Metalloproteinase inhibitor TIMP-1 affects hepatocyte cell cycle via HGF activation in murine liver regeneration


  • Conflict of interest: Nothing to report.


Liver regeneration depends on timely restoration of cellular mass while orchestrating structural matrix remodeling. Matrix metalloproteinases (MMPs) and their endogenous inhibitors (TIMPs) are known to regulate the extracellular matrix (ECM) turnover and, more recently, the processing of growth factors and cytokines. We have previously demonstrated that TIMP-1 inhibits preneoplastic hepatocyte proliferation by attenuating growth factor bioavailability. In the present study, we examined the role of TIMP-1 in de novo hepatocyte cell division during liver regeneration. Comprehensive real-time reverse-transcriptase polymerase chain reaction analyses of regenerating livers revealed significant inductions in the messenger RNA of TIMP-1, TIMP-3, TIMP-4, MMP-2, MMP-9, MMP-13, MMP-14, and MMP-24, while MMP-15 expression was significantly reduced. Induction of TIMP-1 occurred during the peak of hepatocyte DNA synthesis. Studies using genetically altered mice revealed that TIMP-1 loss of function accelerated hepatocyte cell cycle progression. This finding was demonstrated by earlier expression of cyclin D1, proliferating cell nuclear antigen, and phosphorylated histone H3, which mark the G1-S, S, and M phase, respectively. Conversely, TIMP-1 gain of function delayed cell cycle progression. MMP activity was increased in the absence of Timp-1. Examination of hepatocyte growth factor (HGF), and its receptor Met, both of which provide a mitogenic signal for hepatocyte division, showed increased HGF activity in Timp-1−/−–regenerating livers. HGF is released from the ECM and is proteolytically processed to its active form. Active HGF was elevated in Timp-1−/− mice, leading to increased immunostaining of phosphorylated Met as well as activation of a downstream effector, p38. In conclusion, TIMP-1 is a novel negative regulator of HGF activity during liver regeneration. (HEPATOLOGY 2005.)

Understanding the molecular mechanisms of liver regeneration will benefit human therapies for acute hepatic failure and liver transplantation.1 The liver has a remarkable capacity to regulate its growth and mass and to regenerate upon loss or damage. Differentiated, mature hepatocytes re-enter the cell cycle but stop proliferating upon restoration of liver mass.2, 3 The extracellular matrix (ECM) remodels along with the restoration of cellular mass.4 Cell proliferation is induced in two stages following the two-thirds partial hepatectomy (PH), a classical means of inducing liver regeneration: one is the priming of hepatocytes, in which cytokines trigger G0 to G1 transition; the other is the progression from the G1 phase to the S phase via growth factors. Both of these molecular events are necessary for hepatocyte DNA synthesis and recovery of cellular mass.2, 3 The rapid response to liver injury requires immediate release of multiple factors independent of de novo protein synthesis; a current hypothesis is that pericellular proteolytic events participate in regulating hepatocyte division and reconstitution of cellular mass.2 Metalloproteinases are now recognized for controlling both ECM remodeling and growth factor bioactivity,5–7 yet little is known of their function in organ regeneration.

Hepatocyte priming is initiated by tumor necrosis factor α (TNF-α), which induces nuclear translocation of nuclear factor kappa B8 to trigger interleukin 6 production with subsequent Stat-3 activation.9 Mice deficient in TNF receptor 1 and interleukin 6 have increased mortality following PH.10–12 TNF-α release from the cell surface is mediated by the metalloproteinase called TNF-α–converting enzyme.13, 14 Tissue inhibitor of metalloproteinase 3 (TIMP-3) is the only known inhibitor of TNF-α–converting enzyme.15 We recently reported that TIMP-3 is essential for successful liver regeneration.16 In the absence of TIMP-3, increased hepatic TNF-α accelerates hepatocyte cell cycle progression after PH. However, these mice subsequently succumb to liver failure due to widespread TNF-α-mediated hepatic cell death. Thus, metalloproteinases and their endogenous inhibitors provide important points of control during liver regeneration.

