Regulation of Monocyte Activation by PPARα Through Interaction With the cGAS-STING Pathway

Growing evidence suggests that retinal inflammation plays a pathogenic role in diabetic retinopathy (DR) (1). Diabetes can result in leukocyte activation, adhesion to the retina vascular endothelium, and extravasation into the retinal tissue, leading to a chronic inflammatory state of the retina (2). Leukocyte adhesion to retinal vasculature, referred to as leukostasis, has been observed in patients with diabetes and in animal models of diabetes (3,4). Increased leukocyte adhesion contributes to the blood-retinal barrier breakdown, a principal cause of diabetic macular edema (DME) (5,6). Leukostasis can also result in nonperfused capillaries, which leads to retinal ischemia and neovascularization. Thus, leukostasis is the major contributor to DME and neovascularization (7,8).

As a type of leukocyte, monocytes can differentiate into macrophages or myeloid lineage dendritic cells in tissues and are pivotal in endothelium damage in adaptive immunity-related diseases, such as atherogenesis (9–11). Monocytes from donors with diabetes exhibited increased proatherogenic activities compared with those from control donors (12,13). In addition, monocyte adhesion to the retinal vascular endothelium is an important feature in DR (14,15). However, the regulation of monocyte activation in retinal inflammation and vascular leakage in DR has not been well studied.

The cGAS-STING signaling axis, a major regulator of inflammation, consists of cyclic GMP-AMP synthase (cGAS) and a stimulator of interferon genes (STING) (16). cGAS is a cytosolic DNA sensor that generates 2′,3′-cyclic GMP-AMP (cGAMP) in response to the binding of double-stranded DNA (17). cGAMP binds and activates STING, leading to the activation of a cascade of downstream pathways, including those mediated by TANK-bind kinase 1 (TBK1)/interferon regulatory factor 3 (IRF3), nuclear factor-κB (NF-κB), and Janus kinase (JAK)-signal transducer and activator of transcription (STAT) (18–20). Consequently, these pathways upregulate inflammatory cytokines, such as interleukin 1β (IL-1β) and tumor necrosis factor-α (TNF-α), eliciting innate immune responses. cGAS-STING has been linked with chronic inflammation and cancer progression. However, its association with DR has not been established.

Peroxisome proliferator-activated receptor-α (PPARα) is a ligand-activated transcription factor (21). PPARα activates the expression of genes in lipid metabolism and regulates energy homeostasis (22,23). PPARα also inhibits the transcription of inflammatory response genes (24,25). PPARα-deficient mice show a prolonged response to inflammatory stimuli (26,27). Two independent prospective clinical trials reported that fenofibrate, a PPARα agonist, has robust therapeutic effects on DR in patients with type 2 diabetes (28,29). Fenofibrate has a PPARα-dependent anti-inflammatory effect by inhibiting proinflammatory transcription factors, such as NF-κB (30). In addition, fenofibrate reduces retinal vascular leakage in a PPARα-dependent manner in diabetic animal models, which may contribute to its effect on DME in patients with diabetes (31). Recently, Wang et al. (32) identified that PPARα contributed to the aging-induced inflammation in monocytes. However, the function of PPARα in monocytes in diabetes and DR is unknown.

In the present research, we measured PPARα levels in monocytes from patients with diabetes and diabetic animal models. We also investigated the regulatory role of PPARα in monocyte activation and retinal leukostasis in DR models and its interaction with the cGAS-STING pathway.

The study adhered to the tenets of the Declaration of Helsinki and was approved by the Tianjin Medical University Eye Hospital. Written informed consent was obtained from each patient. Peripheral blood mononuclear cells (PBMCs) were isolated from healthy volunteers and patients with type 2 diabetes by density gradient centrifugation (Ficoll Paque Plus; Sigma-Aldrich, St. Louis, MO). In some experiments, monocytes were purified from the isolated PBMCs by using a positive monocyte isolation kit (MACS; Miltenyi, Bergisch Gladbach, Germany).

