Potential conflict of interest: Nothing to report.
Liver fibrogenesis is associated with the transition of quiescent hepatocytes and hepatic stellate cells (HSCs) into the cell cycle. Exit from quiescence is controlled by E-type cyclins (cyclin E1 [CcnE1] and cyclin E2 [CcnE2]). Thus, the aim of the current study was to investigate the contribution of E-type cyclins for liver fibrosis in man and mice. Expression of CcnE1, but not of its homolog, CcnE2, was induced in fibrotic and cirrhotic livers from human patients with different etiologies and in murine wild-type (WT) livers after periodical administration of the profibrotic toxin, CCl4. To further evaluate the potential function of E-type cyclins for liver fibrogenesis, we repetitively treated constitutive CcnE1−/− and CcnE2−/− knock-out mice with CCl4 to induce liver fibrosis. Interestingly, CcnE1−/− mice were protected against CCl4-mediated liver fibrogenesis, as evidenced by reduced collagen type I α1 expression and the lack of septum formation. In contrast, CcnE2−/− mice showed accelerated fibrogenesis after CCl4 treatment. We isolated primary HSCs from WT, CcnE1−/−, and CcnE2−/− mice and analyzed their activation, proliferation, and survival in vitro. CcnE1 expression in WT HSCs was maximal when they started to proliferate, but decreased after the cells transdifferentiated into myofibroblasts. CcnE1−/− HSCs showed dramatically impaired survival, cell-cycle arrest, and strongly reduced expression of alpha smooth muscle actin, indicating deficient HSC activation. In contrast, CcnE2-deficient HSCs expressed an elevated level of CcnE1 and showed enhanced cell-cycle activity and proliferation, compared to WT cells. Conclusions: CcnE1 and CcnE2 have antagonistic roles in liver fibrosis. CcnE1 is indispensable for the activation, proliferation, and survival of HSCs and thus promotes the synthesis of extracellular matrix and liver fibrogenesis. (HEPATOLOGY 2012;56:1140–1149)
Liver fibrosis is a chronic wound-healing process leading to liver scarring by excessive accumulation of extracellular matrix (ECM) proteins, including collagen. The main producers of liver collagen are myofibroblasts derived from activated hepatic stellate cells (HSCs). Additionally, other cell types, such as portal fibroblasts and bone marrow derived cells, may contribute to ECM production. Liver fibrosis develops on the basis of chronic liver injury induced, for example, by chronic viral hepatitis B or C infection, excessive alcohol abuse, or fatty liver disease frequently associated with obesity.1 Although immune cells play an essential role in the modulation of liver fibrosis, its pathogenesis implicitly involves the injury and proliferation of HSC, hepatocytes, and, potentially, other cell species. Upon liver damage, dying hepatocytes stimulate remnant hepatocytes to reenter the cell cycle to restore original liver mass and function.2 Liver injury also stimulates HSC activation through complex mechanisms. This involves the conversion of a resting, vitamin A–storing cell into a proliferating HSC without vitamin A droplets, but is capable of producing proinflammatory cytokines and ECM components such as collagen.3
The transition from quiescent (i.e., G0) cells into the active phase of the cell cycle is predominantly controlled by E-type cyclins and their associated kinase, cyclin-dependent kinase 2 (Cdk2).4 In mammals, two E-cyclins are known, termed cyclin E1 (CcnE1) and cyclin E2 (CcnE2), respectively.5, 6 Despite their anticipated essential function for developmental and regenerative processes, the single genetic inactivation of CcnE1, CcnE2, or Cdk2 does not affect viability or development in mice.7–10 However, fibroblasts deficient for both E-cyclins are unable to reenter the cell cycle from G0.9 We recently demonstrated that CcnE1 and CcnE2 play antagonistic roles in the regenerating liver after partial hepatectomy (PH).11 Accordingly, CcnE2−/− livers show increased, prolonged CcnE1/Cdk2 activity, resulting in earlier and sustained DNA synthesis, hepatomegaly, and excessive endoreplication of hepatocytes, whereas the ablation of CcnE1 provoked only a moderate delay of hepatocyte proliferation. Earlier work using rat HSCs indicated that HSC activation is associated with increased gene expression of CcnE, cyclin D, and induction of polyploidy.12 However, the precise role of E-type cyclins for the activation and proliferation of HSCs, and subsequent liver fibrogenesis, has remained elusive.
