A mouse model of accelerated liver aging caused by a defect in DNA repair


  • Siobhán Q. Gregg,

    1. Department of Microbiology and Molecular Genetics, University of Pittsburgh School of Medicine, Pittsburgh, PA
    2. University of Pittsburgh Cancer Institute, Pittsburgh, PA
    Search for more papers by this author
  • Verónica Gutiérrez,

    1. Department of Microbiology and Molecular Genetics, University of Pittsburgh School of Medicine, Pittsburgh, PA
    2. University of Pittsburgh Cancer Institute, Pittsburgh, PA
    Search for more papers by this author
  • Andria Rasile Robinson,

    1. University of Pittsburgh Cancer Institute, Pittsburgh, PA
    2. Department of Human Genetics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA
    Search for more papers by this author
  • Tyler Woodell,

    1. University of Pittsburgh Cancer Institute, Pittsburgh, PA
    Search for more papers by this author
  • Atsunori Nakao,

    1. Department of Surgery, Thomas E. Starzl Transplantation Institute, University of Pittsburgh, Pittsburgh PA
    Search for more papers by this author
  • Mark A. Ross,

    1. Department of Cell Biology and Physiology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • George K. Michalopoulos,

    1. Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Lora Rigatti,

    1. Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Carrie E. Rothermel,

    1. Department of Cell Biology and Physiology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Irene Kamileri,

    1. Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology-Hellas, Crete, Greece
    2. Department of Biology, University of Crete, Crete, Greece
    Search for more papers by this author
  • George A. Garinis,

    1. Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology-Hellas, Crete, Greece
    2. Department of Biology, University of Crete, Crete, Greece
    Search for more papers by this author
  • Donna Beer Stolz,

    1. Department of Cell Biology and Physiology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Laura J. Niedernhofer

    Corresponding author
    1. Department of Microbiology and Molecular Genetics, University of Pittsburgh School of Medicine, Pittsburgh, PA
    2. University of Pittsburgh Cancer Institute, Pittsburgh, PA
    • University of Pittsburgh Cancer Institute, 5117 Centre Avenue, Hillman Cancer Center, Office 2.6, Pittsburgh, PA 15213
    Search for more papers by this author
    • fax: (412) 623-7761

  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the National Institute of Environmental Health Sciences (ES016114), the Ellison Medical Foundation (AG-NS-0303-05), and the University of Pittsburgh Claude D. Pepper Center (P30AG024827; to L.J.N.), the American Federation of Aging Research MSTAR Program (to T.W.), and the Pittsburgh Center for Kidney Research Pilot Program (P30DK079307; to D.B.S. and L.J.N). This project used the UPCI Cell and Tissue Imaging and Animal Facilities and was supported, in part, by award P30CA047904.


The liver changes with age, leading to an impaired ability to respond to hepatic insults and increased incidence of liver disease in the elderly. Therefore, there is critical need for rapid model systems to study aging-related liver changes. One potential opportunity is murine models of human progerias or diseases of accelerated aging. Ercc1−/Δ mice model a rare human progeroid syndrome caused by inherited defects in DNA repair. To determine whether hepatic changes that occur with normal aging occur prematurely in Ercc1−/Δ mice, we systematically compared liver from 5-month-old progeroid Ercc1−/Δ mice to old (24-36-month-old) wild-type (WT) mice. Both displayed areas of necrosis, foci of hepatocellular degeneration, and acute inflammation. Loss of hepatic architecture, fibrosis, steatosis, pseudocapillarization, and anisokaryosis were more dramatic in Ercc1−/Δ mice than in old WT mice. Liver enzymes were significantly elevated in serum of Ercc1−/Δ mice and old WT mice, whereas albumin was reduced, demonstrating liver damage and dysfunction. The regenerative capacity of Ercc1−/Δ liver after partial hepatectomy was significantly reduced. There was evidence of increased oxidative damage in Ercc1−/Δ and old WT liver, including lipofuscin, lipid hydroperoxides and acrolein, as well as increased hepatocellular senescence. There was a highly significant correlation in genome-wide transcriptional changes between old WT and 16-week-old, but not 5-week-old, Ercc1−/Δ mice, emphasizing that the Ercc1−/Δ mice acquire an aging profile in early adulthood. Conclusion: There are strong functional, regulatory, and histopathological parallels between accelerated aging driven by a DNA repair defect and normal aging. This supports a role for DNA damage in driving aging and validates a murine model for rapidly testing hypotheses about causes and treatment for aging-related hepatic changes. (HEPATOLOGY 2012)

Aging is characterized by the progressive loss of homeostatic reserve in all tissues, leading to a decreased ability to respond to stress, functional decline, and dramatically increased risk of morbidity and mortality. 1 Chronic liver disease and cirrhosis are the twelfth leading cause of death in the United States, with 25,192 (1.1%) deaths each year.2 Mortality resulting from liver disease is not limited to the United States, as the World Health Organization (WHO) estimates that 20 million individuals have cirrhosis and/or liver cancer worldwide, and that 1-2 million die annually as a result of liver failure. Many liver diseases are more prevalent in the elderly, including alcohol-induced cirrhosis, viral hepatitis-induced cirrhosis, diabetic-associated chronic liver disease, hepatocellular carcinoma, and biliary cirrhosis,3, 4 illustrating that the aged liver is less able to cope with a variety of stressors.

