Impaired coupling of adipose tissue expansion and vascularization is proposed to lead to adipocyte hypoxia and inflammation, which in turn contributes to systemic metabolic derangements. Pigment epithelium-derived factor (PEDF) is a powerful antiangiogenic factor that is secreted by adipocytes, elevated in obesity, and implicated in the development of insulin resistance. We explored the angiogenic and metabolic role of adipose-derived PEDF through in vivo studies of mice with overexpression of PEDF in adipocytes (PEDF-aP2). PEDF expression in white adipocytes and PEDF secretion from adipose tissue was increased in transgenic mice, but circulating levels of PEDF were not increased. Overexpression of PEDF did not alter vascularization, the partial pressure of O2, cellular hypoxia, or gene expression of inflammatory markers in adipose tissue. Energy expenditure and metabolic substrate utilization, body mass, and adiposity were not altered in PEDF-aP2 mice. Whole body glycemic control was normal as assessed by glucose and insulin tolerance tests, and adipocyte-specific glucose uptake was unaffected by PEDF overexpression. Adipocyte lipolysis was increased in PEDF-aP2 mice and associated with increased adipose triglyceride lipase and decreased perilipin 1 expression. Experiments conducted in mice rendered obese by high-fat feeding showed no differences between PEDF-aP2 and wild-type mice for body mass, adiposity, whole body energy expenditure, glucose tolerance, or adipose tissue oxygenation. Together, these data indicate that adipocyte-generated PEDF enhances lipolysis but question the role of PEDF as a major antiangiogenic or proinflammatory mediator in adipose tissue in vivo.
- pigment epithelium-derived factor
- insulin sensitivity
adipocytes secrete proteins and metabolites that play major roles in the regulation of metabolism, appetite, energy expenditure, and inflammation. Indeed, defects in adipocyte secretory function contribute to type 2 diabetes and cardiovascular disease. This has led to major efforts to identify the adipocyte secretome (1, 13, 23, 28), the actions of these secreted proteins, and their involvement in obesity-associated pathologies.
Several years ago, we identified pigment epithelium-derived factor (PEDF) as a protein, secreted from adipocytes, that links obesity to insulin resistance (13). Levels of PEDF in adipocytes and serum are increased in obesity and reduced by weight loss in mice. PEDF administered to cultured cells (20) and lean mice causes insulin resistance, and neutralizing PEDF in obese mice enhances insulin's actions (13). PEDF's actions extend to modulation of lipid metabolism, where it interacts with adipose triglyceride lipase (ATGL) to increase lipolysis and reduce fatty acid oxidation (5, 10), which in turn contributes to ectopic lipid deposition and is associated with proinflammatory signaling. The finding that PEDF induces prominent features of the metabolic syndrome is notable because PEDF levels are elevated in human obesity and diabetes in association with cardiometabolic dysfunction (9, 25, 26, 37). Moreover, a common genetic variant in the gene locus encoding PEDF contributes to adiposity and obesity-related insulin resistance (4).
Angiogenesis, and thus vascularization of adipose tissue, is at least partly regulated by the balanced release of pro- and antiangiogenic proteins secreted from adipocytes and cells of the stromal vascular fraction (7). Adipose tissue expansion accompanied by insufficient neovascularization is linked to an inadequate supply of nutrients, growth factors, and oxygen, which is proposed to induce sequelae, including hypoxia, adipocyte cell death, chronic low-grade inflammation, fibrosis, and insulin resistance (21, 24, 39). Improving the vascular network in adipose tissue by overexpressing the dominant angiogenic factor vascular endothelial growth factor (VEGF) protects mice against obesity-associated metabolic derangements. This observation indicates that metabolic imbalances of obesity can be reversed by enhancing adipose tissue vascularity and thus reducing hypoxia (19, 34, 35). PEDF was identified originally as a neurotrophic factor secreted by retinal pigment epithelial cells but has since been shown to regulate diverse functions, including stem cell maintenance, miRNA inhibition, cell survival and differentiation, migration, and invasion (11). Most prominently, PEDF is a powerful inhibitor of angiogenesis. This is evidenced by the observation that Serpinf1−/− mice (i.e., PEDF-null) demonstrate an angiogenic phenotype characterized by substantial stromal vascularity and epithelial cell hyperplasia in the retina, prostate, and pancreas (17). Because adipocyte PEDF expression and secretion are increased in obesity (12), we postulate that PEDF disrupts the local balance between pro- and antiangiogenic factors in adipose tissue, which contributes to the metabolic and proinflammatory derangements of obesity.
