Angiogenesis-Independent Endothelial Protection of Liver: Role of VEGFR-1

See allHide authors and affiliations

Science  07 Feb 2003:
Vol. 299, Issue 5608, pp. 890-893
DOI: 10.1126/science.1079562


The vascular endothelium was once thought to function primarily in nutrient and oxygen delivery, but recent evidence suggests that it may play a broader role in tissue homeostasis. To explore the role of sinusoidal endothelial cells (LSECs) in the adult liver, we studied the effects of vascular endothelial growth factor (VEGF) receptor activation on mouse hepatocyte growth. Delivery of VEGF-A increased liver mass in mice but did not stimulate growth of hepatocytes in vitro, unless LSECs were also present in the culture. Hepatocyte growth factor (HGF) was identified as one of the LSEC-derived paracrine mediators promoting hepatocyte growth. Selective activation of VEGF receptor–1 (VEGFR-1) stimulated hepatocyte but not endothelial proliferation in vivo and reduced liver damage in mice exposed to a hepatotoxin. Thus, VEGFR-1 agonists may have therapeutic potential for preservation of organ function in certain liver disorders.

The vascular endothelium is a highly versatile system and, in addition to its well-established function of nutrient and gas exchange between tissues and blood, it plays multiple homeostatic roles (1). Furthermore, the endothelium has an inductive effect on liver (2) and pancreas (3) development before the establishment of a blood flow. Vascular endothelial growth factor–A (VEGF-A) (4), a major regulator of normal and pathological angiogenesis, binds to two tyrosine kinase receptors, VEGFR-1 (Flt-1) (5) and VEGFR-2 (KDR/Flk-1) (6, 7). VEGFR-2 is the major mediator of the mitogenic, angiogenic, and permeability-enhancing effects of VEGF-A (8). However, many conflicting reports about the function of VEGFR-1 exist. This receptor has been implicated in the inhibition of VEGF-dependent endothelial mitogenesis and chemotaxis by several mechanisms (9–11). Other studies have shown that VEGFR-1 mediates monocyte chemotaxis (12), recruitment of endothelial cell progenitors (13), and survival of hematopoietic stem cells (14). VEGFR-1 activation also has been reported to result in collateral vessel growth through recruitment of bone marrow–derived cells (15). Thus, the importance of VEGFR-1 signaling in the vascular endothelium is largely unclear.

We sought to investigate the effects of VEGFR activation on parenchymal cell proliferation and survival. To achieve sustained systemic levels of VEGF, we injected Chinese hamster ovary (CHO) cells expressing VEGF165 or control CHO cells into the legs of nude mice (16). We observed substantially increased liver sizes in the CHO-VEGF groups. The liver/brain ratio (that is, the relative liver mass) of the CHO-VEGF group (4.73 ± 0.39) was significantly increased compared with that of the CHO–dihydrofolate reductase (CHO-DHFR) (3.18 ± 0.25; P < 0.0001) and CHO–Hakata antigen (CHO-HAg) (3.00 ± 0.45; P < 0.0001) controls. This reflects an increase in relative liver masses of 49% and 59%, respectively.

Histological analysis of the livers of CHO-VEGF–injected animals revealed that a large number of hepatic cells displayed mitotic figures (16). We observed mitotic activity in both parenchymal and nonparenchymal cells (Fig. 1, B and D). Hepatocyte mitoses in the livers of CHO-DHFR control animals were very rare (Fig. 1, A and C). In contrast, the livers of the CHO-VEGF group showed at least five mitotic figures per 10 high-power fields. Immunohistochemistry for Flk-1 demonstrated a normal pattern of sinusoidal and nonsinusoidal endothelial staining in the livers of CHO-DHFR animals. In the CHO-VEGF livers, the sinusoids demonstrated more complex and extensive branching (17).

Figure 1

Systemic VEGF increases liver size and induces mitosis in endothelial cells and hepatocytes. Representative liver sections from CHO-DHFR (A and C) and CHO-VEGF (B and D) animals are shown. (A) and (B) were stained with hematoxylin and eosin, and (C) and (D) demonstrate immunoperoxidase labeling of BrdU-labeled cells, indicating DNA synthesis. The CHO-DHFR liver is histologically normal, whereas livers from the CHO-VEGF animals show a striking increase in hepatocyte mitotic figures (arrows) and hyperplasia of sinusoidal lining cells (arrowheads). Scale bar, 25 μm.

