Investigating the Role of PPARβ/δ in Retinal Vascular Remodeling Using Pparβ/δ-Deficient Mice

Peroxisome proliferator-activated receptor (PPAR)β/δ is a member of the nuclear receptor superfamily of transcription factors, which plays fundamental roles in cell proliferation and differentiation, inflammation, adipogenesis, and energy homeostasis. Previous studies demonstrated a reduced choroidal neovascularization (CNV) in Pparβ/δ-deficient mice. However, PPARβ/δ’s role in physiological blood vessel formation and vessel remodeling in the retina has yet to be established. Our study showed that PPARβ/δ is specifically required for disordered blood vessel formation in the retina. We further demonstrated an increased arteriovenous crossover and wider venous caliber in Pparβ/δ-haplodeficient mice. In summary, these results indicated a critical role of PPARβ/δ in pathological angiogenesis and blood vessel remodeling in the retina.


PPARβ/δ Is Highly Expressed in the Mouse Retina
Genetic disruption of Pparβ/δ and specific PPARβ/δ antagonists have previously been shown to attenuate laser-induced CNV in mice [21]. Surprisingly, our study showed no detectable PPARβ/δ protein in the choroid/retinal pigment epithelium (RPE) fraction of C57BL/6 mouse eyes (Figure 1a). Retinal pigment epithelium-specific protein 65 kDa (RPE65) was only observed in the choroid/RPE fraction, verifying the isolation procedure. Consistently, retinal Pparβ/δ mRNA level was around 5-fold higher than that in choroid/RPE compartment (Figure 1b). Furthermore, the gene expression of Pparβ/δ remained unchanged in both retina ( Figure S1a) and choroid/RPE ( Figure S1b) compartments of C57BL/6 mice subjected to laser-induced CNV. Considering the high PPARβ/δ levels in normal mouse retina, an essential role of PPARβ/δ in fatty acid oxidation (FAO) [22,23], and the emerging role of FAO in EC metabolism and angiogenesis [13], we went on exploring the association between retinal Pparβ/δ expression and changes in the body's metabolic status. C57BL/6 mice were subjected to overnight fasting followed by refeeding with normal chow diet. Our study showed that retinal Pparβ/δ level was significantly higher in refed mice as compared to that in mice undergoing fasting (Figure 1c), which confirmed what was previously seen in liver and kidney [24]. The mouse retina is avascular at birth and the retinal vasculature is fully established within the first few weeks of life [25]. As FAO is an aerobic process, we next analyzed the expression of Pparβ/δ in fully vascularized postnatal day (P)21 retina and compared it to that in avascular P2 retina. As expected, Pparβ/δ was expressed at a much lower level in avascular hypoxic retina at P2 (Figure 1d).
formation of normal retinal vasculature [26]. The resulting central avascular retina becomes hypoxic once the mice are returned to room air at P12 [26]. Pparβ/δ expression was highly induced in the retina of mice subjected to 5 days of hyperoxia treatment as compared to that in age-matched C57BL/6 mice that had been kept under normoxic condition during this period (Figure 1e). Carnitine palmitoyltransferase 1A (Cpt1a) is a rate-limiting enzyme for FAO, and it has been reported to regulate angiogenesis [27]. Cpt1a is also a well-established target of PPARβ/δ [28]. Our data showed that Cpt1a was significantly induced in the retina of mice exposed to hyperoxia (Figure 1f). Similar to what we had observed in the developing retina, Pparβ/δ expression in hypoxic mouse retina at P13 was significantly lower than that in the retina of age-matched control mice that had been kept under normoxic condition ( Figure 1g). As expected, Cpt1a expression was also suppressed in hypoxic mouse retina ( Figure 1h). Together, these data showed a tight nutritional and environmental regulation of retinal Pparβ/δ levels. Figure 1. Peroxisome proliferator-activated receptor (PPAR)β/δ is highly expressed in mouse retina. (a) Representative Western blot of PPARβ/δ, retinal pigment epithelium-specific protein 65 kDa (RPE65) and glyceraldehyde-3-phosphate dehydrogenase (GADPH) in the retina (n = 4) and choroid/RPE (n = 4) compartments of adult C57BL/6 mice. Relative gene expression of Pparβ/δ in (b) the retina (n = 10) and choroid/RPE (n = 3) compartments of C57BL/6 mice; (c) the retina of fasted (n = 4) and re-fed (n = 4) adult C57BL/6 mice; (d) the retina of P2 (n = 4) and P21 (n = 4) C57BL/6 mice, as determined by quantitative real-time polymerase chain reaction (RT-qPCR) analysis. Relative expressions of (e) Pparβ/δ and (f) Carnitine palmitoyltransferase 1A (Cpt1a) in P12 hyperoxic retina of C57BL/6 mice subjected to oxygen-induced retinopathy (OIR) (n = 4) as compared to those in agematched normoxic retina (n = 4), as determined by RT-qPCR analysis. Relative expressions of (g) Pparβ/δ and (h) Cpt1a in P13 hypoxic retina of C57BL/6 mice subjected to OIR (n = 4) as compared to those in age-matched normoxic retina (n = 4), as determined by RT-qPCR analysis. Gene expressions are quantified relative to the housekeeping gene, Gadph, as detailed in the Materials and Methods. Data are expressed as mean ± standard error of the mean (SEM). Unpaired, two-tailed t-test was used for statistical analysis; *** p < 0.001, * p < 0.05.
