Thrombospondin-1 is a novel negative regulator of liver regeneration after partial hepatectomy via TGF-β1 activation in mice

Abstract

The matricellular protein thrombospondin-1 (TSP-1) is prominently expressed during tissue repair. TSP-1 binds to matrix components, proteases, cytokines, and growth factors and activates intracellular signals via its multiple domains. TSP-1 converts latent transforming growth factor-β1 (TGF-β1) complexes into biologically active form. TGF-β plays significant roles in cell-cycle regulation, modulation of differentiation, and induction of apoptosis. Although TGF-β1 is a major inhibitor of proliferation in cultured hepatocytes, the functional requirement of TGF-β1 during liver regeneration remains to be defined in vivo. We generated a TSP-1-deficient mouse model of a partial hepatectomy and explored TSP-1 induction, progression of liver regeneration, and TGF-β-mediated signaling during repair process following hepatectomy. We show here that TSP-1-mediated TGF-β1 activation plays an important role in suppressing hepatocyte proliferation. TSP-1 expression was induced in endothelial cells as an immediate early gene in response to partial hepatectomy. TSP-1 deficiency resulted in significantly reduced TGF-β/Smad signaling and accelerated hepatocyte proliferation via downregulation of p21 protein expression. TSP-1 expression in endothelial cells was induced by reactive oxygen species (ROS) and modulated TGF-β/Smad signaling and proliferation in hepatocytes in vitro, suggesting that the immediately and transiently produced ROS in the regenerating liver were the responsible factor for TSP-1 induction.

Conclusions

We have identified TSP-1 as an inhibitory element in regulating liver regeneration via TGF-β1 activation. Our work defines TSP-1 as a novel immediate early gene that could be a potential therapeutic target to accelerate liver regeneration.

Cell proliferation is part of the wound healing response and plays a central role in regeneration after tissue damage. It is crucial to advance our understanding of the molecular mechanisms underlying tissue regeneration and to develop a novel strategy to enhance the regenerative process. Such knowledge in turn would yield clinical benefits, such as decreased morbidity and mortality. Partial hepatectomy is a well-established model system in rodents for studying the molecular mechanisms of liver regeneration. Partial hepatectomy triggers activation of the immediate early genes (genes that are rapidly but transiently activated) within approximately the first 4h (1), and thereby hepatocytes re-enter the cell-division cycle. Immediate early genes encode proteins that regulate later phases in G1 and play an important role in cell growth in the regenerating liver (1–2). The process of liver regeneration after hepatectomy is coordinated by both pro-and anti-proliferative factors. Transforming growth factor-β1 (TGF-β1) is a potent inhibitor of mitogen-stimulated DNA synthesis in cultured hepatocytes (3). Therefore, it has been thought that TGF-β1 is a potent candidate to limit or stop liver regeneration following hepatectomy (4). Since TGF-β is synthesized and secreted as a latent complex, the important step in regulating its biological activity is the conversion of the latent form into the active one. However, the contribution of TGF-β to the liver’s regenerative response after hepatectomy is still poorly understood. The TGF-β1 mRNA induction occurs within 4h, and levels of TGF-β1 remain elevated until 72h after hepatectomy (5–6). In sharp contrast, in the model of complete lack of TGF-β signaling using hepatocyte-specific TGF-β type II receptor knockout mice, the lack of TGF-β signaling does not result in prolonged hepatocyte proliferation; rather, only transiently upregulated proliferation of hepatocyte is shown in the later phase after hepatectomy, with a peak at ~36h (7). These differences raise an open question about whether locally activated TGF-β1 is indeed essential for the inhibition of hepatocyte proliferation in vivo. Furthermore, the time course of locally activated TGF-β1 and its activation mechanism following hepatectomy still remain largely unknown.

The matricellular protein thrombospondin-1 (TSP-1) was first shown as a component of the α-granule in platelets and can act as a major activator of latent TGF-β1 (8–9). TSP-1 is induced in response to tissue damage or stress and plays a role as a transient component of extracellular matrix during tissue repair (8, 10–11). However, the roles of TSP-1 and of TSP-1/TGF-β1 interdependence during liver regeneration have not yet been addressed. We hypothesize that the initiation of local TGF-β activation occurs much earlier following hepatectomy, and TSP-1 plays a critical role in this process. Here, using a TSP-1-deficient mouse model, we investigated whether TSP-1 was a suitable molecular target for accelerating liver regeneration after partial hepatectomy.

