Inhibition of placental growth factor activity reduces the severity of fibrosis, inflammation, and portal hypertension in cirrhotic mice


  • Potential conflict of interest: ThromboGenics NV developed PlGF inhibitors for antiangiogenic treatment under a license from VIB and K. U. Leuven. Jean Marie Stassen is the Senior Director of Research & Development at ThromboGenics NV.


Placental growth factor (PlGF) is associated selectively with pathological angiogenesis, and PlGF blockade does not affect the healthy vasculature. Anti-PlGF is therefore currently being clinically evaluated for the treatment of cancer patients. In cirrhosis, hepatic fibrogenesis is accompanied by extensive angiogenesis. In this paper, we evaluated the pathophysiological role of PlGF and the therapeutic potential of anti-PlGF in liver cirrhosis. PlGF was significantly up-regulated in the CCl4-induced rodent model of liver cirrhosis as well as in cirrhotic patients. Compared with wild-type animals, cirrhotic PlGF−/− mice showed a significant reduction in angiogenesis, arteriogenesis, inflammation, fibrosis, and portal hypertension. Importantly, pharmacological inhibition with anti-PlGF antibodies yielded similar results as genetic loss of PlGF. Notably, PlGF treatment of activated hepatic stellate cells induced sustained extracellular signal-regulated kinase 1/2 phosphorylation, as well as chemotaxis and proliferation, indicating a previously unrecognized profibrogenic role of PlGF. Conclusion: PlGF is a disease-candidate gene in liver cirrhosis, and inhibition of PlGF offers a therapeutic alternative with an attractive safety profile. (HEPATOLOGY 2011;)

Chronic liver disease can be defined as a complex pathophysiological process of progressive destruction and regeneration of liver parenchyma, leading to fibrosis, cirrhosis, and increased risk of hepatocellular carcinoma. A profound alteration of the hepatic angioarchitecture due to induction of long-term structural vascular changes is underlying this remodeling process. Hepatic angiogenesis occurs during the progression of several chronic liver diseases, including hepatitis B/C, biliary cirrhosis, alcoholic cirrhosis, and nonalcoholic steatohepatitis. The resulting neovasculature is mainly located in the fibrotic areas of the liver and induces the formation of arterio-portal and porto-venous systemic anastomoses.1

Preclinical studies of this phenomenon have demonstrated that angiogenic inhibitors interfere with the progression of fibrosis. In human and experimental liver fibrosis, neovascularization seems to be a process strictly related to progressive fibrogenesis.2 In this context, studies in experimental models of cirrhosis have shown that treatment with angiogenic inhibitors such as neutralizing monoclonal anti–vascular endothelial growth factor receptor (VEGFR) antibody, TNP-470, and adenovirus expressing the extracellular domain of Tie2 decreased liver fibrosis.3, 4 Other parallels between fibrosis and angiogenesis have been postulated, such as the promotion of different subpopulations of hepatic stellate cells (HSCs; angiogenic versus fibrogenic phenotypes), and of hepatic inflammation as a process linking angiogenesis and fibrogenesis.2, 5 Consequently, multitargeted therapies acting against both angiogenesis and inflammation have been shown to be beneficial in inhibiting the progression of fibrosis to cirrhosis. The validity of the latter approach was demonstrated in cirrhotic rats in which sunitinib and sorafenib, two inhibitors of tyrosine kinase receptors (RTKs) that target the platelet-derived growth factor and vascular endothelial growth factor (VEGF) signaling pathways, produced a reduction in the degree of hepatic angiogenesis, fibrosis, and inflammation, as well as a significant decrease in portal pressure.6, 7 Conversely, inhibition of angiogenesis can worsen fibrogenesis in specific conditions, as was demonstrated by administration of integrin inhibitors.8

Moreover, important questions arise not only to the class of angiogenic inhibitors that can be used successfully, but also with respect to the safety, especially considering potential application in patients with critically ill portal hypertension and cirrhosis. Many of the currently available multitargeted therapeutic strategies are associated with toxicities, thereby limiting their use in critically ill patients. Recent preclinical studies suggest that therapies targeting placental growth factor (PlGF) activity may possess such a safety profile.9, 10 PlGF is a member of the VEGF family and a specific ligand for VEGFR1 that was originally discovered and isolated from the human placenta. The human transcript for PlGF generates four isoforms (PlGF-1 to −4), PlGF-2 being the only one present in mice.11 Unlike VEGF, PlGF plays a negligible role in physiological angiogenesis and is not required as a survival signal for the maintenance of quiescent vessels in healthy tissues. Furthermore, studies in transgenic mice revealed that the angiogenic activity of PlGF is restricted to pathological conditions.12, 13 In contrast to VEGF inhibitors, a monoclonal anti-PlGF antibody (αPlGF) has been shown to reduce pathological angiogenesis in various spontaneous cancers and other disease models without affecting healthy blood vessels, resulting in no major side effects in mice and humans.9, 10, 14, 15

Based on the aforementioned considerations, PlGF might be an attractive therapeutic target for cirrhosis, but nearly nothing is known about its pathogenetic role in this disorder, nor its therapeutic potential. Here, we demonstrate that anti-PlGF antibody treatment might be considered as a novel potential therapy for cirrhosis due to its multiple mechanisms of action against angiogenesis, inflammation, and hepatic fibrosis. We also provide mechanistic insight into the fibrogenic role of PlGF by demonstrating its biological effect on HSCs. Importantly, all these results were obtained in the absence of the adverse effects that are usually associated with antiangiogenic therapies based on VEGF blockade.