Although there are no definitive biomarkers of the aged liver, there are several hallmark structural changes. The volume of the liver decreases with age. 5 This decrease is gender-specific, with an average decrease of 6.5% in males and 14.3% in females.6 Hepatocytes are quiescent cells, until stimulated to proliferate after injury or surgical resection.7 The capacity for hepatocyte regeneration after partial hepatectomy (PH) decreases with age.8 During the aging process, the nuclear size and shape of hepatocytes becomes altered.9 Nuclear invaginations and larger nuclei are commonly associated with age in liver samples.10 Nuclear polyploidy of hepatocytes increases with age, rising from 7% to 10% in young mice to approximately 30% in old mice.9 Finally, lipofuscin, insoluble oxidized proteins, accumulates in the cytoplasm of hepatocytes.11, 12

Liver sinusoidal endothelial cells (LSECs) line the hepatic microvasculature, express little to no basement membrane, and exhibit numerous fenestrations necessary for filtering lipopoteins from the portal blood supply. A reduction in the number of these pores, combined with the thickening of the endothelium, called pseudocapillarization, is commonly seen in aged livers, 13 as is nonalcoholic fatty liver disease or steatosis.14 The underlying cause for fatty liver disease is not known, but some possibilities include diabetes mellitus, oxidative stress, mitochondrial dysfunction, and metabolic and hormonal imbalances.15 All of these age-related liver changes in humans are recapitulated in the mouse.16-20

Aging research is challenging because of the incredible variability in humans. Mice offer an attractive alternative, because genetic and environmental variables can be controlled. Nevertheless, the time and expense required to generate and maintain mice >3 years of age is daunting. One approach to overcome this is to use mouse models of human progerias, or accelerated aging. Werner syndrome, Cockayne syndrome, XFE progeroid syndrome, and trichothiodystrophy are examples of human diseases of accelerated aging. 21 A common thread linking these disorders is that they all result from mutations in genes involved in genome maintenance. This strongly suggests that DNA damage can promote age-related degenerative changes.

Excision repair cross-complementation group 1/xeroderma pigmentosum group F (ERCC1-XPF) is a structure-specific endonuclease involved in the repair of several types of DNA lesions, including interstrand cross-links, double-strand breaks, and bulky, helix-distorting lesions. 22, 23 ERCC1 and XPF are obligate binding partners and stabilize one another in vivo.24 Mutations in XPF that cause reduced expression of ERCC1-XPF cause a progeroid syndrome in humans, characterized by dramatically accelerated aging of most organ systems.24Ercc1−/− and Xpf−/− mice have an identical phenotype, and mimic the human progeroid syndrome. 25, 26 These mice die in the 4th week of life with aging-like degenerative changes, including osteoporosis, neurodegeneration, bone marrow hypoplasia, epidermal atrophy, sarcopenia, and liver and kidney dysfunction.24, 27 Liver dysfunction is life limiting in Ercc1−/− mice.28 Mice with reduced, but not ablated, ERCC1-XPF expression (Ercc1−/Δ) live longer than Ercc1−/− mice (30- versus 4-week maxiumum lifespan)29, 30 The Ercc1−/Δ mice are healthy into adulthood (8 weeks), then begin to show numerous progressive symptoms associated with aging.30 The aim of this study was to systematically compare the liver of 4-5-month-old Ercc1−/Δ mice to that of old WT mice (i.e., 2+ years old) to determine whether the progeroid mice would offer an accurate model of natural aging that could be used to accelerate research on the biology of aging-related liver changes and therapeutic interventions to extend healthspan.


ALT, alanine aminotransferase; AST, aspartate transaminase; CD, cluster of differentiation; CDC, Centers for Disease Control and Prevention; DAPI, 4′,6-diamidino-2-phenylindole; ECM, extracellular matrix; ERCC1, excision repair cross-complementation group 1; GH, growth hormone; H&E, hematoxylin and eosin; IGF-1, insulin-like growth factor 1; IHC, immunohistochemistry; LFTs, liver function tests; LPO, lipid peroxidation; LSECs, liver sinusoidal endothelial cells; PBS, phosphate-buffered saline; PH, partial hepatectomy; qPCR, quantitative polymerase chain reaction; SA β-gal, senescence-associated β-galactosidase; SEM, scanning electron microscopy; α-SMA, alpha smooth muscle actin; TEM, transmission electron microscopy; WHO, World Health Organization; WT, wild-type; XPF, xeroderma pigmentosum group F.