In the present study, we aimed to investigate the contribution of E-type cyclins for liver fibrosis in vivo using constitutive CcnE1−/− and CcnE2−/− knock-out mice and derived primary HSCs. Our current work demonstrates that CcnE1, but not CcnE2, is essential for HSC survival, proliferation, and liver fibrogenesis.
Human liver samples were available from routine liver biopsies or from explanted cirrhotic livers resulting from transplantation, as described recently.13 The study protocol was approved by the local ethics committee (ethics committee of University Hospital Aachen, RWTH Aachen, Aachen, Germany), and the study was conducted according to the principles expressed in the Declaration of Helsinki.
Housing and Breeding of Mice.
Animal husbandry and procedures were approved by the authority for environment conservation and consumer protection of the state of North Rhine–Westfalia (LANUV, Germany) and the University Hospital Aachen Animal Care Facility's guidelines. For our study, we used mice of male gender with constitutive deletion of CcnE1 (CcnE1−/−) and CcnE2 (CcnE2−/−) and wild-type (WT) littermates from heterozygous breeding couples, as recently reported.9, 11
Isolation and Fluorescence Activated Cell-Sorting Analysis of HSCs.
HSCs were prepared from adult male mice (weighing approximately 25 g), according to the collagenase method,14 as described in the Supporting Materials.
Data are expressed as mean ± standard deviation of the mean. Statistical significance was determined by two-way analysis of variance, followed by a Student t test.
Increased Expression of the Cell-Cycle Mediator, CcnE1, in Human and Murine Liver Fibrosis.
E-type cyclins CcnE1 and CcnE2 control the transition of quiescent cells into the S phase and subsequent cell proliferation.4 We hypothesized that liver fibrogenesis involves the cell proliferation of hepatic cell populations and determines overall CcnE expression in liver biopsies from patients with liver fibrosis of different etiologies (see Supporting Table 1). CcnE1 mRNA expression was significantly up-regulated in patients with advanced hepatic fibrosis (F3) and liver cirrhosis (F4), compared to healthy control livers (F0) or patients with mild (F1) fibrosis (Fig. 1A). In contrast, CcnE2 was not aberrantly expressed in liver fibrosis at any stage (Fig. 1B). Immunostaining of liver biopsies confirmed the overexpression of CcnE1 in liver cirrhosis (Fig. 1C). Detailed analysis revealed the substantial nuclear expression of CcnE1 in nonparenchymal cells (NPCs), but also in hepatocytes of cirrhotic livers, which was not observed in healthy liver samples (Fig. 1D).
We next investigated the involvement of CcnE1 in experimental liver fibrosis in WT mice subjected to repetitive CCl4 injections. In agreement with the human samples, mRNA expression of CcnE1, but not of CcnE2, was induced in murine liver after CCl4 treatment and correlated with fibrosis progression (Fig. 2A). This was associated with increased protein expression of the fibrosis marker, alpha smooth muscle actin (α-SMA; expressed in activated HSCs), along with the cell-cycle mediator, proliferating cell nuclear antigen (PCNA) (Fig. 2B). Costaining of CcnE1 and α-SMA in fibrotic livers revealed an accumulation of CcnE1-expressing cells in areas of fiber formation (Fig. 2C). Using confocal laser scanning microscopy, we demonstrated nuclear CcnE1 localization predominantly in α-SMA-expressing cells (Fig. 2D). These data demonstrated that liver fibrogenesis in humans and mice involves increased cell-cycle activity, potentially driven by CcnE1, and suggested that CcnE1 is especially induced in HSCs during this process.
Liver Regeneration After CCl4-Mediated Acute Liver Injury Depends on CcnE1.
Single CCl4 administration triggers approximately 20-fold CcnE1 messenger RNA (mRNA) expression in WT mice within 48 hours (Fig. 2A). We recently reported that genetic ablation of CcnE2 results in the overexpression of CcnE1 and excessive hepatocyte proliferation after PH.11 In agreement with our earlier findings, CcnE1 mRNA induction was approximately 5-fold higher in CcnE2−/− mice, compared to WT animals, 48 hours after single CCl4 administration (Fig. 2A).