Male db/db mice, their heterozygous littermates (db/m), male Akita mice, OVE26 mice, STING knockout (KO) mice, PPARα global KO (PPARα^−/−) mice, C57BL/6J mice (wild-type [WT] control for PPARα^−/− mice), and FVB mice (nondiabetic control of OVE26 mice) were from Jackson Laboratory (Bar Harbor, ME). Male Brown Norway (BN) rats were from Charles River Laboratories (Wilmington, MA). All experiments were performed following the guidelines of the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and approved by the Oklahoma University Health Sciences Center Institutional Animal Care and Use Committee.

The transgenic PPARα^flox/flox mouse line was generated using chicken β-actin promoter and a floxed stop cassette through a contracted service with Cyagen Biosciences (Santa Clara, CA). The transgenic PPARα^flox/flox mouse line was crossbred with CX3CR1-Cre mice (Jackson Laboratory) to generate monocyte-specific PPARα transgenic mice (PPARα^MCTg) overexpressing PPARα.

PPARα^flox/flox mice for KO were designed by inserting loxP sites flanking exon 4 of the PPARα gene in Ingenious Targeting Laboratory (Ronkonkoma, NY). To generate monocyte-specific PPARα KO mice (PPARα^MCKO), we crossbred PPARα^flox/flox mice with CX3CR1-Cre mice.

BN rats received an intraperitoneal (i.p.) injection of streptozotocin (STZ; 50 mg/kg) after overnight fasting. Blood glucose levels were measured 3 days after the injection. Diabetes was induced in mice with daily i.p. injections of STZ at a dose of 55 mg/kg for 5 consecutive days. Only the animals with glucose levels >350 mg/dL were used as diabetic animals. Animals were fed control rodent chow (5053; LabDiet/TestDiet, Ft. Worth, TX) or the chow containing 0.04% (for rats) and 0.014% (for mice) fenofibrate (Sigma-Aldrich) for 8 weeks (rats) and 12 weeks (mice) from the onset of diabetes.

The assay was performed as described previously (33). Briefly, anesthetized mice were perfused with warm PBS, and then adherent leukocytes in the vasculature were stained by perfusion with FITC-conjugated concanavalin-A (ConA; Vector, Burlingame, CA). Extra Con A was washed off by PBS perfusion. Adherent leukocytes were counted in the whole retina under a Zeiss AxioObserver Z1 epi-fluorescent microscope (Carl Zeiss, Göttingen, Germany). Representative images in flat-mounted retinas were collected and presented.

Monocyte lysate was resolved by SDS-PAGE and immunoblotted with primary antibodies against human PPARα (Affinity; Changzhou, Jiangsu, China), mouse PPARα (Abcam, Waltham, MA), STING (Proteintech Group), cGAS (Cell Signaling Technology, Danvers, MA), and β-actin (Sigma-Aldrich). The band intensities were semiquantified by densitometry using ImageJ software (National Institutes of Health, Bethesda, MD).

Total RNA was isolated from monocytes using TRIzol reagent (Thermo Fisher, Waltham, MA). The quantitative RT-PCR (qRT-PCR) results were normalized against levels of hypoxanthine phosphoribosyl transferase (HPRT) mRNA. The primer sequences were as follows: human PPARα–forward: 5′-ATGGTGGACACGGAAAGCC-3′, human PPARα–reverse: 5′-CGATGGATTGCGAAATCTCTTGG-3′, human HPRT–forward: 5′-CCTGGCGTCGTGAT-TAGTGAT-3′, and human HPRT–reverse: 5′-AGACGTTCAGTCCTGTCCATAA-3′.

To quantify protein levels of PPARα and STING in monocytes, flow cytometry was used as previously described (34). Briefly, after fixation, cells were incubated with a panel of antibodies against CD11b-BV510 (BioLegend, San Diego, CA) and PPARα (Novus, Centennial, CO), or a panel of antibodies against CD11b conjugated with FITC (101205; BioLegend) and STING (19851-1-AP; Proteintech, Rosemont, IL). The stained cells were counted by Stratedigm S1300Ex (Stratedigm, San Jose, CA), and data were analyzed by FlowJo software (BD Biosciences, Franklin Lakes, NJ).