Materials and Methods

Mutant mice and animal studies

TSP-1-null mice were kindly provided by Dr. Jack Lawler (Beth Israel Deaconess Medical Center, Boston, MA, USA) (12). Male wild-type and TSP-1-null mice, at 8 to 12 wks old (C57BL/6 background), were used for the experiments. The two anterior lobes (median and left lateral lobes), which comprise 70% of the liver weight, were resected, while the caudate and right lobes were left intact. This study was approved by the Institutional Animal Care and Use Committee.

Immunostaining and western blotting

For histological analyses, liver samples (the same lobe from each mouse) were either directly frozen in OCT compound (Tissue-Tek, Sakura Finetek) or fixed overnight in 4% paraformaldehyde in PBS, pH 7.2, and dehydrated in a graded alcohol series and embedded in paraffin. Then the materials were sectioned at a thickness of 5 μm. Immunofluorescence and immunohistochemical staining was performed as described previously (13). The negative control staining was performed without addition of primary antibody. Immunostained slides were viewed under a Leica DM 5500B microscopic system. A minimum of 10 different images were randomly selected and the data shown were as representative of results. Western analysis were performed as described previously (13). The same lobe from each mouse was used for protein isolation and subsequent analysis. Image J 1.40 software was used for densitometric analysis.

Assessment of BrdU incorporation

Mice received an intraperitoneal injection of 5-bromo-2-deoxyuridine (BrdU, 100 mg/kg; Roche Applied Science) 2h prior to sacrifice. Six random visual high-power fields (0.64 mm² per field) per mouse were evaluated to determine the number of BrdU-positive nuclei in hepatocytes and nonparenchymal cells. Nonparenchymal cells were defined as cells with smaller, irregularly shaped nuclei compared with larger, circular nuclei of hepatocytes, as previously described (14). All BrdU-positive cells, from both cell types, were summed at each time point.

Assessment of apoptotic index

TUNEL analysis was performed using an in situ apoptosis detection kit (Roche). Six visual high-power fields (0.64 mm² per field) per mouse were evaluated to determine the number of TUNEL-positive nuclei.

Antibodies

The antibodies used for analyses were summarized in Supplementary Table 1. The amount of active and total TGF-β1 in liver samples was determined using an Elisa kit (Quantikine® TGF-β1 Immunoassay, from R&D) according to the manufacturer’s instructions.

Real-time PCR

Real-time PCR was performed as described previously (13). The primers used were summarized in Supplementary Table 2.

Lipid peroxidation assay

The liver tissue content of malondialdehyde (MDA) was measured by the thiobarbituric acid reduction method using a commercially available kit (Cayman Chemical #10009055). Values were obtained after 30-min incubation at 90°C under acidic conditions.

In vitro assay using human umbilical vein endothelial cells (HUVECs) and mouse primary hepatocytes

HUVECs were used at passages 3–6. For analysis of reactive oxygen species (ROS), hydrogen peroxide (H₂O₂) (Fisher Scientific) and N-acetylcysteine (NAC) (Calbiochem) were used as a ROS inducer and a ROS scavenger, respectively. To examine the effects of H₂O₂ on TSP-1 expression, HUVECs were seeded on the 0.1% gelatin coated culture plates and incubated overnight. Without change of medium, H₂O₂ was applied at final concentrations of 0.01, 0.05, and 0.1 mM, and incubated for 10-min. For immunocytochemistry, HUVECs were plated into Lab-Tek Permanox slides precoated with 0.1% gelatin and incubated overnight. Then the cells-with or without pretreatment with 30 mM NAC for 60-min were treated with 0.1 mM H₂O₂ for 10-min.

To examine the effects of HUVEC-derived TSP-1 on TGF-β/Smad signaling and proliferation in primary hepatocyte cultures, primary hepatocytes were isolated from 8- to 12-wk-old adult wild-type mouse livers using collagenase perfusion as described (15). Isolated hepatocytes were plated on type I collagen (10 μg/ml)-coated dishes in Williams’ E medium supplemented with 5 μg/ml insulin, 5 μg/ml transferrin, 10 ng/ml EGF, 10⁻⁵ M aprotinin, 10⁻⁵ M dexamethasone, 10⁻³ M nicotinamide, and 10% FBS, and incubated at 37°C for 24h. To examine the effect of HUVEC-derived TSP-1 on TGF-β/Smad signaling in hepatocytes, the conditioned media from HUVECs (treated with 1.0 mM H₂O₂ for 2h) were added to primary hepatocytes with or without pretreatment of 5 μM LSKL or SLLK peptide (GenScript) (16–17), cultured for additional 4h, and the cells were used for the analysis. To examine the effect of HUVEC-derived TSP-1 on hepatocyte proliferation, the conditioned media from HUVECs were added to primary hepatocytes, cultured for an additional 24h, and the cells were used for the analysis.

