Integrative Biology

Liver, but not adipose tissue PEDF gene expression is associated with insulin resistance



Recent studies linked circulating pigment epithelium-derived factor (PEDF) to obesity-associated insulin resistance, but the main source of circulating PEDF is unknown. We aimed to investigate liver and adipose tissue PEDF gene expression in association with obesity and insulin resistance.

Design, subjects and methods:

Three (two cross-sectional and one longitudinal) independent cohorts have been studied, for adipose tissue (n=80 and n=30) and liver gene expression (n=32 and n=14). Effects of high glucose and cytokines on HepG2 cell line were also investigated. PEDF gene expression and circulating PEDF were analyzed using real-time PCR and ELISA, respectively.


In a first cohort of subjects, PEDF relative gene expression was higher in subcutaneous (SC) than in omental (OM) adipose tissue (P<0.0001) being also higher in mature adipocytes compared with stromo-vascular cells (P<0.0001). However, OM PEDF relative gene expression was decreased in morbidly obese subjects (P=0.01). Both OM PEDF and OM PEDF receptor (PEDFR) correlated positively with lipogenic and lipolytic genes, and with genes implicated in the lipid vacuole formation. Circulating PEDF levels were not associated with fat PEDF gene expression. In the second cohort, SC PEDF was decreased in subjects with type 2 diabetes and did not change significantly after weight loss. We next explored circulating PEDF in association with markers of liver-related insulin resistance injury (alanine aminotransferase, r=0.59, P=0.001). Interestingly, liver PEDF gene expression increased with obesity and insulin resistance in men, being significantly associated with fasting glucose and glycated hemoglobin in two independent cohorts. In fact, high glucose led to increased PEDF in HepG2 cells, while inflammatory stimuli present in the adipose tissue environment downregulated PEDF.


Liver, but not adipose tissue, might be the source of increased circulating PEDF linked to insulin resistance.


Adipose tissue is well known for its essential role as energy storage depot but also for secreting adipokines, which affect other tissues such as brain, liver, muscle, β cells, gonads, lymphoid organs and systemic vasculature.1, 2

Pigment epithelium-derived factor (PEDF) is a 50-kDa protein first identified in the conditioned medium of human retinal pigment epithelial cells as a neurotrophic factor.3 Sequence analysis of the human PEDF gene showed that it is a member of the serpin (serine protease inhibitor) family,4 but lacks a serine-reactive loop and thus has no function on protease inhibition.5 PEDF was also identified as forming part of the secretome in primary cultures of human subcutaneous (SC) adipose-derived stem cells (derived from four individual female donors6 or from a 5-year-old male donor7).

Recent observations showed that PEDF was upregulated during adipogenesis and is mostly produced by mature adipocytes.8, 9 PEDF has anti-angiogenic, anti-oxidant, and anti-inflammatory effects and its transcript was highly expressed in a broad range of human fetal and adult tissues including the liver.10 In fact, the expression level of PEDF in porcine liver has been shown to be associated with body muscularity and obesity.11

Increased serum levels of PEDF have been found in type 2 diabetes mellitus (T2DM).12, 13, 14 However, contradictory results have been reported on the possible role of PEDF in the development of obesity-related metabolic disorders. In rodents, recombinant PEDF administration reduced insulin sensitivity and promoted ectopic fat deposition.8 On the other hand, other authors suggested that elevated PEDF in metabolic syndrome might act as a counter system against atherosclerosis15 with its protective anti-inflammatory16 and antioxidant properties.17 Recently, we have described that circulating PEDF levels were increased in obese subjects, decreasing after weight loss.9 For this reason, we hypothesized increased PEDF gene expression in adipose tissue and liver in association with obesity and type 2 diabetes. To test the effects of inflammation and hyperglycemia on adipocytes and hepatocytes, in vitro experiments in human adipocytes and in human hepatocarcinoma cell line (HepG2) have also been performed.

Materials and methods

Patient recruitment

In the first study cohort, eighty paired SC and omental (OM) abdominal adipose tissue samples from the same location were obtained during elective surgical procedures (cholecystectomy (n=4), surgery of abdominal hernia (n=8) and gastric by-pass surgery (n=68)), washed, fragmented, and immediately flash-frozen in liquid nitrogen before being stored at −80 °C. The subjects (19 men and 61 women), who were invited to participate at the Endocrinology Service of the Hospital Universitari de Girona Dr Josep Trueta (Girona, Spain), had a body mass index (BMI) of between 22.0 and 69.0 kg m−2. All subjects were non-diabetic individuals of Caucasian origin and reported that their body weight had been stable for at least 3 months before the study. Liver and renal diseases were specifically excluded by biochemical work-up. Human subjects classified in three groups according to obesity. The criteria used were BMI (lower 30 kg m−2 for non-obese, between 30 and 40 kg m−2 for obese and higher than 40 kg m−2 for morbidly obese subjects). All subjects gave written informed consent, validated and approved by the ethical committee of the Hospital Universitari Dr Josep Trueta (Comitè d’Ètica d’Investigació Clínica, CEIC), after the purpose of the study was explained to them.