VEGF has mitogenic effects on certain nonendothelial cell types (18). Therefore, we investigated whether VEGF induces DNA replication in freshly isolated mouse hepatocytes (16) (fig. S1). VEGF failed to induce an increase in [3H]thymidine incorporation. Likewise, neither the receptor-selective VEGF mutants Fltsel and KDRsel, which specifically activate VEGFR-1 and VEGFR-2, respectively (19), nor the naturally occurring receptor-specific agonists PlGF (VEGFR-1) (9) and VEGF-E (VEGFR-2) (20) induced hepatocyte proliferation, whereas hepatocyte growth factor (HGF) and epidermal growth factor (EGF) resulted in the expected stimulation (fig. S1). This is consistent with in situ ligand binding studies showing that VEGF binding sites are localized to endothelial cells, but not hepatocytes, in liver sections (21). Thus, the hepatocyte growth-promoting effects of VEGF require the action of an endothelial cell–derived paracrine mediator(s).

We established cultures of purified primary sinusoidal endothelial cells (LSECs) (16) (fig. S2) in isolation or in coculture with primary hepatocytes in a transwell format. As expected (22), in both isolated and transwell cocultures, VEGF-A, KDRsel, and VEGF-E induced a significant increase in [3H]thymidine incorporation in primary LSECs (Fig. 2, A and B) and ERK1/2 phosphorylation (fig. S1). In contrast, Fltsel and PlGF were indistinguishable from the negative control in LSEC proliferation (Fig. 2, A and B) and ERK1/2 phosphorylation (fig. S2). In agreement with previous studies (23), HGF had a minimal effect on LSEC proliferation (Fig. 2, A and B). In the transwell format, VEGF, Fltsel, and PlGF resulted in hepatocyte proliferation comparable to that induced by HGF (Fig. 2C). Thus, VEGFR-1 activation in LSECs results in the release of paracrine factors by a nonmitogenic signal transduction pathway. KDRsel and VEGF-E resulted in less pronounced and consistent hepatocyte stimulation, indicating that VEGFR-2 activation is less efficient at triggering paracrine signals, at least in LSECs (Fig. 2C).

Figure 2

VEGFR-2 versus VEGFR-1 stimulation of primary liver endothelial cells. (A) Wild-type VEGF and the VEGFR-2 agonists KDRsel and VEGF-E induce [3H]thymidine uptake in primary cultures of LSECs. In contrast, the VEGFR-1 selective agonists Fltsel and PlGF fail to promote LSEC proliferation. Ligands were added at a concentration of 10 ng/ml, except HGF, which was given at 50 ng/ml. (B) In transwell LSEC/hepatocyte cocultures, VEGF, KDRsel, and VEGF-E induced, as expected, [3H]thymidine incorporation in LSECs, whereas the VEGFR-1 agonists are ineffective. The concentration of ligands is the same as in (A). (C) Primary hepatocytes in the cocultures demonstrate stimulation of [3H]thymidine incorporation to a level comparable to HGF-treated cells when incubated with PlGF or Fltsel. In contrast, incubation with KDRsel or VEGF-E resulted in little or no stimulation of hepatocyte proliferation. Error bars represent SD; *P ≤ 0.004. (D) VEGFR-1 and VEGFR-2 selective agonists induce expression of distinct and overlapping genes. Shown is a representative experiment of Taqman analyses of 12 distinct gene transcripts in LSECs treated for 24 hours with VEGF (10 ng/ml), KDRsel, or Fltsel and normalized to control, untreated cells arbitrarily set to a value of 1. Expression profiles indicate that HGF and IL-6 induction by VEGF is selectively mediated by VEGFR-1, whereas expression of HB-EGF or CTGF is responsive to both VEGFR-1– and VEGFR-2–mediated signals. See bar graph coding in (D).