Data presentation and statistical analysis

All experiments were performed in triplicate, and the data shown are representative of results consistently observed. Data are expressed as the mean ± SD. Data analysis was performed with SPSS 12.0.1 for Windows (SPSS Inc., Chicago, IL). Statistical analyses were performed using Student’s t test or ANOVA followed by Bonferroni’s multiple comparison tests when appropriate. A P value of < .05 was considered significant.

Results

Partial hepatectomy induces an immediate and prominent induction of TSP-1 mRNA and protein in the regenerating liver

An intact liver in adult mice expresses nearly undetectable levels of TSP-1 mRNA (12). We first determined whether partial hepatectomy could trigger TSP-1 induction in the regenerating liver. TSP-1 mRNA was immediately induced, with a peak at 3h following hepatectomy, in wild-type mice by real-time PCR (Figure 1A). TSP-1 protein was also induced, reaching a peak at ~6h (Figure 1B). Those mRNA and protein levels returned to basal levels by 24h (Figure 1, A and B). Thus, partial hepatectomy induced immediate and transient TSP-1 expression in the initial phase of liver regeneration. Secondary minor inductions of TSP-1 mRNA and protein were found to peak at 48h and 72h, respectively (Figure 1, A and B).

Figure 1. An immediate and significant induction of TSP-1 mRNA and protein in response to partial hepatectomy.

(A) Real-time PCR analysis of TSP-1 mRNA expression after 70% partial hepatectomy (70% PH) (n=8 per time point) vs. sham operation (n=3 per time point). Data were normalized to the amount of 18S ribosomal RNA serving as the internal control. *, P < .05 vs. sham-operated mice; **, P < .01 vs. sham-operated mice.

(B) Western blot analysis of TSP-1 protein expression in the regenerating liver. Heat shock cognate protein 70 (HSC70) served as a loading control.

(C) Double immunofluorescence staining of TSP-1 (red) and GPIIb/IIIa (green) at 0h in wild-type (TSP1^+/+) and TSP-1-null (TSP1^−/−) liver. Note that the distribution of TSP-1 in the intact wild-type liver shows only in platelets, as evidenced by co-localization of TSP-1 with the platelet marker GPIIb/IIIa (yellowish-dots in the merged image). Scale bar=50 μm.

(D) Immunofluorescence staining for TSP-1 in wild-type liver at 0, 6 and 72h after hepatectomy. Scale bar=50 μm.

We next determined the cellular source of TSP-1 by immunostaining. In the intact liver, the expression of TSP-1 protein was detectable only in platelets with GPIIb/IIIa expression by double immunofluorescence staining (Figure 1C). The tissue distribution of TSP-1 protein localized in the sinusoid at 6h and 72h after hepatectomy (Figure 1D), suggesting that cells localized in the sinusoid (e.g., endothelial cells, Kupffer cells, and hepatic stellate cells [HSCs]) are responsible for newly synthesized TSP-1 in the regenerating liver. Double immunofluorescence staining revealed that TSP-1 protein predominantly co-localized with PECAM-1/CD31 (an endothelial cell marker) at 6h in the regenerating liver (Figure 2A). In contrast, TSP-1 protein at 6h did not co-localized with either F4/80 (a Kupffer cell marker) or α-smooth muscle actin (α-SMA, a marker for myofibroblasts such as activated hepatic stellate cells [HSCs]) (Figure 2A). The activation peak of HSCs is at 72h after hepatectomy (18), and many α-SMA-positive cells were observed (Supplementary Figure 1). At 72h, however, TSP-1 protein did co-localize with PECAM-1/CD31 and α-SMA, but not with F4/80 (Figure 2B). Indeed, it is known that activated HSCs express TSP-1 and thereby activate the TGF-β signaling pathway in vitro (19). These results suggest that endothelial cells are the major source of TSP-1 expression in the initial phase at 6h, whereas endothelial cells and activated HSCs participate in secondary TSP-1 expression at 72h. As noted above, immediate early genes are genes that are rapidly but transiently (within approximately the first 4h) activated in response to hepatectomy (1–2). Thus, TSP-1 produced by endothelial cells is a novel candidate immediate early gene in the initial response to partial hepatectomy.