The mouse model of oxygen-induced retinopathy (OIR) is able to reliably reproduce the defining characteristics of retinopathy of prematurity in human. Exposing P7 pups to 75% of oxygen for 5 continuous days not only causes the regression of immature retinal vessels but also prevents the formation of normal retinal vasculature [26]. The resulting central avascular retina becomes hypoxic once the mice are returned to room air at P12 [26]. Pparβ/δ expression was highly induced in the retina of mice subjected to 5 days of hyperoxia treatment as compared to that in age-matched C57BL/6 mice that had been kept under normoxic condition during this period (Figure 1e). Carnitine palmitoyltransferase 1A (Cpt1a) is a rate-limiting enzyme for FAO, and it has been reported to regulate angiogenesis [27]. Cpt1a is also a well-established target of PPARβ/δ [28]. Our data showed that Cpt1a was significantly induced in the retina of mice exposed to hyperoxia (Figure 1f). Similar to what we had observed in the developing retina, Pparβ/δ expression in hypoxic mouse retina at P13 was significantly lower than that in the retina of age-matched control mice that had been kept under normoxic condition ( Figure 1g). As expected, Cpt1a expression was also suppressed in hypoxic mouse retina ( Figure 1h). Together, these data showed a tight nutritional and environmental regulation of retinal Pparβ/δ levels.
In our hand, around 18% of Pparβ/δ −/− mice survived into adulthood. Pparβ/δ-haplodeficient mice were used in most cases to study the functional role of PPARβ/δ in retinal angiogenesis because their phenotype corresponded to some human retinal diseases (see below). Retinal vasculature was visualized in posterior eyecup whole mounts by immunofluorescence staining with a vascular endothelial cell marker, CD31. Quantitative analysis showed no difference in vascular density between Pparβ/δ +/− and wild-type littermate controls ( Figure 2a). Next, to mimic new blood vessel formation in vivo, aorta rings were prepared from P10 Pparβ/δ +/− and wild-type littermate controls. There was no difference in vessel outgrowth from explanted aortic rings between Pparβ/δ +/− and wild-type counterparts (Figure 2b). This observation was further supported using choroidal angiogenesis assay in which vessel outgrowth from Pparβ/δ +/− and wild-type choroid explants were analyzed and showed no difference (Figure 2c). mice were used in most cases to study the functional role of PPARβ/δ in retinal angiogenesis because their phenotype corresponded to some human retinal diseases (see below). Retinal vasculature was visualized in posterior eyecup whole mounts by immunofluorescence staining with a vascular endothelial cell marker, CD31. Quantitative analysis showed no difference in vascular density between Pparβ/δ +/− and wild-type littermate controls ( Figure 2a). Next, to mimic new blood vessel formation in vivo, aorta rings were prepared from P10 Pparβ/δ +/− and wild-type littermate controls. There was no difference in vessel outgrowth from explanted aortic rings between Pparβ/δ +/− and wildtype counterparts (Figure 2b). This observation was further supported using choroidal angiogenesis assay in which vessel outgrowth from Pparβ/δ +/− and wild-type choroid explants were analyzed and showed no difference ( Figure 2c).  To confirm these observations, wild-type aortic rings and choroid explants were treated with a novel PPARβ/δ antagonist, 10h, which specifically antagonizes the agonist-mediated transcriptional activity of PPARβ/δ and not the ligand activation of the two other PPAR isotypes [31]. The 10h-mediated PPARβ/δ inhibition was confirmed by evaluating the expression of PPARβ/δ target genes in human retinal microvascular ECs. Our results demonstrated a significant reduction in Angiopoietin-like 4 (ANGPTL4), CPT1A, and Pyruvate Dehydrogenase Kinase 4 (PDK4) gene expression levels ( Figure 3a). However, no change in vessel outgrowth was observed in 10h-treated aortic rings ( Figure 3b) and choroid explants ( Figure 3c) as compared to those treated with dimethyl sulfoxide (DMSO) vehicle control. Together, our data showed that normal blood vessel formation was not affected by Pparβ/δ-haplodeficiency and in the presence of the specific PPARβ/δ antagonist 10h.
To confirm these observations, wild-type aortic rings and choroid explants were treated with a novel PPARβ/δ antagonist, 10h, which specifically antagonizes the agonist-mediated transcriptional activity of PPARβ/δ and not the ligand activation of the two other PPAR isotypes [31]. The 10hmediated PPARβ/δ inhibition was confirmed by evaluating the expression of PPARβ/δ target genes in human retinal microvascular ECs. Our results demonstrated a significant reduction in Angiopoietin-like 4 (ANGPTL4), CPT1A, and Pyruvate Dehydrogenase Kinase 4 (PDK4) gene expression levels ( Figure 3a). However, no change in vessel outgrowth was observed in 10h-treated aortic rings ( Figure 3b) and choroid explants ( Figure 3c) as compared to those treated with dimethyl sulfoxide (DMSO) vehicle control. Together, our data showed that normal blood vessel formation was not affected by Pparβ/δ-haplodeficiency and in the presence of the specific PPARβ/δ antagonist 10h. Scale bar: 200 µ m. Data are expressed as mean ± SEM. Unpaired, two-tailed t-test was used for statistical analysis; *** p < 0.001.