αPlGF, anti-PlGF antibody; αSMA, α-smooth muscle actin; αVEGFR, anti-VEGFR antibody; BrdU, bromodeoxyuridine; ERK, extracellular signal-regulated kinase; HSC, hepatic stellate cell; IgG1, immunoglobulin G1; mRNA, messenger RNA; PAS, periodic acid-Schiff; PDGFRA, platelet-derived growth factor receptor-α; PlGF, placental growth factor; RTK, tyrosine kinase receptor; RT-PCR, reverse-transcription polymerase chain reaction; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor.

Materials and Methods

Experimental Models of Cirrhosis.

All experiments were performed in 8-week-old male PlGF wild-type (PlGF+/+) mice (50% Sv129/50% Swiss), matched PlGF-knockout mice (PlGF−/−) of the same genetic background (Vesalius Research Center Leuven, Belgium), and male Wistar rats (Charles River, Saint Aubin les Elseuf, France). Cirrhosis was induced by way of CCl4 application (see Supporting Information Methods).

Human Samples.

Hepatic expression of PlGF and serum PlGF levels were assessed in liver specimens and blood samples from patients with alcoholic hepatitis, chronic hepatitis C, nonalcoholic steatohepatitis, and normal liver specimens. For PlGF immunohistochemistry, biopsy samples were obtained from patients with hepatitis C. The demographic and clinical characteristics of the patients included in the study are further represented in the Supporting Information Methods and in Supporting Information Tables 2 and 3.

PlGF Inhibition Studies.

The effect of PlGF deficiency in cirrhosis was first studied in PlGF−/− mice. CCl4 and saline (n = 8 in each group) were administered to PlGF+/+ and PlGF−/− mice. After 25 weeks of CCl4 treatment, animals were sacrificed and experiments were performed. For the therapeutic study, control (n = 5) and CCl4-treated mice (n = 9) were treated with 25-mg/kg intraperitoneal injections of αPlGF (ThromboGenics NV, Leuven, Belgium) that were administered twice weekly on days 0 and 3 from week 12 until week 20 of the CCl4 treatment. To eliminate the possibility of passive immunization, a group of matched control (n = 5) and a group of CCl4-treated mice (n = 7) were injected with mouse immunoglobulin G1 (IgG1) (ThromboGenics NV) at the same dose and times as mice in the αPlGF groups. The dosing schedule of αPlGF was based on previous published pharmacokinetic studies that were performed in mice.9, 10 To provide therapeutic data for end-stage cirrhotic mice, αPlGF was administered at the same dosage as described above, but was given from week 18 to week 25 of the CCl4 treatment.

Hemodynamic studies, vascular corrosion casting, histology (Sirius Red, periodic acid-Schiff–diastase), immunohistochemistry (CD31, α-smooth muscle actin), immunofluorescence (PlGF and vascular cell adhesion molecule 1), cytology (phalloidin), antibodyarray assay, statistical analysis, and all other methods are described in the Supporting Information Methods.


Enhanced PlGF Expression in CCl4-Treated Rodents and Patients with Cirrhosis.

Changes in the expression of PlGF that occur in the setting of cirrhosis were investigated in experimental models of cirrhosis in mice and rats as well as in patients with cirrhosis. After treating mice with CCl4, hepatic PlGF protein levels increased after 4 weeks and remained elevated during 16 weeks of treatment (P < 0.05 versus control mice) (Fig. 1A). Increased hepatic PlGF expression was also detected via western blot analysis of rats with established cirrhosis. As seen in Fig. 1B, there was an approximately four-fold increase in PlGF protein levels in cirrhotic rat livers compared with control livers (4.2±1.4 versus 0.7 ± 1.1 relative densitometric units, respectively; P < 0.05).

Figure 1.

Enhanced PlGF expression in CCl4-treated rodents and in patients with cirrhosis. (A) The hepatic PlGF protein levels of cirrhotic mice were quantified by enzyme-linked immunosorbent assay. PlGF concentrations were significantly higher in the CCl4-treated samples than in the controls, with a maximum PlGF level occurring after 4 weeks. In contrast, PlGF was undetectable in the livers of the controls. The black horizontal lines in the boxes represent the median value. Outliers are either represented by °16 (mild) or *20 (extreme). #P < 0.05 compared with controls. (B) Western blot analysis of PlGF expression in the livers of control (n = 10) and cirrhotic rats (n = 10). Total protein extracts (30 μg) that were immunoblotted with αPlGF showed increased levels of PlGF in cirrhotic animals. Ponceau S staining was used as a normalization control. WB, western blotting. (C) The PlGF mRNA levels (top panel) were evaluated via reverse-transcription polymerase chain reaction (RT-PCR) using total RNA isolated from the livers of patients with cirrhosis (n = 6) and without cirrhosis (n = 6). The expression of the housekeeping gene (HPRT) was used as normalization control. A representative result of three samples for each group is shown. bp, base pairs; −RT, negative RT-PCR control. (D) Dot plot of enzyme-linked immunosorbent assay reactivities with an αPlGF monoclonal antibody (clone 37203) in the serum from patients with cirrhosis and healthy controls. Dots represent means of duplicate values. The central horizontal line represents the median value. (E) Correlation of PlGF serum levels and hepatic venous pressure gradient in patients with cirrhosis. Dots represent the means of duplicate values (r = 0.386, P < 0.05).