In the present study, we tested the hypothesis that adipose-generated PEDF can reduce adipose tissue vascularity and in turn induce adipocyte hypoxia, inflammation, and insulin resistance. We generated transgenic mice to test whether adipocyte-derived PEDF reduces adipose tissue vascularity, thus promoting local tissue hypoxia and inflammation. Simultaneously, we examined the metabolic consequences of PEDF overexpression on adipocyte and systemic substrate metabolism.
RESEARCH DESIGN AND METHODS
The Monash University School of Biomedical Sciences Animal Ethics committee approved all procedures. Transgenic mice expressing the hemagglutinin (HA)-tagged human SerpinF1 cDNA under the control of the aP2 promoter were generated by cloning full-length human Serpinf1 cDNA (amplified using the 5′-ATC TGG GAA TTC TGC AGC ACG ATG TCT GGT TAC CCA TAC GAT GTT CCA GAT TAC GCT GTG TCT GAA GAA GAG GCG GCT-3′ forward primer and 5′-TA TAT AGC GGC CGC TTA GGG GCC CCT GGG GTC CAG AAT-3′ reverse primer). A HA tag was attached to the NH2 terminus of PEDF. A Bluescript 2 SK+ vector containing the aP2 promoter was purchased from Addgene (plasmid no. 11424), and the HA-PEDF was inserted into the aP2 vector at the EcoR1 and BamHI sites (Fig. 1A). The transgene was isolated by agarose gel electrophoresis and purified (Promega Wizard SV Gel and PCR Clean Up System). The transgene was injected into the pronucleus of C57BI/6J zygotes and transferred into pseudopregnant female CD1 mice (Monash Animal Services). Founders were identified by PCR analysis of genomic DNA from tail tips of the offspring carrying the HA-PEDF-aP2 transgene. Mice were crossed with C57BI/6J mice for four generations. For phenotyping, mice with HA-PEDF-aP2 were hemizygous for the integrated transgene, and littermates were used as controls [wild type (WT)].
Male mice were housed under controlled temperature (22°C) and lighting (12-h light-dark cycle) and had ad libitum access to food and water. Mice were fed a low-fat diet (19.6% calories from protein, 4.6% from fat, and 4.8% from crude fibre and 14.3 MJ/kg digestible energy; Specialty Feeds) until they were euthanized at 20 wk of age. Some mice were fed a high-fat diet for 20 wk prior to assessment of whole body metabolism and adipose tissue oxygenation (19.5% protein, 36% fat, 4.7% crude fibre, 4.7% acid detergent fibre, 22.8 MJ/kg digestible energy; SF03-002, Specialty Feeds).
Body composition, whole body in vivo metabolism, and glucose metabolism.
d-glucose or 0.7 U/kg insulin (Actrapid; Novo Nordisk) and assessment of blood glucose (Accucheck) from the tail vein at 15-min intervals. Plasma lipids were assessed in 4-h-fasted mice at 1100.
Measurement of plasma free fatty acids (FFA) (NEFA C; Wako Pure Chemical Industries, Osaka, Japan), triglycerides (GPO-PAP Reagent; Roche Diagnostics), and glucose (Accuchek; Roche) was performed according to the manufacturers' instructions. Plasma insulin was determined by ELISA (Department of Physiology, Monash University Core).