Materials and Methods


Animal Care and Experimentation.

Experiments involving mice were approved by the University of Pittsburgh (Pittsburgh, PA) Institutional Animal Care and Use Committee and in accord with the National Institutes of Health guidelines for the humane care of animals. Ercc1−/Δ mice were bred and genotyped as previously described. 23 All mice used in this study were in an f1 mixed genetic background (FVB/n:C57Bl/6).

Histological Analyses.

6-7- and 20-24-week-old wild-type (WT) and Ercc1−/Δ mice, along with aged (26-36-month-old) WT mice, were sacrificed by CO2 inhalation, and excised livers were fixed with 10% formalin and embedded in paraffin. Tissue sections (6 μm) were cut and stained with hematoxylin and eosin (H&E) to detect changes in liver architecture and Masson's trichrome to detect fibrosis, using standard procedures. Alternatively, liver specimens were cryopreserved by fixation in 2% paraformaldehyde for 2-4 hours, followed by incubation in 30% sucrose at 4°C overnight. Tissues were embedded in OCT and flash-frozen before sectioning on a Microm cryostat (Thermo Fisher Scientific, Waltham, MA).

Liver Perfusion and Processing for Ultrastructural Analysis.

For scanning (SEM) and transmission electron microscopy (TEM), the liver of euthanized animals was cleared of blood by perfusion with phosphate-buffered saline (PBS) at 3 mL/min through the inferior vena cava, as previously described. 31 Samples of liver were fixed with glutaraldehyde and processed as described previously.32 TEM images were captured with a JEOL JEM 1011 microscope, and SEM images were taken with the JEOL JSM 6330F microscope (JEOL USA, Inc., Peabody, MA). Porosity was determined using MetaMorph imaging software (Molecular Devices, Sunnyvale, CA). Fenestrations were manually predetermined, and porosity was expressed as percent of measured total sinusoid area.

Immunofluorescence Analyses.

Staining was performed on deparaffinized liver sections from 20-week-old WT, Ercc1−/Δ mice, and aged (24-36-month-old) WT mice. Paraffin-embedded tissue sections were subjected to heat-induced epitope retrieval by incubation in sodium citrate buffer (10 mM, pH 6.0) for 10 minutes in a microwave, followed by a 30-minute cool down. Sections were permeabilized with PBS-Tween (0.05%) and incubated with rabbit polyclonal to acrolein (1:500, ab37110; Abcam, Cambridge, MA), cluster of differentiation (CD)31 (1:500, 550274; BD Pharmingen, San Diego, CA), F4/80 (1:500, 552958; BD Pharmingen), alpha smooth muscle actin (α-SMA) (1:500, #6198; Sigma, St. Louis, MO), and desmin (1:250, ab15200; Abcam) overnight at 4°C. The secondary antibody, Alexa Fluor 594 goat anti-rabbit immunoglobulin G (1:1000, A11012; Molecular Probes, Eugene, OR), was allowed to incubate for 60 minutes in the dark. Nuclei were stained with Vectashield with 4′,6-diamidino-2-phenylindole (DAPI) mounting medium (Vector Laboratories, Inc., Burlingame, CA). Ten random images per sample were acquired using an Olympus BX51 fluorescent microscope (Olympus, Center Valley, PA, USA). Acrolein fluorescence intensity was quantified using Axio Vision imaging software (Carl Zeiss, Gottingen, Germany). Average fluorescence intensity is shown ± standard error of the mean. All P values reported were derived using a standard Student's t test, with a two-tailed distribution. Differences were considered significant at the 95% confidence interval (P < 0.05).

Immunohistochemical Analyses.

Immunohistochemistry (IHC) for Ki67 (rat anti-mouse Ki67 (TEC-3; Dako Cytomation, Carpinteria, CA) was performed on deparaffinized liver sections from 10-week-old WT and Ercc1−/Δ mice that had undergone partial hepatectomy (PH) surgery. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide. Tissue sections were subjected to heat-induced epitope retrieval by incubation in sodium citrate buffer (10 mM, pH 6.0) for 30 minutes in a decloaker, followed by a 30-minute cool down. Blocking was done with 5% rabbit serum for 20 minutes. Primary antibody was applied for 1 hour. Secondary antibody was applied for 30 minutes, followed by the label antibody (ABC Elite; Vector Laboratories) for 30 minutes. Then, 3,3′-diaminobenzidine chromagen (Dako) was applied for 6 minutes, followed by two rinse steps in distilled water. Hematoxylin was used as a counterstain. Bright field images were collected using an Olympus BX51 fluorescent microscope, and Ki67-positive cells were quantified from five random fields of view from each of 4 mice per group. Mean percentages were plotted ± standard error of the mean (P < 0.05, as calculated with the Student's t test with a two-tailed distribution).

