The integrity of the vascular barrier, which is essential to blood vessel homoeostasis, can be disrupted by a variety of soluble permeability factors during sepsis. Pigment epithelium-derived factor (PEDF), a potent endogenous anti-angiogenic molecule, is significantly increased in sepsis, but its role in endothelial dysfunction has not been defined. To assess the role of PEDF in the vasculature, we evaluated the effects of exogenous PEDF in vivo using a mouse model of cecal ligation and puncture (CLP)-induced sepsis and in vitro using human dermal microvascular endothelial cells (HDMECs). In addition, PEDF was inhibited using a PEDF–monoclonal antibody (PEDF–mAb) or recombinant lentivirus vectors targeting PEDF receptors, including adipose triglyceride lipase (ATGL) and laminin receptor (LR). Our results showed that exogenous PEDF induced vascular hyperpermeability, as measured by extravasation of Evan's Blue (EB), dextran and microspheres in the skin, blood, trachea and cremaster muscle, both in a normal state and under conditions of sepsis. In control and LR–shRNA-treated HDMECs, PEDF alone or in combination with inflammatory mediators resulted in activation of RhoA, which was accompanied by actin rearrangement and disassembly of intercellular junctions, impairing endothelial barrier function. But in ATGL–shRNA-treated HDMECs, PEDF failed to induce the aforementioned alterations, suggesting that PEDF-induced hyperpermeability was mediated through the ATGL receptor. These results reveal a novel role for PEDF as a potential vasoactive substance in septic vascular hyperpermeability. Furthermore, our results suggest that PEDF and ATGL may serve as therapeutic targets for managing vascular hyperpermeability in sepsis.

CLINICAL PERSPECTIVES

  • PEDF, which exerts several important but contradictory influences on the VE barrier, is significantly increased in sepsis. However, the mechanism by which PEDF influences the vasculature following sepsis remains elusive.

  • The present study shows that exogenous PEDF interacts with up-regulated ATGL receptor during sepsis, leading to the activation of RhoA signalling. This signalling is accompanied by actin rearrangement and the disruption of intercellular junctions, resulting in increased vascular permeability.

  • PEDF plays a detrimental role in the maintenance of vascular integrity and may lay the groundwork for developing new therapeutic strategies for the treatment of hyperpermeability in sepsis.

INTRODUCTION

Sepsis is characterized by an elevated level of circulating inflammatory cytokines and dysregulated coagulation [1,2]. Sepsis causes millions of deaths globally each year and the frequency of sepsis cases is increasing [3]. Over the past 30 years, new therapies for sepsis, such as interrupting the initial cytokine cascade or interfering with dysregulated coagulation, have met with failure [4]. Thus, our understanding of the pathophysiology that underlies sepsis in humans is inadequate.

It is now recognized that greater attention should be focused on the progressive subcutaneous and body cavity oedema that typically develops in septic patients. The accumulation of parenchymal and interstitial fluids impairs organ function not only by increasing the distance required for oxygen diffusion, but also by compromising microvascular perfusion due to increased interstitial pressure [5]. Tissue oedema has been generally attributed to the widespread increase in vascular permeability in sepsis [6], which is closely related to microvascular endothelial dysfunction [7]. However, the mechanism by which endothelial dysfunction contributes to hyperpermeability in sepsis has not been clarified. Resolving this issue would improve our understanding of the pathophysiology of sepsis and might lead to new therapeutic approaches.

In our previous study, proteomics analysis was applied to compare the changes in serum proteins in patients with sepsis followed by purpura fulminans (PF) to those in healthy individuals [8]. Consistent with previous studies, pigment epithelium-derived factor (PEDF) was found to be significantly increased in septic patients [9]. PEDF is an endogenously produced protein that is widely expressed throughout the human body and has multiple biological activities, including anti-inflammatory, antioxidant and neuro-protective properties [10]. PEDF effectively inhibits vascular endothelial growth factor (VEGF)-driven angiogenesis and vascular permeability by regulating the intracellular proteolysis of VEGF receptors [11]. However, recent reports have raised the possibility that PEDF may act as an important endogenous vasoactive substance [1214]. Indeed, PEDF was shown to activate mitogen-activated protein kinase (MAPK) pathways to regulate endothelial pro-apoptotic and anti-migratory activities [12]. Moreover, PEDF activates peroxisome proliferator-activated receptor-γ (PPAR-γ), which in turn leads to the overexpression of p53 and apoptosis of endothelial cells [13]. Other studies have shown that a high dose of PEDF increases VEGF mRNA levels in granulosa cells and is not effective at reducing ovarian hyperstimulation syndrome-induced leakage [14]. However, the precise mechanisms underlying the effects of PEDF in the vasculature have not been clarified.

The aim of the present study was to investigate the role of PEDF in vascular permeability during sepsis and to identify the molecular mechanism underlying its effects. We employed exogenous soluble PEDF as well as a monoclonal antibody (mAb) and recombinant lentivirus vectors to inhibit PEDF and its receptors and then examined the effect of PEDF on the vasculature and endothelium under normal and septic conditions, both in vivo and in vitro.

MATERIALS AND METHODS

Antibodies, recombinant proteins and inflammatory mediators

The following antibodies were used: anti-human PEDF antibody (Sino Biological Inc.); anti-mouse PEDF antibody (R&D Systems); anti-mouse CD31 antibody (Chemicon); anti-ZO-1 (zonula occludens-1) antibody (Cell Signaling Technology); anti-VE-cadherin antibody, anti-67KDa laminin receptor (LR) antibody, anti-calnexin antibody (Abcam); anti-ATGL (adipose triglyceride lipase) antibody (GeneTex); anti-RhoA antibody, anti-ROCK1/2 (Rho-associated coiled-coil protein kinase) antibody, anti-ACTB (actin beta unit) antibody, anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) antibody (Proteintech); anti-VDAC (voltage-dependent anion channel) antibody (Santa Cruz Biotechnology Inc.); anti-histone H3 antibody (Bioss); horseradish peroxidase (HRP)-conjugated goat-anti-rabbit/mouse antibody, TRITC (tetramethylrhodamine isothiocyanate)–phalloidin (Sigma–Aldrich); Cy3-conjugated goat anti-hamster antibody and FITC-conjugated goat anti-rabbit IgG antibody (Jackson ImmunoResearch). The following recombinant proteins and inflammatory mediators were used: recombinant human SERPINF1(serpin peptidase inhibitor, clade F)/PEDF protein (Sino Biological Inc.); lipopolysaccharide (LPS), C3 transferase (Sigma–Aldrich); and recombinant human tumour necrosis factor (TNF)-α (Peprotech).

Septic patients and blood sample collection

From January 2011 to December 2012, blood samples (5 ml) were obtained from healthy individuals (n=12) and post-burn septic patients (n=12) admitted to the Institute of Burn Research, Southwest Hospital, according to previously defined inclusion criteria [15,16]. The exclusion criteria consisted of age <18 years, pregnancy, the presence of chronic hepatic failure, chronic renal failure or haematological disease or a do-not-resuscitate status. Blood samples were clotted for 30 min followed by centrifugation at 3000 g for 10 min. Serum was then freshly frozen at −80°C until use. The study protocol was in accordance with the Helsinki Declaration and was reviewed and approved by the Research Ethics Board of Southwest Hospital. Informed consent was obtained from all patients participating in the study.

Sepsis mouse model

A total of 90 BALB/c male mice (6–8 weeks of age; 20–25 g weight) were purchased from the Laboratory Animal Center of Third Military Medical University. Mice were housed in micro-isolator cages and fed a rodent chow diet. Three different sepsis models were used in our experiments, with mice randomly divided into the following five sub-groups for each model: sham vehicle treatment and sepsis at 3, 6, 12 and 24 h (n=6 for each sub-group). Blood samples were collected by eyeball enucleation at the indicated time-point and clotted for 30 min followed by centrifugation at 3000 g for 10 min. Then, serum was freshly frozen at −80°C until examination. Animal care and experimental procedures were performed under the regulations of the Animal Care Center of Third Military Medical University in accordance with the guidelines of the National Institutes of Health in the United States. (The detailed methods employed for the three sepsis models can be found in the Supplementary Materials.)