Figure 2. Tissue distribution of TSP-1 protein in the regenerating liver.

(A and B) TSP-1 expression at 6h (A) and 72h (B) after partial hepatectomy.

Double immunofluorescence staining for TSP-1/PECAM-1(CD31), TSP-1/F4/80, and TSP-1/α-SMA at 6 and 72h in wild-type mice (TSP-1, red; PECAM-1(CD31), F4/80, and α-SMA, green). Scale bar=25 μm.

TSP-1 deficiency accelerates a liver regeneration after partial hepatectomy, but does not affect the termination phase

Since immediate early genes play a significant role in the regulation of cell growth in the regenerating liver (1–2), we next examined the involvement of TSP-1 in the control of liver regeneration. The rates of recovery of liver mass and of cell proliferation after hepatectomy were compared between wild-type and TSP-1-null mice. TSP-1-null mice showed significantly faster recovery of liver:body weight ratio from day 1 to day 7 after surgery compared with controls (P < .05 at 24, 48, 168h, and P < .01 at 72h; Figure 3A). However, no excess liver mass had been gained at day 14 in TSP-1-null mice compared with controls. Next, cell proliferation was evaluated using a BrdU incorporation assay (a marker for the S-phase of the cell cycle). The proliferation peaks of hepatocytes and nonparenchymal cells after partial hepatectomy occurred at ~36–48h and 72h, respectively (2, 4, 14). Although only a few BrdU-positive hepatocytes were detectable at 24h in wild-type mice, TSP-1-null mice showed a significantly increased number of BrdU-positive hepatocytes (8-fold over controls) (P < .01; Figure 3, B and C). The number of BrdU-positive nonparenchymal cells in TSP-1-null mice significantly increased (2-fold) at 72h compared with controls (P < .01; Figure 3C). Total proliferative activity (of hepatocytes and nonparenchymal cells) in TSP-1-null mice was significantly higher at 24h and 72h compared with controls (P < .01 in both; Figure 3C).

Figure 3. Accelerated liver regeneration with downregulation of p21 protein expression in TSP-1-null mice after partial hepatectomy.

(A) Assessment of restoration of liver mass. Liver:body weight ratio was measured after partial hepatectomy (PH) (n=10 per time point in each group). *, P < .05 vs. TSP-1⁺/⁺ mice; **, P < .01 vs. TSP-1⁺/⁺ mice.

(B and C) Assessment of BrdU incorporation in the regenerating liver.

(B) Immunohistochemistry of BrdU in the regenerating liver. Arrowheads indicate BrdU-positive hepatocyte nuclei (brown) at 24h. Scale bar=50 μm.

(C) The number of BrdU-positive hepatocytes, nonparenchymal cells, and all positive cells (n=10 per time points in each group). **, P < .01 vs. TSP-1⁺/⁺ mice.

(D) Real-time PCR analysis of cyclin A2 (Ccna2) and cyclin D1 (Ccnd1) mRNA expression in the regenerating liver (n=10 per time point in each group). *, P < .05 vs. TSP-1⁺/⁺ mice.

(E) Assessment of p21 protein expression in the regenerating liver.

Left panels: Western blot analysis of p21 protein expression in wild-type vs. TSP-1-null liver. HSC70 was used as a loading control. Right panel: Densitometric analysis of p21 protein expression (n=3). Each p21 intensity was normalized to HSC70, then the intensity of wild-type mice at 0h was set to 1. *, P < .05 vs. TSP-1⁺/⁺ mice.

Cyclins are required for cell cycle progression. The mRNA levels of cyclin A2 and cyclin D1 increase and peak in S-phase and early to mid G1-phase, respectively. Expression levels of Ccna2 mRNA in TSP-1-null mice were significantly higher at 24h (2.3-fold) and 72h (1.5-fold) compared with controls (P < .05 in both; Figure 3D). Although Ccnd1 mRNA levels increased and peaked at 48h in both wild-type and TSP-1-null mice, there was no significant difference between them (Figure 3D). The cyclin-dependent kinase inhibitor p21 plays a critical role in the inhibition of hepatocyte proliferation at the G1/S transition of the cell cycle in vivo (20). Induction levels of p21 protein in TSP-1-null mice significantly diminished at 12h and 24h compared with controls (70% less than that of controls both at 12h and 24h; P < .05 in both), whereas p21 showed at similar levels at 48h in wild-type and TSP-1-null liver (Figure 3E). These results suggest that TSP-1 is a negative regulator of liver regeneration after partial hepatectomy and that TSP-1 deficiency accelerates the S-phase entry of hepatocytes by downregulation of p21 protein expression. However, TSP-1 does not affect the termination phase of liver regeneration after partial hepatectomy.