Detection of Lipids.

Slides with frozen liver tissue were warmed to room temperature, fixed in 2% paraformaldehyde for 10 minutes, rinsed with PBS, and incubated with LipidTox Red Neutral Lipid Stain (Invitrogen, Carlsbad, CA) for 30 minutes at room temperature. Tissues were rinsed in PBS, and nuclei were counterstained with Hoechst dye for 1 minute. Hoechst dye was rinsed off with three PBS washes, and tissue was covered using gelvatol mounting medium. Image acquisition was done on an Olympus BX51 fluorescent microscope.

See Supporting Information for additional experimental procedures.


Histopathological Changes in Progeroid Ercc1−/Δ Mouse Liver Mimic Normal Aging.

To investigate structural changes in the liver of progeroid Ercc1−/Δ mice, we compared liver tissue of 20-24-week-old Ercc1−/Δ mice to WT littermates and to 26-34-month-old WT mice. The relative liver/body weight of Ercc1−/Δ mice was not significantly different from littermate controls at any age. 30 The architecture of the Ercc1−/Δ liver was irregular, with mild to moderate variation in lobular size (Fig. 1A). Pathologic changes detected in both Ercc1−/Δ mice and old WT mice include multifocal degeneration, necrosis and neutrophilic inflammation, a diffuse increase in sinusoidal lining cells, and karyomegaly, with intranuclear inclusions (Fig. 1; Supporting Fig. 1). None of these changes were detected in young adult (20-24-week-old) WT mice. The only abnormalities detected in juvenile (7-week-old) Ercc1−/Δ mice were mild anisokaryosis and intranuclear inclusions. Liver from aged WT mice, like humans, displays mild portal fibrosis (Fig. 1B) and increased neutral lipid accumulation (Fig. 1C). Similar changes were detected in 20-24-week-old Ercc1−/Δ mice, but not age-matched WT mice or 7-week-old Ercc1−/Δ mice. This demonstrates that the Ercc1−/Δ mice undergo progressive degenerative processes, rather than having developmental liver abnormalities.

Figure 1.

Premature onset of age-related pathology in Ercc1−/Δ mouse liver. (A) H&E-stained liver sections from 7-week-, 23-week-, and 34-month-old WT mice and Ercc1−/Δ littermates (10×). Bottom right panel is a high-magnification (40×) image of Ercc1−/Δ liver (23-week-old), illustrating inflammatory infiltrate (black arrows), abnormal nuclei with inclusion (green arrows,) and dead hepatocytes (blue arrows). (B) Masson's trichrome–stained liver sections. Insets (40×) in the bottom right panel show perisinusoidal fibrosis in Ercc1−/Δ and old WT liver samples. (C) LipidTOX stain (Invitrogen, Carlsbad, CA) of liver sections to detect neutral lipids (20×). (D) Immunofluorescence detection of desmin and α-SMA, markers of hepatic stellate cells and their activation, respectively, in frozen liver sections (20×).

Stellate cells are pericytes that reside between hepatocytes and endothelial cells lining the sinusoids in the extracellular space of Disse. Stellate cells store fat and, when activated, produce extracellular matrix (ECM) proteins, contributing to fibrosis. With aging in humans, stellate cell number, but not activity, is reported to increase. 35 Similar results were found in 3-year-old WT mice and progeroid Ercc1−/Δ mice, as measured by desmin and SMA staining, which are markers of stellate cells and their activation, respectively (Fig. 1D). Kupffer cells, resident liver macrophages, are also reported to increase in number and phagocytic activity with aging in humans.36 However, we did not detect increased F4/80 immunostaining, a macrophage marker, in liver sections of old WT or progeroid mice (Supporting Fig. 2).

Pseudocapillarization and Defenestration in Progeroid Ercc1−/Δ and Old WT Mouse Liver.

Age-associated endothelial changes in the liver are prominent and contribute to age-related decline in liver function. 37 Expression of CD31, a marker of endothelium, was dramatically increased in 5-month-old Ercc1−/Δ mice and 2.5-year-old WT mice, compared to young adult WT mice (Fig. 2A), indicative of pseudocapillarization. Pseudocapillarization encompasses many alterations, including thickening of the sinusoidal endothelial cells (LSECs) and increased deposition of basement membrane proteins in the space of Disse.13, 38 TEM of livers from old and progeroid mice revealed a thickening of the basement membrane (Fig. 2B). This was confirmed by detection of the basement membrane protein, laminin, which revealed a dramatically increased level of laminin in old WT and 20-week-old Ercc1−/Δ mouse liver, compared to young adult WT mice (Fig. 2C).

Figure 2.

