonary CD11b⁺ populations, including both CD11b⁺Ly6C^hiLy6G⁻ and CD11b⁺Ly6C^loLy6G⁺ cells (Fig. 6, D - G). We also noted a trend towards increased VEGF expression at baseline in non-exposed STING^−/− mice immune cells (Fig. 6, D - G). Interestingly, we also found a mild yet significant increase in VEGF expression in CD11c⁺ cells, which include macrophages and dendritic cell subsets, of STING^−/− mice exposed to chronic hypoxia (Fig. 6, H and I). Together, these data highlight the important role of the STING-VEGF signaling axis in PH development secondary to chronic hypoxia.

Figure 6: Protection against PH in STING^−/− mice exposed to chronic hypoxia is associated with increased VEGF expression, primarily derived from CD11b⁺ and CD11c⁺ cells.

(A) Multiplex quantification of vascular endothelial growth factor (VEGF) in whole lung cells of control and STING^−/− mice at baseline and when subjected to chronic hypoxia exposure. (B, C, E, F, G, and H) Flow cytometric quantification of VEGFa expression in pulmonary infiltrated (B) αSMA⁺, (C) CD31⁺, (E) CD11b⁺, (F) CD11b⁺Ly6C^hiLy6G⁻, (G) CD11b⁺Ly6C^loLy6G⁺, and (H) CD11c⁺ cells of control and STING^−/− mice from dedicated experimental groups. (D and I) Flow plot representation of VEGFa expression in (D) CD11b⁺ and (I) CD11c⁺ cells in designated groups. Each dot represents an individual mouse (n=4-6/group). Column represents mean ± SEM. Statistical significance was determined by unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant.

Inhibition of VEGF signaling through VEGFR1/2 antagonist SU5416 (Sugen) reverses protection against PH secondary to chronic hypoxia in STING^−/− mice.

To further interrogate the role of the STING-VEGF signaling axis in PH development secondary to chronic hypoxia, we sought to block VEGF signaling in STING^−/− mice by employing VEGFR1/2 antagonist SU5416 (Sugen) in a validated PH model replicating key components of pulmonary arterial hypertension. Although the Sugen/Hypoxia PH model is classically used for mechanistic insights into Group I PH, we felt the model was relevant to these experiments given the role of VEGF described above. Additionally, it allowed us to potentially validate our findings against a gold standard PH model. To this end, 8-10-week-old control and STING^−/− mice received weekly injections of Sugen before and while undergoing chronic hypoxia exposure. Interestingly, Sugen/Hypoxia exposed STING^−/− mice no longer demonstrated protection against PH (Fig. 7A). Similarly, there was no change in RV remodeling between experimental groups (Fig. 7B). Assessment of muscularized pulmonary vessels demonstrated no difference in number of vessel muscularization in Sugen/Hypoxia STING^−/− mice compared to controls, as well as complete versus partial muscularized vessels (Fig. 7, C - E). Likewise, staining with MTC revealed no difference in inflammation or collagen deposition between Sugen/Hypoxia control and STING^−/− mice (Fig. 7, F and J). In addition, there was no gross discernible change to vessel structure between control and experimental mice (Fig. S5). Correlative with VEGF immunosuppressive ability, we observed an overall decrease in pulmonary infiltrated CD11b⁺ and CD11b⁺Ly6C^hiLy6G⁻ cells in STING^−/− mice in the Sugen/Hypoxia cohort, but not CD11b⁺Ly6C^loLy6G⁺ cells, quantified by flow cytometry (Fig. 7, G - I). These data collectively highlight the complex role of VEGF signaling in PH and the potential importance of functional differences in myeloid cells in disease pathogenesis relative to STING expression.

Figure 7: Inhibition of VEGF signaling through VEGFR1/2 antagonist SU5416 reverses protection against PH secondary to chronic hypoxia in STING^−/− mice.

