Sustained telomere length in hepatocytes and cholangiocytes with increasing age in normal liver


  • Potential conflict of interest: Nothing to report.


Telomeres, a validated biomarker of aging, comprise multiple nucleotide repeats capping chromosomes that shorten with each cell cycle until a critical length is achieved, precipitating cell senescence. Only two previous studies focused on the effect of aging in “normal” liver tissue, but these studies were compromised by small sample size, limited age range, tissue derived from individuals with an increased risk of senescence, and the use of liver homogenates. We developed a robust large-volume, four-color quantitative fluorescent in situ hybridization technique to measure telomere length in large numbers of hepatocytes, Kupffer cells, hepatic stellate cells, CD4-positive and CD8-positive lymphocytes, and cholangiocytes. Following validation against the gold standard (Southern blotting), the technique was applied to normal archived paraffin-embedded liver tissue obtained following reperfusion of implanted donor liver. We studied 73 highly selected donors aged 5-79 years with a short medical illness preceding death and no history of liver disease, reperfusion injury, or steatosis and normal graft function 1-year posttransplantation. Cholangiocytes had significantly longer telomeres compared with all other intrahepatic lineages over a wide age range (P < 0.05). Age-related telomere attrition was restricted to sinusoidal cells (i.e., Kupffer cells [P = 0.0054] and stellate cells [P = 0.0001]). Cholangiocytes and hepatocytes showed no age-related telomere shortening. Conclusion: In normal liver and over a broad age range, cholangiocytes have longer telomeres than all other intrahepatic lineages. Age-related telomere length decline is restricted to Kupffer cells and stellate cells. (HEPATOLOGY 2012)

As the median age of populations increases worldwide, so too does that of patients presenting with liver disease. There are many structural and functional changes in all organs with increasing age, including the liver.1, 2 Strong evidence links increased age with impaired liver regeneration and increased risk of fibrosis, hepatocellular carcinoma and death.

Age is an important adverse factor in chronic liver disorders including chronic hepatitis C virus infection,3, 4 non–alcohol-related fatty liver disease,5 primary biliary cirrhosis, alcohol-related hepatitis and hemochromatosis.1 Donor age has an enormous impact on graft survival following liver transplantation.6 The risk of death with liver disease is substantially higher in older patients.7 Impaired liver regeneration with increasing age is demonstrated clinically in acute liver failure8 and acute hepatitis A virus infection9, 10 and rodent models of partial hepatectomy, where restoration of liver mass is slower in older animals.

Telomeres are repetitive non-coding DNA elements (TTAGGG) at either end of the chromosome that shorten with each cell division.11 Mounting in vitro and in vivo evidence suggests that progressive loss of telomeres is an important component of aging.12-14 Telomere shortening eventually reaches a critical point that triggers replicative senescence (irreversible growth arrest). There is a direct correlation between telomere length, the proliferative capacity of somatic cells and aging in normal healthy individuals.15, 16 Telomere length is a validated biomarker of aging.17-20

Real-time polymerase chain reaction (PCR) is the gold standard for measuring telomere length, but using this technique for liver homogenates has limitations, because distinct intrahepatic cell lineages cannot be analyzed separately. Quantitative fluorescence in situ hybridization (Q-FISH) is a reliable indirect measure of telomere length.21, 22 Studies in diseased liver have revealed significant reductions in telomere length in small series of patients with cirrhosis or hepatocellular carcinoma, where small numbers of cells were analyzed.23, 24 Only two studies25, 26 examined the relation between age and telomere length in “healthy” liver. These were limited by small sample size, limited age range, and the use of tissue derived from individuals with an increased risk of senescence. Furthermore, only 64% of cells in liver tissue are hepatocytes,27 hence analysis of telomere length in whole liver homogenates is unlikely to reflect hepatocyte telomere length. The effect of aging on other intrahepatic lineages is unknown.

Our study is the first to examine the effect of aging in normal liver, distinguishing between each intrahepatic lineage, using a large volume Q-FISH in situ approach and archival liver.


DAPI, 4′,6-diamidino-2-phenylindole; PCR, polymerase chain reaction; Q-FISH, quantitative fluorescent in situ hybridization; TBS, Tris-buffered saline.

Materials and Methods


The Norfolk and Norwich Research Ethics Committee approved the use of archived liver tissue.

Selection of Liver Tissue.

Finding normal liver tissue for studies across a wide age range is problematic. Liver biopsy is not performed in healthy individuals, and it would be unethical to subject healthy controls to liver biopsy for research. In other circumstances, investigators elect to use liver obtained at resection for hepatic metastases, particularly colorectal malignancy, using tissue distant from the tumor that appears normal microscopically. However, colorectal malignancy and hepatocellular carcinoma arise with increasing frequency with increased age and are associated with telomere shortening28-30; malignancy generally arises more often in accelerated aging or senescence. It is improbable that liver tissue could be obtained readily across a wide age range in this context.

