Endothelial Cell-Derived Angiopoietin-2 Controls Liver Regeneration as a Spatiotemporal Rheostat

See allHide authors and affiliations

Science  24 Jan 2014:
Vol. 343, Issue 6169, pp. 416-419
DOI: 10.1126/science.1244880


Liver regeneration requires spatially and temporally precisely coordinated proliferation of the two major hepatic cell populations, hepatocytes and liver sinusoidal endothelial cells (LSECs), to reconstitute liver structure and function. The underlying mechanisms of this complex molecular cross-talk remain elusive. Here, we show that the expression of Angiopoietin-2 (Ang2) in LSECs is dynamically regulated after partial hepatectomy. During the early inductive phase of liver regeneration, Ang2 down-regulation leads to reduced LSEC transforming growth factor–β1 production, enabling hepatocyte proliferation by releasing an angiocrine proliferative brake. During the later angiogenic phase of liver regeneration, recovery of endothelial Ang2 expression enables regenerative angiogenesis by controlling LSEC vascular endothelial growth factor receptor 2 expression. The data establish LSECs as a dynamic rheostat of liver regeneration, spatiotemporally orchestrating hepatocyte and LSEC proliferation through angiocrine- and autocrine-acting Ang2, respectively.

Vascular Endothelium and Tissue Regeneration

The vascular endothelium is increasingly being recognized to play a role during organogenesis and tissue regeneration. Hu et al. (p. 416) found that rapid down-regulation of endothelial-derived Angiopoietin-2 following partial hepatectomy releases an endogenous transforming growth factor β1–driven paracrine proliferative brake on hepatocytes. Later, recovery of endothelial Angiopoetin-2 expression facilitates angiogenesis in the regenerating liver in a vascular endothelial growth factor receptor 2–dependent manner. Thus, the vascular endothelium may help to orchestrate tissue regeneration through the control of inhibitory and stimulatory pathways in parenchymal and nonparenchymal cells.

The vascular endothelium is considered a passive cell population that acts in response to exogenous cytokines. However, recent work has shown that the endothelium can actively function as gatekeeper of tissue homeostasis. Endothelial cell-derived angiocrine signals orchestrate organogenesis during development (1, 2) and promote liver and lung regeneration in the adult (3, 4). Liver regeneration is a prototypic example of the intricate cross-talk between parenchymal cells and stromal cells (58). Liver sinusoidal endothelial cells (LSECs) have been shown to exert protective functions on hepatocytes (9) and promote hepatocyte proliferation during liver regeneration (3).

In order to systematically analyze the mechanisms of LSEC-regulated angiocrine growth control during liver regeneration, we isolated LSEC from sham-operated and two-thirds partial hepatectomized mice 1 day after surgery and performed transcriptomic gene expression analyses. Ninety-three genes were significantly up-regulated in LSEC upon partial hepatectomy (PHx) (table S1). Only nine genes were significantly down-regulated. Among the most strongly down-regulated LSEC genes was Angiopoietin-2 (Ang2) (Fig. 1A and table S2). Ang2 is a contextual antagonist of the vascular receptor tyrosine kinase Tie2 and is expressed at low levels in resting endothelial cells (10, 11). Angiogenic or inflammatory endothelial activation leads to the up-regulation of Ang2 (1214). The rapid down-regulation of LSEC Ang2 after PHx was consequently counterintuitive and prompted us to systematically study the role of LSEC-derived Ang2 during liver regeneration.

Fig. 1 Dynamics of Ang2 expression in LSECs after hepatectomy.

(A) Heat map representation of significantly changed LSEC genes from sham operated and hepatectomized mice one day after surgery (n = 4 mice). Details of identified genes are listed in tables S1 and S2. (B) Temporal kinetics of Ang2 expression during liver regeneration by means of quantitative PCR analysis of mRNA from whole-liver lysate (mean ± SD, n = 4 mice, **P < 0.01).

Quantitative polymerase chain reaction (PCR) analysis of Ang2 in liver lysates confirmed the rapid Ang2 down-regulation after PHx. One day after PHx, Ang2 mRNA levels were down-regulated to 18% of liver lysates from sham-operated mice. Ang2 expression thereafter steadily recovered to normal levels at day 8, when the liver restores its normal mass (Fig. 1B). The temporal pattern of Ang2 expression corresponded to the well-established pattern of hepatocyte and nonparenchymal cell proliferation after PHx (15). In mice, liver regeneration after PHx occurs rapidly by means of hepatocyte hyperplasia and hypertrophy to reach a proliferation peak as early as 48 hours after hepatectomy (16). Thereafter, hepatocyte proliferation steadily declines to baseline (15). In contrast, nonparenchymal cells, including LSECs, reach a proliferation peak 4 days after PHx, which is concomitant with the gradual recovery of Ang2. We therefore hypothesized that LSEC-derived Ang2 may negatively control hepatocyte proliferation and that Ang2 down-regulation after PHx may contribute to hepatocyte proliferation by releasing an angiocrine growth regulatory brake.