For progression from the G1 phase to the S phase, hepatocyte growth factor (HGF), via signaling through its receptor Met, is considered the central stimulus.2, 3 Mice with deletions of either gene die in utero and have defects in hepatic development.17, 18 HGF is also a potent mitogen for hepatocytes in culture and induces the expression of immediate early genes seen during liver regeneration.19 After PH, plasma HGF increases within 1 hour in rats,20 but its messenger RNA is not synthesized until 3 to 6 hours, suggesting an alternate mechanism exists for the initial rise in plasma HGF.21 HGF is produced as a single polypeptide bound to hepatic ECM and must be released and activated by cleavage into a two-chain peptide.22–24 ECM-bound HGF may serve as a store of readily available growth factor that can be released upon matrix remodeling.25–27 The major ECM cleaving enzymes, matrix metalloproteinases (MMPs), are transcriptionally induced during rat liver regeneration. Gelatinases (MMP-2 and MMP-9) are activated within 6 to 12 hours,28–30 during the window of growth factor stimulus.2 In addition, the peak induction of metalloproteinase inhibitor TIMP-1 is reported to coincide with the timing of hepatocyte DNA synthesis.28, 29 We investigated whether gain or loss of TIMP-1 function affects hepatocellular proliferation and growth factor bioavailability. Our data reveal that genetic alterations in TIMP-1 lead to shifts in the timing of hepatocyte cell cycle progression by impacting HGF bioactivity.


MMP, matrix metalloproteinase; TIMP, tissue inhibitor of metalloproteinase; ECM, extracellular matrix; HGF, hepatocyte growth factor; PH, partial hepatectomy; TNF-α, tumor necrosis factor α; RT-PCR, reverse-transcriptase polymerase chain reaction; PCNA, proliferating cell nuclear antigen; HRP, horseradish peroxidase; pH3, phosphohistone H3; uPA, urokinase plasminogen activator.

Materials and Methods


Sense and antisense TIMP-1 transgenic (C57BL/6) and Timp-1−/− mice have been described.31, 32 Male mice (ages 9-11 weeks) and their age- and sex-matched wild-type littermates were used. Mice were anesthetized with isofluorane, and a 70% hepatectomy was performed.33 Surgeries were performed consecutively on experimental and control pairs before 11 A.M. Mice were sacrificed; liver tissue was then collected and was either flash-frozen for biochemical analysis or fixed in 4% formaldehyde and paraffin-embedded for histology. Animal care was provided according to the guidelines of the Canadian Council for Animal Care.

Quantitative TaqMan Real-Time Reverse-Transcriptase Polymerase Chain Reaction.

Total RNA was isolated by homogenization of frozen tissue as previously described.31 Quantitative reverse-transcriptase polymerase chain reaction (RT-PCR) was performed as previously described34 using the ABI prism 7700 sequence-detection system (TaqMan, Applied Biosystems, Mississuaga, Canada). Briefly, oligonucleotide primers and fluorescently labeled TaqMan probes were designed using Primer Express 1.0 software (Applied Biosystems). To control against amplification of genomic DNA and to ensure that the PCR signal was generated from complementary DNA, primers were placed within different exons, close to intron–exon boundaries. BLASTN (NCBI, Bethesda, MD) searches were conducted on all primer–probe nucleotide sequences to ensure gene specificity, and the identity of PCR products was confirmed via direct sequencing of amplicons. GAPDH RNA was used as an endogenous control to normalize for differences in the amount of total RNA in each sample. The ABI Prism 7700 measured the cycle–cycle changes in fluorescence in each sample and generated a kinetic profile of DNA amplification over the 40-cycle PCR reaction. The cycle number (CT) at which amplification entered the exponential phase was determined and used as an indicator of the amount of target RNA. Standard curves for CT versus input RNA were prepared. All quantitative real-time RT-PCR data presented in the results are shown as relative values, where the maximal value of each gene was taken as 1.0. Sense-specific RT-PCR for TIMP-1 has been previously described.35

Gelatin Zymography.

Frozen liver tissue was homogenized in buffer (50 mmol/L Tris HCl (pH 7.6), 150 mmol/L NaCl, 0.1% SDS, 1% Deoxycholate Na, 1% Nonident P40, 1% Triton-X 100), and the protein concentration was determined using the Dc Biorad assay (Bio-Rad Laboratories, Hercules, CA). Triplicate samples at each time point were pooled (40 μg total) and resolved on 10% SDS-PAGE gel containing 0.3% gelatin. Gels were placed in 2.5% Triton-X 100 for 30 minutes; incubated for 24 to 48 hours in buffer containing 50 mmol/L Tris [pH 7.5], 5 mmol/L CaCl2, and 40 mmol/L NaN3; and stained with Coomassie blue R-250.