(A) RVSP measurement of C57BL/6J (control) and global STING deficient (STING^−/−) mice exposed to normoxia or chronic hypoxia exposure accompanied by SU5416 (Sugen) injection. (B) Fulton Index (RV/LV+S) of mice from dedicated experimental groups. (C – E) Quantification and image representation of muscularized pulmonary vessels (small, medium, large, complete, and partial) from IHC staining for αSMA on formalin-fixed lung sections of mice from designated experimental groups. Scale bar = 200μm at 10x magnification. (F) Inflammation score from MTC-stained lung sections of mice from each experimental group. (G – I) Flow cytometric quantification of pulmonary infiltrated (G) CD11b⁺, (H) CD11b⁺Ly6C^hiLy6G⁻, and (I) CD11b⁺Ly6C^loLy6G⁺ of control and STING^−/− mice from different experimental groups. Each dot represents an individual mouse (n=5-8/group). (J) Representative images of MTC-stained lung sections. Scale bar = 200μm at 10x magnification. Column represents mean ± SEM. Box violin shows mean, mode, and interquartile range of data set. Statistical significance was determined using unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant. P values from (C) were calculated from total vessels (small+medium+large) of mice from experimental groups.

STING inhibition in human PBMC-differentiated myeloid cells results in increased VEGF secretion.

Given these in vivo data, we sought to validate the findings further in human primary cells; experimental design illustrated in Fig. 8A. Peripheral blood mononuclear cells (PBMCs) were first collected from healthy donors and differentiated into human MDSCs, following a well-validated and established protocol⁴⁹. After phenotypic validation with flow cytometric markers (Fig. 8B, gating strategy) and functional validation with T cell suppression assay, the differentiated cells were confirmed to be MDSCs (CD11b⁺CD33⁺HLA-DR⁻), with the majority of the cell population to be monocytic like (CD11b⁺CD33⁺HLA-DR⁻CD14⁺CD15⁻). Myeloid-derived cells’ ability to suppress CD4⁺ and CD8⁺ T cells, with increased MDSC:T coculture ratio resulting in increased suppression, was re-confirmed (Fig. 8, C and D). Upon differentiation, myeloid-derived cells were stimulated with either vehicle, diABZI (STING agonist), or H151 (STING antagonist), washed, and then cocultured with human pulmonary arterial endothelial cells (PAECs). Supporting our in vivo findings, we found significantly decreased VEGF expression in both CD11b⁺CD33⁺HLA-DR⁻ and CD11b⁺CD33⁺HLA-DR⁻CD14⁺CD15⁻ cells upon treatments with diABZI (STING agonist) compared to vehicle. Conversely, inhibition of STING with H151 (STING antagonist treatment) showed significantly increased VEGF expression in the two cell populations, CD11b⁺CD33⁺HLA-DR⁻ and CD11b⁺CD33⁺HLA-DR⁻CD14⁺CD15⁻, compared to vehicle. Together, these data support the important role of STING-VEGF signaling axis in immune cell contribution to PH development.

Figure 8: STING inhibition in human PBMC-differentiated myeloid cells results in increased VEGF secretion.

(A) Experimental scheme. PBMC was collected from human blood and cultured in hGM-CSF for 7 days. Differentiated myeloid cells were treated with either STING agonist (diABZI) or STING antagonist (H151) for 12h, then cultured with PAEC for 12h at 1:1 myeloid:PAEC ratio. Suspended cells were collected for flow cytometry. (B) Gating strategy for differentiated myeloid cells from human PBMC (C and D) Suppressive capability of differentiated myeloid cells on human (C) CD4 and (D) CD8 T cells. T cells and PBMC were from the same donor (male). Each dot represents a technical replicate (E and F) Flow cytometric quantification of VEGF expression in human PBMC-differentiated (E) CD11b⁺CD33⁺HLA-DR⁻ and (F) CD11b⁺CD33⁺HLA-DR⁻CD14⁺CD15⁻. Technical duplicates from 2 human donors (1 male, 1 female) (n=2-3 technical replicate/donor) were displayed. Column represents mean ± SEM. Statistical significance was determined by unpaired two-tailed Welch T’s test. P values are shown on graph. P<0.05 was considered significant.

Finally, to better understand the STING-VEGF related mechanistic pathways in disease context, we performed single-cell RNA sequencing (scRNAseq) on the lungs of control and STING^−/− mice exposed to normoxia or chronic hypoxia. Full annotation of pulmonary cell populations shown in Fig. S6. Gene ontology (GO) enrichment analysis of differentially expressed genes in VEGF signaling pathway demonstrated upregulation of VEGF-mediated genes, including apoptosis-related ATF3, FOS, and EGR1 in control mice exposed to hypoxia, but not hypoxic STING^−/− mice (Fig. S7, A - C). Further analysis confirmed that downregulation of these VEGF-mediated genes in hypoxic STING^−/− mice was preserved in pulmonary endothelial cells but not pulmonary smooth muscle cells and pulmonary fibroblasts (Fig. S7D). Together, the data suggest a contribution of the STING-VEGF signaling pathway to PH pathogenesis, through regulation of downstream apoptosis genes in effector pulmonary resident cells.

