State Key Laboratory of Cell Biology, Shanghai Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
Address reprint requests to: Lijian Hui, Ph.D., State Key Laboratory of Cell Biology, Shanghai Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 320 Yue-yang Road, 200031 Shanghai, China. E-mail: firstname.lastname@example.org; fax: +86-21-5492132.
Potential conflict of interest: Nothing to report.
L.H.'s laboratory is funded by the National Science Foundation of China (31071238), the Ministry of Science and Technology of China (2012CB945001 and 2011ZX09307-302-01), and the Chinese Academy of Sciences (the Hundred Talents Program). E.M.C. is supported by the Canadian Institutes for Health Research, the Canada Foundations for Innovation, and the Natural Sciences and Engineering Research Council of Canada. E.M.C. holds a CSL-Behring Research Chair and a Canada Research Chair in Endothelial Cell Biology and is an adjunct scientist with the Canadian Blood Services (CBS). The authors thank Dr. V. Factor (National Cancer Institute, National Institutes of Health, Bethesda, MD) for A6 antibody and Dr. X. Wang (Inner Mongolia University, Inner Mongolia, China) for antibodies against CK19.
Hepatocytes possess a remarkable capacity to regenerate and reconstitute the parenchyma after liver damage. However, in the case of chronic injury, their proliferative potential is impaired and hepatic progenitor cells (HPCs) are activated, resulting in a ductular reaction known as oval cell response. Proapoptotic and survival signals maintain a precise balance to spare hepatocytes and progenitors from hyperplasia and cell death during regeneration. Survivin, a member of the family of inhibitor of apoptosis proteins (IAPs), plays key roles in the proliferation and apoptosis of various cell types. Here, we characterized the in vivo function of Survivin in regulating postnatal liver development and homeostasis using mice carrying conditional Survivin alleles. Hepatic perinatal loss of Survivin causes impaired mitosis, increased genome ploidy, and enlarged cell size in postnatal livers, which eventually leads to hepatocyte apoptosis and triggers tissue damage and inflammation. Subsequently, HPCs that retain genomic Survivin alleles are activated, which finally differentiate into hepatocytes and reconstitute the whole liver. By contrast, inducible ablation of Survivin in adult hepatocytes does not affect HPC activation and liver homeostasis during a long-life period. Conclusion: Perinatal Survivin deletion impairs hepatic mitosis in postnatal liver development, which induces HPC activation and reconstitution in the liver, therefore providing a novel HPC induction model. (Hepatology 2013; 58:2109–2121)
tumor necrosis factor–related weak inducer of apoptosis.
The liver possesses the remarkable capacity to restore damaged hepatic tissues after injury.[1-4] During normal hepatocyte turnover or partial hepatectomy-regeneration, this is achieved by replication of highly differentiated parenchymal hepatocytes in response to loss of liver mass. When the proliferative capacity of mature liver cells is blocked or compromised, a population of small portal zone cells with a high nuclear cytoplasmic ratio and an ovoid nucleus, known as “oval cells,” proliferate extensively and differentiate into hepatocytes and bile duct cholangiocytes to restore damaged liver tissue.[5-7] Oval cells are regarded as the facultative hepatic progenitor cells (HPCs) in adult livers.[1-4]
Chemically injured animal models are commonly used to trigger oval cell activation. For example, feeding mice with a diet supplemented with 3,5-diethoxycarbonyl-1,4-dihydro-collidine (DDC) is a protocol for oval cell induction.[8-10] However, chemicals cause severe, unpredictable toxicity to both hepatocytes and nonparenchymal cells. Chemical injury fails to distinguish damaged cells from nondamaged cells during tissue repair, specifically newly derived hepatocytes from activated HPCs. Recent progresses have developed genetic mouse models for HPC activation. Liver-specific deletion of β-catenin or DDB1[12, 13] caused HPCs induction and reconstitution during hepatocyte turnover in aged mice. However, the ductular reaction eventually engenders liver tumors in β-catenin or DDB1 models. Other studies showed that liver-specific overexpression of tumor necrosis factor–related weak inducer of apoptosis (TWEAK) and fibroblast growth factor 7 (FGF7) stimulated HPC proliferation. Nevertheless, mature hepatocytes in these models retain proliferation capacity, which makes it difficult to analyze the hepatic differentiation and reconstitution of HPCs. Animal models that selectively target hepatocytes for cell death and proliferation arrest, but spare progenitor cells, would bring additional insight in characterizing HPCs.