Cell culture and plasma membrane/tissue protein extraction

Human dermal microvascular endothelial cells (HDMECs) were cultured at 37°C in 5% CO2 in endothelial cell medium (ECM, ScienCell) containing 10% FBS (ScienCell). The culture medium was replaced every 2–3 days. After reaching 70%–80% confluence, the cells were harvested with 0.05% trypsin (ScienCell) and passaged at a ratio of 1:3.

Plasma membrane protein was extracted using the Plasma Membrane Protein Extraction Kit (BioVision). Tissue protein was extracted using the Total Protein Extraction Assay (KeyGEN BioTECH). (The detailed methods can be found in the Supplementary Materials.)

ELISA analysis

PEDF and VEGF levels were measured using commercially available sandwich ELISA kits (USCN) according to the manufacturer's instructions. Plates were read in an ELISA plate reader (Thermo Fisher Scientific) within 15 min after stopping the reaction. The intra-assay imprecision was <5% and the inter-assay imprecision was <8%. All samples were routinely analysed in triplicate. (The detailed methods can be found in the Supplementary Materials.)

RhoA pull-down assays

Evaluation of RhoA activity was performed using RhoA pull-down assays [17]. Glutathione sepharose-coupled rhotekin Rho-binding domain (GST–RBD) beads (20 mg/pull-down) were applied (Upstate Biotechnology). Briefly, cells were starved for 2 h and left unstimulated or stimulated prior to rapid lysis in pull-down lysis buffer. Cells were scraped immediately, transferred into a microfuge tube and centrifuged at 13000 g for 3 min at 4°C. A 50 μl of sample of the whole cell lysate was retained and the remaining supernatant was incubated with the GST–RBD beads at 4°C with rotation for 1 h. The beads were washed three times in pull-down lysis buffer and then boiled for 5 min in Laemmli sample buffer. Proteins were resolved by Western blotting (WB).

Western blotting and immunoprecipitation

Cells were washed three times with PBS (4°C) and lysed in radioimmunoprecipitation assay (RIPA) buffer (ProteinSimple). Cell lysates and tissue protein samples were heated to 95°C in SDS loading buffer and the protein concentration was determined using the RCDC (reducing agent compatible and detergent compatible) protein detection assay (Bio-Rad). Proteins separated by SDS/PAGE were transferred on to PVDF membranes (Millipore), blocked in 5% skim milk and incubated with primary and secondary antibodies. Bands were visualized by chemiluminescence using the ChemiDoc™ XRC+ Imaging System (Bio-Rad) and ECL detection reagents (GE Healthcare).

For immunoprecipitation, cells were washed three times with PBS (4°C) and lysed in RIPA buffer at 4°C overnight. Samples were lysed and centrifuged at 12000 g at 4°C for 15 min. The supernatant was collected and transferred to new tubes. The samples were pre-cleared with protein A/G beads (Santa Cruz Biotechnology Inc.) for 1 h at 4°C, followed by centrifugation at 15000 g at 4°C for 1 min. The supernatant was then collected. PEDF was immunoprecipitated from 500 μg of extract using 3 μg of anti-PEDF antibody followed by 50 μl of protein A/G beads. The beads were washed three times with PBS (4°C), boiled in 30 μl of SDS loading buffer and resolved by WB.

Immunofluorescence staining

Immunocytochemistry was performed using a standard procedure. In brief, cells were washed three times with PBS (37°C), fixed in 4% paraformaldehyde for 20 min and permeabilized with 0.1% Triton X-100 for 15 min. Cells were incubated for 1 h in 5% normal goat serum, followed by immunofluorescence staining.

Sites of leakage in mouse tracheal blood vessels and the cremaster muscle were visualized using 100-nm extravasated green fluorescent polymer microspheres (Duke Scientific Corp.) as described previously [18]. In brief, anaesthetized mice received an intravenous injection of 20 μl (4×1011 particles) containing fluorescent microspheres. Two minutes later, the intravascular microspheres were removed from the bloodstream by vascular perfusion with 1% paraformaldehyde. The trachea and cremaster muscle were dissected and fixed in 4% paraformaldehyde for 30 min (trachea for 2 h), washed six times with PBS and incubated with 5% normal goat serum containing 0.1% Triton X-100 for 1 h. The tissue was immunostained with CD31 antibody, followed by staining with Cy3-conjugated antibody to visualize CD31 on venules. Tracheas were rinsed in PBS, flattened and mounted with the luminal surface facing up in Vectashield (Vector Laboratories). Samples were examined under a Zeiss LSM 700 confocal microscope.

Measurement of Evan's Blue leakage into the skin

A modified Miles assay was used to evaluate Evan's Blue (EB) dye leakage into the skin, as previously described [19]. Mice were anaesthetized and injected through the caudal vein with 0.5% EB (100 μl/20 g; Sigma–Aldrich), which was allowed to circulate for 30 min. PEDF (1000 ng in 50 μl/site) or PBS was injected intradermally. After 30 min, the dorsal skin was excized, homogenated and incubated in dimethylformamide (Sigma–Aldrich) for 24 h at 55°C. The amount of EB in each tissue sample was quantified by spectrometry (Thermo Fisher Scientific) at 620 nm.

FITC–dextran permeability assay in vivo

FITC–dextran (70 kDa; 10 mg/20 g; Sigma–Aldrich) was administered systemically via caudal vein injection [20]. One hour later, blood was collected by eyeball enucleation in anaesthetized animals and then transferred to microcentrifuge tubes, placed on ice for 1 h and centrifuged at 15000 g for 10 min. The supernatant was then collected. Serum samples (60 μl, 1:3 diluted in PBS) were added into each fluorescence plate well and read at 490/510 nm. Measurements were recorded as relative fluorescence units of serum.

Assessment of cell permeability in vitro

Transwell tracer experiments were performed using the 24-well Transwell system (0.4 μm of pore size, 6.5 mm diameter; Costar). HDMECs (1×105) were plated in Transwell chambers, grown for 3 days and serum-starved for 4 h. To assess endothelial permeability, the paracellular flux and transendothelial electrical resistance (TER) were measured after stimulation. To measure paracellular flux, the medium from the basolateral domain was collected at the indicated times after FITC–dextran (40 kDa; Sigma–Aldrich) addition for 1 h. TER was measured using a Millicell ERS-2 Volt-Ohm Meter (Millicell). At the indicated times, resistance (Ω) was obtained from each insert and multiplied by the membrane area (Ω × cm2) to obtain values of TER. The resistance value of an empty culture insert (no cells) was subtracted as background.

Lentivirus vector-mediated PEDF receptor silencing

Four patatin-like phospholipase domain-containing protein 2 (PNPLA2; ATGL) and ribosomal protein SA (RPSA; LR) targeted double-stranded oligonucleotides were cloned into the pLKD–CMV (cytomegalovirus)–PuroR–U6–shRNA vector (Neuron Biotech Co.). A scrambled shRNA-expressing virus was used as a control [21]. The production of lentivirus was accomplished as previously described [22]. PNPLA2–shRNA or RPSA–shRNA lentivirus was added directly to the medium. To determine the optimum level of ATGL or LR expression, an immunoblotting technique was adopted to measure the multiplicity of infection (MOI) and the duration of infection for each lentivirus.

Statistical analysis

Statistical analysis between groups was performed using a two-tailed one-way ANOVA followed by Tukey's post-hoc analysis. Statistical analyses were performed using Statistical Product and Service Solutions (SPSS 19.0; SPSS Inc.). Differences were considered statistically significant at a P<0.05.

RESULTS

The serum PEDF level was significantly elevated in septic patients and mice

To determine whether the serum level of PEDF was increased during sepsis, serum samples from septic patients and mice were collected at the indicated times and analysed by ELISA and WB. The baseline clinical data of the healthy controls and septic patients are presented in Table 1. Age, gender and body mass index (BMI) did not differ between septic patients and normal controls (P>0.05). Owing to the lack of a proper internal control for serum proteins, we assumed both equal loading of proteins and equal loading volumes to compare changes in serum PEDF level. In addition, the serum samples were pooled to dilute potential individual differences in WB analysis.

Table 1
Patient demographics*

Abbreviations: dTBSA, total surface area of deep second-degree or third-degree burns; TBSA, total burn surface area.