Age-related thickening of the sinusoidal endothelium in Ercc1−/Δ mouse liver. (A) Immunofluorescence detection of CD31, a surface marker of endothelial cells in frozen liver sections (20×). (B) TEMs (12,000×), of liver illustrating thickening of the basement membrane in sinusoids of mutant and old WT mice, compared to young WT. Insets show enlargement of a portion of the original image, and arrows indicate deposition of basement membrane and pseudocapillarization of the sinusoidal endothelium. (C) Immunoblotting to detect the ECM protein, laminin, in liver lysates from 20-week-old Ercc1−/Δ mice, control littermates, and old WT (2-3-year-old) mice (n = 3 per group). Densitometry was used for quantification, and averages are listed as a ratio compared to 20-week-old WT mice.

With advanced age, LSECs also lose their fenestrations, which is necessary for lipoprotein filtration, endocytosis, and immunological functions. 37 SEM of murine liver sections revealed a significant reduction in fenestration of the sinusoids in old (26-30-month-old) WT mice, compared to young (20-week-old) WT animals (Fig. 3A). Micrographs from 20-week-old Ercc1−/Δ mice looked more similar to those from the old WT mice than to their WT littermates, with significantly reduced fenestrations, decreased porosity, and loss of sieve plates (Fig. 3A,B).

Figure 3.

Premature defenestration in Ercc1−/Δ mouse liver. (A) SEM (10,000×) of liver, illustrating defenestration in sinusoids of mutant and old WT mice. Insets show enlargement of a portion of the original image. (B) Quantification of porosity shown in (A). Three SEM images per mouse (n = 3 mice per group) were quantified, and the percent open area of the sinusoid is represented ± standard error of the mean. Asterisks indicate significant differences. (C) Serum cholesterol and (D) triglyceride levels in WT and Ercc1−/Δ mice of various ages. Average ± standard error of the mean are plotted (n = 6 for 7-week-old mice; n = 9 for 21-week-old and old WT mice). Asterisks indicate significant differences, as calculated by a paired Student's t test.

Age-related LSEC defenestration can lead to impaired hepatic clearance of atherogenic lipoproteins, contributing to hypertriglyceridemia, hypercholesteremia, and vascular disease. 39, 40 Therefore, we measured serum cholesterol levels in the mice. Cholesterol levels were normal in young (7-week-old) Ercc1−/Δ mice (Fig. 3C). By 21 weeks of age, when defenestration was apparent, serum cholesterol was significantly elevated in Ercc1−/Δ mice, compared to WT littermates. In contrast, serum triglycerides were significantly lower in 21-week-old, compared to littermate controls (Fig. 3D). Serum lipoproteins were not significantly altered in old WT mice, compared to young WT mice.

Functional Changes in Progeroid Ercc1−/Δ Mouse Liver Mimic Normal Aging.

Under normal conditions, hepatocytes are quiescent cells, but can be stimulated to proliferate in response to parenchymal damage. 41 In an experimental setting, proliferation is induced by partial hepatectomy (PH), which leads to compensatory hyperplasia. Hepatocyte proliferation in response to PH is significantly decreased in aged rodents, compared to young animals.8 Thirty percent of the liver was surgically resected from 10-week-old WT and Ercc1−/Δ mice. Ercc1−/Δ mice did not survive a standard PH, which removes 70% of the liver. Thus, the surgery was modified to only remove 1 lobe or 30% of the liver. At 48 hours post-resection, mice were euthanized and liver tissue was collected for analysis. Previous experiments, in which WT mice were euthanized every 12 hours post-PH, from 12 to 72 hours, indicated that 48 hours was the time point at which proliferation was the highest in 10-week-old WT mice (data not shown). IHC staining for the proliferation-associated antigen, Ki67, revealed a significant decrease in the number of proliferating cells in the liver from Ercc1−/Δ mice, compared to WT littermates (Fig. 4A). Therefore, regenerative capacity of the liver is prematurely reduced in Ercc1−/Δ mice.

Figure 4.

Premature loss of liver function in Ercc1−/Δ mice. (A) Representative liver sections from 10-week-old WT and Ercc1−/Δ mice immunostained for the proliferation marker, Ki67, 48 hours post-PH. The fraction of proliferating hepatocytes is graphed (five random fields of view analyzed from n = 4 mice per genotype; *P < 0.05, Student's two-tailed t test). (B) Serum chemistries, reflecting liver function. Plotted are the average ± standard error of the mean for 6 mice per age and genotype. WT animals are graphed in red and Ercc1−/Δ mice in blue. Darker colors indicate increasing age. Asterisks indicate significant differences.