To examine this hypothesis, we performed PHx in wild-type (WT) and Ang2-deficient mice. Although Ang2-deficient mice showed impaired postnatal retinal angiogenesis characterized by a chaotic and incomplete vascular plexus (fig. S1) (17), the mice developed normally and had a similar life span as that of WT mice. Livers of Ang2-deficient mice were normal in size and showed the same hierarchical macrovasculature and microarchitecture of liver sinusoids as seen in WT mice (figs. S2 and S3 and movies S1 and S2). Two days after PHx at the peak of hepatocyte proliferation, liver regeneration in Ang2-deficient mice was enhanced as evidenced by significantly more Ki67+ hepatocytes and larger liver mass as compared with that of WT mice (Fig. 2, A to C). This pattern of increased hepatocyte proliferation was reverted during the angiogenic phase (days 4 to 8). In contrast to the steady increase of liver mass in WT mice, hepatocyte proliferation was lower in Ang2-deficient mice during the angiogenic phase of liver regeneration (Fig. 2, A to C). Similar results were obtained in WT mice treated with Ang2-neutralizing antibody (fig. S4, A to D).

Fig. 2 Genetic ablation of Ang2 dynamically alters hepatocyte proliferation during liver regeneration.

(A) Quantitation of Ki67-positive proliferating hepatocytes in liver sections of WT and Ang2-deficient mice after PHx (mean ± SD, n = 4 to 6 mice, *P < 0.05, **P < 0.01). (B) Representative images of Ki67-positive hepatocytes (brown stained large nuclei). (C) Liver-to-body-weight ratios of WT and Ang2-deficient mice after PHx (mean ± SD, n = 4 to 6 mice, *P < 0.05).

To unravel the mechanism of Ang2-regulated hepatocyte proliferation, we examined possible direct effects of Ang2 on hepatocytes. First, the exclusive expression of Ang2 and its receptor Tie2 in LSEC was confirmed (figs. S5, A and B, and S6A). Next, we directly stimulated hepatocytes with hepatocyte growth factor (HGF) and Ang2. In contrast to HGF, which induced strong c-Met, AKT, signal transducers and activators of transcription 3 (STAT3), c-Jun N-terminal kinase (JNK), and extracellular signal–regulated kinase (ERK) phosphorylation, Ang2 failed to trigger any intrinsic signaling in hepatocytes (fig. S5, C and D). Moreover, EdU incorporation revealed no direct effect of Ang2 on hepatocyte DNA synthesis (Fig. 3A and fig. S7).

Fig. 3 LSEC-derived Ang2 regulates hepatocyte proliferation indirectly by modulating the expression of TGFβ1 in LSEC.

(A) Primary WT mouse hepatocytes were stimulated with different cytokines and the percentage of EdU+ hepatocytes was determined (mean ± SD, n = 3 mice, *P < 0.05) (fig. S7, representative images). (B) TGFβ1, HGF, HB-EGF, IL-6, and TNFα mRNA expression was measured with quantitative PCR of liver lysates 1 day after PHx (mean ± SD, n = 4 mice, *P < 0.05). (C) TGFβ1, HGF, and HB-EGF expression was quantified with quantitative PCR of freshly isolated LSEC 1 day after PHx (mean ± SD, n = 4 mice, *P < 0.05). (D and E) Protein levels and phosphorylation states of SMAD, c-MET, and cyclin D1 in liver lysates of WT and Ang2-deficient mice were analyzed with immunoblotting. (F) Primary WT mouse hepatocytes were stimulated with conditioned medium from WT or Ang2-deficient LSECs supplemented with or without TGFβ1 neutralizing antibody, hepatocytes were pulsed with EdU, and the percentage of EdU+ hepatocytes was determined (mean ± SD, n = 4 mice, *P < 0.05, **P < 0.01).

To identify Ang2-regulated LSEC-derived hepatotropic cytokines, we performed a candidate-based screen of established hepatocyte growth factors. Expression of HGF, tumor necrosis factor–α (TNFα), heparin-binding epidermal growth factor (EGF)–like growth factor (HB-EGF), and interleukin-6 (IL-6) was not altered in livers of Ang2-deficient mice 1 day after PHx. In contrast, expression of the potent inhibitor of hepatocyte proliferation transforming growth factor–β1 (TGFβ1) was reduced in Ang2-deficient mice (Fig. 3B). The detailed temporal analysis of Ang2 and TGFβ1 expression identified a close relationship between Ang2 and TGFβ1 (fig. S8).