Healthy controls (n=12)Post-burn sepsis (n=12)
Age (years) 42±11.54 50.33±11.86 
Gender (% of male) 50 58.33 
BMI (kg/m223.29±1.59 21.86±1.83 
TBSA burned (%) – 75.75±14.47 
dTBSA (%) – 29.92±9.74 
Time after burn injury (days) – 11.48±4.32 
Time after sepsis diagnosis (days) – 2.67±0.98 
Healthy controls (n=12)Post-burn sepsis (n=12)
Age (years) 42±11.54 50.33±11.86 
Gender (% of male) 50 58.33 
BMI (kg/m223.29±1.59 21.86±1.83 
TBSA burned (%) – 75.75±14.47 
dTBSA (%) – 29.92±9.74 
Time after burn injury (days) – 11.48±4.32 
Time after sepsis diagnosis (days) – 2.67±0.98 

*The data are presented as the mean±S.D.

The results of both the ELISA (6.26±1.94 compared with 17.29±3.33 μg/ml; Figure 1A) and the WB analysis (Figure 1B) showed that septic patients had significantly elevated serum PEDF levels compared with controls. In post-burn septic mice, the serum PEDF level was significantly elevated from 6 h after establishment of the model and remained elevated up to 24 h post-injury; no significant difference in serum PEDF level was observed at 3 h post-injury (Figure 1C). However, in both cecal ligation and puncture (CLP)- and LPS-induced septic mice, serum PEDF level was significantly elevated starting 3 h after establishment of the model and remained elevated up to 24 h after injury (Figures 1D and 1E). The level of circulating PEDF peaked at 12 h in both post-burn septic and CLP-induced septic mice and peaked at 3 h in LPS-induced septic mice (Figures 1C–1E). To verify these results, serum samples collected from septic mice at the peak level of expression were pooled and further analysed by WB and the results demonstrated similar elevations in serum PEDF among treated mice in comparison with sham controls (Figure 1F). Taken together, these results demonstrate that serum PEDF level is significantly increased following sepsis.

Serum PEDF level was significantly elevated in septic patients and mice

Figure 1
Serum PEDF level was significantly elevated in septic patients and mice

(A) Serum level of PEDF in septic patients. Blood samples were taken from post-burn septic patients and healthy individuals. PEDF level was measured using an ELISA system to detect human PEDF. The data are shown as the mean±S.D., n=12 in each group. (B) Serum PEDF level was measured using WB analysis. Owing to the lack of an internal control for serum proteins, we assumed equal loading of proteins and volumes, n=12 in each group. (CE) Serum level of PEDF in three sepsis mouse models. Blood samples were taken at the indicated time points. PEDF level was measured using an ELISA system to detect mouse PEDF. The data are shown as the mean±S.D. as compared with the control group, n=6 in each group. (F) The peak level of PEDF in septic mice was measured using WB analysis, n=6 in each group. *P<0.05, **P<0.01.

Figure 1
Serum PEDF level was significantly elevated in septic patients and mice

(A) Serum level of PEDF in septic patients. Blood samples were taken from post-burn septic patients and healthy individuals. PEDF level was measured using an ELISA system to detect human PEDF. The data are shown as the mean±S.D., n=12 in each group. (B) Serum PEDF level was measured using WB analysis. Owing to the lack of an internal control for serum proteins, we assumed equal loading of proteins and volumes, n=12 in each group. (CE) Serum level of PEDF in three sepsis mouse models. Blood samples were taken at the indicated time points. PEDF level was measured using an ELISA system to detect mouse PEDF. The data are shown as the mean±S.D. as compared with the control group, n=6 in each group. (F) The peak level of PEDF in septic mice was measured using WB analysis, n=6 in each group. *P<0.05, **P<0.01.

Exogenous PEDF led to increased extravasation of tracer in mice

CLP is the most widely used animal model of sepsis because it most closely mimics sepsis in humans [23]. This fact, combined with the clear increase in serum PEDF noted in the CLP mouse model (Figure 1D), led us to select the 12-h time point after CLP for subsequent experiments.

To assess the effect of PEDF on the vasculature in vivo, we injected exogenous PEDF and evaluated the loss of tracheal vessel integrity by monitoring the extravasation of 100-nm of microsphere particles (green). Extravasation was not detected in control mice but was detected in PEDF mice (Figures 2A and 2B); a similar disruption in vasculature integrity was observed in the cremaster muscle microvasculature following PEDF injection (Supplementary Figure S1). In CLP septic mice, exogenous PEDF aggravated the barrier disruption observed in both the tracheal vasculature (Figures 2A and 2B) and the cremaster muscle microvasculature (Supplementary Figure S1). Moreover, blockade of PEDF using a PEDF–mAb prevented the vascular damage previously observed in the trachea (Figures 2A and 2B) and cremaster muscle (Supplementary Figure S1) of CLP septic mice.

Exogenous PEDF led to increased extravasation of tracer in mice

Figure 2
Exogenous PEDF led to increased extravasation of tracer in mice

(A) Twelve hours after PEDF (10 μg/20 g, the dose was based on the serum concentration of PEDF in sepsis) or PEDF–mAb (2 μg/20 g) injection, the trachea vasculature of normal or septic mice was stained for CD31 (red). Green represents microspheres in the extravascular space, n=3 in each group. Scale bars=50 μm. (B) Quantification of the data in (A) using ImagePro Plus 6.0 software. The data are shown as the mean±S.D. (C) Twelve hours after PEDF or PEDF–mAb injection, FITC–dextran (70 kDa; 10 mg/20 g) was injected into the normal or CLP septic mice. Blood samples were collected 1 h later and relative fluorescence units of serum were measured. The data are shown as the mean±S.D., n=3 in each group. (D) Systemic (intravenous) administration of EB in different groups of mice was followed 2 h later by intradermal injection of PBS or PEDF (1 μg/50 μl). EB dye leakage in dorsal skin was assessed after 30 min, n=3 in each group. Scale bars=1 cm. (E) Quantification of extravasated EB by dimethylformamide extraction from skin samples in (D). The data are shown as the mean±S.D. *P<0.05, **P<0.01.

Figure 2
Exogenous PEDF led to increased extravasation of tracer in mice

(A) Twelve hours after PEDF (10 μg/20 g, the dose was based on the serum concentration of PEDF in sepsis) or PEDF–mAb (2 μg/20 g) injection, the trachea vasculature of normal or septic mice was stained for CD31 (red). Green represents microspheres in the extravascular space, n=3 in each group. Scale bars=50 μm. (B) Quantification of the data in (A) using ImagePro Plus 6.0 software. The data are shown as the mean±S.D. (C) Twelve hours after PEDF or PEDF–mAb injection, FITC–dextran (70 kDa; 10 mg/20 g) was injected into the normal or CLP septic mice. Blood samples were collected 1 h later and relative fluorescence units of serum were measured. The data are shown as the mean±S.D., n=3 in each group. (D) Systemic (intravenous) administration of EB in different groups of mice was followed 2 h later by intradermal injection of PBS or PEDF (1 μg/50 μl). EB dye leakage in dorsal skin was assessed after 30 min, n=3 in each group. Scale bars=1 cm. (E) Quantification of extravasated EB by dimethylformamide extraction from skin samples in (D). The data are shown as the mean±S.D. *P<0.05, **P<0.01.

Under homoeostatic conditions, most of the FITC–dextran tracer (70 kDa) was retained within the blood and skin blood vessels limited the passage of dextran larger than 70 kDa [24,25]. The decrease in blood fluorescence of FITC–dextran suggested that tracer shifted into the interstitial space after PEDF injection (Figure 2C). In addition, blood fluorescence further declined in septic mice injected with PEDF compared with septic mice that were not injected (Figure 2C). However, fluorescence was notably increased in septic mice following PEDF–mAb injection, as compared with septic mice and then recovered to near-normal level (Figure 2C). In the EB dye assay, circulating dye that escaped into the skin was increased 1.9-fold after local injection of PEDF compared with the control group injected with PBS (Figures 2D and 2E). In CLP septic mice, leakage of EB dye was increased 1.4-fold after PEDF injection (Figures 2D and 2E). Notably, septic mice with PEDF–mAb injection exhibited significantly less EB leakage compared with the septic group that was not injected (Figures 2D and 2E). Combined, these data demonstrate that an elevated level of circulating PEDF induces an increase in extravasation, and blocking endogenous PEDF with a PEDF–mAb prevented the increased extravasation observed during sepsis in vivo. Thus, PEDF may be responsible for an increase in vascular permeability in sepsis.