Immunohistochemistry and adipocyte size.
Epididymal adipose tissues were embedded in paraffin, cut into 4-μm sections, and stained with haematoxylin and eosin. High-resolution images were captured (Aperio ScanScope AT Turbo, Aperio, CA), and Aperio ImageScope software was used to determine the cross-sectional area of the adipocytes (600 adipocytes from 3 areas/mouse).
Endothelial cell labeling.
Mice received 0.1 mg of rhodamine-tagged lectin-1 (Vector Laboratories, Burlingame, CA) via the tail vein. After 4 min, anesthetized mice (isoflurane) were perfusion fixed by transcardial infusion of saline, followed by 1% paraformaldehyde for 5 min. Adipose tissue was dissected and postfixed in 10% PBS-buffered formalin for 24 h. Tissues were embedded in paraffin, cut into 5-μm sections, and mounted. Images were obtained using a Zeiss AxioImager Z1 fluorescence microscope (Carl Zeiss, Oberkochen, Germany). AxioVision software was used to count blood vessels.
Assessment of adipose Po2.
Anesthetized mice (2.5% isoflurane) were maintained at ∼37°C through the experiments. A MouseOx pulse oximeter sensor (Starr Life Sciences) was attached to the hindleg throughout the procedures to measure heart rate, respiratory rate, and percentage of saturation of hemoglobin with oxygen (So2). An OxyLite fiber-optic probe (Oxford Optronix, Oxford, UK) was inserted centrally ∼1–2 mm into each fat pad to measure local temperature, and a Clark electrode with a 50-μm tip (Unisense, Aarhus, Denmark) was used to measure Po2 in the retroperitoneal and epididymal fat pads. Data were digitized using custom software running in LabView 7.1 (Universal Analysis, Circulatory Control Laboratory, Department of Physiology, University of Auckland). Po2 was measured in four evenly spaced locations within each fat pad, starting proximally and moving distally. The Clark electrode was placed 1–2 mm below the surface and held in position until a stable reading was obtained over 30 s and the average Po2 recorded. Upon conclusion of Po2 measurements, blood was taken from the descending aorta and analyzed for pH, Pco2, Po2, So2 and glucose using an i-STAT CG8+ cartridge for blood oximetry (Abbott Point of Care, Princeton, NJ).
Assesment of hypoxia.
Mice received an intraperitoneal injection of pimonidazole hydrochloride (Hypoxyprobe-1; Hypoxyprobe) in saline (60 mg/kg body wt) and were euthanized 30 min later by cervical dislocation under isoflurane anaesthesia. Epididymal and retroperitoneal fat pads were excised and fixed in PBS-buffered formalin for 24 h. Tissues were embedded in paraffin, cut into 4-μm sections, and mounted on slides (3 sections/sample) using Vectashield Mounting Media for Fluorescence (Vector Laboratories).
Total RNA was extracted from tissues using Qiazol extraction reagent (Qiagen). Reverse transcription of 1 μg of mRNA was performed (iScript cDNA Synthesis Kit; Bio-Rad Laboratories, Hercules, CA), and gene products were determined by real-time quantitative RT-PCR (Realplex Mastercycler; Eppendorf) using the TaqMan Universal PCR Master Mix and TaqMan Gene Expression Assays. The relative quantification was calculated using the ΔΔCT method.