PPARβ/δ Mediates Blood Vessel Remodeling
Although there was no difference in retinal vessel density between Pparβ/δ +/− and wild-type control mice at P10, a closer look at the retinal microvasculature revealed an increased incidence of crossover of the radial arteries and veins and of their side branches in the Pparβ/δ +/− mice (Figure 4a), a feature that is closely associated with branch RVO in human [3,32,33]. We further observed a wider caliber of radial veins but not radial arteries in Pparβ/δ +/− mice (Figure 4b), which is also observed in human patients with branch RVO [34]. Consistent with the observation in Pparβ/δ +/− mice, no change in retinal vessel density was observed in adult Pparβ/δ −/− mice; however, there was a significant reduction in pericyte coverage (Figure 4c). To support these observations, we investigated the expression of a panel of vascular markers that are involved in angiogenesis and blood vessel remodeling. As expected, there was no change in the expression of the genes for key angiogenic markers, including Vegf (Figure 4d

PPARβ/δ Is Specifically Required for Pathological Retinal Angiogenesis
As our data thus far had demonstrated an altered Pparβ/δ expression in the retina under the hypoxic and hyperoxic conditions, we investigated whether intra-retinal/pre-retinal neovascularization, as observed in PDR, was affected in Pparβ/δ +/− mice subjected to OIR. Consistent with our observation in neonatal and adult mice, there was no significant difference in physiological neovascularization between Pparβ/δ +/− and wild-type animals as demonstrated by percentage of avascular area at P17 (Figure 5a). However, the area occupied by disordered neovascular growth (tufts), either as small isolated neovascular tissue or as large continuous areas lying on the surface of the retina, was significantly reduced in Pparβ/δ +/− mice (Figure 5b). This data suggests that PPARβ/δ was specifically required for disordered blood vessel formation in the mouse retina.

PPARβ/δ is Specifically Required for Pathological Retinal Angiogenesis
As our data thus far had demonstrated an altered Pparβ/δ expression in the retina under the hypoxic and hyperoxic conditions, we investigated whether intra-retinal/pre-retinal neovascularization, as observed in PDR, was affected in Pparβ/δ +/− mice subjected to OIR. Consistent with our observation in neonatal and adult mice, there was no significant difference in physiological neovascularization between Pparβ/δ +/− and wild-type animals as demonstrated by percentage of avascular area at P17 (Figure 5a). However, the area occupied by disordered neovascular growth (tufts), either as small isolated neovascular tissue or as large continuous areas lying on the surface of the retina, was significantly reduced in Pparβ/δ +/− mice (Figure 5b). This data suggests that PPARβ/δ was specifically required for disordered blood vessel formation in the mouse retina.