PEDF disrupted ZO-1-based junctional complexes and rearranged actin in vitro

Treatment of HDMECs with PEDF decreased the TER in a dose- and time-dependent manner and stimulation with PEDF at a dose of 1000 ng/ml led to a significant decrease in TER at 24 and 48 h (Figure 3A). We further analysed the effect of PEDF on paracellular permeability by measuring the leakage of FITC–dextran tracer (40 kDa) across cultured endothelial cell monolayers. Stimulation with PEDF (1000 ng/ml) caused an increase in fluorescent leakage under normal or LPS- (1000 ng/ml)/TNF-α- (500 ng/ml) induced inflammatory conditions (Figure 3B).

PEDF disrupted ZO-1-based junctional complexes and rearranged actin

Figure 3
PEDF disrupted ZO-1-based junctional complexes and rearranged actin

(A) Temporal changes in TER were measured across a microvascular endothelial monolayer grown on a Transwell insert treated with vehicle (unstimulated; Control) and PEDF alone. PEDF was used at concentrations of 100, 200, 500 and 1000 ng/ml. Analyses were performed in triplicate and the data are shown as the mean±S.D. (B) Endothelial cells were incubated with vehicle (unstimulated: Control), PEDF (1000 ng/ml), LPS (1000 ng/ml), TNF-α (500 ng/ml), LPS + PEDF or TNF-α + PEDF for 24 h. Paracellular permeability was measured following the addition of FITC–dextran (40 kDa) for 1 h. Analyses were performed in triplicate and the data are shown as the mean±S.D. (C) Endothelial cells were incubated using the conditions described in (B). Cells were stained for F-actin (red). White asterisks indicate small gaps between the cells. White and blue arrows indicate central and peripheral F-actin fibres in cells respectively. Scale bars=10 μm. (D) Morphometric analysis of the data in (C) using ImagePro Plus 6.0 software. The data are shown as the mean±S.D. (E) Different cell groups were stained for ZO-1 (green). Right panels show magnification of boxed area in left panels. Representative images are shown from three separate experiments. Scale bars=10 μm. *P<0.05, **P<0.01.

Figure 3
PEDF disrupted ZO-1-based junctional complexes and rearranged actin

(A) Temporal changes in TER were measured across a microvascular endothelial monolayer grown on a Transwell insert treated with vehicle (unstimulated; Control) and PEDF alone. PEDF was used at concentrations of 100, 200, 500 and 1000 ng/ml. Analyses were performed in triplicate and the data are shown as the mean±S.D. (B) Endothelial cells were incubated with vehicle (unstimulated: Control), PEDF (1000 ng/ml), LPS (1000 ng/ml), TNF-α (500 ng/ml), LPS + PEDF or TNF-α + PEDF for 24 h. Paracellular permeability was measured following the addition of FITC–dextran (40 kDa) for 1 h. Analyses were performed in triplicate and the data are shown as the mean±S.D. (C) Endothelial cells were incubated using the conditions described in (B). Cells were stained for F-actin (red). White asterisks indicate small gaps between the cells. White and blue arrows indicate central and peripheral F-actin fibres in cells respectively. Scale bars=10 μm. (D) Morphometric analysis of the data in (C) using ImagePro Plus 6.0 software. The data are shown as the mean±S.D. (E) Different cell groups were stained for ZO-1 (green). Right panels show magnification of boxed area in left panels. Representative images are shown from three separate experiments. Scale bars=10 μm. *P<0.05, **P<0.01.

We next investigated the mechanism by which PEDF contributed to increased paracellular permeability. Staining of HDMECs for F-actin revealed a profound change in cell morphology. In particular, cells in the control group demonstrated confluency with few intercellular gaps and long thin F-actin fibres in the central portion of cells (Figure 3C, white arrow). However, in cells stimulated with PEDF or LPS/TNF-α, the intercellular space (asterisk) was significantly increased and central F-actin bundles (white arrow) were obvious (Figures 3C and 3D). In cells co-stimulated with PEDF + LPS or PEDF + TNF-α, these stress fibres were thicker (white arrow) and dense wider peripheral F-actin (blue arrow) appeared. Moreover, nearly all the cell–cell junctions were disrupted with considerable gaps in the monolayer (asterisk in Figures 3C and 3D).

Tight junctions play an important role in the function of the vascular endothelial (VE) barrier and ZO-1 is the principal protein involved in the assembly of tight junction. Because ZO-1 protein is linked to the cytoskeletal protein F-actin, we evaluated the subcellular distribution of ZO-1 in HDMECs. As illustrated in Figure 3(E), ZO-1 fluorescence was concentrated at cell–cell borders in the normal group, whereas in cells treated with PEDF, LPS or TNF-α, ZO-1 staining was weak and localized in a discontinuous fashion around the cell membrane. Moreover, disassembly of ZO-1 along the cell membrane was more severe in cells co-stimulated with PEDF + LPS or PEDF + TNF-α (the enlarged area in Figure 3E). These results provide clear evidence that PEDF contributes to increased paracellular permeability by altering the distribution of ZO-1 and F-actin in conditions of sepsis.

RhoA signalling activation mediated PEDF-induced permeability

RhoA has been implicated as a potent regulator of endothelial barrier function and can stimulate the formation of F-actin stress fibres [26]. Although total RhoA expression was unchanged, the activated RhoA level was significantly increased following treatment with PEDF, LPS or TNF-α. Furthermore, cells treated with PEDF + LPS or PEDF + TNF-α demonstrated greater activation of RhoA compared with cells treated with LPS or TNF-α alone (Figure 4A). However, ROCK was not activated following PEDF stimulation (Supplementary Figure S2).

RhoA signalling activation mediated PEDF-induced permeability in vitro

Figure 4
RhoA signalling activation mediated PEDF-induced permeability in vitro

(A) Endothelial cells were incubated under the conditions described in Figure 3(B). Cells were rapidly lysed and GTP-bound RhoA was affinity-purified with GST–RBD beads. The levels of GTP-bound RhoA and total cell lysate RhoA were measured by WB. (B) Endothelial cells were incubated with vehicle (unstimulated), PEDF (1000 ng/ml), TNF-α (500 ng/ml) or TNF-α + PEDF for 24 h. C3 transferase (10 μg/ml) was added for the final 4 h of the 24-h PEDF or TNF-α + PEDF incubation. The levels of GTP-bound RhoA and total cell lysate RhoA were measured by WB. (C) Paracellular permeability was assessed using FITC-dextran with the conditions described in (B). After 24 h, paracellular permeability was measured with the addition of FITC-dextran for 1 h. Analyses were performed in triplicate and the data are shown as the mean±S.D. (D) Endothelial cells were stained for F-actin (red). White asterisks indicate small gaps between the cells. White and blue arrows indicate central and peripheral F-actin fibres in cells respectively. Scale bars=10 μm. (E) Cells were stained for ZO-1 (green). Right panels show magnification of boxed areas in left panels. Representative images are shown from three separate experiments. Scale bars=10 μm. *P<0.05, **P<0.01.

Figure 4
RhoA signalling activation mediated PEDF-induced permeability in vitro

(A) Endothelial cells were incubated under the conditions described in Figure 3(B). Cells were rapidly lysed and GTP-bound RhoA was affinity-purified with GST–RBD beads. The levels of GTP-bound RhoA and total cell lysate RhoA were measured by WB. (B) Endothelial cells were incubated with vehicle (unstimulated), PEDF (1000 ng/ml), TNF-α (500 ng/ml) or TNF-α + PEDF for 24 h. C3 transferase (10 μg/ml) was added for the final 4 h of the 24-h PEDF or TNF-α + PEDF incubation. The levels of GTP-bound RhoA and total cell lysate RhoA were measured by WB. (C) Paracellular permeability was assessed using FITC-dextran with the conditions described in (B). After 24 h, paracellular permeability was measured with the addition of FITC-dextran for 1 h. Analyses were performed in triplicate and the data are shown as the mean±S.D. (D) Endothelial cells were stained for F-actin (red). White asterisks indicate small gaps between the cells. White and blue arrows indicate central and peripheral F-actin fibres in cells respectively. Scale bars=10 μm. (E) Cells were stained for ZO-1 (green). Right panels show magnification of boxed areas in left panels. Representative images are shown from three separate experiments. Scale bars=10 μm. *P<0.05, **P<0.01.