TGF-β/Smad signaling is activated by TSP-1 in response to partial hepatectomy

To address the possible mechanisms underlying this accelerated liver regeneration in TSP-1-null mice, we examined TGF-β/Smad signaling. TGF-β1 mRNA levels in both wild-type and TSP-1-null mice increased after hepatectomy by real-time PCR, and those levels in TSP-1-null mice were significantly upregulated at 3h and 6h compared with controls (P < .05 at 3h and P < .01 at 6h; Figure 4A). In sharp contrast, the levels of active TGF-β1 in TSP-1-null liver were significantly lower than controls at 6h after partial hepatectomy, whereas the levels of total TGF-β1 did not show any significant differences between them (Figure 4B). Furthermore, the levels of phosphorylated Smad2 (pSmad2, C-terminal Ser465/467) protein as a downstream mediator of active TGF-β1 significantly diminished at 6h and 12h in TSP-1-null mice compared with controls (to 16% at 6h and 69% at 12h vs. controls, respectively; P < .01 at 6h, and P < .05 at 12h), as determined by western blotting (Figure 4C). Using immunofluorescence staining, we confirmed the significantly decreased number of nuclear localized pSmad2-positive cells at 6h in TSP-1-null mice compared with controls (P < .01; Figure 4D). A secondary, minor induction of pSmad2 at 72h was also significantly attenuated in TSP-1-null mice compared with controls (Figure 4C).

Figure 4. Significantly decreased TGF-β/Smad signal transduction and cell death in TSP-1-null mice after partial hepatectomy.

(A) Real-time PCR analysis of TGF-β1 mRNA expression after partial hepatectomy (PH) (n=8 per time point in each group). *, P < .05 vs. TSP-1⁺/⁺ mice; **, P < .01 vs. TSP-1⁺/⁺ mice.

(B) The levels of active and total TGF-β1 in wild-type and TSP-1-null liver at 6h after partial hepatectomy (n=6 per time point in each group). **, P < .01 vs. TSP-1⁺/⁺ mice.

(C-E) Effects of TSP-1 deficiency on pSmad2 expression in the regenerating liver.

(C) Upper panels: Western blot analysis of pSmad2 and total Smad2 in wild-type and TSP-1-null liver.

Lower panel: Densitometric analysis of pSmad2 protein expression (n=3). Each pSmad2 intensity was normalized to total Smad2, then the intensity of wild-type mice at 0h was set to 1. *, P < .05 vs. TSP-1⁺/⁺ mice. **, P < .01 vs. TSP-1⁺/⁺ mice.

(D) Assessment of pSmad2 nuclear localization. Left panel: Immunofluorescence staining for pSmad2 at 6h in wild-type and TSP-1-null liver. Right panel: Analysis of pSmad2-positive nuclei (n=5 in each group). Arrowheads indicate pSmad2 (red)/4′,6-diamidino-2-phenylindole (DAPI, blue) double-positive nuclei (purple). **, P < .01 vs. TSP-1⁺/⁺ mice. Scale bar=25 μm.

(E) Real-time PCR analysis of PAI-1 mRNA expression in wild-type and TSP-1-null liver after PH (n=8 per time point in each group) and in wild-type liver after sham operation (n=3 per time point). **, P < .01 vs. TSP-1⁺/⁺ mice.

(F) Assessment of TUNEL-positive cell death in the regenerating liver. Left panel: TUNEL staining at 6h after PH in wild-type and TSP-1-null liver. Right panel: Analysis of TUNEL-positive cells (n=5 per time point in each group). Arrowheads indicate TUNEL (red)/DAPI (blue) double-positive nuclei (purple). **, P < .01 vs. TSP-1⁺/⁺ mice. Scale bar=50 μm.

Plasminogen activator inhibitor-1 (PAI-1) is one of the downstream targets of TGF-β1 in hepatocytes (21). Although intense inductions of PAI-1 mRNA at 6h after hepatectomy were observed in both wild-type mice and TSP-1-null mice by real-time PCR, the induction level in TSP-1-null mice was significantly diminished (to 37% of controls; P < .05 at 6h) (Figure 4E).