Discussion
Abnormal blood vessel formation is a defining feature of many blinding eye diseases, including PDR, nAMD, and RVO. Current targeted treatment options for these diseases focus on a single angiogenic pathway VEGF. However, a substantial number of patients are not responsive to the treatment or may develop resistance over time [35,36]. It is not surprising as angiogenesis is a highly complex and dynamic process involving extensive interactions between different types of cells, growth factors, and extracellular matrix components [37]. The inhibition of one angiogenic pathway may be compensated by the activation of alternative angiogenic pathways [5,6]. Furthermore, accelerated fibrovascular progression and potential neural toxicity associated with long-term anti-VEGF treatment pose a significant concern. There is an urgent need for alternative or complementary treatments to anti-VEGF therapeutics.
Although ECs are traditionally believed to rely on anaerobic glycolysis for adenosine triphosphate (ATP) production [38], FAO has been shown to modulate EC, especially stalk cell, proliferation, and angiogenesis [39]. Targeting EC metabolism may, therefore, offer an alternative strategy to control abnormal blood vessel formation. The PPAR transcription factors play key roles in cell metabolism [40], and all three PPAR isotypes are reported to regulate angiogenesis [17]. However, PPARβ/δ's role in new blood vessel formation, especially those in the eye, is less well characterized.
A previous study reported a reduced lesion size in Pparβ/δ −/− mice subjected to laser-induced CNV [21]. Surprisingly, our study could not detect PPARβ/δ in the RPE/choroid fraction of C57BL/6 mouse eyes. Furthermore, the expression levels of PPARβ/δ was not affected in retina and choroid/RPE fractions of eyes subjected to laser-induced CNV. It is likely that PPARβ/δ activation instead of its expression is involved the development of CNV. Further experiments should be carried out to evaluate the expression of PPARβ/δ target genes, such as Angptl4, in laser-treated mouse eyes. As the focus of this study was PPARβ/δ, we have not explored the expression of other PPAR isotypes in the eyes following nutritional or hyperoxic challenges. However, hypoxia was reported to suppress the expression of PPARα in intestinal epithelial cells [41] and PPARγ in proximal renal tubular cells [42]. On the other hand, PPARα is induced in mouse heart and liver during fasting [43]. Based on these observations, PPARα and PPARγ are likely to be regulated by metabolic modifications in the eye.
Since PPARβ/δ was expressed in mouse retina at a relatively high level and its expression was affected by the body's metabolic status, we went on to evaluate the role of PPARβ/δ in retinal angiogenesis. Our study suggests that Pparβ/δ-haplodeficiency had no impact on normal blood vessel formation as demonstrated in the retina of neonatal and adult mice, as well as in ex vivo angiogenesis assays. This observation was supported by treatment of wild-type aortic ring and choroid explants with the PPARβ/δ antagonist, 10h. In line with this observation, the expression of key angiogenic factors, including VEGF and TGFβ, were not affected in the retina of Pparβ/δ-null mice. Similar as in other vital organs, the arteries in the retina are responsible for taking oxygen and nutrient-rich blood to the retina, whereas veins carry the oxygen and nutrient depleted blood back to the lung and heart. Arteriovenous malformation disrupts this critical process. Although there was no change in retinal vascular density, Pparβ/δ +/− mice demonstrated an increased incidence of arteriovenous crossover in the retina, a feature that is commonly associated with hypertensive retinopathy and RVO in human [26,44,45]. Due to the low survival rate of Pparβ/δ −/− mice, Pparβ/δ-haplodeficient mice were used in most experiments for this study. To circumvent high lethality rate, mouse models with ocular cell-specific knockout of Pparβ/δ could be used to study the tissue-specific functions of PPARβ/δ.
We further observed an increase in venous caliber in Pparβ/δ +/− mice. Retina venous dilation is closely associated with a range of macro-and micro-vascular disorders, such as cerebrovascular diseases, including stroke [46][47][48][49], coronary heart disease [50,51], diabetic retinopathy, and diabetic nephropathy [46,52,53]. As changes in retinal vasculature could be visualized non-invasively and directly in vivo, and these changes are believed to serve as 'markers' of systemic disorders [54], it may be worth evaluating blood pressure, as well as the presence of other vascular complications, in PPAR-deficient mice in the future. Furthermore, external stimuli, such as hypoxia and oxidative stress, have been shown to contribute to the development of both macro- [55] and micro- [56] vascular complications. Considering the role of PPARβ/δ in ocular vessel remodeling, it is worth exploring its role in other vascular beds, including the heart and the brain. PPARβ/δ has previously been implicated in regulating vascular permeability [57]. In the retina, pericytes play a key role in maintaining blood vessel integrity and permeability. Our study demonstrated a reduced pericyte coverage in Pparβ/δ −/− retina. This observation was further supported by reduced gene expression of Pdgfrβ, which plays an important role in pericyte recruitment, in Pparβ/δ −/− retina. Nevertheless, future studies verifying the change of PDGFRβ at the protein level would be necessary to confirm this observation. The role of PPARβ/δ in pericyte signaling and function, pericyte-EC interaction, as well as the crosstalk between 9 of 15 PPARβ/δ and PDGFRβ, also warrants further investigation. Finally, Pparβ/δ deletion in host tissue specifically inhibits the formation of pathological neovascularization tufts but has no impact on normal blood vessel regrowth. Previous studies indicate different metabolic profile in physiological and pathological vessels [58]. Considering an important role of PPARβ/δ in energy metabolism, it will be interesting to see whether it is involved in the metabolic switch associated with the progression of pathological neovascularization. Together, our data demonstrated a critical role of PPARβ/δ in retinal blood vessel remodeling and pathological angiogenesis.