Tissues were homogenized (PRO Scientific, Oxford, CT) in ice-cold RIPA buffer (65 mM Tris, 150 mM NaCl, 1 M dithiothreitol, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and 10% glycerol) with protease inhibitors (Complete Protease Inhibitor Cocktail; Roche, Basel, Switzerland) and phosphatase inhibitors (PhosSTOP Phosphatase Inhibitor Cocktail; Roche). Equal amounts of solubilized proteins were resolved by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and blocked in 2.5% bovine serum albumin. Images of the membranes were captured immediately posttransfer to ensure equal protein loading. Membranes were probed with polyclonal antibodies overnight at 4°C, washed, and exposed to secondary antibodies. The PEDF antibody was provided by E. Duh (John Hopkins University), ATGL (no. 2138) and hormone-sensitive lipase (HSL; no. 4107) were from Cell Signaling Technology, perilipin 1 (P1873) was from Sigma, and comparative gene identification (CGI)-58 (sc-100468) was from Santa Cruz Biotechnology. Immunoreactive protein bands were visualized using enhanced chemiluminescence (Chemiluminescent Peroxidase Substrate-3; Sigma-Aldrich) and imaged on a ChemiDoc MP Imaging System (Bio-Rad Laboratories). Band intensity was quantified using densitometry in Image Lab software (Bio-Rad Laboratories).
Protein secretion ex vivo.
Epididymal adipose tissue was cut into 4 × ∼25-mg pieces and incubated for 4 h in in three volumes of protein-free medium (EX-CELL 325 Protein-Free CHO Serum-Free Medium; SAFC Biosciences) supplemented with 8 mM glucose and 500 nM adenosine. The medium was collected and subjected to SDS-PAGE and immunoblot analysis for PEDF.
GraphPad Prism 6.0 software (GraphPad Software) was used for statistical analysis. A Student t-test or two-way ANOVA with post hoc Bonferroni analysis was used to assess effects between groups (as indicated in results). Statistical significance was established at two-tailed P < 0.05, and data are reported as means ± SE.
Generation of transgenic mice with adipose tissue-specific expression of PEDF.
PEDF-aP2 mice were born at the expected Mendelian ratio, were fertile, and showed no gross abnormalities. PEDF mRNA expression (Fig. 1B) and protein content (Fig. 1D) were increased in white adipose tissue (WAT) of mice, and protein expression in brown adipose tissue tended (P = 0.09) to be greater than that in WT mice (Fig. 1D). The increased expression was in adipocytes and not cells of the stromal vascular fraction (Fig. 1E). Importantly, the 2.5-fold increase in WAT PEDF is of similar magnitude to increases induced in obesity (13), providing physiological relevance to subsequent experiments. The aP2 promoter fragment used for adipose-specific transgene expression has been used extensively to examine adipocyte function. However, there are reports that it drives the expression of target genes in other cell types. PCR analysis revealed robust HA expression in adipose tissues and low expression in the heart and kidney (Fig. 1C). The extra-adipocyte expression could be due to integration of the transgene into these tissues or adipocyte contamination. Irrespective of this, immunoblot analysis indicated no significant differences in PEDF protein content in the heart or kidney, or any other nonadipocyte tissue, in PEDF-aP2 compared with WT mice (data not shown). Thus, overexpression of PEDF was restricted largely to adipocytes. PEDF secretion was increased 1.7-fold in adipose tissue explants obtained from PEDF-aP2 compared with WT mice (Fig. 1F). The 2.5-fold increase in adipocyte PEDF content and 1.7-fold increase in secretion did not coincide with a significant increase in plasma PEDF levels in mice (Fig. 1G).
Adipocyte-specific overexpression of PEDF and angiogenesis.
The growth and expansion of adipose tissue is angiogenesis dependent. The primary aim of these experiments was to determine whether PEDF is a major antiangiogenic factor in adipocytes that reduces local vascularization and leads to the proposed sequalae of adipose tissue hypoxia and inflammation. Mice were administered rhodamine-tagged lectin-1, which labels endothelial cells. Vascular density was determined in sections of epididymal WAT from lean WT and PEDF-aP2 mice by immunofluorescence imaging. The number of blood vessels per section and the average vascular density were not significantly different in WT compared with PEDF-aP2 mice (Fig. 2A). In addition, we did not detect a difference in the levels of the endothelial cell marker Cd31, determined by quantitative PCR, between the two genotypes (Fig. 2B).