In conclusion, these experiments demonstrate a concrete contribution of STING to PH pathogenesis, associated with VEGF signaling and independent of type I IFN. Furthermore, these data shed novel light on the STING-VEGF signaling axis in an established PH disease model, advancing our understanding in development of disease-modifying therapeutics.

Discussion

Pulmonary hypertension is a major global health issue, with Group 3 PH prevalence on the rise, affecting up to 60% of patients with chronic interstitial lung diseases^2,3. Thus, there is an urgent need to identify novel disease mechanisms that can be therapeutically exploited. Here, we report a novel role for cytosolic-DNA sensing protein STING in PH development. Using multiple non-redundant PH models, we confirmed that STING activation results in RVSP elevation, RV remodeling, and immune cell signature changes. Moreover, we provide evidence that STING-associated type I IFN signaling plays a minor role, if any, in PH pathogenesis, with a larger contribution derived from STING-dependent VEGF signaling. Overall, our results identify STING as an important regulator of PH development and progression, thus offering badly needed disease-modifying targeted pathways.

We acknowledge that due to the complex nature of PH, alteration in a single signaling pathway is unlikely to account for all disease characteristics. However, genetic “experiments” of nature, such as mutations described in SAVI patients, may potentially inform disease pathology. Of interest, in a SAVI mouse model with canonical N153S mutation, there is an IRF3-independent mechanism resulting in STING-associated vasculopathy⁵⁰. Intriguingly, an immunophenotype of SAVI mice with V154 mutation occurs even in the absence of IFN signaling⁵¹. Supporting our findings, previous studies focusing on organ-specific role of STING have shown that its deletion in various cells can dampen inflammation in a tissue-dependent disease-specific context⁵², including gut⁵³, kidney^54,55, and brain⁵⁶. Relevant to primary lung disease, a recent study supports an important immunoregulatory function of STING in lung fibrosis, a common risk factor for PH⁵⁷. This report similarly showed that the protective role of STING is independent of type I IFN and is associated with dysregulated neutrophils⁵⁸. This observation is consistent with our findings demonstrating a type I IFN-independent role of STING-alteration in PH²⁶. In slight contrast to our findings, previous evaluation of PH in response to hypoxia in IFNAR1^−/− mice was found to be protective against disease development in the chronic hypoxia model²⁵. A possible explanation for this apparent discrepancy is that the importance of type I IFN signaling is more profound in the early stage of disease development, in which inflammation resolution relies more heavily on innate immune response. With disease progression and chronic inflammation established, persistent IFN signaling may fuel pathogenic inflammation; future studies will be necessary to elucidate the distinction of timing involved in this pathway’s activation.

As noted, STING plays an important role in a variety of lung diseases, as well as in other disease states classically associated with chronic inflammation. For instance, in the study of PH-relevant pathways in the evolution of cancer⁵⁹, STING has been shown to amplify MDSC recruitment to tumor site post-radiation, contributing to radiation resistance and metastasis⁶⁰. In addition, STING expressed in tumor cells contributes to anti-tumor immunity⁶¹, highlighting the pivotal role of STING in suppressing immune response. Related, STING^−/− mice are likewise rescued from injury in several inflammatory models such as silica-induced fibrosis and sepsis^62,63, as well as autoimmune diseases like lupus or rheumatoid arthritis^64,65. Our study provides the first evidence – to our knowledge – of STING relevant to PH. It is worth noting that alleviation in right ventricular remodeling, lower level of fibrosis, and muscularized vessels were observed in STING^−/− mice exposed to bleomycin or chronic hypoxia, highlighting the potential relevance of STING in cells that take part in vessel muscularization and remodeling, such as myofibroblasts. The change in Fulton Index of STING^−/− mice, specifically, has significant translational potential, as patients with even mild preservation of right heart function have been shown to perform significantly better with slower disease progression^66,67. Future studies regarding STING should thus focus on the specific mechanism preventing right-heart failure in end-stage PH patients, in light of our findings and other studies demonstrating that STING^−/− mice were protected from cardiovascular diseases^68,69.