To determine whether PlGF was also overexpressed in human liver cirrhosis, we measured PlGF messenger RNA (mRNA) and protein levels in livers of patients with cirrhosis. A prominent up-regulation of hepatic PlGF mRNA levels was observed in patients with and without cirrhosis (3.5 ± 0.9 versus 0.9 ± 0.2 relative densitometric units, respectively; P < 0.05) (Fig. 1C). In addition, PlGF immunostaining in human hepatitis C virus livers showed a stage-dependent increase in expression, correlating with the progression of fibrosis, with the highest PlGF levels detected in F4 fibrosis grade samples (P ≤ 0.001 versus F0 and F1) (see Supporting Information Fig. 1 for fibrosis grading). This increase in PlGF protein expression was observed in hepatocytes and nonparenchymal cells localized in fibrotic areas (Supporting Information Fig. 1). In agreement with this result, serum PlGF levels in patients with cirrhosis were at least two-fold higher than those in healthy subjects, and in some individuals, these levels reached values that were three-fold higher than those of controls (Fig. 1D). Interestingly, a direct significant correlation was found between PlGF serum levels and hepatic venous pressure gradient in patients with biopsy-proven alcoholic hepatitis, a common cause of acute-on-chronic liver failure (Fig. 1E).

Beneficial Effects of PlGF Deficiency and αPlGF Treatment on Portal Hypertension.

In a prevention study protocol (see Materials and Methods), we investigated the protective effect of PlGF gene deficiency against the development of the splanchnic hemodynamic alterations in cirrhotic mice. As demonstrated in Table 1, cirrhotic PlGF−/− mice (denoted as CCl4 PlGF−/− in Table 1) exhibited a 36.8% reduction in mesenteric artery blood flow and a 17% decrease in pulse rate, both significantly different from the values observed in wild-type cirrhotic mice (denoted as CCl4 PlGF+/+ in Table 1; P < 0.01 and P < 0.001, respectively). These hemodynamic changes resulted in a significantly reduced lower portal pressure in CCl4-treated PlGF−/− mice compared with wild-type cirrhotic animals (−27%). No differences were found in mean arterial pressure or spleen weight between the two CCl4-treated experimental groups.

Table 1. Splanchnic and Hemodynamic Changes in CCl4 Mice in the Prevention and in the Therapeutic Study (Week 12 to Week 20)
Prevention StudyControl PlGF+/+CCl4 PlGF+/+Control PlGF−/−CCl4 PlGF−/−% Change CCl4 PlGF+/+ Versus CCl4 PlGF−/−
Mean arterial pressure, mm Hg96.1 ± 2.792 ± 6.9113.1 ± 6.284 ± 3.4NS
Portal pressure, mm Hg3.8 ± 0.511.7 ± 0.84.3 ± 1.68.5 ± 0.6*−27
Spleen weight, g/10 g body weight0.03 ± 0.0030.04 ± 0.0040.03 ± 0.0040.06 ± 0.004NS
Heart rate, beats/minute402 ± 18528 ± 15543 ± 20440 ± 12−17
Mesenteric artery flow, mL/minute0.75 ± 0.051.44 ± 0.070.72 ± 0.060.91 ± 0.07−37
Therapeutic StudyControl IgG1CCl4 IgG1Control αPlGFCCl4 αPlGF% Change CCl4 IgG1 Versus CCl4 αPlGF
  • Abbreviations: αPlGF, anti-PlGF antibody; NS, not significant; PlGF, placental growth factor.

  • Data are expressed as the mean ± SEM.

  • *

    P < 0.05 CCl4 PlGF−/− versus CCl4 PlGF+/+.

  • P < 0.001 CCl4 PlGF−/− versus CCl4 PlGF+/+.

  • P < 0.01 CCl4 PlGF−/− versus CCl4 PlGF+/+.

  • §

    P < 0.001 CCl4 αPlGF versus CCl4 IgG1.

Mean arterial pressure, mm Hg107.1 ± 6.599.9 ± 1.6104.1 ± 4.199.0 ± 2.3NS
Portal pressure, mm Hg4.7 ± 0.712.8 ± 0.384.1 ± 0.69.5 ± 0.5§−26
Spleen weight, g/10 g body weight0.03 ± 0.0030.04 ± 0.0070.03 ± 0.0030.03 ± 0.002NS
Heart rate, beats/minute419 ± 17444 ± 13448 ± 29463 ± 13NS
Mesenteric artery flow, mL/minute0.75 ± 0.041.58 ± 0.10.86 ± 0.040.95 ± 0.03D−40

To determine whether or not the beneficial effect of PlGF gene deficiency had therapeutic potential, a therapeutic study was set up (see Materials and Methods) in which the effect of αPlGF or IgG1 injection was evaluated in control and CCl4-treated mice (application from week 12 to week 18, Table 1). Similar hemodynamic changes as in the prevention study could be observed, showing now that αPlGF treatment can partially reverse the portal hypertensive syndrome (Supporting Information Results). When αPlGF was administered to mice with end-stage cirrhosis (week 18 to week 25 of CCl4 treatment), we did not observe a significant effect on portal pressure, although a nonsignificant decrease in mesenteric artery flow in these animals was detected (Table 2), likely because the disease had advanced to an irreversible stage.