PEDF gene expression was also studied in the liver in a sample of consecutive obese subjects (n=32, 8 obese and 24 morbidly obese). Liver biopsies were analyzed by a single pathologist expert in hepatic pathology. Hematoxylin and eosin, Masson’s trichrome and reticulin stains were performed in each liver sample. Histological features of steatosis, lobular inflammation, hepatocellular ballooning and fibrosis were scored using the scoring system for NAFLD.18 Steatosis was graded 0–3 (0 is <5%; 1 is 5–33%, 2 is 33–66% and 3 is >66%), lobular inflammation was graded 0–3 based on inflammatory foci per × 20 with a × 20 ocular (0 is none, 1 is <2 foci, 2 is 2–4 foci and 3 is >4 foci) and hepatocellular ballooning was graded 0–2 (0 is none, 1 is few ballooning cells and 2 is prominent ballooning). The features were combined to assess the NAFLD activity score (NAS), going from 0 to 8. NAS≥5 were diagnosed as non-alcoholic steatohepatitis, NAS=0–2 were considered as non-diagnostic of steatohepatitis and NAS=3–4 were considered as indeterminate. Fibrosis was graded 1–4 and was not included in the activity score. All subjects were of Caucasian origin and reported that their body weight had been stable for at least 3 months before the study. Only a few number of participants were under treatment (fibrates (n=3) and statins (n=6)). They had no systemic disease other than obesity and all were free of any infections in the previous month before the study. Liver disease (specifically tumoral disease and HCV infection) and thyroid dysfunction were specifically excluded by biochemical work-up. All subjects gave written informed consent after the purpose of the study was explained to them. The Hospital Ethics Committee approved the protocol.

The second study cohort was composed of a group of morbidly obese female patients with T2DM (T2DM group, n=13) and a control group of healthy lean age-matched females (C group, n=17). In the obese type 2 diabetic group, eight subjects were treated with metformin, two with glitazones, one subject with sulfonylurea derivates and four with insulin. This part of the study was performed at the General University Hospital of Charles University Prague 1st School of Medicine, Czech Republic. All patients had stable body weight for at least 3 months before the study beginning. Written informed consent was obtained from all participants before the enrollment. The study was approved by the local Ethics Committee and was performed according to the rules proposed in the Declaration of Helsinki.

Subjects were studied in the post-absorptive state. BMI was calculated as weight (in kg) divided by height (in m) squared. Blood pressure was measured in the supine position on the right arm after a 10-min rest; a standard sphygmomanometer of appropriate cuff size was used and the first and fifth phases were recorded. Values used in the analysis are the average of three readings taken at 5-min intervals.

Hepatic gene expression levels of PEDF were also assessed in an independent third cohort of 14 Caucasian and morbidly obese men (age 36.5±10.6; BMI 44.2±4.3; being 10 non-diabetic and 4 non-treated type 2 diabetic participants) who were invited to participate at the Clínica Universidad de Navarra (Pamplona, Spain). All participants were candidates to gastric bypass. Although an intraoperative liver biopsy can be performed in obese patients undergoing bariatric surgery, this procedure is not clinically justified in lean subjects. The study was approved from an ethical and scientific standpoint, by the hospital’s ethical committee responsible for research, and written informed consent of all participants was obtained.

Analytical determinations

For cohort 1, serum glucose concentration and insulin concentrations, total serum cholesterol, high-density lipoprotein (HDL) cholesterol, serum triglycerides and serum PEDF. For cohort 2, serum glucose and insulin concentrations, HbA1c, total serum cholesterol, HDL cholesterol, serum triglycerides, serum PEDF, C-reactive protein (CRP), adiponectin, leptin interleukin-6 (IL-6), IL-8 and tumor necrosis factor-alpha (TNF-α). For cohort 3, serum glucose and insulin concentrations, total serum cholesterol, HDL cholesterol, serum triglycerides and CRP. All these determinations were detailed in Supplementary Information.

PEDF expression during differentiation of human preadipocytes

Isolated preadipocytes from lean and obese subjects, and visceral and SC adipose tissue (SP-F-1 (n=1); SP-F-3 (n=1); OP-F-1 (n=2); Zen-Bio Inc., Research Triangle Park, NC, USA) were plated and differentiated to adipocyte as previously reported.9 This procedure is reported in Supplementary Information. To evaluate the effects of anti-adipogenic conditions, preadipocytes were treated with lipopolysaccharides (LPS) (10 ng ml−1) stimulated macrophage conditioned medium (LPS-MCM, 5%) during the whole differentiation process. LPS-MCM was obtained as previously reported.19

PEDF expression in HepG2 cells

To explore the effects of hyperglycemia and inflammation on hepatocyte PEDF mRNA, the HepG2 cell line was used. HepG2 cells were incubated with basal DMEM media (5.5 mM glucose, 10% FBS, and 1% of sodium pyruvate and penicillin-streptomycin) under high glucose (50 mM) and insulin (10 μM) concentrations and/or under inflammatory conditions (5% LPS-MCM) during 48 h.