Adhesion analyzed by fluorescence microscopy: Primary mouse brain microvascular endothelial cells (mECs) were isolated and purified by puromycin from 3- to 4-week-old mice following a published protocol (35). mEC purity was verified by immunofluorescence using an antibody against CD31 (Supplementary Fig. 1). mECs at passages 1–2 or primary human retinal capillary ECs (HRCECs; Cell Systems, Kirkland, WA) were seeded in 6-well plates and allowed to reach confluence. ECs were incubated with or without 2 ng/mL TNF-α for 6 h at 37°C. Freshly isolated PBMCs were incubated with the EC monolayer for 6 h at 37°C. After nonadherent cells were removed, adherent cells were fixed with 4% paraformaldehyde, stained with DAPI, and visualized under a fluorescent microscope. Digital images from five random high-power fields (HPFs) were taken using the Biotek Cytation1 microscope (Agilent, Santa Clara, CA). Adhered cells per HPF were counted using the ImageJ software. The result was presented as adherent cell numbers per field.

Adhesion analyzed by flow cytometry: Briefly, PBMCs were labeled with 2′,7′-bis (2-carboxyethyl)-5(6)-carboxy-fluorescein acetoxymethyl ester (BCECF/AM; Thermo Fisher) and then were incubated with the TNF-α–treated EC monolayer for 6 h at 37°C. After nonadherent cells were removed, the adherent cells were trypsinized, resuspended, and transferred into test tubes (Falcon 5 mL Round Bottom Polystyrene Test Tube) for flow cytometry analysis. Next, cells were loaded on FACS LSR II flow cytometer (BD Biosciences, San Jose, CA) and analyzed using FlowJo (BD Biosciences). The result was presented as the percentage of monocytes in the total cell suspension, including ECs and PBMCs.

Monocytes were seeded into the upper chamber of 6.5-mm polycarbonate membrane inserts with 3.0 μm-diameter pores (Corning, Corning, NY) and incubated for 6 h. Cells were fixed with 70% methanol and stained with hematoxylin-eosin. After cells on the top were removed, membranes were air dried for 1 h, separated with the Transwell inserts, and mounted on a glass slide. Cells migrating through the membrane were imaged for five random HPFs using an Olympus BX43 microscope (Olympus, Tokyo, Japan). The migrated cells of each membrane were counted using ImageJ software and averaged from the five images.

After being incubated with 20 mg/L FITC-conjugated dextran (40 kDa; Molecular Probes, Life Technologies, Carlsbad, CA) for 30 min at 37°C, cells were washed and fixed in 4% paraformaldehyde. Average fluorescence intensities and the percentages of dextran-positive cells were determined by flow cytometry.

Cytosolic and nuclear fractions were isolated using the Cell Fractionation kit (Thermo Fisher) according to the manufacturer’s protocol. The cytosolic mtDNA was quantified by measuring mitochondrial gene cytochrome c oxidase 1 using real-time PCR and normalized by a chromosome gene, β-actin. The following primers were used: cytochrome c oxidase 1–forward: 5′-TCGGAGCCCCAGATATAGCA-3′ and reverse: 5′-TTTCCGGCTAGAGGTGGGTA-3′; and β-actin–forward: 5′-ACCTTCTACAATGAGCTGCG-3′ and reverse: 5′-CTGGATGGCTACGTACATGG-3′.

Analysis of the metabolic profile in monocytes was performed via a Seahorse XFe96 Flux Analyzer (Agilent, Santa Clara, CA). A Seahorse XF Cell Mito Stress Test kit (Agilent) was used to record the oxygen consumption rate (OCR; oligomycin 1 μmol/L, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone [FCCP] 2 μmol/L, rotenone and antimycin A [RAA] 1 μmol/L). The extracellular acidification rate (ECAR) was studied by a Seahorse XF Glycolysis Stress Test kit (Agilent) with glucose 10 mmol/L, oligomycin 1.0 μmol/L, and 2-deoxy-d-glucose 50 mmol/L. Mitochondrial function and glycolysis were calculated and compared according to the manufacturer’s instructions.

All experiments were performed at least three times independently. Quantitative data are presented as mean ± SD and were analyzed by Student t test for comparison of two groups and by ANOVA for comparison of more than two groups.

Data and resources are available upon request. For original data, please contact [email protected].