HGF is a key hepatocyte mitogen (24) produced by nonparenchymal cell types in the liver (16, 23), and we observed strong up-regulation of HGF mRNA by in situ hybridization within the sinusoidal lining cells of CHO-VEGF mice (fig. S3). Therefore, we tested whether HGF may be one of the paracrine mediators of VEGF activity in LSEC-hepatocyte cocultures. Addition of a polyclonal antibody to human HGF, which was able to achieve a partial (∼50%) neutralization of murine HGF, significantly inhibited the increase in [3H]thymidine incorporation induced by VEGF (30 ± 2%; P < 0.001), Fltsel(29 ± 2.4%; P < 0.02), and PlGF (30 ± 1.3%; P < 0.006) (16). Taking into account the partial neutralization provided by this antibody, as much as ∼60% of the activity may be attributed to HGF. This less-than-complete inhibition likely reflects the involvement of additional mediators. To define these factors, we measured the levels of RNA transcripts for a number of growth factors potentially involved in hepatocyte proliferation in primary LSECs after exposure to VEGF, Fltsel, or KDRsel (Fig. 2D). The genes examined include basic and acidic fibroblast growth factors (25), EGF, transforming growth factor–α (TGF-α) (26), interleukin-6 (IL-6) (27), connective tissue growth factor (CTGF) (28), and heparin-binding EGF (HB-EGF) (29). The most striking result was the 5.5-fold induction of HGF in the cultures treated with VEGF and Fltsel but not with KDRsel. Further experiments verified that endothelial cells truly account for such HGF expression (fig. S3). IL-6 also appeared to be a selective target of VEGFR-1 and was induced 3.3-fold. HB-EGF and CTGF were induced to equivalent levels by VEGF, Fltsel, and KDRsel and therefore may represent overlapping targets of VEGFR-1 and VEGFR-2 signals. Although the expression levels of other transcripts including EGF, TGF-β, and acidic and basic fibroblast growth factors were substantial, they were not increased by any of the treatments. Thus, gene expression studies are consistent with the involvement of multiple mediators.

To investigate the mechanism of liver growth, we counted proliferating cells, as assessed by bromodeoxyuridine (BrdU) immunohistochemistry, in liver sections 10 days after adenovirus (Av)–mediated delivery of KDRsel and Fltsel(16) (fig. S4). In agreement with the in vitro findings, Av-Fltsel promoted a significant increase in hepatocyte proliferation compared with Av-KDRsel and Av-LacZ (Fig. 3B). Conversely, KDRsel induced the greatest proliferation of sinusoidal cells (Fig. 3A). The Av-Fltsel–treated livers showed few proliferating sinusoidal cells and in this respect were almost indistinguishable from the Av-LacZ controls (Fig. 3B).

Figure 3

Proliferation of hepatocytes versus sinusoidal cells in response to selective VEGFR activation. Quantitative analysis of proliferating hepatocytes (A) and sinusoidal cells (B) was performed after BrdU immunohistochemistry of liver sections from animals treated with Av-LacZ, Av-KDRsel, or Av-Fltsel 10 days after Av administration. The VEGFR-1 selective agonist Fltsel results in the greatest hepatocyte proliferation but fails to induce sinusoidal endothelial cell proliferation. KDRsel is a potent mitogen for sinusoidal cells and also results in increased hepatocyte proliferation. Analysis was performed as described in (16). Values are means ± SEM. Level of significance was assessed by unpaired ttests; P values are indicated.