Survivin is a unique member of the family of inhibitor of apoptosis proteins (IAPs), which contains a single N-terminal baculovirus IAP repeat domain. As an essential regulator of mitosis, Survivin localizes to kinetochores by associating with Aurora B, INCENP, and Borealin to constitute the chromosomal passenger complex (CPC).[17, 18] It was thus proposed that Survivin contributes to mitosis by facilitating CPC assembling in kinetochores and microtubule stabilization. In line with this, Survivin depletion causes mitotic arrest and the formation of polyploid cells. Survivin is also a key player in antagonizing cell death through directly inhibiting active caspases. The capacity of Survivin to inhibit active caspases likely involves cooperation with hepatitis B X-interacting protein and X-linked IAP.[20, 21]
Consistent with its prominent role in cell-cycle progression and cell survival, Survivin is expressed in highly proliferating cells, such as embryonic cells and adult stem cells. Survivin is indispensable for ontogenesis, because Survivin-deficient mice died in utero as a result of dys-regulated microtubule organization and cellular degeneration. Using conditional alleles, it has been demonstrated that tissue-specific depletion of Survivin impaired the proliferation and maturation of hematopoietic cells, neuronal cells, endothelial cells, cardiomyocytes, and pancreatic beta-cells.[24-29] In addition, Survivin is considered to play a key role in tumor onset and recurrence. A recent study showed that Survivin promotes liver carcinogenesis through enhancing the survival of initiated cancer cells. Interestingly, the in vivo function of Survivin in postnatal liver development and liver homeostasis remains largely undefined.
Using mouse lines carrying conditional alleles, we found that in neonatal livers, but not in adult livers, hepatocytes lacking Survivin are blocked in mitosis and display increased genome ploidy and enlarged cell size. Moreover, loss of Survivin-deficient hepatocytes dramatically activates HPCs that retain Survivin alleles during postnatal liver development. HPCs differentiate into hepatocytes and eventually reconstitute the liver. Our study demonstrates that Survivin is essential for hepatocyte proliferation during postnatal liver development. Furthermore, mice with liver-specific Survivin deletion provide a novel model to study HPC induction and differentiation.
Materials and Methods
Alb-cre; Survivinf/f (svvΔli) and Mx-cre; Survivinf/f (svvΔli*) were generated by crossing svvf/f mice to Alb-Cre mice and Mx-Cre mice, respectively. Liver samples were collected at indicated time points for various assays.
Mouse Hepatocyte Isolation
Mouse hepatocytes were isolated after a modified two-step perfusion and collagenase digestion. Cell viability was determined using Trypan blue exclusion assay. HPCs were purified by Nycodenz density-gradient centrifugation, as previously described.
Liver tissue samples were fixed by paraformaldehyde and embedded in paraffin. Sections were subjected to hematoxylin and eosin (H&E), immunohistochemistry (IHC), Sirius Red, and TdT-mediated dUTP nick end labeling (TUNEL) staining.
Cryosections and Immunofluorescence Staining
Liver tissue samples were snap-frozen in liquid nitrogen and embedded in optimal cutting temperature tissue-freezing media. Frozen sections were subjected to immunofluorescent (IF) staining.
Data were subjected to the Student t test. P < 0.05 was considered statistically significant. Data are presented as mean ± standard deviation.
Further detailed description of the materials and methods used are provided in the Supporting Information.