To determine whether RhoA contributed to PEDF-induced disruption of cell junctions and hyperpermeability, the effects of C3 transferase (RhoA inhibitor) on PEDF-induced actin rearrangement, ZO-1 redistribution and FITC–dextran tracer leakage were assessed. Indeed, RhoA activity returned to the basal level after treatment with C3 transferase (Figure 4B). Correspondingly, the increase in FITC–dextran leakage was also blocked by treatment with C3 transferase (Figure 4C). Moreover, C3 transferase abolished the effects of PEDF, including F-actin rearrangement and loss of ZO-1 along cell–cell junctions (Figures 4D and 4E). Taken together, these results support the importance of RhoA signalling during PEDF-induced hyperpermeability, which results from the rearrangement of actin and the disruption of ZO-1-based junctional complexes during sepsis.

The dynamics of PEDF-R expression at the plasma membrane in sepsis

Considering that PEDF is thought to exert its biological action by binding to the cell surface receptors ATGL or LR, we hypothesized that PEDF activated RhoA signalling by interacting with ATGL or LR.

To identify the receptor that mediates RhoA activation in sepsis, we first investigated whether there was variation in the expression of ATGL and LR during sepsis. Plasma membrane proteins were extracted and VE–cadherin, VDAC, calnexin, histone and GAPDH were used as marker proteins for the plasma membrane, mitochondria, endoplasmic reticulum, nucleus and cytosol respectively. The differential expression of subcellular fractions confirmed the purity of the extracted plasma membrane proteins (Figure 5A). Then, we assessed the expression of ATGL and LR in the total cell lysate and in the plasma membrane fraction before and after inflammatory stimulation (Figure 5B). The results demonstrated that ATGL expression was significantly increased in both the total cell lysate and the membrane fraction after stimulation, whereas LR expression remained unchanged.

The expression of PEDF–R–ATGL was increased in sepsis

Figure 5
The expression of PEDF–R–ATGL was increased in sepsis

(A) The purity of membrane proteins was assessed. VE-cadherin, VDAC, calnexin, histone and GAPDH were used as marker proteins for the plasma membrane, mitochondria, endoplasmic reticulum, nucleus and cytosol respectively. (B) Endothelial cells were incubated with vehicle (unstimulated; Control), LPS (1000 ng/ml) or TNF-α (500 ng/ml) for 24 h. Cells were rapidly lysed and the expression of ATGL and LR in the total cell lysate or plasma membrane fraction was measured by WB. (C) PEDF binding to ATGL/LR was evaluated in the presence of LPS/TNF-α. Cells were maintained in LPS and TNF-α for 24 h, followed by stimulation with PEDF for 6 h. Total cell lysates were immunoprecipitated with anti-PEDF antibody and blotted with ATGL, LR or PEDF antibody. (D) Expression of PEDF receptors in vivo. Twelve hours after establishing the CLP-induced sepsis model, heart, liver, lung, kidney and intestine tissues were rapidly lysed. The expression of ATGL and LR in total tissue lysate was measured by WB.

Figure 5
The expression of PEDF–R–ATGL was increased in sepsis

(A) The purity of membrane proteins was assessed. VE-cadherin, VDAC, calnexin, histone and GAPDH were used as marker proteins for the plasma membrane, mitochondria, endoplasmic reticulum, nucleus and cytosol respectively. (B) Endothelial cells were incubated with vehicle (unstimulated; Control), LPS (1000 ng/ml) or TNF-α (500 ng/ml) for 24 h. Cells were rapidly lysed and the expression of ATGL and LR in the total cell lysate or plasma membrane fraction was measured by WB. (C) PEDF binding to ATGL/LR was evaluated in the presence of LPS/TNF-α. Cells were maintained in LPS and TNF-α for 24 h, followed by stimulation with PEDF for 6 h. Total cell lysates were immunoprecipitated with anti-PEDF antibody and blotted with ATGL, LR or PEDF antibody. (D) Expression of PEDF receptors in vivo. Twelve hours after establishing the CLP-induced sepsis model, heart, liver, lung, kidney and intestine tissues were rapidly lysed. The expression of ATGL and LR in total tissue lysate was measured by WB.

We next investigated the interactions between PEDF and these two receptors. Stimulation with LPS or TNF-α did not lead to expression of PEDF in HDMECs and the concentration of PEDF in the medium was unchanged (Supplementary Figure S3). This finding suggested that, during the process of immunoprecipitation, the concentration of PEDF remained the same. However, treatment with LPS or TNF-α resulted in increased ATGL–PEDF association, whereas the association of LR–PEDF was not altered (Figure 5C). Moreover, in conditions of sepsis, the expression of ATGL was increased in five tissues, including the heart, liver, lung, kidney and intestine, whereas LR expression remained unchanged (Figure 5D). Together, these results show that expression of the PEDF-R–ATGL is significantly increased and the association of PEDF and ATGL is strengthened under conditions of sepsis.

PEDF regulated permeability through the plasma membrane receptor ATGL in sepsis

To more directly address the role of ATGL in the course of PEDF-induced hyperpermeability during sepsis, we established ATGL–/LR–shRNA-treated HDMECs. Total protein was extracted and WB was performed to confirm the knockdown effects of ATGL–and LR–shRNA (Figure 6A). RhoA activation induced by PEDF and PEDF + TNF-α was inhibited in ATGL–shRNA-treated HDMECs (Figure 6B). FITC–dextran leakage was decreased (Figure 6C) and similar effects on ZO-1 distribution were observed (Figure 6D). However, in LR–shRNA-treated HDMECs, there were no significant differences in RhoA activity or fluorescent leakage when compared with scrambled control cells (Figures 6E and 6F) and PEDF treatment still led to the loss of junctional ZO-1 expression (Figure 6G). These results demonstrate that PEDF induces hyperpermeability in sepsis through the plasma membrane receptor ATGL, but not LR.

PEDF regulated paracellular permeability through the plasma membrane receptor ATGL in sepsis

Figure 6
PEDF regulated paracellular permeability through the plasma membrane receptor ATGL in sepsis

(A) HDMECs were infected with scrambled shRNA (scrambled) and recombinant lentivirus vectors to silence ATGL/LR (shRNA). Knockdown efficiency was confirmed by WB. (B and E) Normal cells and ATGL–/LR–shRNA-treated cells were either left untreated or stimulated with PEDF or TNF-α + PEDF for 24 h. The amount of GTP-bound RhoA and total cell lysate RhoA were measured by WB. (C and F) Paracellular permeability to FITC–dextran was evaluated using the conditions described in (B and E). Paracellular permeability was measured following the addition of FITC–dextran for 1 h. Analyses were performed in triplicate and the data are shown as the mean±S.D. (D and G) Different cell groups were stained for ZO-1 (green). Right panels show magnification of boxed areas in left panels. Representative images are shown from three separate experiments. Scale bars=10 μm. *P<0.05, **P<0.01.

Figure 6
PEDF regulated paracellular permeability through the plasma membrane receptor ATGL in sepsis

(A) HDMECs were infected with scrambled shRNA (scrambled) and recombinant lentivirus vectors to silence ATGL/LR (shRNA). Knockdown efficiency was confirmed by WB. (B and E) Normal cells and ATGL–/LR–shRNA-treated cells were either left untreated or stimulated with PEDF or TNF-α + PEDF for 24 h. The amount of GTP-bound RhoA and total cell lysate RhoA were measured by WB. (C and F) Paracellular permeability to FITC–dextran was evaluated using the conditions described in (B and E). Paracellular permeability was measured following the addition of FITC–dextran for 1 h. Analyses were performed in triplicate and the data are shown as the mean±S.D. (D and G) Different cell groups were stained for ZO-1 (green). Right panels show magnification of boxed areas in left panels. Representative images are shown from three separate experiments. Scale bars=10 μm. *P<0.05, **P<0.01.

DISCUSSION

In the present study, we found that serum PEDF level was significantly elevated in sepsis and exogenous PEDF induced markedly increased extravasation of EB, dextran and microspheres in mouse skin, blood, trachea and cremaster muscle both in normal and in the septic condition in vivo. Moreover, the expression of PEDF receptor ATGL was up-regulated in the presence of inflammatory mediators, which was accompanied by the enhanced association between PEDF and ATGL. The arrest of extravasation appeared to be specific to the PEDF–mAb in vivo and ATGL–shRNA in vitro. Our results showed that PEDF exerted this effect by binding to the ATGL receptor, which led to the activation of RhoA, actin rearrangement and intercellular junction disruption. These data provide a mechanism that explains how PEDF-induced hyperpermeability in sepsis is regulated at the molecular level.