As shown in Fig. 1A–C, both the protein and mRNA levels of PPARα were significantly decreased in monocytes from patients with diabetes relative to subjects without diabetes, as shown by Western blot and qRT-PCR analyses (Fig. 1A–C). PPARα levels were also downregulated in monocytes from type 1 diabetic models, including STZ-induced diabetic rats (Fig. 1D and E), OVE26 mice (a genetic type 1 diabetic model) (Fig. 1F and G), STZ-induced diabetic mice (Fig. 1H and I), and a type 2 diabetic model, db/db mice (Fig. 1J and K). The CD11b⁺/PPARα⁺ monocyte count was significantly decreased in diabetic rats. Conversely, fenofibrate, an agonist of PPARα, significantly increased the CD11b⁺/PPARα⁺ monocytes population in diabetic rats (Fig. 1L and M). Similarly, the numbers of CD11b⁺/PPARα⁺ monocytes in STZ-diabetic mice fed fenofibrate chow were higher than in diabetic mice fed regular chow (Fig. 1N and O). The results suggested that fenofibrate attenuated diabetes-induced downregulation of PPARα in monocytes.

As shown by the leukostasis assay, the retinas from the STZ-induced diabetic mice showed increased leukocytes adherent to retinal capillaries relative to nondiabetic retinas, and fenofibrate oral treatment reduced diabetes-induced leukocyte adhesion in the retina (Fig. 2A and B). In vitro leukocyte adhesion assays by cell counting (Fig. 2C and D) and flow cytometry (Fig. 2E and F) both demonstrated that monocytes isolated from diabetic mice showed enhanced adhesion to HRCECs compared with monocytes from nondiabetic mice. Monocytes from diabetic mice fed fenofibrate showed decreased adherence to HRCECs compared with those from diabetic mice fed regular chow (Fig. 2C–F). These results suggested that PPARα activation alleviates leukostasis in the diabetic retina.

The phagocytic activity was compared on monocytes from STZ-induced diabetic mice and age-matched nondiabetic mice fed fenofibrate chow or regular chow. As shown by flow cytometry, relative to the isolated nondiabetic monocytes, diabetic monocytes showed enhanced phagocytosis activity, and the diabetes-induced phagocytosis was attenuated by fenofibrate treatment (Fig. 2G and H).

Monocyte migration was measured using a Transwell assay. Compared with monocytes from nondiabetic mice, STZ-diabetic monocytes showed increased migration rates. Fenofibrate chow significantly alleviated diabetes-induced monocyte migration (Fig. 2I and J). However, fenofibrate did not affect the activities of monocytes under nondiabetic conditions (Fig. 2).

Diabetic PPARα global KO (PPARα^−/−) mouse monocytes demonstrated potent adhesion activity to HRCECs relative to diabetic WT monocytes (Fig. 2K and L). Phagocytosis and migration activities were also enhanced in diabetic PPARα^−/− monocytes compared with diabetic WT cells (Fig. 2M and N). The deficiency of PPARα did not change monocyte activities under nondiabetic conditions. Taken together, these results indicate that PPARα attenuates the activation of monocytes in diabetic conditions.

The effects of fenofibrate on PPARα^−/− monocytes and WT monocytes were compared. Fenofibrate suppressed monocyte adhesion (Fig. 3A–D), phagocytosis (Fig. 3E and F), and migration (Fig. 3G) in diabetic WT mice. However, fenofibrate had no inhibitory effect on activation of diabetic PPARα^−/− monocytes (Fig. 3A–G), suggesting that the effect of fenofibrate on diabetes-activated monocytes was through a PPARα-dependent mechanism.

As a product of lipid peroxidation, 4-hydroxy-2-nonenal (4HNE) is commonly used as a stressor of diabetes in vitro. 4HNE significantly induced cell adhesion (Fig. 3H and I), phagocytosis (Fig. 3J), and migration (Fig. 3K) in primary monocytes from WT mice. Fenofibric acid (FA), an active metabolite of fenofibrate, significantly attenuated the activation of monocytes induced by 4HNE but not under the unstressed control condition (Fig. 3H–L).