Western Blot Analysis.

Frozen tissue was homogenized in buffer (1% Triton-X 100, 500 mmol/L Tris-HCl [pH 7.6], 200 mmol/L NaCl, 10 mmol/L CaCl2, orthovanadate, PMSF, pepstatin A, leupeptin, aprotinin) for all Western blot analyses—except for HGF, for which 1% SDS replaced 1% Triton-X 100. Protein concentration was determined, and 40 μg of protein was separated via SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were blocked with 5% milk in TBST (20 mmol/L Tris [pH 7.5], 500 mmol/L NaCl, 0.05% Tween-20) and probed using primary antibodies against cyclin D1 (Oncogene, Cambridge, MA), proliferating cell nuclear antigen (PCNA) (Novacastra, Newcastle upon Tyne, UK), or HGF, followed by mouse secondary horseradish peroxidase (HRP) (Cell Signaling, Beverly, MA) or rabbit secondary-HRP (Cell Signaling). Electrochemiluminescence-developing reagents were obtained from Pierce (Rockford, IL). Signals were quantified using an ImageQuant program (Molecular Dynamics, Piscataway, NJ). Membranes were subsequently stripped and reprobed with mouse anti–α-tubulin antibody (Oncogene) followed by mouse secondary-HRP (Cell Signaling) antibody for loading control.

Immunohistochemical Analysis.

Five-micrometer liver sections were placed on Superfrost/Plus microscope slides. Tissues were deparaffinized in toluene, rehydrated in water/ethanol, and stained with hemotoxylin-eosin. Antibodies used were against phosphorylated Histone H3 (Upstate Biotech, Lake Placid, NY), phosphorylated Met (Cell Signaling) or Met (Santa Cruz Biotechnology, Santa Cruz, CA). Histomorphometry was performed using blind counts of five random microscopic fields ata magnification of 40×.

HGF Enzyme-Linked Immunosorbent Assay.

A 200-μg protein sample was analyzed in triplicate using an anti-rat HGF enzyme-linked immunosorbent assay kit (Institute of Immunology, Tokyo, Japan). Standard curves were generated according to the manufacturer's instructions, and protein loading was optimized to generate readings within the linear range of the standard curve.

Statistical Analysis.

Data are represented as the mean ± SEM. Statistical significance was determined using the Student t test and ANOVA.


Transcriptional Induction of TIMPs Before the Peak of Hepatocyte DNA Synthesis.

PH triggers mitotically inactive hepatocytes to exit the G0 state and synchronously enter the cell cycle. Hepatocytes undergo one to two divisions, with the peak of DNA synthesis occurring at approximately 48 hours in mice,3 resulting in restoration of hepatocyte cell number within 96 hours.2 This is followed by re-establishment of structural architecture, including new synthesis of ECM proteins with the completion of the regenerative process 8 to 10 days after PH.3 First, we characterized the expression of TIMPs and several MMPs via real-time TaqMan RT-PCR to determine their temporal relation to DNA synthesis (Fig. 1). Over the proliferative period—96 hours—TIMP-3 was significantly induced within 6 hours after PH, and TIMP-1 and TIMP-4 were significantly induced by 48 hours (Fig. 1A–D). Of the MMPs examined (MMP-2, -3, -9, -13, -14, -15, -16, -17, and -24), MMP-2 was induced at 48 hours, MMP-9 at 1 hour and then again at 48 hours, MMP-13 at 6 hours, MMP-14 at 48 hours, and MMP-24 at 48 hours. MMP-15 showed a significant decline at 24 hours (Fig. 1E–M). The above genes that showed significant alterations as a function of liver regeneration were then assessed in livers of sham-operated mice. TIMP-1 and -3 and MMP-9, -13, -14, and -24 did not change significantly up to 72 hours after sham surgery, whereas MMP-2 was significantly elevated 48 hours after sham surgery (data not shown). MMPs are produced as latent enzymes, and gelatinase (MMP-2, MMP-9) activity can be visualized using gelatin zymography. Pro–MMP-9 levels increased within 1 hour, and its active form was visible by 6 hours, whereas MMP-2 remained mainly in its “pro” form with little increase (Fig. 1N). Overall, PH led to significant changes in TIMP and MMP expression during the phase of hepatocyte proliferation.