To evaluate the effect of CcnE1 for the onset of liver fibrosis, we compared the immediate proliferative response 48 hours after single CCl4 treatment in WT, CcnE1−/−, and CcnE2−/− mice by measuring bromodeoxyuridine (BrdU) incorporation (specific for DNA synthesis) and the general proliferation marker, cell-cycle–specific protein encoded by the MKI67 gene (Ki-67). Of note, CCl4 induced a similar level of necrotic liver injury in each group, as evidenced by the comparable induction of alanine aminotransferase (ALT) activity (Fig. 3B and Supporting Fig. 1A). WT and CcnE2−/− livers revealed a similar proliferative response with substantial DNA synthesis of hepatic cells, as evidenced by strong BrdU incorporation (Fig. 3C and Supporting Fig. 1B) and extensive Ki-67 expression. Under these conditions, CcnE1 was localized in the hepatocytes of WT and CcnE2−/− mice (Supporting Fig. 1C). However, elevated CcnE1 levels, as observed in the CcnE2−/− liver, did not result in enhanced hepatocyte proliferation after toxic injury. In contrast, CcnE1 deficiency resulted in a remarkable reduction of hepatocyte proliferation and DNA synthesis after acute CCl4 treatment (Fig. 3C,D). Thus, CcnE1 plays an important role for liver regeneration after CCl4-mediated toxic liver injury.
CcnE1 Is Essential for CCl4-Mediated Liver Fibrosis.
We next investigated the consequences of chronic CCl4 treatment in CcnE1−/− mice, in comparison to WT controls. Repeated injections of CCl4for 4 weeks induced prominent liver fibrosis with septum formation in WT animals, as evidenced by Sirius red staining. In contrast, fiber formation was barely observed in CcnE1−/− mice (Fig. 4A,B), suggesting a functional role of CcnE1. Additionally, WT mice revealed a substantial up-regulation of collagen type I α1 (COL1A1) and α-SMA mRNA expression 4 weeks after CCl4 treatment, which were only moderately induced in CcnE1−/− livers (Fig. 4C,D). The significant difference in α-SMA expression between WT and CcnE1−/− livers was further confirmed by western blotting analysis, suggesting that loss of CcnE1 protects against liver fibrosis in vivo (Fig. 4E).
Furthermore, PCNA expression was significantly expressed in fibrotic WT livers, but faintly detectable in CcnE1−/− mice (Fig. 4E). Detailed analysis of Ki-67 expression in liver sections revealed that ablation of CcnE1 did not significantly affect the proliferation of hepatocytes, which was moderate, but similar, in WT and CcnE1−/− liver (Supporting Fig. 2A,B). However, proliferation of NPCs was significantly reduced in the CcnE1−/− liver, hinting at a cell-type–specific function of CcnE1 during liver fibrosis.
Loss of CcnE2 Triggers CcnE1 Overexpression and Accelerated Liver Fibrogenesis During Chronic CCl4 Treatment.
Several studies have suggested that CcnE1 and CcnE2 might have overlapping functions. To evaluate the role of CcnE2 for liver fibrosis, we treated WT and CcnE2−/− mice with CCl4 for 2 and 4 weeks, respectively. Surprisingly, after 4 weeks of treatment, we detected comparable fiber formation and COL1A1 expression in CcnE2−/− mice and WT controls (Supporting Fig. 3A-C), indicating that CcnE2—in contrast to CcnE1—is not essentially involved in fibrosis progression.