Table 2. Splanchnic and Hemodynamic Changes in CCl4 Mice in the Late Therapeutic Setting with αPlGF (Week 18 to Week 25 of CCl4 Treatment)
Therapeutic StudyCCl4 IgG1CCl4 αPlGFP Value
  1. Abbreviations: αPlGF, anti-PlGF antibody; NS, not significant.

  2. Data are expressed as the mean ± SEM.

Mean arterial pressure, mm Hg98.8 ± 3.496.9 ± 3.1NS
Portal pressure, mm Hg11.5 ± 0.410.2 ± 0.50.08
Spleen weight, g/10 g body weight0.04 ± 0.0060.04 ± 0.004NS
Mesenteric artery flow, mL/minute1.26 ± 0.180.90 ± 0.040.09

Hepatic Inflammation Induced by CCl4 Treatment Is Significantly Attenuated in PlGF−/− Mice and After αPlGF Treatment.

Because studies performed in cirrhotic rats have shown that angiogenic inhibitors such as sunitinib effectively decrease the severity of necroinflammation in cirrhotic livers,7 we investigated whether suppression of PlGF activity affected chronic hepatic inflammation. Periodic acid-Schiff staining with diastase digestion (PAS-diastase) was used to visualize macrophage cell accumulation in the livers of PlGF+/+ and PlGF−/− mice. The livers of PlGF+/+ mice that were chronically treated with CCl4 showed a significant increase in PAS-diastase positivity compared with control PlGF+/+ mice (data shown in legend Fig. 2). Notably, the increase in macrophages associated with cirrhosis was significantly reduced in CCl4-treated PlGF−/− mice (Fig. 2A,B). Likewise, PlGF-blockage by αPlGF reduced macrophage accumulation in CCl4-treated mice compared with IgG1-CCl4–treated mice (Fig. 2C,D).

Figure 2.

The severity of the hepatic necroinflammation induced by CCl4 treatment is significantly attenuated in PlGF−/− mice as well as in PlGF+/+ mice treated with αPlGF. The number of ceroid pigment-containing macrophages in the liver was significantly increased after 25 weeks of CCl4 treatment. These cells formed clusters and predominated in the centrilobular and portal connective tissues. After PAS-diastase staining, these macrophages stain pink. Arrows indicate PAS-diastase–positive macrophages. Data from the control animals are not displayed in the histograms (control PlGF+/+, 5.61 ± 0.47; control PlGF−/−, 4.93 ± 0.07; control IgG1, 6.3 ± 0.43; control αPlGF, 6.58 ± 0.42). Deficiency of PlGF (B) was associated with a significant reduction in PAS diastase-positive macrophages compared with PlGF+/+ mice (A) (−41.8%, 7.7 versus 13.3 cells per field; *P < 0.05). A similar reduction was seen after αPlGF treatment (D) compared with IgG1 treatment (C) (10.1 versus 16.0 cells per microscope field; &P < 0.05). Original magnification ×100.

To further understand the link between PlGF blockade and the reduction in inflammatory infiltrate, the expression of proinflammatory adhesion molecules in the vasculature of cirrhotic mice was analyzed in absence or in presence of PlGF activity. We demonstrated that blockade of PlGF activity decreases the neovasculature expressing vascular cell adhesion molecule 1. Also, PlGF contributes to the recruitment of hepatic inflammatory infiltrate by its chemotactic properties on monocytes (Supporting Information Results and Supporting Information Fig. 2).

Inhibition of PlGF Diminishes Intrahepatic/Splanchnic Neoangiogenesis and Arteriogenesis in Cirrhotic Animals.

To investigate whether PlGF stimulated angiogenesis during cirrhosis, we performed CD31 immunostaining of various tissues (Fig. 3 and Supporting Information Fig. 3). Compared with cirrhotic wild-type mice, CCl4-treated PlGF−/− mice exhibited significant reductions in hepatic, mesenteric, and colonic vascular density (44%, 37%, and 64%, respectively, P < 0.05) (Supporting Information Fig. 3). In agreement with these results of the prevention study, we found that αPlGF treatment (Fig. 3) also reduced hepatic, mesenteric (data not shown) and colonic neoangiogenesis (with 28%, 34%, and 51%, respectively, with respect to the corresponding IgG1-CCl4 mice, P < 0.05).

Figure 3.