Figure 1.

Expression profiling of MMPs and TIMPs in regenerating liver. Real-time TaqMan RT-PCR analysis of RNA from livers before (0 h) and after (1-96 h) PH for the indicated TIMP (A–D) and MMP (E-M) genes. Each bar represents the mean ± SEM of 4 to 6 mice. (N) Gelatin zymography of pooled liver protein homogenates (n = 3 mice per time point). *P < .05. TIMP, tissue inhibitor of metalloproteinase; MMP, matrix metalloproteinase; PH, partial hepatectomy.

Genetic Modulation of TIMP-1 Alters Progression of Hepatocyte Cell Cycle.

To test whether TIMP-1 can effect hepatocyte cell cycle during liver regeneration, we used TIMP-1 transgenic and TIMP-1 deficient mice. Our previously described TIMP-1 sense (overexpressing) and antisense (underexpressing) transgenic mice31 provided a means of testing the effects of gain of function and loss of function, respectively, with TIMP-1 null mice complementing the loss of function studies. The liver is the primary organ altered in these TIMP-1 transgenic mice. To show this directly, we quantified hepatic TIMP-1 RNA levels. Overexpressing transgenic mice had significantly elevated (2-8fold) TIMP-1 in both resting (0 hours) and regenerating livers (1-96 hours) via TaqMan RT-PCR (Fig. 2A). For the underexpressing TIMP-1 transgenics, we used a sense-strand-specific RT-PCR strategy (Fig. 2B), which revealed a greater than 90% reduction of hepatic TIMP-1 transcript (Fig. 2B and 2C).

Figure 2.

Altered hepatic TIMP-1 expression in transgenic mice. (A) Real-time TaqMan RT-PCR of hepatic TIMP-1 RNA in TIMP-1 overexpressing mice and their wild-type littermates. (B) Semiquantitative sense-specific RT-PCR on pooled RNA from livers of TIMP- underexpressing mice and wild-type littermates. β-Actin served as RNA loading control. (C) Densitometric quantification of TIMP-1 levels in panel B following normalization against β-actin (n = 3 mice per time point). *P < .05 WT, wild-type; TIMP-1 Over, tissue inhibitor of metalloproteinase 1 overexpressing; PH, partial hepatectomy; TIMP-1 Under, tissue inhibitor of metalloproteinase 1 underexpressing.

To analyze the hepatocyte cell cycle in TIMP-1 transgenic mice, we determined the temporal expression of cell cycle markers during the proliferative phase of liver regeneration. Cyclin D1, PCNA, and phosphohistone H3 (pH3) are established markers of the G1-S, S, and M phases, respectively. To minimize any variations arising from the metabolic state, PH was performed in the morning, concurrently on age- and sex-matched littermates, and liver tissue was collected at 24-hour intervals. Western blot analyses of three independent experiments, each encompassing 5 transgenic and 5 wild-type littermates, showed a gradual increase in cyclin D1 levels, which—when quantified and normalized against tubulin—revealed a peak at 48 hours in wild-type mice (Fig. 3A–B,E–F). Although we observed some biological variation, the peak of PCNA in wild-type mice was typically observed at 72 hours (Fig. 3C–D,G–H). We found that the timing of G1-S transition, as well as the S phase, was accelerated in TIMP-1 underexpressing mice as indicated by earlier peaks of cyclin D1 at 24 hours and PCNA at 48 hours (see Fig. 3A,C,E,G). Inversely, maximal expression of these markers was delayed in TIMP-1 overexpressing mice by at least 24 hours (see Fig. 3B,D,F,H). These kinetics indicated earlier cell cycle progression in TIMP-1 compromised livers and a delay of this event with TIMP-1 overexpression.

Figure 3.