Although our analysis indicated that vascular density was not compromised by PEDF overexpression, we could not completely discount the possibility of subtle differences in microvascular perfusion and, therefore, tissue oxygenation. Accordingly, adipose tissue Po2 was assessed directly in epididymal and retroperitoneal WAT of lean WT and PEDF-aP2 mice. Arterial So2, Po2, and Pco2 were all similar in WT compared with PEDF-aP2 mice (Table 1). Tissue Po2 averaged 32 mmHg in the epididymal fat of WT mice and was similar to that of PEDF-aP2 mice (Fig. 2C). Although retroperitoneal Po2 tended to be higher than epididymal Po2 (18%, P = 0.08, main effect), there was no significant effect of PEDF overexpression (Fig. 2C). Hypoxia has been defined as Po2 < 10 mmHg. Using this criterion, 3% of all Po2 measurements were considered hypoxic, and there was no significant between-group difference. Similarly, mRNA expression of the hypoxia-inducible gene Hif1α was not significantly altered by PEDF overexpression (Fig. 2D). Finally, local hypoxia in WAT was not detected by pimonidazole staining (hydroxyprobe-1) in either WT or PEDF-aP2 mice (not shown).
The balance between pro- and antiangiogenic factors regulates blood vessel formation. We assessed the expression of several prominent modulators of angiogenesis in WAT. Notably, these could be derived from adipocytes or other inflammatory and stromal cells present in adipose tissue (6). The expression of the proangiogenic factors vascular endothelial growth factor A (Vegfa), angiopoietin-like 4, angiopoietin-1, and leptin was similar in the two genotypes, whereas platelet-derived growth factor was greater in PEDF-aP2 mice (Fig. 2E). Taken collectively, these data suggest that angiogenesis is well balanced in normally expanding tissues in response to mild PEDF overexpression.
Since previous reports have shown that PEDF activates proinflammatory signaling in muscle and liver (12, 20), we postulated that PEDF overexpression would increase local inflammation in adipose tissue. There was no evidence of increased macrophage recruitment to adipose tissue as indicated by Cd68 expression (Fig. 2F). Expression of the prominent proinflammatory cytokines IL-6, TNFα, and IL-1β and the chemokine CCL2 was also not affected significantly by overexpression of PEDF (Fig. 2F).
Overexpression of PEDF does not affect whole body energy metabolism or insulin sensitivity.
In light of previous findings suggesting that the adipocyte is a primary source of circulating PEDF (13, 20), we examined whether adipose-specific PEDF overexpression impacted adipocyte biology and whole body energy homeostasis. The body mass of PEDF-aP2 mice was similar to WT mice throughout development (Fig. 3A), with body masses tracking out to 28 wk (WT: 53 ± 3 g vs. PEDF-aP2: 56 ± 5 g). Lean mass and fat mass were assessed by DEXA and were not significantly different between groups (Fig. 3B). Histological analysis of WAT cross-sections revealed no significant difference in the mean adipocyte area (WT: 3.76 ± 0.58 μm2 vs. PEDF-aP2: 3.84 ± 0.48 × μm2 103), whereas the frequency distribution of the adipocyte area revealed the presence of more intermediate-sized adipocytes and fewer small adipocytes in mice overexpressing PEDF than in WT mice (Fig. 3C). Whole body oxygen consumption (V̇o2) and the respiratory exchange ratio were not influenced significantly by PEDF overexpression, indicating no alterations in total energy expenditure or the rates of carbohydrate and fatty acid oxidation (Fig. 3, D and E). Physical activity and food intake did not differ significantly between the groups (data not shown).