However, differences were found after 2 weeks of CCl4 treatment. WT livers showed minor collagen expression, whereas in CcnE2−/− livers, the first signs of bridging fibers were detectable (Fig. 5A,B). Additionally, quantitative reverse-transcription polymerase chain reaction (qRT-PCR) showed significantly increased hepatic COL1A1 expression associated with significant CcnE1 mRNA up-regulation in CcnE2−/− livers, compared to controls (Fig. 5C,D). In line with these findings, we detected increased proliferation of hepatocytes and NPCs in the CcnE2−/− liver, as evidenced by Ki-67 and PCNA expression analysis (Fig. 5E and Supporting Figure 3D,E). One subpopulation of these highly proliferating CcnE2−/− cells were most likely activated HSCs, because α-SMA mRNA and protein expression was also significantly increased in CcnE2−/− livers (Fig. 5E and Supporting Fig. 3F). Thus, our findings implicate that accelerated fibrosis induction in CcnE2−/− mice might depend on the enhanced proliferation and activation of HSC. However, despite accelerated fibrogenesis, liver injury in CcnE2−/− mice was not significantly increased, compared to WT or CcnE1−/− mice (Supporting Fig. 3G).
Cell-Cycle Progression, Myofibroblast Transdifferentiation, and Survival of HSCs Depends on CcnE1 Expression Level.
We next aimed to define the cellular mechanisms leading to accelerated fibrogenesis in CcnE2−/− mice and fibrosis protection in CcnE1−/− livers. Our first results indicated that HSCs might be the CcnE-dependent effector cell population. HSCs are quiescent in healthy livers (i.e., G0 and CcnE inactive), but start to proliferate (G0-G1/S-phase transition and CcnE dependent) upon profibrotic stimulation.
To elucidate the function of E-type cyclins in activated HSCs, we isolated primary HSCs from WT, CcnE1−/−, and CcnE2−/− knock-out mice and investigated cell-cycle progression, transdifferentiation, and survival over a period of 10 days (days 1-10). Of note, CcnE1−/− livers revealed a normal frequency of resident HSCs (Supporting Fig. 4A). Primary analysis of HSCs was performed by fluorescence-activated cell-sorting (FACS) analysis of DNA content and immunofluorescence staining of Ki67 and α-SMA serving as markers for cell-cycle activation and myofibroblast differentiation, respectively. As expected, the total number of living WT HSCs increased continuously within the observation period of 10 days, whereas the number of CcnE1−/− HSC remained constant at low levels (Supporting Fig. 4B,C).
In agreement with these findings, WT HSCs revealed the marked occurrence of a 4n cell population after 10 days, indicating continuous cell-cycle progression (G2/M phase; Fig. 6A) with a tendency to form polyploid cells, which is in agreement with earlier observations.12 These cells were characterized by the expression of α-SMA and Ki-67 (Fig. 6A and Supporting Fig. 5A), indicating that they proliferate and transdifferentiate into myofibroblasts. In sharp contrast, CcnE1−/− HSCs did not show 2n/4n conversion throughout the 10-day observation time, demonstrating G1 cell-cycle arrest of these cells. Instead, we observed a large sub-G1 population of apoptotic cells with reduced DNA content (<2n) resulting from DNA degradation (Fig. 6B) and low total cell numbers throughout the observation period. Thus, quiescent ex vivo isolated CcnE1−/− HSCs have a defect in entering the cell cycle and are prone to excessive cell death.
Using CcnE2−/− HSCs, completely opposite effects were observed, showing already highly polyploid cells after isolation undergoing a further, time-dependent increase in DNA synthesis and polyploidization (Fig. 6C). The complete data, including all investigated time points, are shown in Supporting Fig. 6 and demonstrates that in WT cells, Ki-67 expression started at day 4 after seeding, whereas transdifferentiated (i.e., α-SMA-positive) myofibroblasts were first detected after 7 days. After 10 days, the majority of HSCs were activated and reached confluence (Supporting Fig. 5A). CcnE2−/− HSCs showed accelerated transactivation, starting day 3 after seeding, with overall stronger Ki-67 expression pointing at an enhanced cell-cycle activity of these cells (Supporting Fig. 6C,F).
mRNA quantification revealed substantial α-SMA induction—and thus transactivation—after 7 days in WT HSCs, but already after 3 days in CcnE2−/− cells (Fig. 6D). Importantly, overall α-SMA expression in CcnE2−/− HSCs significantly exceeded WT levels at all time points investigated. In contrast, overall α-SMA levels in CcnE1−/− HSCs were lower, compared to WT cells, and especially lacked induction after 7 days. These findings suggested that CcnE1 is essential for HSC transactivation. To further test our hypothesis, we measured the expression of platelet-derived growth factor receptor beta (PDGF-Rβ), which is usually induced when HSCs are activated and start to transdifferentiate into myofibroblasts.15 Though PDGF-Rβ was up-regulated in WT HSCs, no increase was evident in CcnE1−/− cells (Fig. 6E). Interestingly, loss of CcnE2 resulted in an approximately 5-fold up-regulation of basal PDGF-Rβ expression, suggesting that quiescent CcnE2−/− HSCs are already primed for accelerated activation.