An important component of sepsis is VE dysfunction, which is characterized by increased microvascular leakage due to a combination of endothelial damage, loss of intercellular junctional integrity and remodelling of the cellular cytoskeleton [7]. The endothelium is critical to the vascular leak and subsequent shock that contribute to mortality among patients with sepsis [27]. Yet, the underlying mechanisms associated with endothelial dysfunction and vascular hyperpermeability during sepsis are not clear.

PEDF, an endogenously produced protein that belongs to the superfamily of serine protease inhibitors, is one of the most potent inhibitors of angiogenesis [28]. Serum PEDF level is found to be elevated in patients with metabolic syndrome, diabetes mellitus, atherosclerosis or polycystic ovary syndrome [2931]. The elevated PEDF in diabetic patients is found to be associated with microvascular complications and poor vascular health [32]. Recently, PEDF was found to be significantly elevated in patients with sepsis [8,9], although its role in vascular permeability remains elusive. The conclusions from previous studies suggest that PEDF exerts several important but contradictory influences on the VE barrier [1113]. In the current study, we found that exogenous PEDF led to an increase in paracellular permeability in a dose- and time-dependent fashion, which was consistent with a previous study that PEDF when used alone at high concentrations did not block VEGF and caused a decrease in TER level inversely [33]. In addition, PEDF further increased the monolayer permeability induced by inflammatory mediators. Therefore, elevated serum PEDF may act as a potent pro-permeability molecule during the condition of sepsis.

It is well known that the concentration of serum VEGF is elevated in sepsis [34] and this was also observed in the patients with sepsis in our study (Supplementary Figure S4). Although our analysis supports the notion that PEDF is detrimental to the endothelial barrier, it is important to address the failure of PEDF to inhibit VEGF-induced vascular hyperpermeability [35] in sepsis. Thus, it remains unclear whether the elevated serum level of PEDF in sepsis is insufficient to counter-regulate VEGF or whether PEDF triggers other pathways that increase vascular permeability under conditions of sepsis. Considering that PEDF exerts its biological actions by binding to cell-surface receptors, we hypothesized that these receptors may be selectively activated during sepsis. There are two main PEDF cell-surface receptors: ATGL (encoded by PNPLA2) and LR (encoded by RPSA). PNPLA2 possesses phospholipase A2 (PLA2) activity that is enhanced upon PEDF binding, leading to the release of bioactive non-esterified fatty acids (NEFAs) and lysophosphatidic acids (LPAs), which can function as second messengers or interact with G protein-coupled receptors [36,37]. NEFA and LPA may also enhance the ability of neuronal cells to survive and differentiate, but may cause irreversible damage to endothelial and tumour cells [38]. Moreover, it has been suggested that PEDF mediates anti-angiogenic, anti-tumorigenic and neurotrophic activity through its interaction with the ATGL receptor [3942]. The other receptor LR has also been reported to mediate the anti-angiogenic effect of PEDF by inducing endothelial cell apoptosis and reducing their migration [43]. In the present study, we observed that the expression of ATGL was significantly up-regulated both in vivo and in vitro and the association between PEDF and ATGL was concomitantly enhanced. Moreover, PEDF-induced hyperpermeability was virtually abolished in ATGL deficient cells, which suggests that the altered expression of ATGL may favour an increase in PEDF-mediated vascular permeability.

Disruption of the cytoskeleton (microtubules, microfilaments and intermediate filaments) and cell junctions plays a pivotal role in increasing vascular permeability [44], and Rho family GTPases regulate both actin-based cytoskeletal organization and integrity of intercellular junctions. Furthermore, Rho GTPases have been implicated in intracellular signalling induced by several vasoactive substances, such as thrombin, histamine, TNF-α, LPS, LPA and VEGF [45]. In the past decade, the small GTPase RhoA has been implicated as a potent regulator of endothelial barrier function. Activation of RhoA by vasoactive substances results in a loss of integrity of the endothelial barrier, both in vitro and in vivo [46,47] and it was reported that RhoA activation is abolished in the presence of chemoattractants in Atgl−/− macrophages [48]. In our study, RhoA was significantly activated by PEDF, whereas PEDF-induced hyperpermeability could be restored by treatment with the RhoA inhibitor C3 transferase. Moreover, RhoA activation was significantly abolished in the presence of PEDF in ATGL–shRNA-treated cells, confirming the critical role of RhoA in PEDF-induced permeability through its interaction with ATGL. However, as the downstream effector of RhoA, ROCK was not activated after PEDF stimulation. Indeed, previous literature has demonstrated that ROCK does not affect TNF-α-induced hyperpermeability [17], which suggests that ROCK may be not be the downstream effector of RhoA in PEDF-induced hyperpermeability during sepsis. However, further studies are needed to determine the intermediate effectors linking PEDF–ATGL binding and RhoA activation as well as the downstream signalling pathways involved.

In summary, the present study demonstrates that elevated levels of serum PEDF in sepsis induce increased extravasation both in normal conditions and in sepsis in vivo. Exogenous PEDF increases paracellular permeability by targeting the plasma membrane ATGL receptor, leading to actin rearrangement and endothelial junction disruption through activating the RhoA pathway in vitro. Importantly, this damage could be blocked by treatment with PEDF–mAb in vivo or ATGL–shRNA in vitro, which may provide new potential therapeutic strategies for the treatment of hyperpermeability in sepsis.

AUTHOR CONTRIBUTION

Ting He and Jiongyu Hu conceived the study, designed the experiments and drafted the paper. Ting He performed the experiments, acquired and analysed the data. Guangning Yan, Lingfei Li and Qiong Zhang helped to perform the experiments and prepare the paper. Dongxia Zhang participated in the study design and co-ordination. Bing Chen and Yuesheng Huang were involved in critically revising the paper for important intellectual content and have given their final approval of the version to be published. All authors have read and approved the final version of this manuscript.

FUNDING

This work was supported by the National Natural Science Foundation of China [grant number 81401595 (to T.H.) and 81270030 (to J.H.)]; the State Key Laboratory of Trauma, Burns and Combined Injury Research Foundation [grant number 2012ZZ(II)05 (to J.H.) and SKLZZ201103 (to Y.H.)]; and the Specific Project of Health, Ministry of Health of China [grant number 201202002].

Abbreviations

     
  • ATGL

    adipose triglyceride lipase

  •  
  • BMI

    body mass index

  •  
  • CLP

    cecal ligation and puncture

  •  
  • EB

    Evan's Blue

  •  
  • GAPDH

    glyceraldehyde-3-phosphate dehydrogenase

  •  
  • GST–RBD

    glutathione sepharose-coupled rhotekin Rho-binding domain

  •  
  • HDMEC

    human dermal microvascular endothelial cell

  •  
  • LPA

    lysophosphatidic acid

  •  
  • LPS

    lipopolysaccharide

  •  
  • LR

    laminin receptor

  •  
  • mAb

    monoclonal antibody

  •  
  • PEDF

    pigment epithelium-derived factor

  •  
  • PNPLA2

    patatin-like phospholipase domain-containing protein 2

  •  
  • RIPA

    radio immunoprecipitation assay

  •  
  • RPSA

    ribosomal protein SA

  •  
  • TER

    transendothelial electrical resistance

  •  
  • TNF

    tumour necrosis factor

  •  
  • VDAC

    voltage-dependent anion channel

  •  
  • VE

    vascular endothelial

  •  
  • VEGF

    vascular endothelial growth factor

  •  
  • WB

    Western blotting

  •  
  • ZO-1

    zonula occludens-1

References

References
1
Levy
 
M.M.
Fink
 
M.P.
Marshall
 
J.C.
Abraham
 
E.
Angus
 
D.
Cook
 
D.
Cohen
 
J.
Opal
 
S.M.
Vincent
 
J.L.
Ramsay
 
G
SCCM/ESICM/ACCP/ATS/SIS.
 
2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference
Crit. Care Med.
2003
, vol. 
31
 (pg. 
1250
-
1256
)
[PubMed]
2
Bone
 
R.C.
Balk
 
R.A.
Cerra
 
F.B.
Dellinger
 
R.P.
Fein
 
A.M.
Knaus
 
W.A.
Schein
 
R.M.
Sibbald
 
W.J.
 