Efficient Deletion of Survivin in Hepatocytes of svvΔli Mice
Mice homozygous for floxed Survivin (svv) alleles were crossed with mice that carry Cre recombinase under the control of the rat albumin promoter (Alb-Cre) to specifically delete Survivin in hepatocytes perinatally (Fig. 1A). Alb-Cre, svvf/f (svvΔli) mice were born indistinguishable from their littermates (svvf/f) with the expected Mendelian frequency. SvvΔli mice had normal body weight (Fig. 1B). The floxed svv alleles were only weakly deleted in neonatal livers, but efficiently deleted in adult livers, as shown by genotyping polymerase chain reaction (PCR; Fig. 1C). To monitor deletion efficiency during postnatal liver development, total liver RNA was isolated from svvΔli livers. Consistent with previous reports, quantitative PCR (qPCR) revealed that messenger RNA (mRNA) levels of Survivin progressively decreased in control livers. Notably, expression of Survivin was almost unchanged in postnatal day 1 svvΔli livers (Fig. 1D). Survivin mRNA levels were reduced by 70% at the age of 1 week, and were further decreased by 80%-90% at the age of 2 and 3 weeks, indicating a progressive and efficient deletion of Survivin in livers (Fig. 1D). Moreover, IHC staining showed that Survivin protein was undetectable in svvΔli livers at 2 weeks of age (Fig. 1E).
Survivin Is Required for Mitotic Transition During Postnatal Liver Development
Although svvΔli mice did not appear ill, their livers were markedly enlarged, with 30% increase of liver/body weight ratios (Fig. 2A). Histological inspection revealed that svvΔli hepatocytes underwent progressive hypertrophy with 3-fold increase in diameter, compared to controls (Fig. 2B). Single hepatocytes prepared by collagenase perfusion were highly polyploid (Fig. 2C,D), suggesting impairments in their cell cycle. Interestingly, the development of bile ducts and biliary epithelial cells seemed not to be affected in svvΔli livers (Supporting Fig. 1).
Previous studies have shown that Survivin is highly expressed in the G2/M phase and is required for mitotic transition through enhancing histone H3 phosphorylation.[17, 18] We analyzed hepatocyte mitosis by IHC staining of phosphorylated H3S10 (p-H3S10). The numbers of p-H3S10-positive hepatocytes were dramatically decreased in 2-week-old svvΔli mice (Fig. 2E). Additionally, expression levels of mitotic phase cyclins, such as cyclin A1 and cyclin B2, were significantly decreased in svvΔli livers (Fig. 2F). Collectively, these data demonstrate that during postnatal liver development, svvΔli hepatocytes are blocked in mitosis, leading to genome polyploidy and cell hypertrophy. Almost all hepatocytes showed hypertrophy in svvΔli livers, because each hepatocyte was proliferating during postnatal development.
Progressive Death of Survivin-Deficient Hepatocytes in Young Mice
We characterized the mitotically arrested svvΔli hepatocytes over an extended period of time. At 2-3 months of age, hepatocytes containing apoptotic bodies were evident in svvΔli livers, as shown by morphological changes and TUNEL (Fig. 3A,B). Quantification of TUNEL-positive cells showed that cell death was initiated at the age of 8 weeks and continued for several weeks (Fig. 3B). Moreover, levels of cleaved caspase-3 were significantly increased in svvΔli livers (Fig. 3C). Expression levels of proapoptotic gene Bak1 were also increased, whereas antiapoptotic genes, such as XIAP, Bruce, and ML-IAP, were reduced in svvΔli livers (Fig. 3D).
Dead hepatocytes triggered infiltration of inflammatory cells in parenchyma of svvΔli livers (Fig. 3A), resulting in elevated levels of several inflammatory cytokines (Fig. 3E). Consistent with increased hepatocyte death and inflammation in svvΔli livers, serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, and total bile acid were significantly elevated in svvΔli mice (Fig. 3F). These data indicate that mitotically arrested svvΔli hepatocytes undergo cell death, which consequently triggers liver damage and dysfunction.
Activation of HPCs in svvΔli Livers
Previous studies showed that HPCs are expanded in number to repopulate the injured liver parenchyma when the proliferative capacity of hepatocytes is impaired.[1-4] Remarkably, in 8- to 12-week-old svvΔli livers, expansion of small ductular cells with a high ratio of nuclear/cytoplasm was observed (Fig. 4A). The expansion of small epithelial cells in svvΔli livers generated cell cords that bridged the portal tracts and delineated the polygonal architecture of hepatic lobules (Fig. 4A). Morphologically, extensive hyperplasia of small epithelial cells fits the description of facultative mouse HPCs.