TIMP-1 alters cell cycle progression. (A,B) Western blot showing the time course of cyclin D1 in livers of TIMP-1underexpressing and overexpressing transgenic mice compared with their respective wild-type littermates. (C,D) Western blot showing the time course of PCNA in livers of TIMP-1 transgenic and wild-type littermates. Tubulin served as a protein-loading control. The figure shows densitometric quantification of (E,H) cylcin D1 and (F,G) PCNA, normalized to tubulin levels, from three independent sets of time course experiments comparing TIMP-1 transgenic mice and their respective controls. *P < .05. WT, wild-type; TIMP-1 Under, tissue inhibitor of metalloproteinase 1 underexpressing; TIMP-1 Over, tissue inhibitor of metalloproteinase 1 overexpressing; PCNA, proliferating cell nuclear antigen; PH, partial hepatectomy.

We reasoned that altered cell cycle progression would be reflected in the timing of mitosis. We examined hepatocyte cell division by monitoring histone pH3, which becomes phosphorylated during chromatin condensation. Immunohistochemical detection of pH3 revealed punctate nuclear chromatin staining and mitotic chromosomes (Fig. 4A). Histomorphometry indicated maximal pH3 positivity at 72 hours in wild-type littermates (Fig. 4B–C). The peak of pH3 was earlier in TIMP-1 underexpressing mice (see Fig. 4B), and delayed in overexpressing mice (see Fig. 4C), suggesting that altered timing of cell cycle progression in TIMP-1 transgenics culminated in altered hepatocyte replication.

Figure 4.

TIMP-1 expression shifts the timing of hepatocyte mitosis. (A) Immunohistochemistry against pH3 on liver sections of wild-type and TIMP-1 transgenic mice. Bar = 40 μm. (B,C) Histomorphometry of pH3-positive nuclei from three independent time course experiments. A total of 500 nuclei were scored per section at a magnification of 40×. *P < .05. TIMP-1 Under, tissue inhibitor of metalloproteinase 1 underexpressing; WT, wild-type; TIMP-1 Over, tissue inhibitor of metalloproteinase 1 overexpressing; PH, partial hepatectomy.

Given the availability of TIMP-1 null mice, we next confirmed our observations associated with TIMP-1 loss of function. TIMP-1 null mice were backcrossed into C57BL/6 strain more than six times, because our TIMP-1 transgenics are of this background. At 24 hours after PH, both cyclin D1 and PCNA protein levels were far greater in Timp-1−/− than in wild-type livers (Fig. 5A–B). Altogether, these data suggest that genetic modulation of TIMP-1 directly affects the kinetics of hepatocyte cell cycle progression and cell division during the process of liver regeneration. We then determined if there were alterations in MMP activity in Timp-1−/− mice by incubating liver homogenates with a fluorescently labeled gelatin peptide substrate. MMP activity increased between 1 and 6 hours after PH, declining thereafter in wild-type livers. In contrast, basal MMP activity was significantly increased in Timp-1−/− livers and failed to decline at 24 hours (Fig. 5C).

Figure 5.

TIMP-1 null mice have accelerated cell cycle progression and increased MMP activity. The figure shows densitometric quantification of (A) cyclin D1 and (B) PCNA following Western blotting of TIMP-1 null liver homogenates (n = 3 mice per time point, per genotype). (C) Cleavage of a fluorescent gelatin peptide substrate by liver extracts from wild-type and Timp-1−/− mice. *P < .05. WT, wild-type; PCNA, proliferating cell nuclear antigen; PH, partial hepatectomy.

TIMP-1 Expression Affects Hepatic HGF Processing but Not Its Transcription.

A possible mechanism for TIMP-1-mediated alteration of hepatocyte cell cycle progression is through metalloproteinase-dependent alterations in growth factor bioactivity. HGF is a major stimulus for hepatocyte G1-S transition.2 After PH, this ECM-bound growth factor becomes available before its new synthesis occurs.21 To test whether TIMP-1 affects HGF bioactivity, we monitored HGF levels, processing, and production. Enzyme-linked immunosorbent assay measurements of liver homogenates showed significant increases in HGF with decreasing TIMP-1 expression in both TIMP-1 underexpressing transgenic and Timp-1−/− mice (Fig. 6A). Strikingly, in Timp-1−/− mice, HGF levels were found elevated even at 0 hours, which represented resting livers before hepatectomy (see Fig. 6A). HGF levels in the plasma, kidney, spleen, and lung did not differ between wild-type and TIMP-1overexpressing and underexpressing mice (data not shown).