Definitions for sepsis and organ failure and guidelines for the use ofinnovative therapies in sepsis. The ACCP/SCCM consensus conference committee
American College of Chest Physicians/Society of Critical Care Medicine. Chest
1992
, vol. 
101
 (pg. 
1644
-
1655
)
3
Dellinger
 
R.P.
Levy
 
M.M.
Rhodes
 
A.
Annane
 
D.
Gerlach
 
H.
Opal
 
S.M.
Sevransky
 
J.E.
Sprung
 
C.L.
Douglas
 
I.S.
Jaeschke
 
R.
 
Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock, 2012
Intensive Care Med.
2013
, vol. 
39
 (pg. 
165
-
228
)
[PubMed]
4
Angus
 
D.C.
 
The search for effective therapy for sepsis: back to the drawing board?
JAMA
2011
, vol. 
306
 (pg. 
2614
-
2615
)
[PubMed]
5
Angus
 
D.C.
van der Poll
 
T.
 
Severe sepsis and septic shock
N. Engl. J. Med.
2013
, vol. 
369
 (pg. 
840
-
851
)
[PubMed]
6
Bansch
 
P.
Nelson
 
A.
Ohlsson
 
T.
Bentzer
 
P.
 
Effect of charge on microvascular permeability in early experimental sepsis in the rat
Microvasc. Res.
2011
, vol. 
82
 (pg. 
339
-
345
)
[PubMed]
7
Aird
 
W.C.
 
Phenotypic heterogeneity of the endothelium: structure, function, and mechanisms
Circ. Res.
2007
, vol. 
100
 (pg. 
158
-
173
)
[PubMed]
8
He
 
T.
Hu
 
J.Y.
Han
 
J.
Zhang
 
D.X.
Jiang
 
X.P.
Chen
 
B.
Huang
 
Y.S.
 
Identification of differentially expressed serum proteins in infectious purpura fulminans
Dis. Markers
2014
, vol. 
2014
 pg. 
698383
 
[PubMed]
9
Nakamura
 
T.
Yamagishi
 
S.I.
 
PEDF and septic shock
Curr. Mol. Med.
2010
, vol. 
10
 (pg. 
312
-
316
)
[PubMed]
10
Becerra
 
S.P.
Notario
 
V.
 
The effects of PEDF on cancer biology: mechanisms of action and therapeutic potential
Nat. Rev. Cancer
2013
, vol. 
13
 (pg. 
258
-
271
)
[PubMed]
11
Cai
 
J.
Chen
 
Z.
Ruan
 
Q.
Han
 
S.
Liu
 
L.
Qi
 
X.
Boye
 
S.L.
Hauswirth
 
W.W.
Grant
 
M.B.
Boulton
 
M.E.
 
γ-Secretase and presenilin mediate cleavage and phosphorylation of vascular endothelial growth factor receptor-1
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
42514
-
42523
)
[PubMed]
12
Konson
 
A.
Pradeep
 
S.
D'Acunto
 
C.W.
Seger
 
R.
 
Pigment epithelium-derived factor and its phosphomimetic mutant induce JNK-dependent apoptosis and p38-mediated migration arrest
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
3540
-
3551
)
[PubMed]
13
Ho
 
T.C.
Chen
 
S.L.
Yang
 
Y.C.
Liao
 
C.L.
Cheng
 
H.C.
Tsao
 
Y.P.
 
PEDF induces p53-mediated apoptosis through PPAR gamma signaling in human umbilical vein endothelial cells
Cardiovasc. Res.
2007
, vol. 
76
 (pg. 
213
-
223
)
[PubMed]
14
Chuderland
 
D.
Ben-Ami
 
I.
Kaplan-Kraicer
 
R.
Grossman
 
H.
Ron-El
 
R.
Shalgi
 
R.
 
The role of pigment epithelium-derived factor in the pathophysiology and treatment of ovarian hyperstimulation syndrome in mice
J. Clin. Endocrinol. Metab.
2013
, vol. 
98
 (pg. 
E258
-
E266
)
[PubMed]
15
Yizhi
 
P.
Jing
 
C.
Zhiqiang
 
Y.
Xiaolu
 
L.
Gaoxing
 
L.
Jun
 
W
 
Diagnostic criteria and treatment protocol for post-burn sepsis
Crit. Care
2013
, vol. 
17
 pg. 
406
 
[PubMed]
16
Orban
 
C.
 
Diagnostic criteria for sepsis in burn patients
Chirurgia
2012
, vol. 
107
 (pg. 
697
-
700
)
[PubMed]
17
McKenzie
 
J.A.
Ridley
 
A.J.
 
Roles of Rho/ROCK and MLCK in TNF-a-induced changes in endothelial morphology and permeability
J. Cell. Physiol.
2007
, vol. 
213
 (pg. 
221
-
228
)
[PubMed]
18
Murakami
 
M.
Nguyen
 
L.T.
Zhuang
 
Z.W.
Moodie
 
K.L.
Carmeliet
 
P.
Stan
 
R.V.
Simons
 
M.
 
The FGF system has a key role in regulating vascular integrity
J. Clin. Invest.
2008
, vol. 
118
 (pg. 
3355
-
3366
)
[PubMed]
19
Sun
 
Z.
Li
 
X.
Massena
 
S.
Kutschera
 
S.
Padhan
 
N.
Gualandi
 
L.
Sundvold-Gjerstad
 
V.
Gustafsson
 
K.
Choy
 
W.W.
Zang
 
G.
, et al 
VEGFR2 induces c-Src signaling and vascular permeability in vivo via the adaptor protein TSAd
J. Exp. Med.
2012
, vol. 
209
 (pg. 
1363
-
1377
)
[PubMed]
20
Ibla
 
J.C.
Khoury
 
J.
 
Methods to assess tissue permeability
Methods Mol. Biol.
2013
, vol. 
1066
 (pg. 
81
-
88
)
[PubMed]
21
Baczyk
 
D.
Drewlo
 
S.
Proctor
 
L.
Dunk
 
C.
Lye
 
S.
Kingdom
 
J.
 
Glial cell missing-1 transcription factor is required for the differentiation of the human trophoblast
Cell Death Differ.
2009
, vol. 
16
 (pg. 
719
-
727
)
[PubMed]
22
Kutner
 
R.H.
Zhang
 
X.Y.
Reiser
 
J.
 
Production, concentration and titration of pseudotyped HIV-1-based lentiviral vectors
Nat. Protoc.
2009
, vol. 
4
 (pg. 
495
-
505
)
[PubMed]
23
Doi
 
K.
Leelahavanichkul
 
A.
Yuen
 
P.S.
Star
 
R.A.
 
Animal models of sepsis and sepsis-induced kidney injury
J. Clin. Invest.
2009
, vol. 
119
 (pg. 
2868
-
2878
)
[PubMed]
24
Stan
 
R.V.
Tse
 
D.
Deharvengt
 
S.J.
Smits
 
N.C.
Xu
 
Y.
Luciano
 
M.R.
McGarry
 
C.L.
Buitendijk
 
M.
Nemani
 
K.V.
Elgueta
 
R.
, et al 
The diaphragms of fenestrated endothelia: gatekeepers of vascular permeability and blood composition
Dev. Cell
2012
, vol. 
23
 (pg. 
1203
-
1218
)
[PubMed]
25
Egawa
 
G.
Nakamizo
 
S.
Natsuaki
 
Y.
Doi
 
H.
Miyachi
 
Y.
Kabashima
 
K.
 
Intravital analysis of vascular permeability in mice using two-photon microscopy
Sci. Rep.
2013
, vol. 
3
 pg. 
1932
 
[PubMed]
26
Szulcek
 
R.
Beckers
 
C.M.
Hodzic
 
J.
de Wit
 
J.
Chen
 
Z.
Grob
 
T.
Musters
 
R.J.
Minshall
 
R.D.
van Hinsbergh
 
V.W.
van Nieuw Amerongen
 
G.P.
 
Localized RhoA GTPase activity regulates dynamics of endothelial monolayer integrity
Cardiovasc. Res.
2013
, vol. 
99
 (pg. 
471
-
482
)
[PubMed]
27
Lee
 
W.L.
Slutsky
 
A.S.
 