To validate the cellular sources of the analyzed cytokines, we examined the differential cytokine expression pattern in freshly isolated hepatic cells, including hepatocytes, LSECs, Kupffer cells, and stellate cells. TGFβ1 and HB-EGF were predominately expressed by LSECs. On the contrary, HGF, IL-6, TGFβ2, and TGFβ3 were primarily expressed by stellate cells (fig. S6B) (1821). Consistent with the TGFβ1 expression in whole liver lysates, TGFβ1 expression in LSECs of Ang2-deficient mice decreased by 50% compared with that of WT LSECs (Fig. 3C). Enzyme-linked immunosorbent assay analyses of liver lysates from hepatectomized mice similarly showed reduced TGFβ1 levels in Ang2-deficient mice (fig. S9A).

TGFβ1 controls hepatocyte proliferation by binding to its type II receptor (TGFBR2) and then recruiting type I receptor (TGFBR1) (22). This activates downstream SMAD signaling, leading to reduced Cyclin D1 and Cyclin E expression and hepatocyte cell cycle arrest (23). Although the expression of TGFBR1 and TGFBR2 was rapidly down-regulated after PHx (fig. S9B) (24), their expression was similar in hepatectomized livers of WT and Ang2-deficient mice (fig. S9C). Corresponding to the reduced TGFβ1 in Ang2-deficient mice, SMAD2 and SMAD3 phosphorylation in liver lysates of Ang2-deficient mice was significantly reduced as compared with WT mice (Fig. 3D). Serum HGF levels were similar in WT and Ang2-deficient mice (fig. S10). Yet, Ang2-deficient mice showed stronger phosphorylation of c-Met in the liver (Fig. 3E). In line with reduced TGFβ1-TGFBR activity, Cyclin D1 expression levels were significantly elevated in Ang2-deficient livers after PHx (Fig. 3E).

To validate the LSEC-controlled Ang2/TGFβ1 hepatocytes growth regulatory loop, we stimulated cultured hepatocytes with conditioned media from WT or Ang2-deficient LSECs. Hepatocyte exposed to conditioned medium from Ang2-deficient LSEC showed significantly higher EdU incorporation. Addition of a TGFβ1-neutralizing antibody further increased EdU incorporation (Fig. 3F), validating that Ang2 indirectly regulated hepatocyte proliferation by controlling LSEC-derived TGFβ1 expression.

In contrast to enhanced liver regeneration in Ang2-deficient mice during the early inductive phase (days 0 to 3), liver regeneration was significantly impaired in Ang2-deficient mice at later stages (days 4 to 8) (Fig. 2A). Cleaved Caspase-3 staining revealed that hepatocyte apoptosis was a rare event during later stages of liver regeneration (fig. S11). The later angiogenic phase of liver regeneration after PHx (days 4 to 8) is characterized by expansion of the nonparenchymal compartment (LSECs, Kupffer cells, and stellate cells) at a time when most hepatocyte proliferation is completed (15). We consequently hypothesized that the impaired liver regeneration in Ang2-deficient mice during the later angiogenic phase could have resulted from changes in stromal cell proliferation. Indeed, Ang2-deficient mice showed a significant reduction of nonparenchymal cell proliferation (fig. S12). To further distinguish which nonparenchymal cell population was affected, mice were pulsed for 2 hours with EdU 4 days after PHx. Subsequently, nonparenchymal cells were isolated, immunostained, and analyzed. This analysis revealed that LSEC proliferation was dramatically reduced in the absence of Ang2 (Fig. 4A). Hepatocyte-derived vascular endothelial growth factor (VEGF) is known to act as a potent stimulus for LSEC proliferation (25). However, serum VEGF levels were similar in WT and Ang2-deficient mice 2 and 4 days after PHx (fig. S13). Candidate-based gene expression screening revealed that VEGF receptor 2 (VEGFR2) expression in LSEC was decreased by 60% in Ang2-deficient mice (Fig. 4B). Expression of the transcription factor Id1, which acts downstream of VEGFR2, was unchanged in the absence of Ang2 (fig. S14). However, the expression of Wnt2, a hepatotropic angiocrine factor whose expression is controlled by VEGFR2 signaling (3), was markedly reduced in Ang2-deficient LSECs during the angiogenic phase (fig. S14).

Fig. 4 Ang2 deficiency inhibits LSEC proliferation and contributes to impaired liver regeneration during the angiogenic phase of liver regeneration.