Gene expression analyses

RNA was prepared from both fat biopsies and cellular debris using the RNeasy Lipid Tissue Mini Kit (QIAgen, Gaithersburg, MD, USA). The integrity of each RNA sample was checked by an Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). Total RNA was quantified by means of a spectrophotometer (GeneQuant, GE Health Care, Piscataway, NJ, USA) or with the bioanalyzer, and 3 μg of RNA was then reverse transcribed to cDNA using the High Capacity cDNA Archive Kit (Applied Biosystems, Darmstadt, Germany) according to manufacturer’s protocol.

Gene expression was assessed by real-time PCR using the LightCycler 480 Real-Time PCR System (Roche Diagnostics, Barcelona, Spain), using SybrGreen and TaqMan Technology suitable for relative gene expression quantification. The reaction was performed following manufacturer’s protocol in a final volume of 25 μl. The cycle program consisted of an initial denaturing of 10 min at 95 °C, then 40 cycles of 15 s denaturizing phase at 92 °C and 1 min annealing and extension phase at 60 °C.

Cyclophilin A (PPIA; Hs99999904_m1, RefSeq. NM_021130.3) was used as endogenous control for the target genes in each reaction. Gene expression analysis and the primer sequences used are described in Supplementary Information.

PEDF gene expression in stromo-vascular fraction and isolated adipocytes

To evaluate the predominant fraction expressing PEDF, 5 g of visceral adipose tissue from 27 subjects and SC adipose tissue from 22 subjects from cohort 1 were aseptically isolated and all visible connective tissue was removed. A previously described method for isolating stromo-vascular cells (SVCs) and adipocytes from adipose tissue was used,20 which was briefly explained in Supplementary Information.

Statistical methods

Statistical analyses were performed using the SPSS 13.0 software (Chicago, IL, USA). Before statistical analysis, normal distribution and homogeneity of the variances were evaluated using Levene’s test. Unless otherwise stated, descriptive results of continuous variables are expressed as mean±s.d. for Gaussian variables. Parameters that did not fulfill a normal distribution were logarithmically transformed to improve symmetry for subsequent analyses. Paired and unpaired t-tests were used to compare groups with respect continuous variables. Levels of statistical significance were set at P<0.05.

In the second cohort, anthropometric, biochemical and hormonal results are expressed as means±s.e.m. Differences of relative gene expression and serum parameters within obese patients before and after treatment vs healthy subjects were evaluated using a one-way ANOVA followed by the Holm–Sidak method or ANOVA on Ranks followed by Dunn’s test as appropriate. Differences between patients with obesity before vs after treatment were evaluated using paired t-test or Wilcoxon Signed-Rank test as appropriate. Statistical significance was assigned to P<0.05. In both cohorts, Spearman or Pearson correlation test and multiple regression analysis were used to calculate the relationships between relative expression of selected genes in adipose tissue and other parameters.


PEDF gene expression analysis in adipose tissue

First cohort

Clinical and biochemical variables of the subjects in the first study cohort are summarized in Table 1. PEDF relative gene expression was higher in SC adipose tissue than in visceral (OM) adipose tissue (0.95±0.34 vs 0.58±0.26, P<0.0001) and correlated with each other (r=0.32, P=0.003).

Table 1 Anthropometrical and biochemical characteristics of study subjects and associations of PEDF gene expression in visceral and subcutaneous adipose tissue

PEDF gene expression was significantly higher in mature adipocytes vs SVCs (1.42±0.51 vs 0.68±0.53, P<0.0001). In the SVC fraction, PEDF gene expression was not significantly different between visceral or SC fat depots (0.56±0.66 vs 0.83±0.23, P=0.210). In contrast, in adipocytes, PEDF gene expression was significantly higher in SC than in visceral adipose tissue (1.72±0.44 vs 1.16±0.41, P=0.002).

OM PEDF relative gene expression differed according to obesity status, being significantly decreased in morbidly obese subjects (Figure 1a). OM PEDF gene expression significantly and positively correlated with lipogenic genes such as fatty-acid synthase (FASN), acetyl-CoA carboxylase (ACC1) and Spot14 (THRSP), and with gene markers for insulin action (IRS-1) and antioxidant capacity (STAMP2), with lipolytic genes such as PRKACA, with genes implicated in the lipid vacuole formation, CIDEC/FSP27 and Perilipin A (PLIN1), and PPARγ (Table 1). PEDF gene expression in SC adipose tissue was also significantly and negatively associated with BMI, and positively associated with FASN, ACC1, THRSP, IRS-1, PRKACA, CIDEC/FSP27 and PLIN1 gene expression (Table 1). Lipogenic and adipogenic genes in both SC and visceral adipose tissue were significantly decreased in association with obesity, mainly in morbidly obese subjects (Supplementary Table 1).

Figure 1

(a) Relative PEDF gene expression according to obesity status; (b) circulating PEDF levels according to obesity status. (c) PEDF gene expression in SC and OM adipose tissue; (d) PEDF gene expression in SVCs and mature adipocytes. (e) PEDF serum concentrations and (f) PEDF mRNA expression in SC adipose tissue of control group and obese type 2 diabetic group at baseline and after 3 weeks of very low-calorie diet (VLCD).

No drug or surgery group effects were found to influence PEDF gene expression in VAT, SAT and liver.