To further monitor liver function and integrity, liver function tests (LFTs) were performed on the plasma of Ercc1−/Δ and WT mice at multiple ages (Fig. 4B). Alanine aminotransferase (ALT) was significantly elevated in old WT mice, compared to young mice, as previously reported. 42 ALT was also significantly elevated in Ercc1−/Δ mice, compared to age-matched WT mice, even at 7 weeks of age. ALT levels more than doubled as the Ercc1−/Δ mice aged. Aspartate transaminase (AST) was also significantly elevated in old WT and 21-week-old Ercc1−/Δ mice, compared to young adult WT mice. Serum albumin is known to decrease with age in humans and rodents.43, 44 Albumin was significantly reduced in the plasma of 21-week-old, but not 7-week-old, Ercc1−/Δ mice (Fig. 4B). Thus, a very similar pattern of time-dependent changes in LFTs, indicative of chronic, progressive tissue degeneration and loss of organ function, are observed in Ercc1−/Δ and old WT mice. However, the changes are dramatically accelerated in the Ercc1−/Δ mice.

Increased Cellular Senescence in Ercc1−/Δ and Old WT Liver.

To further investigate the mechanism driving the loss of regenerative capacity in Ercc1−/Δ and old WT mice, we asked whether hepatocytes were undergoing cellular senescence. Liver sections were stained to detect senescence-associated β-galactosidase (SA-β-gal) activity, a marker of cellular senescence (Fig. 5A). SA-β-gal staining was increased in livers of 20-week-old Ercc1−/Δ and old WT mice, compared to 20-week-old WT mice. p16INK4a is another established marker of senescent cells. 45 p16INK4a expression was increased 1.4- and 2.0-fold in 20-week-old Ercc1−/Δ and 2-year-old WT mice, respectively, compared to young WT mice (Fig. 5B). Increased cell size, nuclear size, and the nuclear:cytoplasmic ratio are also associated with cell senescence and old age.46-48 The nuclear size of hepatocytes was heterogeneous in Ercc1−/Δ and old WT mice, compared to 7- and 23-week-old WT mice (Fig. 5C). Hepatocyte nuclei of Ercc1−/Δ mice were significantly larger than that of WT littermates at 7 weeks of age and increased by 21 weeks of age (Fig. 5C). Nuclear size was also significantly increased in old WT mice, compared to young WT mice. Electron micrographs revealed a highly irregular surface area of the hepatocyte nuclei from Ercc1−/Δ and old WT, but not young, WT mice (Fig. 5D).

Figure 5.

Premature senescence of Ercc1−/Δ mouse liver. (A) SA-β-gal histochemical stain on flash-frozen liver sections. (B) Immunoblotting detection of the senescence marker, p16INK4a, in liver lysates of a 20-week-old WT mouse, an Ercc1−/Δ littermate, and a 26-month-old WT mouse. Tubulin was used as a loading control to calculate the fold increase in p16 expression relative to 20 week-old WT mice. (C) DAPI staining of hepatocyte nuclei, illustrating increased nuclear size and heterogeneity in old WT (34-month-old) and adult Ercc1−/Δ mouse liver (1,000×). Quantification of nuclear size using Metamorph imaging software and images of DAPI-stained frozen liver sections. Plotted are the average nuclear areas ± standard error of the mean calculated from 10 random fields from 5 mice per group. Asterisks indicate significant differences. (D) TEMs of representative hepatocyte nuclei (12,000×).

Increased Oxidative Damage in Ercc1−/Δ Liver Is Also Found With Natural Aging.

Oxidative stress has been implicated in driving age-dependent cellular senescence. 49 To determine whether oxidative stress is elevated in Ercc1−/Δ mice and with old age, we measured lipid peroxidation (LPO) in liver. The level of lipid hydroperoxides, measured by enzyme-linked immunosorbence assay, was elevated in the livers of 20-week-old Ercc1−/Δ mice, compared to WT littermates (Fig. 6A). A similar elevation was seen in old WT mouse livers, compared to young WT. This increase in LPO was confirmed by the immunodetection of acrolein, a stable product of LPO. Ercc1−/Δ and old WT mice had significantly increased acrolein, compared to adult WT mice (Fig. 6B).

Figure 6.

Increased oxidative damage in livers of old and progeroid mice. (A) Detection of lipid hydroperoxides, a by-product of LPO, in liver lysates from 20-week-old WT and Ercc1−/Δ littermate mice versus old WT mice (26-34 months of age). Box plot represents the mean (in colored square), along with the 25th and 75th percentile for each group (n = 5 mice). Average lipid hydroperoxide levels for groups were as follows: 20-week-old WT (4.8 μM/g liver), 20-week-old Ercc1−/Δ (7.3 μM/g liver; P = 0.12, one-tailed Student's t test), and old WT (7.2 μM/g liver). (B) Immunofluorescence detection of acrolein, a by-product of LPO. Staining intensity was quantified using AxioVision imaging software. Plotted are the averages ± standard error of the mean determined from analysis of 10 random fields of liver sections from 2 mice per group. Asterisks indicate significant differences. (C) Representative images illustrating lipofuscin accumulation in 20-week-old WT and Ercc1−/Δ mice and 26-34-month-old WT mouse liver. AxioVision software was used for quantification of 10 random fields measured from 5 mice per group. Plotted is the average fluorescence intensity for each group ± standard error of the mean. Asterisks indicate significant differences.