Cell death is also implicated as a mechanism of TGF-β-mediated cell growth inhibition. TUNEL-positive cells as a marker for cell death are immediately and transiently detectable after hepatectomy (22). We determined whether deficiency of TSP-1 affected cell death in the regenerating liver. Although the number of TUNEL-positive cells in wild-type liver transiently increased at 6h after hepatectomy, TSP-1-null liver showed a significant reduction compared with controls (P < .05 at 6h; Figure 4F).

These results suggest that TSP-1-mediated active TGF-β1 plays a pivotal role in TGF-β/Smad signal transduction after partial hepatectomy.

Deficiency of TSP-1 accelerates STAT3 and PI3K/Akt signals, not Erk1/2 signal in the early phase after partial hepatectomy

There is in vitro evidence that TSP-1 downregulates phosphorylated Akt (Ser473) expression via its receptor CD47 in HUVECs (23). Indeed, signaling pathways such as PI3K/Akt, STAT3, and Erk1/2, are important for cell survival and/or proliferation after hepatectomy (24). Therefore, we next examined whether the deficiency in TSP-1 affected the activation of these signaling pathways in the early phases post hepatectomy. TSP-1-null mice showed earlier and more intense phosphorylation of STAT3 (Tyr705) (6-fold at 1h; P < .01) and Akt (Ser473) (4.2-fold at 1h; P < .01) in the early stage after hepatectomy compared with controls as determined by western blotting (Figure 5). In contrast, levels of phosphorylated Erk1/2 did not show any remarkable differences between the two groups (Figure 5).

Figure 5. TSP-1 deficiency enhances STAT3 and PI3K/Akt but not Erk1/2 signal in the early phase after partial hepatectomy.

Upper panels: Western blot analysis of STAT3, PI3K/Akt, and Erk1/2 signals. HSC70 served as a loading control. Lower Panels: Densitometric analysis of phosphorylated protein expression after partial hepatectomy (PH). Each pSTAT3, pAkt, and pErk1/2 intensity was normalized to HSC70, then the intensity of wild-type mice at 0h was set to 1. *,P < .05 vs. TSP-1⁺/⁺ mice. **, P < .01 vs. TSP-1⁺/⁺ mice.

TSP-1 induction in endothelial cell is associated with reactive oxygen species

Although our findings show that TSP-1 plays a potential role as a negative regulator in the regenerating liver, the mechanism of TSP-1 induction in endothelial cells in response to hepatectomy remains unknown. There is a line of evidence that ROS are produced in the regenerating liver after hepatectomy (22, 25). In wild-type mice, levels of tissue content of MDA as a lipid peroxidation marker for ROS generation were significantly increased at both 3h and 6h and returned to basal levels by 12h after hepatectomy (P < .05 in both; Figure 6A). Next, to determine whether ROS could induce TSP-1 expression in endothelial cells, we performed an in vitro study using HUVECs with the potent ROS inducer H₂O₂. In HUVECs, treatment with H₂O₂ induced TSP-1 protein expression in a dose-dependent manner (Figure 6, B–D). Furthermore, this induction was inhibited by pretreatment with 30 mM NAC, a scavenger of ROS (Figure 6, B–D). Thus, these results indicate that oxidative stress is one factor responsible for TSP-1 induction in endothelial cells.

Figure 6. TSP-1 induction in endothelial cell by ROS.

(A) Assessment for levels of MDA after 70% partial hepatectomy (70% PH) (n=5) and sham operation (n=3). *, P < .05 vs. sham-operated mice.

(B-D) TSP-1 protein expression by H₂O₂,a potent ROS inducer, in HUVECs

(B) Western blot analysis of TSP-1 after treatment of HUVECs with H₂O₂. β-tubulin served as a loading control.

(C) Densitometric analysis of TSP-1 expression from three independent experiments. Each TSP-1 intensity was normalized to β-tubulin, then the intensity of control was set to 1. Note that the TSP-1 protein expression levels after treatment with 0.05 and 0.1 mM H₂O₂ are significantly higher vs. controls, whereas the induction of TSP-1 by treatment with 0.1 mM H₂O₂ is significantly inhibited with a pretreatment using 30 mM NAC. *, P < .05. **, P < .01.