Given the decreased adipogenic (Supplementary Table 1) and PEDF expression levels in morbidly obese subjects and the presence of PEDF in adipocytes in association with adipogenesis, we hypothesized a decreased maturation of adipocytes in obese subjects.

PEDFR relative gene expression was analyzed in cohort 1 (Table 1). PEDFR relative gene expression was higher in SC adipose tissue than in visceral (OM) adipose tissue (1.76±0.65 vs 1.24±0.58, P<0.0001) and correlated with each other (r=0.323, P=0.022). PEDFR gene expression was significantly higher in mature adipocytes vs SVC (3.42±1.43 vs 0.08±0.03, P=0.001, n=49).

OM PEDFR gene expression significantly and negatively correlated with weight (r=−0.294, P=0.007), BMI (r=−0.269, P=0.011), and positively correlated with lipogenic genes such as FASN, ACC1, PPARγ and THRSP, with gene markers for insulin action (IRS-1), with STAMP2, with lipolytic genes such as AKAP1, PRKACA, with genes implicated in the lipid vacuole formation, CIDEC/FSP27 and PLIN1 and with PEDF relative gene expression (Table 1). PEDFR gene expression in SC adipose tissue was also significantly and positively associated with the lipolytic gene PRKACA, with genes implicated in the lipid vacuole formation, CIDEC/FSP27 and PLIN1, with PPARγ and PEDF (Table 1). These gene expression data were obtained in all subjects.

Second cohort

SC PEDF was significantly decreased in morbidly obese T2DM subjects (Figure 1f), which further strenghtens the results obtained in morbidly obese patients without diabetes mellitus. Furthermore, decreased SC adiponectin expression was found in the T2DM group accompanied by a consistent, albeit non-significant, trend to increased expression of selected pro-inflammatory markers.

Serum PEDF

We next hypothesized associations with circulating PEDF, which were analyzed in the last consecutive 30 subjects. The characteristics of these subjects (26 women and 4 men, aged 46.5±10.5 years, BMI 42.5±7.9 kg m−2 and waist-to-hip ratio 0.96±0.09 were similar to the whole cohort). The mean of OM PEDF gene expression was 0.58±0.22, and of SC PEDF gene expression, 1.02±0.36; mean circulating PEDF concentration was 5.11±1.93 μg ml−1. Circulating PEDF increased with obesity status (Figure 1b). No significant correlations were found between circulating PEDF levels and PEDF gene expression (in OM, r=−0.084, P=0.717; in SC depot, r=−0.036, P=0.872; Table 1).

In the second study cohort, we analyzed serum PEDF levels and its SC adipose tissue expression in a group of obese female patients with T2DM (T2DM group) and compared it with a control group of healthy lean age-matched females. Anthropometric, biochemical and hormonal characteristics of these subjects are summarized in Table 2. A possible explanation about the different circulating PEDF concentration could be the different ELISA kit used in each cohort. However, similarly to obese patients without glucose metabolism disorder, circulating PEDF concentrations were significantly increased in obese T2DM individuals (16.7±1.6 vs 10.4±0.5 μg ml−1, P<0.05—Figure 1e) and correlated positively with BMI (r=0.633, P=0.002) in the combined population of both study groups (morbidly obese females and healthy lean age-matched females). Moreover, a strong positive correlation between serum PEDF and parameters of glucose control, such as fasting plasma glucose (r=0.487, P=0.025) and fasting insulin (r=0.698, P<0.001) and circulating pro-inflammatory markers, hsCRP (r=0.609, P=0.004) and IL-6 (r=0.592, P=0.005) as well as negative correlation with adiponectin (r=−0.510, P=0.013) were observed in this group.

Table 2 Anthropometric, biochemical and hormonal parameters of subjects in T2DM and VLCD substudy

Influence of very low-calorie diet on serum PEDF concentrations and SC PEDF expression

Given the association of PEDF with the nutritional status emerged from previous results, in the next part of our study we investigated the influence of short-term caloric restriction on circulating PEDF and SC adipose tissue expression. Three weeks of very low-calorie diet (VLCD) (energy intake 2500 kJ per day) significantly reduced body weight, improved glucose control and the metabolic profile and decreased pro-inflammatory status (as assessed by hsCRP) in obese T2DM patients (Table 2). However, neither serum PEDF levels (14.7±1.1 vs 16.7±1.6 μg ml−1, n.s.) nor its mRNA expression in SC adipose tissue were significantly affected by this procedure (Figure 1e and f).

Liver PEDF

In subjects from the first cohort, liver PEDF gene expression was higher than in SC or visceral adipose tissue (2.09±0.6 vs 1.18±0.38 vs 0.79±0.28, respectively, P<0.0001), being liver PEDF gene expression positively associated with fasting glucose (r=0.37, P=0.035) and glycated hemoglobin (r=0.336, P=0.045). In the last, consecutive 13 participants, fasting insulin was measured. HOMA tended to be positively associated with liver PEDF gene expression (r=0.44, P=0.1). On the contrary, adipose tissue PEDF gene expression was not significantly associated with these parameters.