We next compared CcnE1 mRNA expression levels in WT and CcnE2−/− HSC throughout the transdifferentiation process. Interestingly, CcnE1 expression was significantly elevated in CcnE2−/− HSCs at all time points investigated (Fig. 6F). CcnE1 peak expression in WT cells was found at day 7 after seeding, whereas comparable expression levels were detected in CcnE2−/− HSCs between days 3 and 10. Interestingly, in both groups, maximal CcnE1 expression was detected before the first appearance of transdifferentiated, α-SMA-positive myofibroblasts, suggesting that CcnE1 might be involved in HSC transactivation. We therefore performed expression analysis of HSC-derived profibrotic proteins, which confirmed the accelerated onset of α-SMA and collagen I expression in CcnE2−/− HSC, compared to WT controls (Fig. 7A). Of note, protein data could not be obtained from CcnE1−/− HSCs because of poor survival and thus low protein yields.
To better characterize the findings in CcnE1−/− HSCs, we performed terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) analysis of seeded HSCs from all groups and controls up to 10 days after isolation. These experiments revealed that CcnE1−/− HSCs were more prone to undergo apoptosis, which was not evident in CcnE2−/− cells or controls (Fig. 7B,C). Accordingly, CcnE1 is essential for triggering the proliferation, transdifferentiation, and survival of HSCs.
Liver fibrosis is a chronic wound-healing process leading to liver scarring and directing progressively deteriorating organ function. In this context, chronic liver injury triggers a proliferative response of hepatocytes, but also of nonparenchymal liver cells, including matrix-producing cells such as activated HSCs and myofibroblasts. Therefore, liver fibrogenesis involves the cell-cycle reentry of quiescent cells, such as hepatocytes and HSCs. Surprisingly, little information exists on how cell-cycle mediators, such as cell-cycle–dependent kinases and cyclins, contribute to the progression of liver fibrosis.16 Genetic inactivation of single D-type (e.g., CcnD1-3) and E-type (e.g., CcnE1 and CcnE2) cyclins or their associated kinases (e.g., Cdk2, 4, and 6) did not affect general cellular processes, such as embryonic development, presumably because of overlapping or even redundant functions.17 However, it has been postulated that these cyclins and Cdks may also perform cell-type–specific functions,18 and in line with this hypothesis, we recently described nonredundant functions for CcnE1 and CcnE2 in hepatocytes during liver regeneration after PH.11
In the present study, we found that CcnE1, but not CcnE2, was up-regulated in human patients with advanced liver fibrosis, suggestive of a specific function of CcnE1 during liver fibrogenesis. To address this hypothesis, we treated CcnE1−/− and CcnE2−/− mice acute or chronically with CCl4 and analyzed the effect on the proliferative response of hepatocytes and nonparenchymal liver cells.
Ubiquitous ablation of CcnE1 in this fibrosis model revealed several unexpected findings defining CcnE1 as an essential profibrotic mediator. After acute toxic liver injury, the overall proliferative response of CcnE1-deficient hepatic cells was dramatically impaired. Several studies, including our own work, demonstrated that CcnE1 is dispensable for the proliferation of continuously cycling cells and regenerating hepatocytes after surgical partial liver resection.9, 11, 19 From our present data, it is now evident that the requirement of proliferating hepatocytes for CcnE1 depends on the proliferation stimulus. Though CcnE1 is dispensable for hepatocyte proliferation in a proinflammatory environment (e.g., hepatectomy), it is apparently essential after toxic liver injury (i.e., CCl4) in vivo. In agreement with this hypothesis, we recently observed a prolonged cell-cycle arrest of CcnE1−/− hepatocytes in vivo after treatment with the hepatotoxic agent, diethylnitrosamine (data not shown).