In addition, we demonstrated that MDSCs recruited to the lungs of STING^−/− mice exposed to chronic hypoxia exhibit a decrease in expression of immune checkpoint PD-L1. This corresponds to an increase in number of “exhausted” pulmonary infiltrated CD4⁺ T cells in the same mice with a significant reduction of receptor PD-1 expression. Such findings further support the intricate role of the PD-1/PD-L1 axis acting between MDSCs and T lymphocytes in PH development and progression, a finding that has been reported previously by our group^37,42. STING^−/− mice exhibited a significant decrease in the number of lung CD11b⁺Ly6C^loLy6G⁻ cells, most likely composed of alveolar macrophages, although a significant number of dendritic cell sub-populations cannot be ruled out. Analysis of this cell population yielded no significant changes between groups, but we cannot rule out involvement in STING-mediated PH development based on these data, and additional work is required.

The role of VEGF in PH is complex depending upon the timing of expression and secretion, responsive cell type, and concentration⁷⁰. As well described, treatment with VEGFR1/2 inhibitor, SU5416/Sugen, concurrently with chronic hypoxia exposure, produces irreversible PH in rodents⁷¹. Therefore, the use of the Sugen/hypoxia model in our experiments adds to the strength of the described studies, demonstrating evidence for the importance of VEGF-STING signaling in vascular regeneration. Importantly, in our hands, the source of VEGF secretion derived in large part from pulmonary infiltrated CD11b⁺ and CD11c⁺ cells, while classical vascular VEGF-producing cells such as pulmonary vascular endothelial cells or pulmonary smooth muscle cells exhibited little to no change in VEGF expression. This finding suggests a healthy angiogenic role for cell-specific VEGF that is STING-regulated. While the immunosuppressive effect of VEGF secreted from CD11b⁺ cells has been the focus across various inflammatory models^72,73, other studies have convincingly demonstrated the critical role of VEGF from both CD11b⁺ and CD11c⁺ cells in inflammation resolution and subsequent repair^73-75.

Moreover, experiments in PBMC-differentiated myeloid-derived cells treated with STING antagonist demonstrated an increase in VEGF expression, while treatment with STING agonist resulted in a decrease of VEGF secretion compared to vehicle in the same cell populations. Additionally, our scRNAseq data demonstrated downregulation of VEGF-mediated apoptosis genes (ATF3, EGR1, and FOS) in pulmonary endothelial cells, correlating with protection against PH in STING^−/− mice. As VEGF has been shown to prevent apoptosis in multiple homeostatic and disease contexts^76-78, further investigation of cell-specific STING-VEGF mediated pathways will provide more insight into PH pathogenesis.

Our study has several noteworthy limitations. The causal mechanism for STING activation is mainly unexplored, though cell-free DNA has been described as relevant to PH in several human-cell and animal-based studies^27,29. Further investigations are required to fully uncover the respective contribution of either native or viral DNA to STING activation in PH. Also, the discernment of VEGF secretion between CD11b⁺ and CD11c⁺ cells and the origin of infiltrated myeloid cells remains unknown. Future studies focusing on sub-populations of CD11c⁺ cells (infiltrated dendritic cells or alveolar macrophages) and different cells’ role in STING-VEGF signaling will allow more in-depth understanding of these cell-specific influences of STING in disease development.

In conclusion, these data demonstrate a unique contribution of STING to PH development that is largely independent of type I IFN signaling, instead acting primarily through alteration of VEGF expression in myeloid-derived cells. Thus, targeting these pathways may be of therapeutic value in future clinical trials for treatment of Group 3 PH and beyond.

Highlight.

• STING contributes to PH largely independent of type I IFN signaling.

• Suppression of STING-mediated VEGF secretion from pulmonary infiltrated CD11b⁺ and CD11c⁺ cells advance PH.

• STING-regulated VEGF expression promotes PH.

Non-standard Abbreviations and Acronyms

IFNAR1^−/−

Global interferon alpha and beta receptor subunit 1 deficiency

IHC

Immunohistochemical

IPF

Idiopathic pulmonary fibrosis

IRF3

Interferon regulatory factor 3

MDSC

Myeloid-derived suppressor cell

Mo-MDSC

Monocytic myeloid-derived suppressor cell

MTC

Masson trichrome

PBMC

Peripheral blood mononuclear cell

PD-L1

Programmed death ligand-1

PH

Pulmonary hypertension

PMN-MDSC

Polymorphonuclear myeloid-derived suppressor cell

RVSP

Right ventricular systolic pressure

STING

Stimulator of Interferon Genes

STING^−/−

Global STING deficiency

VEGF

Vascular endothelial growth factor

WT

Wild type