αPlGF treatment diminishes intrahepatic and colonic neo-angiogenesis in cirrhotic mice. Representative images of CD31 immunohistochemistry in the liver (top panels, original magnification −100) and colon (bottom panels, original magnification −400) of IgG1-treated cirrhotic mice (left column) and αPlGF-treated cirrhotic mice (right column). Arrow indicates the presence of CD31-positive endothelial cells in blood vessels. *P < 0.05 versus CCl4 IgG1. &P < 0.01 versus CCL4 IgG1.

Similar results were obtained when evaluating the role of PlGF in angiogenesis on vascular corrosion casts from the splanchnic tissues and livers of cirrhotic mice. In addition, we could demonstrate a normalization of the sinusoidal vessel course on liver casts following αPlGF treatment (Supporting Information Results and Supporting Information Fig. 4), resulting in significant reduction of the hypoxic environment in the liver (Supporting Information Fig. 5). The expression of hypoxia-inducible glycolytic genes in CCl4-cirrhotic livers showed reduced expression upon αPlGF treatment compared with IgG1. This is translated into a significant down-regulation of HIF-1α protein level (P < 0.05).

Because studies of mice with portal hypertension and solid tumors have demonstrated that PlGF has a pleiotropic action on both angiogenesis and arteriogenesis,10, 13 we subsequently investigated the smooth muscle cell content of vessels by anti–α-smooth muscle actin (αSMA) immunostaining. Both PlGF gene deficiency and αPlGF treatment reduced arteriogenesis in visceral peritoneum, as demonstrated by significantly reduced immunostaining for αSMA in the vasculature of these mice (Supporting Information Fig. 6).

Fibrosis Is Decreased in Animals with PlGF Gene Deficiency and After αPlGF Treatment.

To assess the in vivo effects of PlGF gene deficiency and αPlGF treatment on hepatic fibrogenesis, the extent of liver fibrosis was quantified by Sirius Red staining. After 25 weeks of CCl4 administration, CCl4-PlGF+/+ mice exhibited centro-portal fibrotic septae and centro-central fibrotic linkages (Fig. 4A,C). Remarkably, the lack of the PlGF gene in cirrhotic PlGF−/− mice (Fig. 4B) substantially decreased the severity and extent of the fibrotic changes, as illustrated by a 36% reduction in fibrosis score compared with wild-type CCl4-treated mice (39,316 μm2 versus 61,034 μm2 fibrotic area, respectively; P < 0.05). In addition, CCl4-treated wild-type mice given αPlGF for 8 weeks (from week 12 to week 20) also showed less fibrosis compared with IgG1-treated cirrhotic mice (53,676 versus 90,357 μm2 fibrotic area, respectively; P < 0.05) (Fig. 4D). The effect of αPlGF treatment to decrease the extent of fibrosis in cirrhotic mice was further confirmed by macroscopic and stereomicroscopic evaluation, which revealed loss of nodularity after αPlGF treatment (Fig. 4E-H). On the other hand, no changes in the fibrosis score were detected when end-stage cirrhotic mice (week 18 to week 25 of CCl4 treatment) were treated with αPlGF. These results point to a therapeutic window during which the antifibrotic effect of αPlGF can be successful.

Figure 4.

Targeting PlGF inhibition results in reduced fibrosis scores. Histological images of livers from cirrhotic PlGF+/+ mice (A), cirrhotic PlGF−/− mice (B), cirrhotic IgG1-treated mice (C), and cirrhotic αPlGF-treated mice (D) stained with Sirius Red. Original magnification ×100. The histogram represents the computerized quantification of fibrosis scores. *P < 0.05 versus PlGF+/+ and &P < 0.05 versus IgG1. Fibrosis scores were below 11,000 in all noncirrhotic animals (control PlGF+/+, 8,520 ± 309; control PlGF−/−, 8,211 ± 795; control IgG1, 8,339 ± 184; control αPlGF, 10,172 ± 1,034) (not shown in the histograms). Representative liver (E and F) and stereomicroscopic images (G and H) obtained immediately after Batson injection. Mice treated with αPlGF experienced less explicit macroscopic features of cirrhosis (irregular, nodular liver surface and blunt liver edge) than mice treated with IgG1.

Localization and Cellular Source of PlGF in Fibrotic and Cirrhotic Rodent Livers.

To understand why a decrease in PlGF activity was associated with a reduction in fibrosis severity, we studied the intrahepatic expression of PlGF by immunofluorescence in livers of control (rats, n = 10; mice, n = 10) and CCl4-treated rats (n = 10) and mice (n = 10). A PlGF signal was weakly observed in the livers of control animals (Fig. 5A). PlGF-positive cells, however, were quite evident in CCl4-treated animals. The livers of PlGF-deficient mice were totally devoid of PlGF immunoreactivity (data not shown). In an attempt to identify the cellular source of PlGF expression, we measured PlGF protein and mRNA levels in mouse HSCs (Supporting Information Fig. 7). Activation of HSCs was associated with increased αSMA expression, a finding that reached significance from day 8 onward (Supporting Information Fig. 7A), and with a significant PlGF increase in the cell supernatants (Supporting Information Fig. 7B). These data were further confirmed in primary HSCs isolated from control and cirrhotic rats (Supporting Information Fig. 7C). In these cells, an intense up-regulation of PlGF was observed in activated HSCs and, to a lesser extent, in hepatocytes and endothelial cells isolated from cirrhotic rats.