Given the significant expression of PEDF in the liver and its association with hyperglycemia, we analyzed the potential relationship with markers of liver injury in the first cohort. Circulating PEDF was significantly associated with alanine aminotransferase (ALT), a well-known marker of liver injury (r=0.59, P=0.001; Figure 2a) but not with AST (Figure 2b). In agreement with these data, liver PEDF gene expression was also significantly associated with ALT (r=0.42, P=0.04) and tended to be associated with AST (r=0.37, P=0.055). Furthermore, 4 vs 28 participants had an NAS over 3, and liver PEDF gene expression tended to be increased in these participants (2.31±0.11 vs 2.07±0.66, P=0.09). Interestingly, circulating PEDF increased significantly in these subjects (8.82±1.8 vs 6.07±1.7 μg ml−1, P=0.008).

Figure 2

(a) Linear relationship between circulating PEDF and ALT; (b) linear relationship between circulating PEDF and aspartate aminotransferase. (ce) Linear relationship between liver PEDF gene expression and BMI, fasting insulin and HOMA value.

In fact, in agreement with the data found in the first cohort, hepatic PEDF mRNA increased with obesity and insulin resistance (HOMA value) (Figures 2c–e; Table 3) in liver samples of cohort 3. Subjects with HOMA value over the median had higher waist circumference (137.5±1.67 vs 129.5±2.9 cm, P=0.02) and fasting insulin (39.76±9.19 vs 11.9±2.16 mU l−1, P=0.02), and lower insulin sensitivity (QUICKI index 0.281±0.007 vs 0.332±0.008, P=0.001) in parallel to increased liver PEDF gene expression (2.89±0.43 vs 1.50±0.31 RU, P=0.02).

Table 3 Bivariate correlations among PEDF gene expression in liver and anthropometric and biochemical parameters

PEDF and PEDFR in human preadipocytes and during adipocyte differentiation

As expected, FASN and adiponectin gene expression increased during the differentiation process in parallel with the accumulation of lipid droplets in the cytoplasm. PEDF and PEDFR gene expression were higher in differentiated adipocytes than in non-differentiated adipocytes in visceral and SC fat depots, being increased in SC adipocytes in the same manner as other adipogenic genes (FASN and ADIPOQ) (Figure 3a).

Figure 3

(a) PEDF, PEDFR, FASN, adiponectin and IL-6 gene expression under inflammatory conditions LPS (10 ng ml−1)-MCM (5%) at day 14 of adipocyte differentiation *P<0.05 and **P<0.005 compared with differentiated control; #P<0.05 and ##P<0.005 compared with visceral adipocytes. (b) PEDF gene expression in HepG2 under high glucose (50 mM) and insulin (10 μM), and LPS (10 ng ml−1)-MCM (5%) administration *P<0.05 and **P<0.005 compared with vehicle or control treatment (normal glucose: 5.5 mM).

In vitro effects

In support of all these findings, LPS-MCM (5%) decreased significantly PEDF and PEDFR gene expression during adipocyte differentiation (Figure 2a) in parallel to FASN and ADIPOQ. IL-6 followed an inverse pattern (Figure 3a).

PEDF gene expression increased significantly under high glucose and insulin concentrations in HepG2 cells (Figure 2b). Otherwise, IRS-1 and IRS-2 gene expression decreased significantly under high glucose and insulin concentrations (Figure 3b). Similarly to adipocytes, PEDF gene expression decreased significantly in HepG2 treated with LPS-MCM (5%).


In this manuscript, we analyzed for the first time, to our knowledge, PEDF gene expression in human adipose tissue. PEDF gene expression was higher in SC than in visceral adipose tissue and was associated with the expression of several lipogenic genes in both fat depots. These associations and the well-known increased expandability of SC vs visceral adipose tissue, led us to speculate PEDF as a lipogenic-associated factor. Further experiments were necessary to explore the possible relationship between PEDF and lipogenesis.

Supporting this hypothesis, PEDF gene expression was higher in mature adipocytes than in SVC. PEDF gene expression also differed significantly between isolated SC vs visceral adipocytes. In agreement with adipogenic genes (Supplementary Table 1), which decreased in association with obesity, PEDF gene expression decreased in obese subjects, mainly in morbidly obese subjects. Strikingly, in second cohort, PEDF expression from SC adipose tissue of obese T2DM group was significantly decreased compared with healthy lean controls, supporting thus the results from the first cohort of obese non-diabetic subjects. Whether the decreased in PEDF gene expression in SC fat depot might be due to the fact that these subjects are also morbidly obese as in the group 1 is a study limitation. Unfortunately, in this cohort non-diabetic morbidly obese subjects were not evaluated. For this reason, the contribution of obese vs diabetes status could not be assessed.

These surprising data are in contrast to findings in rodents where increased adipocyte PEDF expression was present in several models of obesity.8 PEDF was also shown to be one of the most abundant proteins secreted from adipose cells, with increased secretion during adipogenesis.21 On the contrary, TNF-α and hypoxia downregulated PEDF expression in adipose tissue.21 In our obese diabetic group, the SC adipose tissue expression of TNF-α tended to be higher in T2DM subjects than in controls and might therefore constitute one possible cause of the paradoxically decreased PEDF expression. In fact, in vitro data showed in the current study confirmed that PEDF gene expression decreased significantly under inflammatory conditions in both cell models assessed (adipocytes and HepG2 cell line). However, more elaborate studies analyzing a larger variety of factors, with more thoroughly defined cellular sub-populations and testing specifically the effects of TNF-α, are needed to provide a plausible explanation for these findings.