(D) Immunocytochemistry for TSP-1 protein in HUVECs after treatment with H₂O₂ (TSP-1, green; DAPI, blue). Note that HUVECs after treatment with 0.1 mM H₂O₂ express TSP-1 in their cytoplasm (arrowheads), whereas the induction of TSP-1 is inhibited by pretreatment using 30 mM NAC. Scale bar=50 μm.

(E, F) Assessment of pSmad2 nuclear localization in primary hepatocytes.

(E) Immunofluorescence staining for pSmad2 with or without ROS-treated conditioned media from HUVECs (HUVEC CM). Scale bar=25 μm.

(F) The effect of TSP-1-inhibitory peptide LSKL on pSmad2 induction. Error bars represent standard deviation (n = 5 in each group; field = 0.15 mm²). SLLK, control peptide. **, P < .01.

To further determine whether HUVEC-derived TSP-1 can modulate TGF-β/Smad signaling and proliferation in hepatocytes in vitro, we isolated primary hepatocytes from adult control mice (15). The treatment of conditioned media from HUVECs with primary hepatocytes actually induced pSmad2 (Figure 6E). Furthermore, the pretreatment of primary hepatocytes with TSP-1-inhibitory peptide LSKL (16–17) significantly suppressed conditioned media-induced pSmad2 expression, whereas the control peptide SLLK showed no effects (Figure 6F). It is known that primary hepatocytes lack the ability to proliferate, even though such cells in vivo readily replicate and/or synthesize DNA following partial hepatectomy (26). Although a few proliferative primary hepatocytes were found by Ki67 immunostaining in culture, the treatment of conditioned media from HUVECs with primary hepatocytes significantly reduced the number of Ki67-positive cells (Supplementary Figure 2).

Discussion

In the present study, we have demonstrated the following (Figure 7): (i) TSP-1 is induced in endothelial cells as an immediate early gene by ROS and participates in TGF-β signal transduction in the initial response to partial hepactectomy; and (ii) TSP-1 deficiency results in the significant reduction of TGF-β/Smad signal and this could cause the accelerated S-phase entry of hepatocytes by downregulation of p21 protein expression. Thus, this is the first study providing compelling evidence that local TGF-β activation machinery plays an important role in inhibiting liver regeneration after hepatectomy.

Figure 7. Schematic illustration of the role of TSP-1 in the regenerating liver.

In wild-type mice, newly synthesized ROS in response to partial hepatectomy stimulate endothelial cells to express TSP-1. TSP-1 induced by endothelial cells converts latent TGF-β1 into its active form. Active TGF-β1 suppresses cell-cycle progression in hepatocytes at the G1/S checkpoint. In contrast, TSP-1 deficiency decreases active TGF-β1 levels, in turn results in the acceleration of liver regeneration.

Our study supports the notion that oxidative stress is one factor responsible of TSP-1 induction in the regenerating liver. TSP-1 is the most likely candidate protein induced by oxidative stress in proteomic analysis using brain endothelial cells (27). These findings imply that endothelial cells initially sense locally produced ROS in response to tissue damage and that the subsequent induction of TSP-1 in these cells afterwards initiates tissue remodeling. Indeed, our results revealed that endothelial cell-derived TSP-1 can modulate TGF-β/Smad signaling and proliferation in hepatocytes. Endothelial cells represent the largest population of nonparenchymal cells in the liver. Identification of the functional role of immediate early genes provides the clues for understanding the molecular bases of liver regeneration. One recent study documented that Id-1, a vascular endothelial growth factor-A receptor (VEGFR)-2-mediated transcriptional factor, was induced in endothelial cells at ~48h after hepatectomy; Id-1, in turn, promoted hepatocyte proliferation (28). There has as yet been no report implicating endothelial cells in earlier stages of the regenerating liver (within 24h). We have identified TSP-1 as a novel immediate early gene derived from endothelial cells, showing that the expression level of TSP-1 was immediately upregulated and returned to basal levels by 24h in response to hepatectomy. Our findings and the previous report (28) suggest that endothelial cells may play two distinct roles in hepatocyte proliferation following hepatectomy: one is an anti-proliferative role by activating the TSP-1/TGF-β1 axis within 24h, and the other is a pro-proliferative role by activating VEGFR-2 after 24h. This finding is consistent with the evidence that TSP-1 inhibits activation of VEGFR-2 via its receptor CD47 in endothelial cells (23) and suggests that the reduction of TSP-1 expression may be required for the functional shift in endothelial cells from anti- to pro-proliferative role in hepatocytes. Microvasvular rearrangement is important for tissue remodeling, and the anti-angiogenic action is one of the well recognized functions of TSP-1 (29). However, the expression of CD31 mRNA for monitoring angiogenesis did not show any significant difference between wild-type and TSP-1-null mice at 24, 48, and 72 h after hepatectomy (Hayashi H, and Sakai T. Unpublished data), suggesting that TSP-1 does not affect the vascularization during liver regeneration after hepatectomy.