Figure 5.

PlGF is overexpressed in cirrhotic livers and induces sustained activation of ERK1/2 in activated HSCs. (A) PlGF (red) immunofluorescent staining was performed in normal control, fibrotic, and cirrhotic livers of rat and mice using a PlGF-specific monoclonal antibody. There was a significant increase in PlGF reactivity in the cirrhotic livers (arrows). Original magnification ×100. (B) Expression of VEGFR1 (Flt-1) and VEGFR2 (Flk-1) receptors was evaluated in primary HSCs isolated from cirrhotic livers (n = 5) and in LX-2 cells (n = 5) by conventional RT-PCR. Amplification of α-actin (actin) was used as normalization control. MW, molecular weight marker. (C) Primary HSCs from cirrhotic rats and LX-2 cells were stimulated with PlGF (100 ng/mL) for different time durations (+). Lysates (40 μg of protein) were analyzed via western blotting analysis with specific antibodies targeted against phosphorylated ERK1/2-Thr202/Tyr204 and ERK1/2 (n = 5). wb, western blotting.

In Vitro Characterization of PlGF Signaling in Activated HSC Cells.

Considering the major pathophysiological role that HSCs play in fibrogenesis, the effect of PlGF on rat and human activated HSCs was studied. As shown in Fig. 5B, there was a significant overexpression of VEGFR1 receptors in primary HSCs from cirrhotic rats and in the LX-2 human HSC cell line. Expression of VEGFR2, another member of the VEGF family of RTKs, was less prominent, particularly in HSCs isolated from cirrhotic animals, in which no detectable expression was present. To assess whether PlGF may regulate the expression of profibrogenic genes, LX-2 cells were incubated in the presence or absence of 100 ng/mL PlGF for 24 hours. LX-2 cells treated with PlGF did not show significant changes in mRNA levels of genes that play a major role in fibrogenesis (i.e., collagen-1, transforming growth factor β, metalloproteinase-2, and tissue inhibitor of metalloproteinase-1) compared with untreated cells (data not shown).

We next sought to determine which downstream signaling pathways were up-regulated in activated HSCs in response to PlGF treatment. Fig. 5C shows that treatment of primary HSCs and LX-2 cells with PlGF was associated with a sustained induction of extracellular signal-regulated kinase (ERK) 1/2 phosphorylation lasting for more than 60 minutes, during which the total level of ERK1/2 expression remained constant. The treatment of LX-2 cells with anti-VEGFR1 antibodies inhibited the phosphorylation of ERK1/2 induced by PlGF (Supporting Information Fig. 8).

It has been shown previously that sustained ERK1/2 activation promotes fibroblast chemotaxis and proliferation.16 To assess whether a similar mechanism also occurs in HSCs, we quantified cell chemotaxis in untreated LX-2 cells and in LX-2 cells treated with PlGF. Fig. 6A shows time-lapse microphotographs of LX-2 cell migration. Approximately 35% of the cells showed migration in response to 10 minutes of treatment with 100 ng/mL PlGF (34.6 ± 2 versus 1.3±0% of migrating cells in cultures treated with vehicle only; P < 0.001). To further characterize the role of PlGF as a chemotactic substance, LX-2 cells were subjected to a cell migration assay in a modified Boyden chamber in the presence of a PIGF gradient (Fig. 6B). Only a few cells migrated in the absence of PlGF, whereas a significant (seven-fold) increase in directional migration was observed at a concentration of 50 ng/mL PlGF (P < 0.01). The chemoattractant response of LX-2 cells to PlGF was inhibited by disrupting PlGF-VEGFR1 interaction with anti-VEGFR1 antibody.

Figure 6.

PlGF stimulates chemotaxis and proliferation in LX-2 cells. (A) Representative time-lapse microphotographs of LX-2 cells treated with 100 ng/mL PlGF. Arrows indicate HSCs that migrated in response to treatment (n = 3). Original magnification ×400. (B) LX-2 cells were preincubated with or without αVEGFR1 (5 μg/mL) and then trypsinized and resuspended in chemotaxis medium. In total, 2 × 104 cells were then added to a polycarbonate membrane (8-μm pore size) coated with 1% gelatin in a modified Boyden chamber and exposed to PlGF (50 ng/mL) or PlGF (50 ng/mL) + αVEGFR1 (5 μg/mL) for 4 hours. At the end of the treatment period, cells that had migrated were stained with DiffQuick solution, and the cell number was counted in three random fields. Data points represent the mean ± SEM number of migrating cells/field calculated in three different wells. *P < 0.01 compared with vehicle (n = 3). (C) LX-2 cells were incubated with vehicle, PlGF (100 ng/mL), or PlGF (100 ng/mL) + αVEGFR1 (5 μg/mL) for 5 minutes. F-actin was detected in fixed and permeabilized cells using fluorescein isothiocyanate–labeled phalloidin. In LX-2 cells, PlGF treatment was associated with filopodia formation. αVEGFR1 treatment inhibited cytoskeleton remodeling induced by PlGF (n = 5). (D) Representative figures of a proliferation assay performed in LX-2 cells that were treated with or without PlGF (100 ng/mL) for 24 hours. BrdU incorporation was quantified via flow cytometry. Cells within the oval scatter gate were analyzed (upper left panel). For each panel, the percentage of cells that stained positively for BrdU is indicated (n = 8).