Insulin resistance is associated with elevated plasma PEDF in humans (32, 38), and PEDF causes glucose intolerance and insulin resistance in cultured cells and mice (5, 13, 20). Fasting blood glucose and insulin were similar in WT and PEDF-aP2 mice (Table 2), and adipose-specific PEDF overexpression did not significantly affect glucose tolerance (Fig. 4A). Plasma insulin levels tended to increase in response to the glucose load (P = 0.12, main effect for time) but were similar in the two genotypes (Fig. 4B). Whole body insulin action was assessed by intraperitoneal insulin injection and was not significantly different between WT and PEDF-aP2 mice (Fig. 4C). We postulated that PEDF originating from the adipocytes might exert an autocrine/paracrine influence on insulin sensitivity. Such an effect could occur even in the absence of altered circulating levels of PEDF. Both spontaneous and insulin-mediated glucose uptake were similar in the adipose tissue of PEDF-aP2 mice compared with WT mice (Fig. 4D). Collectively, these results demonstrate that PEDF overexpression in adipose tissue does not impact whole body energy metabolism and glucose metabolism.
Adipose-specific overexpression of PEDF increases adipocyte lipolysis by modulating ATGL and perilipin 1 contents.
PEDF is known to interact with the critical lipolytic protein ATGL to promote lipolysis (5, 10). Spontaneous lipolysis was 33% greater in adipose tissue explants of PEDF-aP2 compared with WT mice, whereas overexpression of PEDF had no significant effect on β-adrenoceptor-mediated lipolysis (Fig. 5A). In support of these findings, plasma glycerol levels were greater in PEDF-aP2 mice than in WT mice (Fig. 5B). There was no significant difference in plasma FFA (Fig. 5C), which may be due to concomitant increases in systemic fatty acid uptake in peripheral tissues. Immunoblot analysis showed that overexpression of PEDF regulates key modulators of adipocyte lipolysis. Consistent with the increased lipolysis, ATGL protein content was greater and perilipin 1 substantially less in PEDF-aP2 compared with WT mice. HSL and CGI-58 expression were similar in the two groups (Fig. 5D).
PEDF overexpression does not impact energy metabolism, glucose tolerance, or adipose tissue Po2 in high-fat-fed mice.
Impaired angiogenesis may be more relevant to pathological tissue growth (7). As such, we assessed the impact of PEDF on tissue Po2 in mice rendered obese by 20 wk of high-fat feeding. Body mass (Fig. 6A), fat mass (Fig. 6B), and whole body energy expenditure (Fig. 6C) were not different between genotypes. Consistent with the observations in mice fed a low-fat diet, PEDF-aP2 mice had increased basal lipolysis but similar β-adrenergic-stimulated lipolysis compared with WT mice (Fig. 6D). Glucose tolerance was not significantly different between PEDF-aP2 and Wt mice (Fig. 6E). Adipose Po2 was less in mice fed a high-fat diet compared with those fed a low-fat diet. However, in fat-fed mice, Po2 was similar between genotypes (Fig. 6F). Approximately 17% of all Po2 measurements were considered hypoxic (Po2 < 10 mmHg), and there was no significant between-group difference. In addition, we did not detect a difference in the levels of the endothelial cell marker Cd31, determined by quantitative PCR (Fig. 6G), or the hypoxia-inducible factor Hif1α between the two genotypes (Fig. 6H). Hence, overexpression of PEDF in adipose tissue did not appear to affect adipose tissue oxygenation in diet-induced obesity.
Adipocyte PEDF expression and secretion are increased in obesity, and PEDF is suggested be a key adipokine in driving obesity-mediated disorders such as diabetes (5, 13, 20, 32). PEDF is also an endogenous inhibitor of angiogenesis (16). Thus, we hypothesized that adipocyte-derived PEDF would reduce adipose tissue vascularity, induce hypoxia and local inflammation, and cause insulin resistance. We can now confidently reject this hypothesis. Indeed, the present study provides evidence that adipose-derived PEDF does not play a major role in regulating adipose tissue vascularity during physiological or pathological tissue growth. Furthermore, overproduction of adipocyte PEDF to levels reported in obesity (13) increases adipocyte lipolysis, but the increased fatty acid mobilization was insufficient to impact energy homeostasis or systemic insulin action. These results indicate that the predominant role of adipocyte-secreted PEDF is that of a positive modulator of spontaneous lipolysis.