TGF-β1 is known to be a potent inhibitor of mitogen-stimulated DNA synthesis in cultured hepatocytes (3). p21 is important for inhibiting hepatocyte proliferation in vivo, especially at the G1/S transition of the cell cycle (20), and the expression of p21 is upregulated by TGF-β1 (30). There is evidence that TGF-β1 mRNA induction occurs within 4h and remains elevated until 72h after hepatectomy (5–6). In contrast, we found the only limited activation of TGF-β signaling in an earlier phase (within 24h), with a peak at ~12h. It is known that TGF-β is secreted as latent forms and they are converted into active TGF-β in response to injury. There are several mechanisms for activation, such as via proteases, integrins (αvβ6, αvβ8), and TSP-1, all of which are likely to be tissue specific (31). While the complete lack of TGF-β-mediated signal in hepatocyte-specific TGF-β type II receptor knockout mice accelerates hepatocyte proliferation in the later phase (~36–48h) after hepatectomy (7), the role of TGF-β signaling in the earlier phase (within 24h) remains to be elucidated. Our present findings provide compelling evidence that locally activated TGF-β1 mediated by TSP-1 as an immediate early gene is critical in the early phase (within 24h) post hepatectomy to initiate the inhibitory effect on hepatocyte proliferation, and this TGF-β signaling has a functional link to the G1/S-phase transition by modulating p21 protein expression. A major downstream target of TGF-β1, PAI-1 (21), is a negative regulator of liver regeneration, and PAI-1-null mice show acceleration of liver regeneration after Fas-mediated massive hepatocyte death (32). The significant downregulation of PAI-1 expression in our TSP-1-null liver may be implicated in the accelerated hepatocyte proliferation after hepatectomy. However, our TSP-1-null model did not show any obvious differences in the termination phase of live regeneration compared with controls like as TGF-β type II receptor knockout mice model (7). Although the molecular mechanisms underlying the termination of liver regeneration remain to be elucidated (4), our and other findings suggest that the orchestrating interactions among positive- and negative-regulators in hepatocyte proliferation would be critical for the termination of liver regeneration (4, 24).

Active TGF-β1 induces hepatocyte cell death. STAT3 and PI3K/Akt signaling pathways are crucial for cell survival (anti-apoptosis) in the acute phase after partial hepatectomy. Our signaling data using TSP-1-null mice are consistent with previous findings showing that STAT3 and PI3K/Akt signaling pathways, but not the Erk1/2 pathway, play a protective role against TGF-β-induced apoptosis in hepatocyte cell lines (33–34). Several in vitro studies have reported that TSP-1 downregulates phosphorylated Akt expression in retina (35) and endothelial cells (23). Another in vitro study showed that the lack of TSP-1 in retinal endothelial cells results in upregulation of phosphorylated Akt expression, but not phosphorylated Erk1/2 (36). Since TSP-1 is a multidomain and multifunctional matricellular protein, our data and these findings suggest that TSP-1 modulates not only TGF-β signal, but also cell survival signals such as STAT3 and PI3K/Akt signals via its multidomain.

In the clinical setting, no established therapeutic strategies to accelerate liver regeneration have been available up to now. The inhibition of TSP-1 function attenuates locally activated TGF-β1 signals and thereby accelerates hepatocyte proliferation; hence TSP-1 could be a novel therapeutic target for accelerating liver regeneration after partial hepatectomy.

List of Abbreviations

α-SMA

α-smooth muscle actin

BrdU

5-bromo-2-deoxyuridine

DAPI

4′, 6-diamidino-2-phenylindole

H₂O₂

hydrogen peroxide

HSC

hepatic stellate cell

HUVEC

human umbilical vein endothelial cell

MDA

malondialdehyde

NAC

N-acetylcysteine

PAI-1

plasminogen activator inhibitor-1

ROS

reactive oxygen species

TSP-1

thrombospondin-1

TGF-β1

transforming growth factor-β1

VEGFR

vascular endothelial growth factor-A receptor