Because cell migration is associated with regulation of the actin cytoskeleton, we next assessed whether PlGF stimulated F-actin reorganization in activated HSCs. In quiescent LX-2 cells, F-actin was found mostly in membrane structures and as unorganized fibers throughout the cell (Fig. 6C, left panel). In contrast, after treatment with PlGF, phalloidin-stained filopodia were present around the cell periphery, indicating that PlGF promotes actin cytoskeleton remodeling (Fig. 6C, middle panel). The treatment of LX-2 cells with anti-VEGFR1 antibodies inhibited cytoskeleton remodeling induced by PlGF (Fig. 6C, right panel). Next, to test whether PlGF could stimulate HSC proliferation, LX-2 cells were cultured in the presence of PlGF, and we assessed the amount of bromodeoxyuridine (BrdU) that was incorporated into the cells using flow cytometry. Medium supplemented with 2% fetal bovine serum was used as a positive control in the proliferation assay. When LX-2 cells were treated with 100 ng/mL PlGF, BrdU uptake was significantly increased (Fig. 6D), indicating that PlGF promotes proliferation of these cells. Treatment of LX-2 cells with anti-VEGFR1 antibody totally blocked the PlGF-induced proliferation (3.2 ± 0.9 versus 20.7±1.3% of BrdU incorporation; P < 0.01) (n = 3).

To gain some initial insight into the signaling mechanisms through which PlGF induces sustained ERK activation, cell migration, and cell proliferation, we analyzed the phosphorylation status of several candidate proteins implicated in the signal transduction. Signal transduction antibody arrays were probed with lysates of LX-2 cells that were treated with or without 100 ng/mL PlGF for 5 minutes and subsequently with anti-phosphotyrosine antibody. Supporting Information Table 1 shows the effect of PlGF on protein tyrosine phosphorylation in HSCs. Bioinformatic analysis of these data is provided in the Supporting Information Results and Supporting Information Fig. 9. Exposure of HSCs to PlGF resulted in a significant increase in the tyrosine phosphorylation of platelet-derived growth factor receptor-α (PDGFRA) and epidermal growth factor receptor. A direct interaction between VEGFR1 and PDGFRA receptors upon PlGF stimulation was confirmed via proximity ligation assay (see Supporting Information Results and Supporting Information Fig. 10).


PlGF stimulates endothelial cell growth, migration, and survival, as well as pathological angiogenesis.9, 10, 17 These proangiogenic and proinflammatory properties of PlGF together with the synergistic effect between inflammation and angiogenesis, as previously demonstrated for other RTK inhibitors in experimental cirrhosis,6, 7 make the inhibition of PlGF activity an attractive therapeutic strategy for the treatment of chronic liver disease.

However, only a few reports demonstrate a role of PlGF in liver disease.7, 13, 18, 19 We previously demonstrated that PlGF is up-regulated in the splanchnic microvasculature of portal-hypertensive mice and showed that PlGF deficiency in mice with partial portal vein ligation is associated with a significant decrease in splanchnic angiogenesis, porto-systemic shunting, and mesenteric artery flow.13 However, the present study is the first to describe a pathological role of PlGF in the context of cirrhosis. We demonstrated in a prevention and therapeutic study that PIGF blockade significantly decreased angiogenesis, arteriogenesis, hepatic inflammation, fibrosis, and portal hypertension in cirrhotic mice. Next, the relevance of these findings in humans was assessed. We showed that the circulating PlGF serum levels and hepatic protein expression were increased in patients with cirrhosis and correlated with the stage of fibrosis. Finally, we explored the cellular effects of PlGF in HSCs, which play a key role in the pathogenesis of fibrosis and portal hypertension.

An important finding of the present study is the association between PlGF blockade and the significant decrease in portal pressure in cirrhotic mice. Although keeping in mind the limitations of translating results in animal models into clinical practice, we found that there was a significant positive correlation between circulating PlGF serum levels and hepatic venous pressure gradient in patients with cirrhosis. Based on such observations, we could speculate that PlGF may also be involved in the pathogenesis of portal hypertension in humans. There is compelling evidence suggesting that the increase in portal blood flow seen in portal hypertension is not only due to splanchnic vasodilation, but also to enlargement of the splanchnic vascular tree caused by angiogenesis.13 Considering this evidence, the significant inhibition of angiogenesis and arteriogenesis in the splanchnic area by αPlGF may therefore contribute to the decrease in portal inflow following therapy.