Figure 6.

Increased mature HGF levels with loss of TIMP-1 function. (A) HGF levels by enzyme-linked immunosorbent assay in liver homogenates, before (0 h) and after (1 and 6 h) PH. (B) Western blot of mature HGF represented by light-chain HGF. (C) Western blot of inactive, single-chain HGF and mature light-chain HGF in resting livers from multiple Timp-1−/− and control mice. The figure shows densitometric quantification of (D) light-chain HGF in panel B normalized against protein levels and of (E) light-chain HGF in panel C normalized against single-chain HGF and protein levels. (F) Real-time TaqMan RT-PCR of HGF RNA in wild-type and TIMP-1 underexpressing livers (n = 3 mice per time point, per genotype). *P < .05. WT, wild-type; TIMP-1 Under, tissue inhibitor of metalloproteinase 1 underexpressing; HGF, hepatocyte growth factor; PH, partial hepatectomy; TIMP-1 Over, tissue inhibitor of metalloproteinase 1 overexpressing; lc-HGF, light-chain HGF; sc-HGF, single-chain HGF.

HGF exists in its inactive form as a single-chain peptide bound to the ECM. HGF activation requires its release from the ECM and cleavage into a two-chain form composed of a light chain and heavy chain. We next determined the presence of the active HGF species following PH via Western blotting. Light-chain HGF first increased at 1 hour after PH in wild-type livers (Fig. 6B). The level of active mature HGF form was significantly diminished in TIMP-1 overexpressing transgenic mice, and inversely, significantly elevated in Timp-1−/− mice (Fig. 6B–C). Furthermore, processed HGF was significantly increased in resting Timp-1−/− livers (Fig. 6D-E). Next, we determined whether these changes could arise from altered HGF transcription. It is known that in rats HGF RNA increases approximately 3 to 6 hours after PH with hepatocyte DNA synthesis occurring at 24 hours. Previous studies have shown that in comparison to rats, mice have an approximately 24-hour delay in the kinetics of hepatocyte replication.3 Our TaqMan RT-PCR analyses revealed an increase in HGF RNA by 24 hours after PH in wild-type livers (Fig. 6F). HGF RNA levels did not significantly deviate from wild-type levels in TIMP-1underexpressing (see Fig. 6F) or overexpressing (data not shown) transgenic mice. Therefore, HGF transcription was not responsible for increases in this growth factor. Altogether, TIMP-1 expression inversely correlated with active HGF levels in regenerating livers, and its effect occurred posttranscriptionally. This suggests that changes in hepatocyte cell cycle progression in TIMP-1 transgenic and knockout mice may arise from alterations in the HGF mitogenic signal preceding the proliferative phase.

Increased HGF Signaling Underlies Accelerated Progression of Cell Cycle With TIMP-1 Loss of Function.

HGF signals by binding its receptor, a tyrosine kinase encoded by the c-MET proto-oncogene, and inducing its phosphorylation and downstream signaling.36 To test whether increased HGF processing led to increased mitogenic signaling, we sought to determine the activation of Met. Even though we used all of the commercially available antibodies, we could not detect Met in liver tissue homogenates, either via immunoprecipitation or phosphorylation-specific immunoblotting. However, we were successful in detecting Met via immunohistochemistry. In accordance with increased light-chain HGF in the resting Timp-1−/− liver, phosphorylated Met showed an apparent distribution along the hepatocyte cell surface (Fig. 7B). This result was not observed in resting livers of wild-type controls (Fig. 7A). Immunohistochemical detection of nonphosphorylated Met confirmed that both wild-type and Timp-1−/− resting livers expressed this receptor (Fig. 7C–D). In contrast to Timp-1−/− mice, Met phosphorylation was only observed in regenerating wild-type livers starting at 1 hour after PH (Fig. 7E-G). To exclude the possible contribution of c-MET transcriptional changes toward altered HGF signaling, we measured Met RNA levels by quantitative TaqMan RT-PCR during liver regeneration. No differences were observed in Met RNA levels in TIMP-1underexpressing (Fig. 7H) or overexpressing (data not shown) mice. HGF has been shown to induce cyclin D1 expression through p38 activation, a downstream effector in the MAPK pathway.37 To test for activation of this mitogenic pathway with TIMP-1 loss of function, we determined p38 phosphorylation spanning the time course when HGF was found altered. At 1 hour after PH, p38 was phosphorylated in wild-type livers (Fig. 7I) and was increased in TIMP-1 underexpressing livers (Fig. 7I). Timp-1−/− livers displayed a marked increase in p38 phosphorylation even in resting livers before hepatectomy (see Fig. 7I); this was consistently observed in the resting livers of multiple mice (Fig. 7J). These data provide direct evidence for increased HGF bioactivity as a consequence of TIMP-1 loss.