We next tested whether VEGF receptor activation may protect the liver from toxic injury such as that induced by carbon tetrachloride (CCl4). We initially established that endogenous VEGF plays an important role in liver repair after CCl4 damage (16) (fig. S5). We administered CCl4 to mice injected with Av-LacZ, Av-Fltsel, or Av-KDRsel. When Av-KDRsel was administered 4 days before the CCl4 , alanine aminotransferase levels were reduced about 45% compared with Av-LacZ (16) (table S1). The protective effect was greater when Av-KDRsel was delivered 8 days before the toxic injury, approaching an 86% reduction. This time course is consistent with the hypothesis that the KDRsel protective effects primarily depend on endothelial proliferation, which may amplify a paracrine survival-factor cascade. Animals in the Av-Fltselgroups also exhibited a marked protection and 65% to 70% reduction in serum alanine aminotransferase levels relative to controls. We observed no significant difference when Av-Fltsel was administered 4 and 8 days before CCl4 (table S1), consistent with a protective mechanism based on the release of survival/mitogenic factors from nonproliferating LSECs. Figure 4illustrates the morphology of livers in the group that received Av vectors 8 days before CCl4. There was extensive confluent perivenular necrosis in the Av-LacZ group involving 30% to 50% of the total hepatocyte mass (Fig. 4, A and D). In Av-KDRselanimals, perivenular hepatocyte necrosis was much less severe. Periportal areas showed changes similar to those animals that had not received CCl4 (Fig. 4, B and E). Av-Fltselanimals had a comparable reduction in hepatocyte necrosis (Fig. 4, C and F). Immunostaining for Flk-1 demonstrated that Av-KDRsel animals had markedly increased vascular density compared with the other groups (Fig. 4E). In contrast, the pattern and density of vessels in Av-Fltsel animals were similar to the LacZ groups, except, as mentioned above, for the reduction in necrotic areas (Fig. 4, D and F).

Figure 4

Both Av-KDRsel and Av-Fltsel reduce liver necrosis in the CCl4acute liver toxicity model, but only Av-KDRsel promotes angiogenesis. Shown are micrographs of representative livers from CCl4-treated animals that were administered Av-LacZ (A), Av-KDRsel (B), and Av-Fltsel (C) 8 days before CCl4. Av-LacZ livers show confluent perivenular hepatocyte necrosis (nc). PT, portal tract; THV, terminal hepatic venule. Necrosis is much reduced in Av-KDRsel and Av-Fltsel groups. The Av-KDRsel livers show prominent endothelial cell hyperplasia around terminal hepatic venules, which is much less pronounced in the Fltsel group. Arrows indicate areas of endothelial hyperplasia. (D to F) Immunostaining for Flk-1 in the same groups verifies endothelial hyperplasia in the Av-KDRsel animals (E). Note the very similar vascular pattern in Av-LacZ (D) and Av-Fltsel groups in the viable areas (F). Scale bar, 100 μm.

Our findings show that LSEC proliferation in response to VEGF-A is sufficient to promote growth and result in liver mass larger than normal through increased blood supply, combined with the paracrine release of key mitogenic signals. Perhaps the most noteworthy conclusion of our study is that, after VEGFR-1 activation, the endothelium is instructed to produce a series of mitogenic/survival factors that can protect parenchymal cells from injury and initiate regeneration. A model is presented in fig. S6. Whether other nonparenchymal cells in the liver, such as stellate and Kuppfer cells, participate in this VEGF-dependent paracrine mechanism remains to be established.

Given that the known dose-limiting side effects of VEGF (hypotension, edema, excessive angiogenesis) are associated with VEGFR-2 activation (30), Fltsel or other VEGFR-1 agonists could form the basis of a therapeutic scheme aimed toward protection of hepatocytes in at least some liver disorders. The addition of a VEGFR-2 agonist or other angiogenic factor at a lower ratio might result in a maximal therapeutic benefit by providing stimulation of angiogenesis.

Finally, although other organs showed some increase in mass after VEGF delivery, the liver exhibited the largest proportion of cells that had undergone DNA synthesis. HGF induction by VEGF or VEGFR-1 agonists is not a general response of endothelial cells; human umbilical vein endothelial cells and other endothelial cells tested did not show such induction (31). Therefore, the paracrine mechanism we describe is, at least in part, vascular bed–specific. Such a restricted induction may have a therapeutic advantage, in view of the pleiotrophic effects of HGF. Previous studies have reported on an angiogenic mitogen with selectivity for a specific type of endothelium (32). Perhaps “keys” more selective than VEGF may trigger the release of tissue-specific growth factors from the endothelium of other organs.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S6

Table S1

  • * Present address: Amgen, Alpenquai 30, Post Office Box 2065, CH-6002 Lucerne, Switzerland.

  • To whom correspondence should be addressed. E-mail: nf{at}


View Abstract

Navigate This Article