Another important finding of this study is the blockade of hepatic fibrosis by targeting PlGF. This finding is in agreement with previous studies demonstrating that several angiogenic inhibitors inhibit the progression of liver fibrosis.3, 6, 7 We demonstrated that hepatic PlGF immunoreactivity was strong in cirrhotic rats and mice. Moreover, activated HSCs were the major source of PlGF production in these rodents, and they exhibited substantial VEGFR1 expression. However, it is intriguing that although the blockade of PlGF in vivo is antifibrogenic, we were unable to find significant changes in the expression of profibrogenic genes when human activated HSCs were treated with PlGF. This discrepancy may be explained considering that PlGF promotes an angiogenic phenotype in HSCs characterized by a sustained ERK1/2 phosphorylation as well as chemotaxis and proliferation. The acquisition of an angiogenic phenotype by HSCs has been described by others in response to PlGF and connected to the enhanced HSCs coverage of sinousoid characteristic of cirrhotic livers.5 All of these changes result in abnormalities in hepatic blood vessels that compromise the regulation of intrahepatic pressure and tissue perfusion. The sacculated and chaotically disorganized appearance of the microvessels in the cirrhotic livers of control mice, as analyzed by the vascular corrosion casts, is consistent with such vessel abnormalization.20 Interestingly, αPlGF treatment resulted in a partial normalization of the three-dimensional architecture of the hepatic blood vessel network and induces a significant decrease of proinflammatory vasculature, which is characterized by the expression of vascular cell adhesion molecule 1. A similar mechanism of vessel normalization induced by αPlGF treatment was recently described in hepatocellular carcinoma nodules.10 Interestingly, a reduction in fibrosis was only demonstrated when mice were treated with αPlGF in the early phase of cirrhosis induced by CCl4 treatment (from week 12 to week 20). No significant beneficial effect was observed following αPlGF therapy in mice with end-stage cirrhosis induced by CCl4 (week 18 to week 25). This observation supports the idea that the acquisition of an angiogenic phenotype by HSCs, in response to PlGF, causes an increase in the HSC population in early phase of cirrhosis that correlates with the degree of fibrosis. However, when the HSC population reaches a critical mass, the therapeutic efficiency of PlGF blockade is limited, because PlGF does not have any effect on the regulation of profibrogenic genes. In agreement with this hypothesis, it has been shown that the expression of angiogenic factors in fibrotic/cirrhotic livers occurs mainly in areas of active fibrogenesis and not in larger bridging septae or in end-stage cirrhotic tissue.21 Therefore, this evidence points to a therapeutic window during which αPlGF treatment is effective at inhibiting and reducing fibrosis.

The sustained ERK activation in response to PlGF in HSCs prompted us to investigate the underlying mechanisms, because VEGFR1 has a relatively weak tyrosine kinase activity. Some authors also have suggested that VEGFR1 could function as a decoy receptor for VEGF-A, thereby amplifying the activity of VEGF.12 However, HSCs did not express detectable levels of VEGFR2, suggesting that VEGFR1's role extends beyond a mere decoy activity. Comparison of the protein tyrosine phosphorylation profile of activated HSCs showed that PlGF induced the phosphorylation of other tyrosine kinase receptors, including PDGFRA and epidermal growth factor receptor. These findings raise the intriguing possibility that upon PlGF activation, VEGFR1 may amplify its own signaling by highjacking other RTKs via a molecular association. In our initial analysis, we identified PDGFRA as a candidate of such molecular cross-talk that may further potentiate sustained ERK activation. A similar cross-talk between VEGFR1 and VEGFR2, whereby PlGF amplifies VEGF-driven angiogenesis, has been documented in endothelial cells.22 VEGFR1 also interacts with low-density lipoprotein receptor, that results in ligand-independent activation of VEGFR1 by LDL.23 However, a molecular cross-talk between VEGFR1 and other types of RTKs, resulting in sustained signaling, has never been documented yet.

Although antiangiogenic agents are frequently used in the treatment of angiogenesis-related diseases, their clinical use has been associated with adverse effects, such as hypertension, proteinuria, thrombosis, and reduced wound healing capacity. These adverse effects warrant some caution to select angiogenic inhibitors for the treatment of patients with cirrhosis who are critically ill. Studies in transgenic mice have shown that loss of PlGF does not affect development, reproduction, or normal postnatal health, but impairs pathological angiogenesis in implanted and spontaneously arising cancer models.10 Moreover, administration of αPlGF is not associated with vascular pruning in healthy organs in mice,9 and is well tolerated in humans, where phase I trials in healthy volunteers and patients with solid tumors have thus far not revealed any major adverse effects.14, 15 The present study confirms the safety profile of αPlGF (Supporting Information Results and Supporting Information Fig. 11). Furthermore, αPlGF did not compensatorily up-regulate the expression of VEGF; such up-regulation has been suggested to represent a possible cause of resistance to antiangiogenic treatment (Supporting Information Results and Supporting Information Fig. 11).

In conclusion, this experimental study characterized the pathophysiological mechanisms and molecular effects that PlGF exerts on murine and human cirrhotic livers and on HSCs. Blockade of the PlGF pathway in cirrhotic mice by monoclonal antibodies or by genetic deficiency of PlGF decreased hepatic and mesenteric angiogenesis, mesenteric arterial blood flow, fibrosis, and inflammation, as well as portal pressure. Also because of its safety profile, αPlGF may be considered as an attractive candidate for treating patients with chronic liver disease.


We thank Julien Dupont and Huberte Moreau for technical assistance, Kin Jip Cheung for compiling the demographic data of the patients, and Susana Kalko for technical assistance with bioinformatic analysis. LX-2 cells were generously supplied by Scott L. Friedman; αPlGF was kindly provided by ThromboGenics NV.