Primary monocytes from age-matched WT mice or PPARα^−/− mice were isolated and treated with 4HNE. PPARα deficiency dramatically exacerbated 4HNE-induced adhesion, phagocytosis, and migration of monocytes compared with those in WT monocytes (Fig. 3L–O). FA significantly reduced adhesion (Fig. 3L and M), phagocytosis (Fig. 3N), and migration (Fig. 3O) in 4HNE-treated monocytes from WT mice, but not in those from PPARα^−/− mice.

To determine whether PPARα, expressed in monocytes or in ECs, plays a significant role in regulating monocyte-EC adhesion, we incubated PPARα^−/− ECs with WT monocytes, WT ECs with PPARα^−/− monocytes, WT EC with WT monocytes, and PPARα^−/− ECs with PPARα^−/− monocytes side by side. As shown in Fig. 4A–D, cocultures of PPARα^−/− ECs with WT monocytes and WT ECs with PPARα^−/− monocytes showed increased adhesion compared with the incubation of WT monocytes with WT ECs, suggesting that PPARα deficiency in either monocytes or ECs promoted monocyte-EC adhesion. Moreover, incubation of PPARα^−/− monocytes with PPARα^−/− ECs showed the highest adhesion increases among the experimental groups.

Using the Cre-lox system, we generated tamoxifen-induced monocyte-specific conditional PPARα KO (PPARα^MCKO) and PPARα transgenic (PPARα^MCTg) mouse lines. Flow cytometry results verified that the CD11b⁺/PPARα⁺ cell count was lower in PPARα^MCKO mice and was higher in PPARα^MCTg mice than in age-matched WT mice (Fig. 4E and F). 4HNE induced a further increase in cell adhesion from PPARα^MCKO monocytes but fewer adherent monocytes from the PPARα^MCTg mice relative to WT mice (Fig. 4G and H). PPARα^MCKO significantly promoted, while PPARα^MCTg decreased 4HNE-induced monocytic phagocytosis (Fig. 4I) and migration (Fig. 4J) compared with that in WT mice. We also induced diabetes in PPARα^MCKO, PPARα^MCTg, and age-matched WT mice. At 12 weeks after diabetes onset, retinal leukostasis was examined and compared between groups. Compared with the diabetic WT group, diabetic PPARα^MCKO mice showed significant increases in leukocytes adherent to retinal capillaries, and PPARα overexpression in PPARα^MCTg monocytes attenuated diabetes-induced retinal leukostasis (Fig. 4K).

The cGAS-STING pathway is crucial for inflammation modulation. However, there is no previous evidence implicating the cGAS-STING pathway in DR. To explore the role of the cGAS-STING pathway in DR and its interaction with PPARα, we examined STING levels in monocytes by flow cytometry (Fig. 5A and B). Diabetic mice showed significantly increased STING⁺ monocytes (CD11b⁺/STING⁺ cells) relative to nondiabetic mice (Fig. 5A and B). Furthermore, cGAS and STING protein levels were dramatically upregulated in diabetic monocytes relative to nondiabetic monocytes (Fig. 5C and D). As mtDNA release into the cytosol is a key inducer of the cGAS-STING activation, we quantified cytosolic mtDNA by qRT-PCR. As shown in Fig. 5E, hyperglycemia induced mtDNA cytosolic leakage in monocytes relative to the nondiabetic group.

To study the role of monocytic cGAS-STING signaling in monocyte adhesion, we cocultured HRCECs with monocytes from WT or STING^−/− mice, with or without 4HNE exposure. As shown in Fig. 5F–H, in the absence of 4HNE, there was no significant difference between WT and STING^−/− monocytes in adherence to HRCECs. 4HNE significantly promoted WT monocyte adhesion. In contrast, 4HNE induced less adhesion in STING^−/− monocytes relative to WT monocytes (Fig. 5F–H). A STING inhibitor, C-176 (Cayman Chemical, no. 25859), was used to determine the role of cGAS-STING signaling in monocyte activation in Akita mice (10 weeks of age) and db/db mice aged 12 weeks. Animals received C-176 by daily i.p. injections at the dose of 750 nmol for 14 days, with corn oil as the vehicle control. C-176 significantly reduced the adherent leukocyte number in the retinas of db/db and Akita mice (Fig. 5I and J). In vitro experiments using isolated monocytes also showed that C-176 blocked the 4HNE-induced WT monocyte adhesion (Fig. 5O and P) and phagocytosis (Fig. 5Q and R). These data indicated that the cGAS-STING pathway plays an essential role in monocyte activation in diabetes.