Angiogenesis is an integral component of tissue health and growth, particularly in tissues that undergo marked remodeling throughout the lifespan, such as adipose tissue (31). Adipocytes and other cells of the adipose stromal vascular fraction secrete many proangiogenic molecules, including fibroblast growth factor, leptin, adiponectin, hepatocyte growth factor, angiopoeitin 2, and, most prominently, VEGF-A, which accounts for the majority of proangiogenic activity in adipose tissue (7, 40). Much less is known regarding the antiangiogenic factors secreted locally and the role they play in the regulation of angiogenesis in expanding adipose tissue. PEDF is a major endogenous inhibitor of angiogenesis, circulating levels of PEDF are tightly correlated with obesity (25, 26, 30, 32, 37, 38), and adipocytes are a major source of PEDF (13, 20). Studies of human primary adipocytes have demonstrated that PEDF secretion is eightfold greater than adiponectin secretion and 140-fold greater than VEGF secretion (20). Moreover, PEDF secretion from adipocytes is increased in obesity (13). Accordingly, we hypothesized that upregulating PEDF secretion would contribute to the development of metabolic dysfunction by blocking angiogenesis during physiological expansion of adipose tissue.
Overexpression of PEDF in adipose tissue was insufficient to influence vascular density or impact oxygen delivery and Po2 in normally developing adipose tissue. Although PEDF can downregulate the expression of VEGF and other proangiogenic factors in other tissues and tumors (2, 11), we observed no such effect in adipose tissue in vivo. Adipose tissue hypoxia in obesity is linked to inadequate angiogenesis (24), and PEDF is upregulated in obesity. We reasoned that PEDF's action might be context dependent and effective only in an obesogenic milieu. Although studies in obese mice demonstrated reduced adipose tissue Po2 compared with lean mice, this effect was no greater in obese PEDF-aP2 mice. Thus, the current study indicates that increasing PEDF in adipose tissue does not modulate angiogenic balance, and therefore vascularity, during adipose tissue expansion in lean or obese mice. The absence of an antiangiogeneic or angiostatic response in PEDF-aP2 mice may be due to compensatory upregulation of proangiogenic factors not examined in our analyses and/or to the fact that the increase in PEDF was insufficient to drive an antiangiogenic response. Although PEDF is unequivocally a powerful endogenous antiangiogenic factor, it is possible that its impact is tissue dependent. In this context, mice with global PEDF deletion display marked heterogeneity in tissue microvessel density, with small increases (1.4-fold) observed in the kidney and marked effects in the retina (5.2-fold) and prostate (3.2-fold) (17). Together, our studies indicate that PEDF does not contribute to reduced vascular density and is unlikely to contribute to adipose tissue dysfunction in obesity via hypoxia-related tissue remodeling.
Although it has been known for some time that spontaneous (basal) lipolysis is increased in obesity (27), the underlying mechanisms remain incompletely understood. ATGL is necessary and sufficient for PEDF-mediated lipolysis (5, 10); PEDF binds to the COOH terminus of ATGL (amino acids 268–504) and promotes its translocation to the lipid droplet for lipolysis (15). PEDF also reduces the expression of the ATGL inhibitor G0/G1 switch gene 2, which would promote triglyceride degradation (14). Our results extend on these studies and show that adipocyte-produced PEDF activates lipolysis concomitant with the novel observations of an increase in ATGL protein content and marked decrease in perilipin 1. Given that ATGL is the major triglyceride lipase (22) and perilipin 1 attenuates basal lipolysis (29), this provides a logical molecular explanation for the increased lipolysis in PEDF-aP2 mice. However, our current findings provide no conclusive support for this hypothesis, which remains speculative. Our finding of increased ATGL in PEDF-aP2 mice contrasts with those of a recent study by Dai et al. (14) that ascribed a role for PEDF in reducing ATGL by promoting ubiquitin-mediated proteasome degradation. These contrasting findings are difficult to reconcile, but the explanation may relate to between-study differences in the magnitude of PEDF expression, the source of PEDF, and/or the duration of PEDF administration. We detected a 2.5-fold change in PEDF content in adipocytes, which may be insufficient to drive proteasome degradation, whereas chronic increases in PEDF, as seen in PEDF-aP2 mice, could modulate other transcriptional/translational regulators of ATGL, which could overcome the acute inhibitory effects of PEDF. Irrespective of this, our data support the concept that the overproduction of PEDF in obesity (26, 36) contributes to the increased rates of spontaneous lipolysis in obesity (27).