Based on the premise that liver donors by their nature are “unselected” and often present following trauma or disease unrelated to aging, intraoperative liver biopsies from implanted donor livers taken immediately after reperfusion were studied (time-zero liver biopsies).

It is unlikely such tissue would express features of accelerated senescence because of the process of organ retrieval or transplantation, but because of theoretical concerns that the duration of warm and cold ischemia might influence subsequent findings, samples studied were selected from a large cohort with careful review to minimise that risk.

Liver tissue was obtained from an archive of paraffin-embedded time-zero liver biopsies stored in the Department of Pathology, Addenbrooke's Hospital, Cambridge, UK. All liver tissue had been fixed in 10% neutral formalin and subjected to standard processing and paraffin-embedding. Tissue was reviewed by a single histopathologist for features of graft injury including steatosis, reperfusion injury and preexisting liver disease.

Time-zero liver samples and subject data were reviewed and sections of adequate size were chosen to reflect normality according to the following criteria: no history of liver or senescence-related disease; a short medical illness preceding death (intracranial hemorrhage in 65%, trauma in 26%); no or minimal reperfusion injury; no steatosis; and normal recipient posttransplantation liver function at 1 year.

Liver sections from 73 subjects aged 5-79 years were selected from over 1,000 cases. Mean cold ischemic time was 675 minutes (SD 155). (Table 1: subject demographics).

Table 1. Demographic Data for Normal Time-Zero Liver Biopsies
Age, YearsSexCold Ischemic Time (Minutes)Cause of Death
  1. Abbreviations: F, female; M, male; RTA, road traffic accident.

15F708Hypoxic brain damage
20M564Intracranial hemorrhage
20M561Drug overdose
21F531Intracranial hemorrhage and brain tumor
26M730Other trauma; accident
28M433Intracranial hemorrhage
29F644Intracranial hemorrhage
29F952Hypoxic brain damage
30F585Intracranial hemorrhage
30M787Intracranial hemorrhage
31M689Intracranial hemorrhage
33F195Intracranial hemorrhage
35F613Trauma RTA; motorbike
37F778Intracranial hemorrhage
37F737Intracranial hemorrhage
37F925Other trauma; accident
38FUncertainIntracranial hemorrhage
38M643Intracranial hemorrhage
40M818Intracranial hemorrhage
41M478Intracranial hemorrhage
42M804Intracranial hemorrhage
43F637Intracranial hemorrhage
45F535Intracranial hemorrhage
45M557Intracranial hemorrhage
47M485Intracranial hemorrhage
48F785Intracranial hemorrhage
48M700Intracranial hemorrhage
48M785Intracranial hemorrhage
49M640Intracranial hemorrhage
50F641Intracranial hemorrhage
51F820Intracranial hemorrhage
51F604Intracranial hemorrhage
52F425Intracranial hemorrhage
53M647Intracranial hemorrhage
54F801Intracranial hemorrhage
54F647Intracranial hemorrhage
55M793Intracranial hemorrhage
58F1104Intracranial hemorrhage
59F721Intracranial hemorrhage
60F473Intracranial hemorrhage
60M685Intracranial hemorrhage
60M675Intracranial hemorrhage
63F670Intracranial hemorrhage
63M741Intracranial hemorrhage
65F621Intracranial hemorrhage
65F498Intracranial hemorrhage
66M679Intracranial hemorrhage
67M648Hypoxic brain damage
67M648Intracranial hemorrhage
69F789Intracranial hemorrhage and brain tumor
74F851Intracranial hemorrhage
75F750Intracranial hemorrhage
76M767Intracranial hemorrhage
77F552Intracranial hemorrhage
79F719Intracranial hemorrhage

To determine whether selection criteria for using time-zero liver were valid, archived liver tissue from patients with hyperoxalosis (n = 5) removed at combined liver and kidney transplantation was studied and compared with age-matched time-zero samples. These livers were processed immediately and were not subjected to ischemic insult prior to processing.