One of the most common markers of the aged liver is the cytoplasmic accumulation of lipofuscin, which is highly oxidized lipid material. 11 Lipofuscin granules, which fluoresce at 488 nm, were 30-fold elevated in livers of 20-week-old Ercc1−/Δ mice, compared to WT littermates (Fig. 6C). A similar increase in lipofuscin was observed in old WT mouse liver, compared to young WT mice.

Transcriptional Changes in Ercc1−/Δ Mouse Liver Parallels Natural Aging.

Gene expression changes were measured by microarray in livers of Ercc1−/Δ mice and WT littermates at 5 and 16 weeks of age. Genes that were significantly up- or down-regulated, compared to WT littermates are shown in Fig. 7A). There was a progressive down-regulation of genes associated with the insulin-like growth factor 1/growth hormone (IGF-1/GH) axis and oxidative phosphorylation in Ercc1−/Δ mice as they age, compared to WT littermates. Additionally, genes involved in DNA damage, oxidative stress response, cell-cycle arrest, and apoptosis were significantly up-regulated in 16-week-old Ercc1−/Δ mice, compared to WT littermates. Quantitative polymerase chain reaction (qPCR) was used to confirm the microarray results (Fig. 7B).

Figure 7.

Age-associated transcriptional reprogramming in liver of Ercc1−/Δ mice. (A) Significant gene-expression changes (P < 0.05, two-tailed t test) in 5- and 16-week-old Ercc1−/Δ livers, as compared to littermate control animals. Note the progressive dampening in the expression of genes involved in oxidative metabolism and the GH/IGF-1 axis, along with the up-regulation in the expression of genes involved in cell-cycle arrest, DNA damage responses, oxidative stress, and detoxification. red color, up-regulated; green color, down-regulated, FC, fold change, wks, weeks. (B) qPCR measurement of messenger RNA levels of a subset of genes identified to be differentially expressed in the livers of 16-week-old Ercc1−/Δ animals, compared to wild-type littermates. For each gene, expression levels in the mutant tissue are plotted relative to those of controls (red dotted line). Error bars indicate standard error of the mean between replicates (n ≥ 4). (C) Spearman's r correlation, reflecting transcriptome similarities between the 5- or 16-week-old Ercc1−/Δ and the 130-week-old mouse livers, where −1.0 is a perfect negative (inverse) correlation, 0.0 is no correlation, and 1.0 is a perfect positive correlation.

The transcriptional changes observed in 16-week-old Ercc1−/Δ mouse liver were comparable to old WT (130-week-old) mice (Fig. 7C). Spearman's r correlation indicates that as Ercc1−/Δ progeroid mice age from 5 to 16 weeks, their gene expression profiles have increasing similarity to old WT mice.


The human liver undergoes numerous characteristic structural and functional changes with increasing age. 13 Structural changes include loss of organ volume, nuclear polyploidy, anisokaryosis, and pseudocapillarization.13, 42, 50 These changes lead to impaired liver function and loss of regenerative capacity. Not surprisingly, transplanted livers from late-age donors have decreased graft acceptance and function.51, 52 However, the underlying cause for these changes remains unknown.

Nonhuman primates have been used to study liver aging, but the commonly used species, Macaca mulatta and Macaca nemestrina, can live more than 20 years. 53, 54 Mice are shorter-lived, and, importantly, age-related changes characteristic of the aged human liver are recapitulated in mice, including pseudocapillarization, increased polyploidy, decreased hepatocyte number, and increased nuclear size.18-20, 55 Thus, mice represent an accurate model system for studying age-associated liver changes in humans. However, the 3-year lifespan of mice still represents a substantial barrier to rapid testing of hypotheses about the causes and treatments of liver aging.

Herein, we establish that the structural, functional, and regulatory changes that occur in the liver of WT mice in their 3rd year of life are recapitulated in the Ercc1−/Δ progeroid mouse strain within their 7-month lifespan. There is a highly significant correlation between the genome-wide expression profile of 4-month-old Ercc1−/Δ mice and 32-month-old WT mice (Fig. 7). Both have elevated LFTs and reduced regenerative capacity, as well as focal necrosis, inflammation, anisokaryosis, and steatosis (Figs. 1 and 4). Markers of cellular senescence (e.g., SA-β-gal and p16) are elevated in livers of progeroid and old mice (Fig. 5). This is supported by the strong up-regulation of genes associated with cell-cycle arrest (Fig. 7), and the observation that livers of Ercc1−/Δ mice, like that of old WT mice, 56-58 have reduced regenerative capacity after PH (Fig. 4). Collectively, these data support the conclusion that age-related decrease of liver function is caused by the dysfunction of hepatocytes, rather than their attrition. Interestingly, there is accumulation of oxidation products (e.g., lipofuscin and LPO) in the livers of progeroid and old mice (Fig. 6), suggesting that oxidative damage may drive hepatocyte dysfunction.