Figure 7.

Altered HGF signaling with loss of TIMP-1 function. (A,B) Basal activation of Met receptor in livers of WT and Timp-1−/− mice prior to hepatectomy detected by immunohistochemistry for phosphorylated Met (p-Met). (C,D) Immunohistochemistry of non-phosphorylated Met in serial sections. (E-G) Time course of Met activation during liver regeneration in WT mice. (A–G) Bars = 40 μm. (A–D) Inset bars = 0.1 mm. (H) Real-time TaqMan RT-PCR of Met RNA in wild-type and TIMP-1 underexpressing livers. (I) Western blot of p38 phosphorylation (p-p38) in regenerating livers. Nonphosphorylated p38 was used as a control for total p38 levels. (J) Activation of p38 in Timp-1−/− livers before hepatectomy detected via Western blotting of p-p38 compared with wild-type livers (n = 3 mice per time point, per genotype). WT, wild-type; TIMP-1 Under, tissue inhibitor of metalloproteinase 1 underexpressing; PH, partial hepatectomy.


Rapid response to liver injury depends on the immediate release of multiple factors before new synthesis can occur.2 Given the newly recognized functions of metalloproteinases in regulating growth factor bioactivity, it is currently speculated that pericellular proteolysis participates in regulating hepatocyte division.2 We have reported that TIMP-3 is essential for liver regeneration.16 In the present study, we tested whether TIMP-1 provides a point of control during liver regeneration. We show that genetic alterations of TIMP-1 directly impact hepatocyte cell cycle progression by altering HGF activation and its downstream signaling. Thus, TIMP-1 is a novel negative regulator of HGF activity during liver regeneration.

To date, detailed TIMP and MMP expression profiling has not been reported after PH in the mouse. Our comprehensive elucidation of all TIMPs and several MMPs identified that three of four TIMPs were transcriptionally induced. TIMP-3 was induced early (6 hours) coinciding with the timing of TNF-α reduction, consistent with its role in regulating TNF-α during liver regeneration. TIMP-1 and TIMP-4 were induced at 48 hours. Previous studies in the rat have alluded to a function of TIMP-1 in hepatocyte proliferation with TIMP-1 induction occurring just before hepatocyte DNA synthesis at 24 hours.29, 30 Although there are considerable interspecies variations in liver regeneration, it is interesting to note that the timing of TIMP-1 expression in the mouse at 48 hours after PH (24 h later than in rats) is consistent with the observation that mice have a longer G1 phase.3 This finding suggests that TIMP-1 expression may be linked to the hepatocyte cell cycle, which we examined using TIMP-1 mutants. Gain of TIMP-1 function delayed cell cycle progression after PH, whereas reduction or loss of TIMP-1 accelerated it. Genetic manipulation of TIMP-1 impacted the three phases of the cell cycle. The role of TIMP-1 in proliferation has been debated because TIMP-1 can have both mitogenic as well as inhibitory effects.38–41 In the current study, TIMP-1 overexpression clearly exerted an inhibitory effect on de novo hepatocyte proliferation. Thus the timing of endogenous TIMP-1 induction may reflect its role in downregulating extracellular mitogenic stimuli after hepatocytes have undergone the necessary division.