The interaction between PPARα and cGAS-STING signaling is previously unknown. Diabetic PPARα^−/− mice showed more prominent STING upregulation in monocytes relative to diabetic WT mice (Fig. 5B). Fenofibrate oral treatment reduced the monocytic STING expression in diabetic WT mice but not in diabetic PPARα^−/− mice (Fig. 5B). Further, PPARα KO resulted in upregulation in monocytic STING levels in nondiabetic mice (Fig. 5B).

PPARα^−/− monocytes also showed significantly higher levels of cGAS and STING than that of WT monocytes (Fig. 5L and M). 4HNE induced further STING upregulation in PPARα^−/− monocytes compared with WT monocytes, and FA reduced the upregulation of STING in monocytes from WT monocytes but not in PPARα^−/− monocytes (Fig. 5K). qRT-PCR showed dramatically elevated cytosolic mtDNA leakage in PPARα^−/− monocytes relative to WT monocytes (Fig. 5N). 4HNE induced a more prominent increase in adherent and phagocytosis in PPARα^−/− monocytes, which was inhibited by C-176 (Fig. 5O–R). This result suggested that PPARα KO resulted in mitochondrial damage and activation of the cGAS-STING pathway, leading to monocyte activation.

To elucidate the mechanism by which PPARα regulates the cGAS-STING pathway and monocyte activation, we measured the metabolism of monocytes using a Seahorse XF96 analyzer. The basal mitochondrial respiration, maximal mitochondrial respiration, mitochondrial ATP generation, and spare mitochondrial respiration were calculated using real-time OCR values. Compared with those in diabetic WT mice, diabetic PPARα^−/− monocytes showed a decreased basal respiration rate (Supplementary Fig. 2A) and ATP generation (Fig. 6A and E), suggesting impaired oxidative phosphorylation in diabetic PPARα^−/− monocytes (Supplementary Table 4). Furthermore, the maximal respiration capacity (Fig. 6D) and spare respiration capacity (Supplementary Fig. 2B) also declined in diabetic PPARα^−/− monocytes, suggesting that PPARα deficiency exacerbated mitochondrial damage in diabetic monocytes. Meanwhile, diabetic PPARα^−/− monocytes showed elevated glycolysis, including increased glycolysis level (Fig. 6I and L), glycolytic capacity (Fig. 6M), and glycolytic reserve (Supplementary Fig. 2H and Supplementary Table 4), suggesting that PPARα deficiency in monocytes resulted in a metabolism switch from mitochondrial oxidative phosphorylation to glycolysis. Fenofibrate effectively preserved the monocytic mitochondrial function in diabetic mice (Fig. 6B, F, and G and Supplementary Fig. 2C and D) and attenuated glycolysis (Fig. 6J, N, and O and Supplementary Fig. 2I). However, fenofibrate showed no effect on monocyte metabolism in diabetic PPARα^−/− mice (Fig. 6C, H, K, and P, Supplementary Fig. 2E–G, J, and K, and Supplementary Table 4), suggesting that PPARα activation preserved the homeostasis of energy metabolism and the protective effect of fenofibrate on monocytic metabolism is PPARα dependent.

4HNE induced a more prominent metabolism switch in PPARα^−/− monocytes relative to WT monocytes, as reflected by declined mitochondrial function (Fig. 7A and B, Supplementary Fig. 3A and B, and Supplementary Table 5) and increased glycolysis (Fig. 7F and G, Supplementary Fig. 3H, and Supplementary Table 5). FA prevented 4HNE-induced metabolism change (Fig. 7C, D, H, and I, Supplementary Fig. 3C, D, and I, and Supplementary Table 5) only in WT monocytes but not in PPARα^−/− monocytes (Fig. 7E and J, Supplementary Fig. 3E–G, J, and K, and Supplementary Table 5) exposed to 4HNE. Taken together, these data indicated that PPARα prevented metabolic profile change and protected mitochondria in monocytes under diabetic conditions, which regulates the cGAS-STING pathway and monocyte activation (Fig. 7K).