(A) Four days after PHx, WT and Ang2-deficient mice were pulsed with EdU, and the percentage of proliferating LSEC was quantified (mean ± SD, n = 3 mice, **P < 0.01). (B) VEGFR2 mRNA of LSECs isolated from WT and Ang2-deficient mice 4 days after hepatectomy was quantified with quantitative PCR (mean ± SD, n = 3 mice, **P < 0.01). (C and D) VEGFR2 levels in control small interfering RNA (siRNA) (si-Ctr)–, Ang2 siRNA (si-Ang2)–, or Tie2 siRNA (si-Tie2)–transfected HUVEC were analyzed by means of immunoblotting. (E) Proliferation of si-Ctr– and si-Ang2–transfected HUVEC was quantified by means of EdU incorporation assay (mean ± SD, n = 3 replicates, *P < 0.05, **P < 0.01) (fig. S16, representative images). (F) Migration of HUVEC after Ang2 knockdown was evaluated by means of wound healing assay (mean ± SD, n = 4 replicates, **P < 0.01) (fig. S17, representative images). (G) Model of endothelial-derived Ang2 function in the control of liver regeneration.

We further validated that Ang2 regulated VEGFR2 expression in cultured human endothelial cells (HUVECs). Ang2 silencing significantly reduced mRNA and protein levels of VEGFR2 (Fig. 4C and fig. S15A). Moreover, Ang2 overexpression led to increased VEGFR2 expression (fig. S15B). Conversely, silencing of the Ang2 receptor Tie2 increased VEGFR2 expression, indicating that Ang2-regulated VEGFR2 expression was Tie2-dependent (Fig. 4D). In line with the Ang2-regulated VEGFR2 expression, Ang2 silencing not only reduced HUVEC proliferation but also inhibited HUVEC migration and delayed wound closure (Fig. 4, E and F, and figs. S16 and S17). Consistently, migration of cultured LSEC from Ang2-deficient mice was similarly impaired (fig. S18).

This study was aimed at elucidating angiocrine mechanisms of liver regeneration. After the identification of Ang2 as a key vascular-derived regulator of liver regeneration, we hypothesized that similar paracrine cross-talk mechanisms may be operative during chronic liver injury. To study this hypothesis, WT and Ang2-deficient mice were treated with CCL4. Chronic liver injury was evident by a rough liver surface, collagen deposits, and elevated serum levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (fig. S19, A, B, and E). Sirius red and cleaved Caspase-3 staining showed similar levels of collagen deposits and hepatocyte apoptosis in both genotypes (fig. S19, B to D). Serum levels of AST, ALT, and alkaline phosphatase (AP) also showed no significant differences (fig. S19E). Nonetheless, hepatocytes of Ang2-deficient mice showed ~2.5 times higher proliferation rates as compared with that of hepatocytes of WT mice (fig. S19, F and G).

LSEC-derived Ang2 is required to spatiotemporally orchestrate the proliferation of hepatocytes and LSEC to efficiently restore liver structure and function after hepatectomy (Fig. 4G): The rapid down-regulation of Ang2 after hepatectomy regulates hepatocyte proliferation in a paracrine (angiocrine) manner by down-regulating LSEC TGFβ1 production, removing an endogenous growth inhibitory mechanism. During the angiogenic phase, recovery of Ang2 expression controls LSEC proliferation by regulating VEGFR2 expression in an autocrine manner. The dynamics of these divergent indirect and direct two-compartment effects are strictly controlled by the temporal regulation of LSEC Ang2 expression during liver regeneration (rapid down-regulation after hepactectomy and gradual recovery during the later angiogenic phase). Previous studies have shown that LSECs support liver regeneration in a stimulatory manner by secreting hepatotropic cytokines, including HGF and Wnt2 (3, 9). In this study, we establish that LSECs control liver regeneration in a much more dynamic manner through stimulatory and inhibitory effects. These data indicate a general mechanism by which signals transmitted from LSECs to hepatocyte play critical and rate-limiting roles in orchestrating liver regeneration. Collectively, the data shed fundamental insights into the role of the endothelium as a gatekeeper and regulator of tissue homeostasis and regeneration.

Supplementary Materials

Materials and Methods

Figs. S1 to S20

Tables S1 to S3

References (2629)

Movies S1 and S2

References and Notes

  1. Acknowledgments: We thank G. Thurston (Regeneron, Tarrytown, NY) for providing Ang2-deficient mice and critical discussions. We also thank A. Budnik and S. Savant (DKFZ Heidelberg, Germany) for important discussions and comments on the manuscript. We gratefully acknowledge the excellent technical support of F. Bestvater and M. Brom of the Microscopy Facility, M. Bewerunge-Hudler of the Microarray Unit, and the DKFZ laboratory animal core facility. This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB-TR23 “Vascular Differentiation,” SFB-TR77 “HCC,” and SFB873 “Stem Cell Biology”), the DKFZ-MOST Cancer Research Cooperation, and the EU FP7 “SYSTEMAGE” (all to H.G.A.). H.G.A. is supported by an endowed chair from the Aventis Foundation. Microarray data were deposited in the Gene Expression Omnibus (GEO) database (accession no. GSE50046).
View Abstract

Stay Connected to Science

Navigate This Article