Circulating PEDF concentrations were not associated with PEDF gene expression in visceral or SC adipose tissue. Of note, circulating PEDF levels were significantly elevated also in the second cohort of obese patients with T2DM and correlated strongly with parameters of glucose homeostasis and insulin resistance which is in accordance with several previous reports in humans.9, 12, 13, 14 The differences on circulating PEDF concentration between the cohort 1 and the cohort 2 might be explained for the different ELISA kit used in each cohort. However, participant origin and sample manipulation in each cohort could also contribute to these differences. In previous studies, weight reduction was associated with decreased circulating PEDF levels in rodents8 as well as in humans.9 In the current study, systemic PEDF tended to decrease after 3 weeks of VLCD, even though the difference did not reach statistical significance. This might be due to the relatively low number of subjects included in the weight reduction program. Interestingly, the already decreased PEDF expression in SC adipose tissue showed a similar, albeit non-significant, trend to further reduction after VLCD. Taken together, these data suggest that adipose tissue might not be the main source of PEDF in obesity and T2DM in humans.

Otherwise, circulating PEDF and liver PEDF gene expression significantly and positively correlated with ALT. Serum ALT levels reflect, with specificity and reasonable sensitivity, liver injury.22 In a recent study, serum PEDF levels in patients with alcoholic liver disease were significantly greater than in non-drinkers.23 For this reason, we focused our analyses on non-drinkers. Interestingly, PEDF gene expression and circulating PEDF were increased in patients with NAS over 3. In agreement with this, the presence of PEDF transcripts in a broad range of adult human tissues, especially in liver, may account for increased levels of circulating PEDF with metabolically induced liver injury.24 In addition, hepatic PEDF gene expression was strongly associated with obesity and insulin resistance. These data led us to suggest that circulating PEDF may reflect hepatic PEDF production. In fact, PEDF gene expression increased significantly in HepG2 cells under hyperglycemia and hyperinsulinemia, but not under inflammatory stimuli. In parallel with this, high glucose and insulin conditions decreased significantly IRS-1 and IRS-2 gene expression. Previous studies have shown that the decreased IRS-2 mRNA levels induced and were associated with insulin resistance in HepG2 cells.25, 26 A previous study also suggested that serum PEDF levels may be elevated in response to circulating advanced glycosylation end products (AGEs) as a counter system against the AGE-evoked vascular cell damage in humans.27 These authors hypothesized that the liver may be a source of PEDF in the circulation. In fact, PEDF expression in a hepatocyte cell line increased under oxidative stress conditions.28 AGE-evoked PEDF overproduction by adipocytes and/or liver may counter-act the deleterious effects of AGEs on cardiometabolic disorders in humans.29 These observations suggest that liver could be one of the main origins of PEDF in circulation, and are in concordance with our in vitro findings, suggesting a protective role of PEDF against the deleterious effects of high glucose levels associated with prediabetes. Interestingly, liver lipogenesis is known to be increased in obese subjects in parallel to decreased gene expression of lipogenic markers in adipose tissue.30 This study supports the association between PEDF and lipogenesis (as mentioned above).

Interestingly, in both SC and OM adipose tissues, PEDF gene expression was significantly associated with its receptor (PEDFR). PEDFR gene expression in OM adipose tissue positively correlated with lipogenic genes, with IRS-1, with lipolytic genes, with genes implicated in the lipid vacuole formation and with PEDF relative gene expression. PEDFR gene expression in SC adipose tissue was also significantly and positively associated with the lipolytic gene PRKACA and with genes implicated in the lipid vacuole formation. This suggests that PEDF could be involved in the opposing actions of lipolysis and lipogenesis. Some studies showed PEDF appears to rapidly induce lipolytic stimulation by interacting directly with the key lipolytic protein adipose triglyceride lipase,31 while other adipokines act more slowly presumably by modifying translational control of lipid droplet-associated proteins such as perilipin A.32

In summary, current findings suggest that adipose tissue is not the main source of circulating PEDF levels in humans. In fact, PEDF gene expression was decreased in visceral adipose tissue from morbidly obese non-diabetic subjects in parallel to lipogenic genes as well as in SC adipose tissue of T2DM patients. The absence of a significant effect of diet-induced weight reduction on PEDF adipose tissue expression further supports the idea of adipose tissue not being the central production and regulatory site of PEDF in obesity and T2DM. Circulating PEDF levels might derive from the liver in association with metabolically induced liver damage. Future studies will be necessary to evaluate the production and regulation of circulating PEDF levels by other tissues, including the liver.


  1. 1

    Tilg H, Moschen AR . Adipocytokines: mediators linking adipose tissue, inflammation and immunity. Nat Rev Immunol 2006; 6: 772–783.

    CAS  Article  Google Scholar 

  2. 2

    Shoelson SE, Herrero L, Naaz A . Obesity, inflammation, and insulin resistance. Gastroenterology 2007; 132: 2169–2180.