Among the growth factors involved in liver regeneration, HGF is important for hepatocyte proliferation. Transgenic HGF-overexpressing mice have an increased liver-to-body mass, have a higher DNA labeling index, and regenerate their livers in half the time,42, 43 whereas liver-specific disruption of Met prevents successful liver regeneration.44, 45 Pro-HGF bound to the ECM is released and activated via proteolytic cleavage.22–27 HGF is classically known to bind the extracellular molecule heparan sulfate. We did not find increased expression of the stromelysin MMP-3, which cleaves this proteoglycan. However, HGF is also reported to bind collagens within the liver ECM.46 Our results indicated that expression of MMP-2, -9, -13, and -14 was induced after PH, and these MMPs are known to process collagens.5 Michalopoulos' group has shown that HGF activation in regenerating rat liver occurs in two phases, a consumptive phase and a productive phase.47 Initially, pro-HGF already present in the liver is activated and used, followed by synthesis of new HGF. We observed that HGF messenger RNA was induced at 24 hours after PH, and this production remained unaltered by reduction or increase of TIMP-1. During the consumptive phase, Timp-1−/− mice had significantly increased MMP activity, mature HGF levels, and signaling as determined by immunostaining of phosphorylated Met and p38.

Liberation of growth factors by proteolysis is an important regulatory mechanism for controlling cell proliferation, because MMPs have been linked to the release of specific molecules from the ECM, cell surface, or binding proteins.7 We have previously shown that TIMP-1 overexpression in the liver inhibits neoplastic hepatocyte proliferation by decreasing MMP-mediated release of insulin-like growth factor II from its binding proteins.39, 48 During liver regeneration, several molecules are possible candidates for MMP-mediated processing, including epidermal growth factor, transforming growth factor β, and fibroblast growth factor.7 Given that organ regeneration is an intricate and highly complex process, it is conceivable that factors in addition to HGF contribute to the altered cell division resulting from genetic modulation of TIMP-1. In addition, although the use of genetically altered mice is a widely accepted means of deciphering gene function in vivo, it remains possible that compensatory mechanisms activated during development can influence observations in adult tissues.

In the tissue microenvironment, proteases of distinct classes function within a larger proteolytic cascade to provide initiation signals in the regenerating liver. Liver regeneration following fas-mediated hepatocyte apoptosis is accelerated in serine protease inhibitor plasminogen activator inhibitor 1−/− in mice.49 Inversely, urokinase plasminogen activator−/− (uPA−/−) mice have transiently impaired liver regeneration following this insult.49 Mature HGF production is delayed by 12 hours along with delayed DNA synthesis in these mice,49 and uPA is known to process pro-HGF into its active two-chain form.22 Other serine proteinases such as HGF activator24 and matriptase50 also activate HGF, and matriptase can additionally activate uPA.50 Urokinase PA is upstream of plasminogen activation into plasmin, and pro-MMPs are in part activated by plasmin.51 As proposed in our model (Fig. 8), we envision the TIMP/MMP proteolytic axis functions in conjunction with serine proteinases. For instance, lack of uPA will result in reduced plasmin and in turn reduce pro-MMP activation, leading to decreased release of ECM-bound HGF. Overall, proteolytic cleavage may facilitate two discreet steps: growth factor release from the ECM and maturation of the growth factor. A precedent for such processing exists for transforming growth factor β.52

Figure 8.

Proposed model of the proteolytic cascade in the regenerating liver leading to the liberation and activation of HGF resulting in hepatocyte cell division. ECM, extracellular matrix; HGF, hepatocyte growth factor; TIMP-1, tissue inhibitor of metalloproteinase 1; MMP, matrix metalloproteinase; uPA, urokinase plasminogen activator; MAPK, mitogen-activated protein kinase; PCNA, proliferating cell nuclear antigen.

In summary, timely progession of the cell cycle is vital for many physiological processes. The maintenance of cell cycle control is also important for preventing the genesis of hyperplasia and neoplastic cells. Checkpoints are in place to ensure that each phase during cell division is appropriately initiated. Signal transduction networks achieve precise timing of the cell cycle by regulating temporal gene expression and protein activity.53 We show that extracellular proteolysis governed by the TIMP/MMP proteolytic axis provides a novel point of control by working upstream of the HGF mitogenic signaling pathway during liver regeneration. The regeneration of this organ is linked to the development of several pathologies including cirrhosis, fibrosis, and hepatocellular carcinomas.1, 54 It will be important to review and investigate the role of TIMPs in these human diseases.