Consistent with classic characteristics of HPCs,[8, 9] the small epithelial cells stained positively with anti–cytokeratin 19 (CK19), anti-Sox9 (SRY [sex determining region Y]-box 9), and A6 antibodies (Fig. 4B,C). In contrast, expression of the CK19, Sox9, and A6 antigens were either restricted to bile duct epithelial cells or undetectable in control livers (Fig. 4B,C). Furthermore, genes that are enriched in HPCs were significantly increased in svvΔli livers (Fig. 4D). Double staining of A6 antigens and p-H3S10 demonstrated that a portion of HPCs underwent proliferation (Fig. 4E). Furthermore, quantification of proliferating cells by Ki67 staining showed that all proliferating cells were HPCs in 8-week-old livers, whereas Ki67-positive hepatocytes were undetectable (Supporting Fig. 2). In line with HPC activation, expression of growth factors essential to HPC proliferation, such as hepatocyte growth factor (HGF), transforming growth factor alpha (TGF-α), and FGF7,[15, 33] were increased in svvΔli livers (Fig. 4F). Taken together, these data suggest that deletion of Survivin in hepatocytes causes cell death and liver damage, which triggers extensive HPC expansion.
Activated HPCs in svvΔli Livers Are Comparable to Those in the DDC Model
We compared HPCs induced in svvΔli mice at the age of 8-12 weeks to those in DDC-treated svvf/f livers. HPCs positive for both A6 antigens and CK19 were induced in periportal areas 3 weeks after DDC treatment (Fig. 5A). CK19-positive HPCs induced in svvΔli and DDC-treated livers had similarly scant cytoplasm, ovoid nuclei (Fig. 5B), and increased collagen deposition in the extracellular matrix (Fig. 5C). HPCs in both models were organized in duct-like structures in the periportal region and extended into the liver lobule. The duct-like structures were usually irregularly shaped with poorly defined lumen. Intriguingly, expression levels of genes enriched in HPCs were comparable in both svvΔli livers and DDC-treated livers (Fig. 5D).
HPCs were purified by Nycodenz density gradient centrifugation, as previously described. Three fractions of nonparenchymal cell populations were obtained, and HPCs were enriched in fraction 2 (F2)-containing small cells of 7-10 μm in diameter. Cell numbers of F2 were significantly increased in both svvΔli livers and DDC-treated livers, with highly elevated expression of epithelial cell adhesion molecule (EpCAM), CD133, and CD44 (Fig. 5E). Importantly, the majority of F2 cells were positive for both A6 antigens and CK19 (Fig. 5F). Collectively, these data suggest that in svvΔli livers, HPCs are activated and form pathological structures comparable to HPCs in DDC-treated livers.
HPCs Reconstitute the Whole Liver Without Carcinogenesis in Aged svvΔli Mice
The expansion and differentiation of HPCs were progressive, yielding regenerative hepatic nodules containing hepatocytes of normal size at periportal regions in livers of 12- to 14-week-old svvΔli mice (Supporting Fig. 3A). Strikingly, hepatocytes in these regenerative nodules were proliferating, as determined by Ki67 staining (Supporting Fig. 3B). In line with reports that floxed alleles were inefficiently deleted by Cre recombinase in HPC-derived regenerative livers,[11, 12] IHC staining revealed that Survivin protein was expressed in the nodules (Supporting Fig. 3C). We microdissected tissues containing hypertrophic hepatocytes and regenerative nodules containing normal-sized hepatocytes from svvΔli liver sections. Though floxed Survivin alleles were completely deleted in liver tissues containing hypertrophic hepatocytes, genotyping by PCR revealed that Survivin was mostly undeleted in regenerative nodules (Supporting Fig. 3D).