Defenestration and pseudocapillarization, which were significantly elevated in livers of Ercc1−/Δ and old WT mice (Figs. 2 and 3), contribute to reduced regenerative capacity. 58 Defenestration is also implicated in dyslipidemia resulting from impaired clearance of lipoproteins from the blood to hepatocytes.39 Serum cholesterol was significantly elevated in progeroid Ercc1−/Δ mice, compared to littermate controls (Fig. 3). In contrast, serum triglycerides were significantly lower in mutant animals, compared to WT mice. Despite the fact that the extent of defenestration was similar in progeroid and old WT mice, serum lipid levels were within normal range in the naturally aged mice. Hence, the dyslipidemia in the Ercc1−/Δ mice may reflect metabolic changes in the mice, rather than a loss of liver-transport mechanisms. Indeed, metabolic profiling of the Ercc1−/Δ mice revealed increased serum high-density lipoprotein, but decreased low-density lipoprotein and very-low-density lipoprotein and increased urinary ketone bodies.59 These changes mimic caloric restriction60 and occur as a consequence of genetic reprogramming, rather than diminished nutrient intake in the Ercc1−/Δ mice (Fig. 7).59, 61 The IGF-1/GH axis is also attenuated in Ercc1−/Δ mice and old WT mice (Fig. 7).24, 61 This drives the metabolic shift and also contributes to impaired liver regeneration.62

Ercc1−/Δ mice age rapidly as a consequence of an engineered mutation in the gene that encodes one subunit of the DNA repair endonuclease, ERCC1-XPF, 63 which is required for multiple DNA repair pathways.22, 23 XFE progeroid syndrome, which occurs in the ERCC1-deficient mouse model, is characterized by accelerated aging of the hematopoietic, hepatobiliary, endocrine, musculoskeletal, and neurological systems.24, 27 Werner syndrome, caused by mutations in WRN, which encodes a DNA repair helicase and exonuclease, is characterized by early-onset cancer, cardiovascular disease, and osteoporosis.64 Both Ercc1−/Δ and WrnΔhelhel mice display early onset of dyslipidemia, steatosis, pseudocapillarization, and defenestration in the liver.65 This demonstrates that DNA repair is critical for protecting against loss of liver homeostasis and function. Importantly, levels of 8-oxo-deoxyguanine, an endogenous oxidative DNA lesion, are greater in old WT mice than young mice,66-68 consistent with the hypothesis that unrepaired DNA damage may contribute to liver aging, even in repair-proficient organisms.

Mice in this study were not exposed to environmental genotoxins. Thus, the degenerative changes observed are attributed to the accumulation of oxidative damage arising as a consequence of normal metabolism. The fact that the onset of degenerative changes are accelerated in DNA repair–deficient Ercc1−/Δ mice, compared to WT, reveals that specifically endogenous DNA damage can drive liver degeneration, likely through driving hepatocyte senescence. 69 In humans, environmental exposures as well as endogenously produced genotoxic by-products undoubtedly contribute to the accumulating burden of DNA damage and thereby age-related liver dysfunction.

It is important to note that many of the age-related endpoints measured were more severe in Ercc1−/Δ mice than in WT mice near the end of their respective lifespans. This includes LFTs, fibrosis, steatosis, pseudocapillarization, and anisokaryosis. A unifying explanation for why the liver of 4-5-month-old Ercc1−/Δ mice looks worse than 2-3-year-old WT mice is that the Ercc1−/Δ mice undergo activation of stress-response pathways, similar to that induced by caloric restriction, which are protective, as the result of accumulated, unrepaired DNA damage (Fig. 7). 24, 61 Thus, the Ercc1−/Δ mice survive longer than expected, given their burden of damage, and, therefore, have extreme aging, comparable to that of supercentenarians.30 Nevertheless, Ercc1−/Δ mice develop the majority of structural and funtional characteristics of old human13 and mouse liver within a matter of 5 months, rather than years. The vast majority of these changes occur in adulthood (after 7 weeks of age), making the changes distinct from developmental abnormalities and degenerative in nature. Ercc1−/Δ mice, therefore, represent an accurate model system for rapidly testing hypotheses about the mechanism underlying age-related liver degeneration and pharmacological interventions aimed at delaying or ameliorating age-associated liver disease.


The authors thank Chelsea Feldman, Cheryl Clauson, and members of the Niedernhofer lab for experimental support and critical reading of the manuscript.