    CAS  Article  Google Scholar 

  3. 3

    Tombran-Tink J, Chader GJ, Johnson LV . PEDF: a pigment epithelium-derived factor with potent neuronal differentiative activity. Exp Eye Res 1991; 53: 411–414.

    CAS  Article  Google Scholar 

  4. 4

    Steele FR, Chader GL, Johnson LV, Tombran-Tink J . Pigment epithelium-derived factor: neurotrophic activity and identification as a member of the serine protease inhibitor gene family. Proc Natl Acad Sci USA 1993; 90: 1526–1530.

    CAS  Article  Google Scholar 

  5. 5

    Becerra SP, Sagasti A, Spinella P, Notario V . Pigment epithelium-derived factor behaves like a noninhibitory serpin. Neurotrophic activity does not require the serpin reactive loop. J Biol Chem 1995; 270: 25992–25999.

    CAS  Article  Google Scholar 

  6. 6

    Zvonic S, Lefevre M, Kilroy G, Floyd ZE, DeLany JP, Kheterpal I et al. Secretome of primary cultures of human adipose-derived stem cells (ASCs): modulation of serpins by adipogenesis. Mol Cell Proteomics 2007; 6: 18–28.

    CAS  Article  Google Scholar 

  7. 7

    Chiellini C, Cochet O, Negroni L, Samson M, Poggi M, Ailhaud G et al. Characterization of human mesenchymal stem cell secretome at early steps of adipocyte and osteoblast differentiation. BMC Mol Biol 2008; 9: 26.

    Article  Google Scholar 

  8. 8

    Crowe S, Wu LE, Economou C, Turpin SM, Matzaris M, Hoehn KL et al. Pigment epithelium-derived factor contributes to insulin resistance in obesity. Cell Metab 2009; 10: 40–47.

    CAS  Article  Google Scholar 

  9. 9

    Sabater M, Moreno-Navarrete JM, Ortega FJ, Pardo G, Salvador J, Ricart W et al. Circulating pigment epithelium-derived factor levels are associated with insulin resistance and decrease after weight loss. J Clin Endocrinol Metab 2010; 95: 4720–4728.

    CAS  Article  Google Scholar 

  10. 10

    Tombran-Tink J, Mazuruk K, Rodriguez IR, Chung D, Linker T, Englander E et al. Organization, evolutionary conservation, expression and unusual Alu density of the human gene for pigment epithelium derived factor, a unique neurotrophic serpin. Mol Vis 1996; 2: 11.

    CAS  PubMed  Google Scholar 

  11. 11

    Ponsuksili S, Murani E, Schellander K, Schwerin M, Wimmers K . Identification of functional candidate genes for body composition by expression analyses and evidencing impact by association analysis and mapping. Biochim Biophys Acta 2005; 1730: 31–40.

    CAS  Article  Google Scholar 

  12. 12

    Jenkins A, Zhang SX, Gosmanova A, Aston C, Dashti A, Baker MZ et al. Increased serum pigment epithelium derived factor levels in Type 2 diabetes patients. Diabetes Res Clin Pract 2008; 82: 5–7.

    Article  Google Scholar 

  13. 13

    Chen HB, Jia WP, Lu JX, Bao YQ, Li Q, Lu FD et al. Change and significance of serum pigment epithelium-derived factor in type 2 diabetic nephropathy. Zhonghua Yi Xue Za Zhi 2007; 87: 1230–1233.

    CAS  PubMed  Google Scholar 

  14. 14

    Ogata N, Matsuoka M, Matsuyama K, Shima C, Tajika A, Nishiyama T et al. Plasma concentration of pigment epithelium-derived factor in patients with diabetic retinopathy. J Clin Endocrinol Metab 2007; 92: 1176–1179.

    CAS  Article  Google Scholar 

  15. 15

    Yamagishi S, Adachi H, Abe A, Yashiro T, Enomoto M, Furuki K et al. Elevated serum levels of pigment epithelium-derived factor in the metabolic syndrome. J Clin Endocrinol Metab 2006; 91: 2447–2450.

    CAS  Article  Google Scholar 

  16. 16

    Yamagishi S, Inagaki Y, Nakamura K, Abe R, Shimizu T, Yoshimura A et al. Pigment epithelium-derived factor inhibits TNF-alpha-induced interleukin-6 expression in endothelial cells by suppressing NADPH oxidase-mediated reactive oxygen species generation. J Mol Cell Cardiol 2004; 37: 497–506.

    CAS  Article  Google Scholar 

  17. 17

    Yamagishi S, Nakamura K, Ueda S, Kato S, Imaizumi T . Pigment epithelium-derived factor (PEDF) blocks angiotensin II signaling in endothelial cells via suppression of NADPH oxidase: a novel anti-oxidative mechanism of PEDF. Cell Tissue Res 2005; 320: 437–445.

    CAS  Article  Google Scholar 

  18. 18

    Kleiner D, Brunt E, Van Natta M, Behling C, Contos M, Cummings O et al. Design and validation of a histological score system for nonalcoholic fatty liver disease. Hepatology 2005; 41: 1313–1321.