RNA Extraction and Quantitative Real-Time PCR
Mouse retinal and choroid/RPE tissues were immediately dissected after enucleation and were snap frozen in liquid nitrogen. Total RNA was extracted from homogenized mouse tissues using NucleoSpin RNA, Mini kit (Macherey-Nagel, Düren, Germany) following manufacturers' instructions. cDNA was synthesized using qScript cDNA Supermix (Quanta BioSciences, Beverly, MA, USA) according to manufacturer's instructions. Quantitative real-time PCR was performed in a total volume of 20 µL containing PrecisionFAST 2× qPCR Mastermix (with SYBR green and low ROX) (Primerdesign, Camberley, UK) using a QuantStudio™ 6 Flex Real-Time PCR System (Thermo Fisher Scientific, USA). Primers sequences used in this study are listed in (Table 2). Primers utilized for human ANGPTL4, CPT1A, and PDK4 were PrimeTime qPCR primers from Integrated DNA Technologies, where their Primer Assay IDs were Hs.PT.58.25480012, Hs.PT.58.2799026, and Hs.PT.58.28212793, respectively. Except for where the expression of Pparβ/ δ in the mouse retina and choroid/RPE was normalized to β-actin, all gene expressions were quantified relative to Gapdh in the quantitative real-time polymerase chain reaction (RT-qPCR) analysis.

Mouse Model of Laser-Induced Choroidal Neovascularization
CNV was induced in wild type mice as described [60], and the eyes were harvested from the mice 21 days post laser. The retinae were immediately dissected and snap frozen for gene expression analysis.

Mouse Model of Oxygen-Induced Retinopathy (OIR)
OIR was induced as described [61]. In brief, P7 mice and their nursing mothers were placed in a 75% oxygen supply chamber for 5 days and exposed to a standard 12 h light-dark cycle. Mice were returned to room air at P12, and the retinae were collected at P17 for immunohistochemistry. The retinae for gene expression analysis were harvested at P12 (immediately after removal of mice from the hyperoxic chamber) and at P13 (24 h after return to room air).

Aortic Ring Assay
The aortic ring assay was performed as described [60]. Aorta from P10 wild type and Pparβ/δ +/− mice were cut into 1mm rings before being embedded in a 96-well plate coated with rat tail collagen I gel (BD Biosciences, USA). Explants were treated with 50ng/mL VEGF and the media changed every other day. Explants from P3 wild type aorta were treated with 100nM of 10h or DMSO vehicle control. At day 10 of culture, the explants were fixed in 4% PFA and stained with Griffonia Simplicifolia isolectin B4 (IB4) (Vector Lab, Burlingame, CA, USA). Vessel outgrowth was imaged using Eclipse Ti-E Inverted Research Microscope (Nikon, Tokyo, Japan). The IB4 + areas were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Choroid Sprouting Assay
The choroid sprouting assay was performed as described [63] using P3 mice. Choroid explants were treated with 100 nM 10 h or vehicle control. After three days, vessel outgrowth was imaged using Eclipse Ti-E Inverted Research Microscope (Nikon, Japan) and quantified using ImageJ software (National Institutes of Health, USA).

Statistical Analysis
All data are shown as mean ± SEM. Four independent repeats were carried out for each experiment unless stated otherwise. One-way ANOVA followed by Turkey's post hoc analysis or unpaired two-tailed Student's t-test was used to determine the statistical significance using GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA). * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001.

Conclusions
While the deletion or inhibition of PPARβ/δ did not affect normal blood vessel formation, PPARβ/δ appeared to be specifically required for disordered blood vessel formation in the retina. Pparβ/δ-haplodeficiency was sufficient to cause an increased arteriovenous crossover and a wider venous caliber in neonatal mouse retina. Finally, complete removal of Pparβ/δ leads to reduced pericyte coverage in adult mice retina. Taken together, these findings highlight a critical role of PPARβ/δ in pathological angiogenesis and blood vessel remodeling in the retina.