Intriguingly, constitutive ablation of CcnE1—but not inhibition of CcnE2—protected from CCl4-mediated liver fibrosis, which was related to impaired cell-cycle activity of nonparenchymal liver cells. Hence, we focused on HSCs because this cell population is central for the process of liver fibrosis as the major source of ECM proteins.20 A recent study suggested that down-regulation of CcnE1 is related to the delayed cell-cycle progression of the human HSC line, LX-2.21 In line with these findings, we demonstrated that complete ablation of CcnE1 induces a dramatic cell-cycle arrest of HSCs and hypersensitivity to apoptosis and overall poor survival. Although the mechanism triggering apoptosis in CcnE1-deficient cells remains elusive, HSC apoptosis clearly acts as an antifibrotic.22 Thus, reduced liver fibrosis in CCl4-treated CcnE1−/− mice is most likely explained by impaired viability and cell-cycle arrest of HSCs after profibrogenic stimulation.
Our study also revealed an unexpected role of CcnE2 for liver fibrogenesis. In line with our earlier studies,11 loss of CcnE2 resulted in accelerated gene activation of CcnE1 in hepatocytes and HSCs, suggesting that CcnE2 is an inhibitor of CcnE1 expression. At present, the underlying mechanism is unclear; however, CcnE2 was shown to be 1.5- to 10-fold more highly expressed, compared to CcnE1, in at least three independent studies.23 We speculate that CcnE2 might sequester, and thus inactivate, transcriptional activators of CcnE1. Besides up-regulating CcnE1, loss of CcnE2 resulted in early liver fibrogenesis and, more important, in accelerated HSC activation and proliferation. Although it is tempting to speculate that this might be a direct consequence of elevated CcnE1 expression, we cannot exclude alternative mechanisms. Indeed, we also observed a 5-fold basal up-regulation of PDGF-Rβ in CcnE2−/− HSC, which indicates that these cells are already primed for activation and proliferation. We thus conclude that CcnE2 does not share overlapping functions with CcnE1 in HSCs, but acts as an antifibrotic.
Based on our experiments, we suggest an essential role of CcnE1 for HSC activation and fibrosis induction (as illustrated in Fig. 7D): In WT cells, the peak of CcnE1 expression occurs before the maximum expression of α-SMA and PDGF-Rβ. We conclude that CcnE1 drives profibrogenic mechanisms, predominantly through the targeting and activation of hitherto quiescent HSCs. Previous work demonstrated that E-type cyclins are dispensable for the continuous proliferation of embryonic fibroblasts—sharing some similarities with hepatic myofibroblasts—whereas they are essential for the exit from quiescence.9 In our present study, the same phenomenon seems to operate in HSCs, except that they rely only on CcnE1 for cell-cycle reentry, because CcnE2 is not induced during liver fibrogenesis. Accordingly, CcnE1-deficient HSCs are unable to normally reenter the cell cycle from G0.
Our results raise the question of whether our findings are model specific and may only apply for the CCl4-fibrosis model. Although we cannot exclude that some of our results (e.g., cell-cycle arrest of CcnE1−/− hepatocytes) are CCl4 specific, our data from ex vivo analyzed primary HSCs describe a general, model-independent biological function showing that CcnE1 is an essential cell cycle and survival factor for HSCs.
In summary, we demonstrate that CcnE1 is a novel key mediator of hepatic fibrosis in mice because it provides essential functions for the proliferation and survival of HSCs. Future work will evaluate whether the targeted inhibition of CcnE1 might be a therapeutic option to prevent liver fibrosis.
The excellent assistance of Sibille Sauer-Lehnen, Carmen C. Tag, and the Core Unit “Q3-Cell Isolation” of the SFB/TRR57 with the isolation of primary HSCs is gratefully acknowledged. We are also grateful to Kanishka Hiththetiya and Christiane Esch for their technical support with the histological analysis of liver samples. Confocal microscopy was performed in the Interdisciplinary Center for Clinical Research (IZKF) Aachen within the Faculty of Medicine at RWTH Aachen University with the kind support of Gerhard Müller-Newen.