The long-term effects of HPC expansion and reconstitution were assessed in aged svvΔli mice. At the age of 10 months, svvΔli mice showed normal liver size (Fig. 6A) with poorly deleted Survivin (Fig. 6B). ALT, AST, and total bilirubin in svvΔli mice reverted to control levels (Fig. 6C), indicating that liver function recovered in aged svvΔli mice. Importantly, histological analysis showed that hepatocyte size in 10-month-old svvΔli mice was the same as that observed in svvf/f livers (Fig. 6D). Propidium iodide (PI) staining also showed a comparable spectrum of hepatocyte ploidy in aged svvΔli and control mice (Fig. 6E). Intriguingly, liver function and hepatocyte size were already recovered in 4-month-old svvΔli mice (Supporting Fig. 4A-C), suggesting that HPC-derived hepatocytes could reconstitute the liver efficiently within 1-2 months.
HPC expansion has been reported to cause tumor formation in aged mice.[11, 12] However, despite the liver damage observed at the age of 8-12 weeks, svvΔli mice survived as long as control mice (Supporting Fig. 5A). Moreover, the liver morphology and function in 18-month-old svvΔli mice remained the same as that in controls (Supporting Fig. 5B,C). Importantly, no tumors have been found in livers of svvΔli mice at the age of 18 months.
Survivin Is Dispensable in Adult Livers
Given that Survivin-expressing HPCs repopulated liver in svvΔli mice, it was not possible to address whether Survivin is required in adult livers. To that end, we generated an inducible mouse line, Mx-Cre; svvf/f. Survivin can be deleted in liver and hematopoietic cells in adult mice through intraperitoneal (IP) injection of poly(IC). To avoid the mortality induced by Survivin deficiency in hematopoietic cells, Mx-Cre; svvf/f mice were gamma-irradiated and transplanted with littermate svvf/f bone marrows (Fig. 7A). After reconstitution of the hematopoietic system, Mx-Cre; svvf/f mice were injected with poly(IC) to induce liver-specific Survivin deletion (referred to as svvΔli*; Fig. 7A). All svvΔli* mice survived poly(IC) treatment and appeared healthy. Six months after poly(IC) treatment, the floxed Survivin gene was efficiently deleted in svvΔli* livers (Fig. 7B). SvvΔli* mice displayed unchanged liver size and serum levels of ALT and AST (Fig. 7C,D). Also, hepatocyte size was not increased, and no ductular reaction was detected in svvΔli* livers (Fig. 7E). Although a slight, but not significant, increase in the number of tetraploid hepatocytes was detected, the spectrum of hepatocyte ploidy was similar in svvΔli* and control mice (Fig. 7F). These results suggested that Survivin is dispensable for the normal function of mature hepatocytes in adult mice.
In this study, we demonstrated that Survivin is essential for postnatal liver development. Survivin deficiency results in hepatocyte mitotic arrest, hypertrophy, and genome polyploidy. Furthermore, mitotically arrested Survivin-deficient hepatocytes undergo cell death, which subsequently stimulates Survivin-positive HPCs to proliferate and reconstitute svvΔli livers. In aged svvΔli mice, the liver is completely repopulated by HPC-derived hepatocytes retaining Survivin and appears normal, in comparison to the control. Depletion of Survivin in adult livers has no obvious effect on liver homeostasis and HPC activation. These data support the notion that liver-specific deletion of Survivin could serve as a novel progenitor induction model during postnatal liver development.
During the course of hepatocyte polyploidization, we detected extensive apoptosis of hepatocytes with increased DNA content in 8- to 12-week-old svvΔli mice. Intriguingly, Survivin was already efficiently deleted in hepatocytes of svvΔli mice by approximately 2 weeks. Therefore, apoptosis in svvΔli livers was not directly caused by Survivin deficiency. It is likely that mitotic defects and polyploidy pose a stress on proliferating hepatocytes and eventually lead to cell death. In support of this hypothesis, depletion of Survivin in adult livers did not cause cell death in svvΔli* livers. This finding would be of particular importance in targeting Survivin for human cancer treatment. Because of its essential role in normal cell survival and division, Survivin targeting has been always viewed with caution.[34, 35] Our results unambiguously demonstrate that Survivin is dispensable for adult liver homeostasis, therefore suggesting that local inactivation of Survivin is safe for liver cancer treatment. We recognize that although Survivin deletion did not affect adult liver function, it is possible that Survivin deficiency might result in impaired proliferation in pathologies, such as partial hepatectomy and tumorigenesis.