The current study identified for the first time that the expression of PPARα was downregulated in monocytes from patients with diabetes and diabetic rodent models. Our results also demonstrated that PPARα in monocytes plays a crucial role in regulating monocyte activation and retinal leukostasis. Using monocyte-specific conditional PPARα KO and transgenic mice, we have shown that PPARα attenuated retinal leukostasis through suppressing diabetes-induced monocyte activation. Deficiency of PPARα in monocytes resulted in a disturbed metabolism profile with declined mitochondrial function and increased glycolysis, which contributed to the monocyte activation in diabetes. We also demonstrated for the first time that PPARα deficiency contributed to the cGAS-STING pathway activation in diabetic monocytes, likely through increasing the cytosolic release of mtDNA in monocytes. These findings suggest that diabetes-induced PPARα downregulation results in mitochondrial damage and mtDNA release, leading to activation of the cGAS-STING pathway in monocytes and leukostasis in the retina. These results identified a novel interaction between PPARα and the cGAS-STING pathway in diabetic complications. This study suggests that attenuation of monocyte activation represents a new mechanism for the therapeutic effect of fenofibrate on DR.

DR is considered a chronic inflammatory disorder (36). Leukocyte adherence to retinal vasculature or leukostasis contributes to retinal inflammation, blood-retinal barrier breakdown, and DME (1,36). Most previous studies of DR focused on retinal cells, such as retinal ECs and Müller cells (31,37). The current study identified an important role of monocyte activation in retinal inflammation in DR. Our previous studies identified that the expression of PPARα is downregulated in the retinas of DR patients and DR animal models (30,38). PPARα in ECs exerts protective effects against DR by reducing reactive oxygen species production and inhibiting inflammation-related signaling pathways (31,34,38). However, the role of PPARα in monocyte activation has not been studied previously. Monocytes are activated in diabetes and adhere to the vascular endothelium (retinal leukostasis). Increased retinal leukostasis promotes endothelium dysfunction (14,36). Here, we provide evidence that PPARα levels in monocytes are significantly decreased in both patients with diabetes and in diabetic animal models. Using PPARα KO and agonist, we studied monocyte adhesion by both in vivo (retinal leukostasis) and in vitro (using TNF-α–pretreated HRCECs) experiments. Because of the methodology, the effects of fenofibrate on monocyte adhesion were different between the in vivo leukostasis assay and the in vitro cell adhesion assay. However, the results from both experiments consistently demonstrated that PPARα confers a negative regulation of monocyte activation. These results suggest that diabetes-induced PPARα downregulation in monocytes plays a significant role in monocyte activation and retinal inflammation and, thus, represents a novel therapeutic target for DR.

The PPARα agonist fenofibrate is the first oral drug demonstrating robust therapeutic effects in DR patients, as reported by two large, longitudinal clinical trials (39,40). Our previous studies have reported the effect of fenofibrate on retinal inflammation and vascular leakage in diabetic models (38,41). Its mechanism of action is still not well understood. Our results demonstrate that fenofibrate treatment attenuated the diabetes-induced monocyte activation in diabetic WT mice but not in diabetic PPARα^−/− mice, suggesting that fenofibrate suppresses monocyte activation and thus alleviates leukostasis through a PPARα-dependent mechanism. This finding has revealed monocytes as a new drug target of fenofibrate and suggests that inhibition of monocyte activation in diabetes is a new mechanism for the therapeutic effect of fenofibrate on DR.

PPARα is a critical transcriptional factor regulating lipid metabolism and inflammation (42). Global PPARα KO mice develop dyslipidemia (43). We have reported that diabetic global PPARα KO mice demonstrated increased retinal leukostasis relative to diabetic WT mice (38). To exclude possible systemic impacts of dyslipidemia on monocyte activation in global PPARα KO mice, we generated monocyte-specific PPARα conditional KO and transgenic mice for this study. Consistent with the phenotypes from the diabetic global PPARα KO mice, monocyte-specific PPARα conditional KO mice also showed exacerbated retina leukostasis and monocyte activation in diabetes compared with diabetic WT mice. In contrast, transgenic overexpression of PPARα in monocytes alleviated diabetes-induced retinal leukostasis and inhibited monocyte activities. These observations demonstrated that monocytic PPARα is involved in the regulation of monocyte activation and retinal leukostasis in DR.