Sepsis and endothelial permeability
N. Engl. J. Med.
2010
, vol. 
363
 (pg. 
689
-
691
)
[PubMed]
28
Nelius
 
T.
Martinez-Marin
 
D.
Hirsch
 
J.
Miller
 
B.
Rinard
 
K.
Lopez
 
J.
de Riese
 
W.
Filleur
 
S.
 
Pigment epithelium-derived factor expression prolongs survival and enhances the cytotoxicity of low-dose chemotherapy in castration-refractory prostate cancer
Cell Death Dis
2014
, vol. 
5
 pg. 
e1210
 
[PubMed]
29
Jenkins
 
A.J.
Fu
 
D.
Azar
 
M.
Stoner
 
J.A.
Kaufman
 
D.G.
Zhang
 
S.
Klein
 
R.L.
Lopes-Virella
 
M.F.
Ma
 
J.X.
Lyons
 
T.J.
 
VADT investigators
Clinical correlates of serum pigment epithelium-derived factor in type 2 diabetes patients
J. Diabetes Complications
2014
, vol. 
28
 (pg. 
353
-
359
)
[PubMed]
30
Cheng
 
Q.
Xia
 
W.
Yang
 
S.
Ye
 
P.
Mei
 
M.
Song
 
Y.
Luo
 
M.
Li
 
Q.
 
Association of serum pigment epithelium-derived factor with high-sensitivity C-reactive protein in women with polycystic ovary syndrome
J. Endocrinol. Invest.
2013
, vol. 
36
 (pg. 
632
-
635
)
[PubMed]
31
Tahara
 
N.
Yamagishi
 
S.
Tahara
 
A.
Takeuchi
 
M.
Imaizumi
 
T.
 
Serum levels of pigment epithelium-derived factor, a novel marker of insulin resistance, are independently associated with fasting apolipoprotein B48 levels in humans
Clin. Biochem.
2012
, vol. 
45
 (pg. 
1404
-
1408
)
[PubMed]
32
Jenkins
 
A.J.
Zhang
 
S.X.
Rowley
 
K.G.
Karschimkus
 
C.S.
Nelson
 
C.L.
Chung
 
J.S.
O'Neal
 
D.N.
Januszewski
 
A.S.
Croft
 
K.D.
Mori
 
T.A.
, et al 
Increased serum pigment epithelium-derived factor is associated with microvascular complications, vascular stiffness and inflammation in type 1 diabetes
Diabet. Med.
2007
, vol. 
24
 (pg. 
1345
-
1351
)
[PubMed]
33
Yang
 
J.
Duh
 
E.J.
Caldwell
 
R.B.
Behzadian
 
M.A.
 
Antipermeability function of PEDF involves blockade of the MAP kinase/GSK/beta-catenin signaling pathway and uPAR expression
Invest. Ophthalmol. Vis. Sci.
2010
, vol. 
51
 (pg. 
3273
-
3280
)
[PubMed]
34
Jesmin
 
S.
Zaedi
 
S.
Islam
 
A.M.
Sultana
 
S.N.
Iwashima
 
Y.
Wada
 
T.
Yamaguchi
 
N.
Hiroe
 
M.
Gando
 
S.
 
Time-dependent alterations of VEGF and its signaling molecules in acute lung injury in a rat model of sepsis
Inflammation
2012
, vol. 
35
 (pg. 
484
-
500
)
[PubMed]
35
Cai
 
J.
Wu
 
L.
Qi
 
X.
Li Calzi
 
S.
Caballero
 
S.
Shaw
 
L.
Ruan
 
Q.
Grant
 
M.B.
Boulton
 
M.E.
 
PEDF regulates vascular permeability by a γ-secretase-mediated pathway
PLoS One
2011
, vol. 
6
 pg. 
e21164
 
[PubMed]
36
Goetzl
 
E.J.
Tigyi
 
G.
 
Lysophospholipids and their G protein-coupled receptors in biology and diseases
J. Cell. Biochem.
2004
, vol. 
92
 (pg. 
867
-
8687
)
[PubMed]
37
Balsinde
 
J.
Winstead
 
M.V.
Dennis
 
E.A.
 
Phospholipase A(2) regulation of arachidonic acid mobilization
FEBS Lett.
2002
, vol. 
531
 (pg. 
2
-
6
)
[PubMed]
38
Notari
 
L.
Baladron
 
V.
Aroca-Aguilar
 
J.D.
Balko
 
N.
Heredia
 
R.
Meyer
 
C.
Notario
 
P.M.
Saravanamuthu
 
S.
Nueda
 
M.L.
Sanchez-Sanchez
 
F.
, et al 
Identification of a lipase-linked cell membrane receptor for pigment epithelium-derived factor
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
38022
-
3803
)
[PubMed]
39
Rose
 
D.P.
Connolly
 
J.M.
 
Omega-3 fatty acids as cancer chemopreventive agents
Pharmacol. Ther.
1999
, vol. 
83
 (pg. 
217
-
244
)
[PubMed]
40
Rose
 
D.P.
Connolly
 
J.M.
 
Antiangiogenicity of docosahexaenoic acid and its role in the suppression of breast cancer cell growth in nude mice
Int. J. Oncol.
1999
, vol. 
15
 (pg. 
1011
-
1015
)
[PubMed]
41
Mukherjee
 
P.K.
Marcheselli
 
V.L.
Serhan
 
C.N.
Bazan
 
N.G.
 
Neuroprotectin D1: a docosahexaenoic acid-derived docosatriene protects human retinal pigment epithelial cells from oxidative stress
Proc. Natl. Acad. Sci. U.S.A.
2004
, vol. 
101
 (pg. 
8491
-
8496
)
[PubMed]
42
Bazan
 
N.G.
 
Neuroprotectin D1 (NPD1): a DHA-derived mediator that protects brain and retina against cell injury-induced oxidative stress
Brain Pathol
2005
, vol. 
15
 (pg. 
159
-
166
)
[PubMed]
43
Bernard
 
A.
Gao-Li
 
J.
Franco
 
C.A.
Bouceba
 
T.
Huet
 
A.
Li
 
Z.
 
Laminin receptor involvement in the anti-angiogenic activity of pigment epithelium-derived factor
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
10480
-
10490
)
[PubMed]
44
Chu
 
Z.G.
Zhang
 
J.P.
Song
 
H.P.
Hu
 
J.Y.
Zhang
 
Q.
Xiang
 
F.
Huang
 
Y.S.
 
p38 MAP kinase mediates burn serum-induced endothelial barrier dysfunction: involvement of F-actin rearrangement and L-caldesmon phosphorylation
Shock
2010
, vol. 
34
 (pg. 
222
-
228
)
[PubMed]
45
Wojciak-Stothard
 
B.
Ridley
 
A.J.
 
Rho GTPases and the regulation of endothelial permeability
Vascul. Pharmacol.
2002
, vol. 
39
 (pg. 
187
-
199
)
[PubMed]
46
Essler
 
M.
Amano
 
M.
Kruse
 
H.J.
Kaibuchi
 
K.
Weber
 
P.C.
Aepfelbacher
 
M.
 
Thrombin inactivates myosin light chain phosphatase via Rho and its target Rho kinase inhuman endothelial cells
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
21867
-
21874
)
[PubMed]
47
Satchi-Fainaro
 
R.
Mamluk
 
R.
Wang
 
L.
Short
 
S.M.
Nagy
 
J.A.
Feng
 
D.
Dvorak
 
A.M.
Dvorak
 
H.F.
Puder
 
M.
Mukhopadhyay
 
D.
Folkman
 
J.
 
Inhibition of vessel permeability by TNP-470 and its polymer conjugate, caplostatin
Cancer Cell
2005
, vol. 
7
 (pg. 
251
-
261
)
[PubMed]
48
Aflaki
 
E.
Balenga
 
N.A.
Luschnig-Schratl
 
P.
Wolinski
 
H.
Povoden
 
S.
Chandak
 
P.G.
Bogner-Strauss
 
J.G.
Eder
 
S.
Konya
 
V.
Kohlwein
 
S.D.
, et al 
Impaired Rho GTPase activation abrogates cell polarization and migration in macrophages with defective lipolysis
Cell. Mol. Life Sci.
2011
, vol. 
68
 (pg. 
3933
-
3947
)
[PubMed]

Author notes

1

These authors contributed equally to this work.

Supplementary data