Extensive hepatocyte apoptosis in svvΔli livers caused loss of liver mass and inflammation. The cell death lasted for more than 1 month, and this level of liver damage should trigger remarkable liver regeneration. Survivin-deficient hepatocytes lost their proliferative capacity and could not restore the liver mass. Instead, HPCs were activated to proliferate and expand in number. Onset of HPC induction occurred at approximately 8 weeks. Interestingly, cell death was still undetectable at the age of 6 weeks, suggesting that HPC induction and cell death are closely coupled events during onset of tissue regeneration. Proliferative HPCs were found to express Survivin, suggesting that they escaped Cre recombinase-mediated excision of floxed alleles, as previously reported.[11, 12] The Alb-Cre transgene uses the promoter and upstream enhancer of the rat albumin gene. It is likely that the rat albumin promoter does not respond identically to that of the mouse.
Notably, the liver was completely repopulated by HPC-derived hepatocytes in aged svvΔli mice. In the DDC model, liver parenchyma was only partially replaced by HPC-derived hepatocytes, which made it difficult to examine the differentiation capacity of the HPCs and their contribution to the restoration of liver parenchyma. Unlike chemicals, genetic ablation of Survivin is restricted to hepatocytes and leaves nonparenchymal cells, including HPCs, intact. Therefore, this model is ideal for analyzing the extracellular signals that mediate cross-talk between the dying hepatocytes and activated HPCs. Indeed, our study indicate that liver growth factors, which play important roles in HPC induction, are increased in svvΔli livers, including HGF, TGF-α, and FGF7.[15, 33]
Ductular reaction by wild-type HPCs has been documented in genetically modified mouse lines. Liver-specific overexpression of TWEAK and FGF7 stimulated HPC activation. However, unlike Survivin-deficient hepatocytes that are mitotically arrested, mature hepatocytes in these models possess proliferation capacity. Liver-specific DDB1 deletion abrogates the proliferative capacity of hepatocytes, resulting in proliferation of DDB1-expressing HPCs. Intriguingly, these HPCs were activated in the microenvironment with minimal tissue damage.[12, 13] Another study showed that hepatocyte turnover was compromised in livers lacking β-catenin, leading to β-catenin-positive HPC proliferation and liver mass reconstitution at the age of 11-18 months. Because of the long period of ductular reaction, it was difficult to define when β-catenin-proficient HPCs were activated to reconstitute the liver. By contrast, in Survivin-deficient livers, hepatocytes undergo cell death within 8-12 weeks of age, thereby providing a clear onset of progenitor activation. Moreover, the repopulation of HPC-derived hepatocytes is fast and complete in Survivin-deficient livers. By the age of 4 months, the whole liver was almost reconstituted by newly generated hepatocytes.
Remarkably, in DDB1-deficient mice, ductular reaction appeared at the age of 4 weeks and lasted for a long time. Liver tumors were eventually found at the age of 16-20 months. In β-catenin-deficient mice, ductular reaction was evident at approximately 11 months of age, and this led to the development of liver tumors at 18-20 months. In both models, prolonged ductular reaction may cause the accumulation of mutations in HPCs in aged livers. However, in our study, no tumors were found in aged svvΔli livers. We observed that HPC-derived hepatocytes reconstituted young svvΔli livers within 1-2 months, and liver function was well maintained thereafter. The different ductular reactions in these models might explain why svvΔli mice did not develop liver tumors. Together, our study provides a new model with a defined process of activation, differentiation, and repopulation of HPCs.
The authors thank P. Huang for technical support. The authors also thank the Animal Core Facility and Cell Biology Core Facility at Shanghai institute of Biochemistry and Cell Biology for experimental support.