It is well known that metabolism has important impacts on the function of immune cells (44). Immune cells increase glycolytic ATP production during activation to meet increased energy demands, as declined mitochondrial function and enhanced glycolysis are associated with immune cell activation (44). Mitochondrial dysfunctions have also been related to severe chronic inflammatory disorders (45). In monocytes from diabetic animals or in monocytes treated with 4HNE, mitochondrial function was decreased, while glycolysis was increased, correlating with monocyte activation. Compared with WT, PPARα^−/− monocytes have more prominent declined mitochondrial function and increased glycolysis in diabetic and nondiabetic conditions (Supplementary Fig. 4 and Supplementary Tables 4 and 5). The results suggest that PPARα deficiency contributes to mitochondrial functional damage in monocytes and that the subsequent metabolic profile changes could contribute to monocyte activation.

The cGAS-STING pathway is a major pathway regulating inflammation (46). The cGAS-STING pathway is activated and plays a role in diabetic wound healing (47). Its implication in DR, however, has not been documented. The role of this pathway in monocyte activation in diabetes is unknown. The current study investigated the role of the cGAS-STING pathway in monocyte activation in DR. Both our in vitro and in vivo results showed increased levels of cGAS and STING in monocytes from a diabetic model and in primary monocytes exposed to a diabetic stressor. Furthermore, a STING inhibitor and STING KO both attenuated monocyte activation under diabetic conditions. These results provide the first evidence that the cGAS-STING pathway mediates monocyte activation in diabetes and contributes to DR. Interestingly, we also found that fenofibrate attenuated STING activation in monocytes under diabetic conditions, while PPARα KO alone is sufficient to induce the cGAS and STING activation in monocytes, suggesting that PPARα participates in the regulation of the cGAS-STING pathway. Our findings also suggest that suppression of the cGAS-STING activation may represent a molecular mechanism for the anti-inflammatory activity of PPARα.

Cytosolic mtDNA is a known activator of the cGAS-STING pathway (48,49). PPARα is well known to regulate mitochondrial function as well as integrity through regulating PGC-1α and Sirt1 (50,51). To identify the mechanism by which PPARα negatively regulates the cGAS-STING pathway, we measured cytosolic mtDNA release. Diabetes or PPARα KO alone increased cytosolic mtDNA levels in monocytes relative to nondiabetic or WT mice. The increased mtDNA release in the cytosol of diabetic and PPARα^−/− monocytes may represent a mechanism for the cGAS-STING pathway activation and monocyte activation induced by diabetes or PPARα deficiency.

In summary, the current study explored the interplay between mitochondria and the innate immune response in DR. We demonstrated a new molecular mechanism through which PPARα regulates the activation of monocytes by directly mediating bioenergy metabolism and protecting mitochondrial integrity and function, especially under diabetic conditions. Our results also revealed a new functional interaction of PPARα with the cGAS-STING pathway. These findings suggest that modulating the cGAS-STING pathway may become a novel therapeutic strategy for the treatment of DR.

Acknowledgments. The authors thank Juping Liu (Eye Institute and School of Optometry, Tianjin Medical University Eye Hospital, Tianjin, China) for providing samples from patients.

Funding. L.D. received funding from a National Natural Science Foundation of China grant (81570872). J.-x.M. received grants from the National Institutes of Health National Eye Institute (EY019309, EY033330, EY030472, EY012231, EY028949, EY032930, and EY032931). This study was also supported by the Diabetic Animal Core and Histology and Image Core of Diabetic COBRE (GM122744) and the Vision Core supported by the National Eye Institute (P30 EY021725).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.D. conducted experiments, acquired data, analyzed data, and wrote the manuscript. R.C. analyzed data and wrote the manuscript. X.M. and W.L. prepared samples. Y.H. and H.L. collected human samples. K.Z. and Y.D. prepared animals. Y.T., X.Z., and X.-r.L. provided reagents. J.-x.M. designed the studies, analyzed data, and wrote the manuscript. J.-x.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.