    Article  Google Scholar 

  19. 19

    Moreno-Navarrete JM, Ortega FJ, Ricart W, Fernandez-Real JM . Lactoferrin increases (172Thr)AMPK phosphorylation and insulin-induced (p473Ser)AKT while impairing adipocyte differentiation. Int J Obes (Lond) 2009; 33: 991–1000.

    CAS  Article  Google Scholar 

  20. 20

    Bunnell BA, Flaat M, Gagliardi C, Patel B, Ripoll C . Adipose-derived stem cells: isolation, expansion and differentiation. Methods 2008; 45: 115–120.

    CAS  Article  Google Scholar 

  21. 21

    Famulla S, Lamers D, Hartwig S, Passlack W, Horrighs A, Cramer A et al. Pigment epithelium-derived factor (PEDF) is one of the most abundant proteins secreted by human adipocytes and induces insulin resistance and inflammatory signaling in muscle and fat cells. Int J Obes (Lond) 2011; 35: 762–772.

    CAS  Article  Google Scholar 

  22. 22

    Wedemeyer H, Hofmann WP, Lueth S, Malinski P, Thimme R, Tacke F et al. ALT screening for chronic liver diseases: scrutinizing the evidence. Z Gastroenterol 2010; 48: 46–55.

    CAS  Article  Google Scholar 

  23. 23

    Sogawa K, Kodera Y, Satoh M, Kawashima Y, Umemura H, Maruyama K et al. Increased serum levels of pigment epithelium-derived factor by excessive alcohol consumptions–Detection and identification by a three-step serum proteome analysis. Alcohol Clin Exp Res 2011; 35: 211–217.

    CAS  Article  Google Scholar 

  24. 24

    Gaemers IC, Stallen JM, Kunne C, Wallner C, van Werven J, Nederveen A et al. Lipotoxicity and steatohepatitis in an overfed mouse model for non-alcoholic fatty liver disease. Biochim Biophys Acta 2011; 1812: 447–458.

    CAS  Article  Google Scholar 

  25. 25

    García-Ruiz I, Solís-Muñoz P, Gómez-Izquierdo E, Muñoz-Yagüe MT, Valverde AM, Solís-Herruzo JA . Protein-tyrosine phosphatases are involved in interferon resistance associated with insulin resistance in HepG2 cells and obese mice. J Biol Chem 2012; 287: 19564–19573.

    Article  Google Scholar 

  26. 26

    Luo Z, Zhang Y, Li F, He J, Ding H, Yan L et al. Resistin induces insulin resistance by both AMPK-dependent and AMPK-independent mechanisms in HepG2 cells. Endocrine 2009; 36: 60–69.

    CAS  Article  Google Scholar 

  27. 27

    Yamagishi S, Matsui T, Adachi H, Takeuchi M . Positive association of circulating levels of advanced glycation end products (AGEs) with pigment epithelium-derived factor (PEDF) in a general population. Pharmacol Res 2010; 61: 103–107.

    CAS  Article  Google Scholar 

  28. 28

    Nakamura K, Yamagishi S, Yoshida T, Matsui T, Imaizumi T, Inoue H et al. Hydrogen peroxide stimulates pigment epithelium-derived factor gene and protein expression in the human hepatocyte cell line OUMS-29. J Int Med Res 2007; 35: 427–432.

    CAS  Article  Google Scholar 

  29. 29

    Yamagishi S, Matsui T, Nakamura K, Ueda S, Noda Y, Imaizumi T . Pigment epithelium-derived factor (PEDF): its potential therapeutic implication in diabetic vascular complications. Curr Drug Targets 2008; 9: 1025–1029.

    CAS  Article  Google Scholar 

  30. 30

    Diraison F, Dusserre E, Vidal H, Sothier M, Beylot M . Increased hepatic lipogenesis but decreased expression of lipogenic gene in adipose tissue in human obesity. Am J Physiol Endocrinol Metab 2002; 282: E46–E51.

    CAS  Article  Google Scholar 

  31. 31

    Chung C, Doll JA, Gattu AK, Shugrue C, Cornwell M, Fitchev P et al. Anti-angiogenic pigment epithelium-derived factor regulates hepatocyte triglyceride content through adipose triglyceride lipase (ATGL). J Hepatol 2008; 48: 471–478.

    CAS  Article  Google Scholar 

  32. 32

    Rydén M, Arner P . Tumour necrosis factor-alpha in human adipose tissue – from signaling mechanisms to clinical implications. J Intern Med 2007; 262: 431–438.

    Article  Google Scholar 

Download references


This work was partially supported by research grants from the Ministerio de Educación y Ciencia (SAF2008-0273). The Prague substudy was supported by research grants from the Ministry of Health of the Czech Republic and General University Hospital in Prague (IGA 10024-4 and MZOVFN2005).

Author information



Corresponding author

Correspondence to J M Fernández-Real.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies this paper on International Journal of Obesity website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Moreno-Navarrete, J., Touskova, V., Sabater, M. et al. Liver, but not adipose tissue PEDF gene expression is associated with insulin resistance. Int J Obes 37, 1230–1237 (2013).

Download citation


  • pigment epithelium-derived factor
  • insulin resistance
  • adipose tissue
  • liver