Recent observations support a prominent role for PEDF in dysregulated lipid metabolism, insulin resistance, and inflammation (5, 8, 13, 15, 20) by direct actions on target tissues and indirectly by increasing lipolysis and in turn inducing lipotoxic stress (3, 33). Surprisingly, the metabolic profile of PEDF-aP2 mice was relatively normal under low-fat feeding and in diet-induced obesity, with no differences in adiposity, energy homeostasis, substrate partitioning, insulin action, or glucose tolerance despite increased adipocyte lipolysis compared with WT mice. The absence of systemic effects was somewhat surprising despite the finding that circulating PEDF was not increased in the transgenic mice. Our data showing that adipocyte PEDF was elevated 2.5-fold (which is of similar magnitude to obese vs. lean mice) (13) and PEDF secretion increased in adipose tissue explants of transgenic mice indicates that PEDF delivery to tissues may have been matched by increased clearance. The apparent mismatch between protein production and plasma concentrations is not without precedence. For example, despite marked adipocyte overproduction of VEGF-A in the doxycycline-inducible adipocyte-specific VEGF-A overexpression mice, plasma concentrations were not different from WT littermates (34). Since PEDF causes insulin resistance in obesity (13) and PEDF expression is not increased in muscle and liver in obesity (13, 15), another interpretation derived from these studies is that cells other than adipocytes secrete PEDF in obesity to modulate metabolic processes. Since PEDF contains a signal peptide, is variably glycosylated, and acts via multiple high-affinity ligands and cell receptors, it is possible that the degree of glycosylation impacts its receptor interactions and capacity to cross membranes. It is possible that the recombinant PEDF used in previous studies (5, 9, 13, 14) and endogenously produced PEDF activate select subsets of receptors, which modulate different metabolic programs.
In summary, our data support the conclusion that increasing adipocyte-produced PEDF in lean mice to levels observed in obese mice does not modulate angiogenesis during physiological or pathological adipose tissue expansion or act as a major inflammatory mediator in vivo. Rather, our studies of PEDF-aP2 mice demonstrate that PEDF is an important modulator of adipocyte lipolysis, which supports the contention that increased PEDF secretion from hypertrophic adipocytes may increase spontaneous lipolysis in obesity.
These projects were supported by funding from the National Health and Medical Research Council (NHMRC) of Australia. M. J. Watt was supported by a Senior Research Fellowship from the NHMRC.
No conflicts of interest, financial or otherwise, are declared by the authors.
T.V.L., M.L.B., M.M., A.A., and M.J.W. performed experiments; T.V.L., A.A., and M.J.W. analyzed data; T.V.L., M.L.B., M.M., A.A., R.G.E., and M.J.W. approved final version of manuscript; R.G.E. and M.J.W. interpreted results of experiments; R.G.E. edited and revised manuscript; M.J.W. conception and design of research; M.J.W. prepared figures; M.J.W. drafted manuscript.
We thank Jose Gonzalez (Monash University) and Monash Animal Service for transgenic mouse production, Elaine Adler (Monash University) for technical assistance, and Elia Duh (John Hopkins University) for the PEDF antibody.
- Copyright © 2014 the American Physiological Society