Six serial 10-μm sections exceeding 1.5 cm in length were cut to stain the major intrahepatic cell lineages. Paraffin sections were deparaffinized with xylene, hydrated through graded ethanol and placed in deionized water. Slides were boiled at 97°C in sodium citrate buffer (pH 6.0) for 30 minutes to enhance target retrieval. Following cooling at room temperature for 20 minutes, slides were transferred into phosphate buffered saline for 5 minutes before fixation in 4% formaldehyde for 5 minutes at room temperature. Enzymatic unmasking was achieved with porcine pepsin solution containing 1 mg/mL pepsin (Sigma, Gillingham, UK) in a 0.84% hydrochloric acid solution (pH 2.0) for 10 minutes at 37°C. Slides were rinsed in deionized water, and 80 μL hybridization mix was added (2.5 μL of 25 μg/mL PNA Cy5-labeled telomere-specific probe [TelC Cy5-oo-(CCCTAA)3 PNA probe with >95% purity; Cambridge Research Biochemicals, Billigham, UK] with 1.5 μL 1 M Tris-Cl [pH 7.2]/10.7 μL MgCl2 [25 mM MgCl2/9 mM citric acid/8.2 mM NaH2PO4 (pH 7.4)/87.5 μL deionized formamide; Sigma, Gillingham, UK]/6.2 μL 10% [wt/wt] blocking reagent [Roche, Welwyn Garden City, UK] /16.6 μL deionized water).

Hybridization was performed at room temperature for 2 hours in the dark after denaturation at 80°C. Slides were washed twice with wash solution A (70 mL formamide/mL 1 M Tris-Cl [pH 7.2]/mL 10% bovine serum albumin stock solution/28 mL deionized water) and three times with wash solution B (15 mL 1 M Tris-Cl [pH 7.2]/15 mL 1.5 M NaCl/120 μL Tween 20 [0.08% final]/120 mL deionized water). Slides were then washed in Tris-buffered saline (TBS) and incubated at room temperature for 45 minutes with appropriate murine monoclonal antibody to distinguish intrahepatic cell lineages: 90 μL primary antibody (Hepar-1 1:100 [DAKO clone OCH1E5] for hepatocytes; CD68 1:25 [DAKO clone PG-M1] for Kupffer cells; smooth muscle actin 1:25 [DAKO clone 1A4] for hepatic stellate cells; CD4 1:25 and CD8 1:25 for T cell subsets; or cytokeratin-19 1:100 [DAKO clone RCK108] for bile duct cells). Each was diluted in 1% goat serum/TBS with 4′,6-diamidino-2-phenylindole (DAPI) 1:500 to identify cell nuclei (Sigma, Gillingham, UK). All antibodies were titrated to optimum concentration. Three 5-minute washes with TBS/Tween were followed by incubation with 90 μL Alexa Fluor 488 donkey anti-mouse immunoglobulin G (H+L) (Invitrogen, Paisley, UK; 1:100 in 1% goat serum/TBS with DAPI 1:500) for 30 minutes at room temperature. Slides underwent three further 5-minute washes with TBS/Tween before dehydration using graded ethanols followed by air-drying at room temperature for 20 minutes. Finally, sections were mounted, and a cover slip was applied with neat fluorescent mounting media (DAKO, Ely, UK).

A control sample (hyperoxalosis) was included with each run to ensure uniformity and reproducibility.

Telomere Length Measurement by Real-Time PCR.

Genomic DNA was extracted from liver tissue using a QIAamp DNA mini kit (Qiagen, Crawley, UK). DNA concentration was adjusted to 5 ng/μL in H2O. Telomere length was measured on an iCycler real-time PCR system (Bio-Rad Laboratories, Hercules, CA). Telomere PCR reaction conditions were 3 minutes at 95°C followed by 40 cycles of 15 seconds at 95°C and 1 minute at 54°C, with 100 nm Tel-A primer and 300 nm Tel-B primer. glyceraldehyde 3-phosphate dehydrogenase PCR was started with 3 minutes at 95°C followed by 40 cycles of 15 seconds at 95°C and 30 seconds at 58°C; primer concentrations were 100 nm for GA81L and 200 nm for GA81R. Telomere PCRs included 100 nM primer Tel A (5′-CGGTTTGTTTGGGTTTGGGTTTGGGTT TGGGTTTGGGTT-3′), 300 nm primer Tel B: (5′-GGCTTGCTTACCCTTACCCTTACCCTTACCCT TACCCT-3′), 10 ng genomic DNA, 0.1 M SYBR green (Sigma-Aldrich Co.) and 1 M Platinum Quantitative PCR Supermix-UDG (Invitrogen) in a 30-μL reaction. Reactions were performed in quadruplicate in 96-well plates. Each plate included four DNA quantity standards (serial dilutions of a reference DNA sample giving final DNA quantities of between 30 and 1.87 ng per reaction), one negative control, and three internal controls represented by three samples of genomic DNA with known telomere lengths (3, 5.5, and 9.5 kb).


DAPI, a blue fluorescent stain that binds strongly to A-T–rich regions in DNA, counterstained nuclei (Fig. 1). Antibodies to Hepar-1, cytokeratin-19, CD4, CD8, CD68, and smooth muscle actin, in conjunction with Cy3-conjugated donkey anti-mouse antibody identified cytoplasm of hepatocytes, cholangiocytes, intrahepatic lymphocyte subsets, Kupffer cells, and stellate cells, respectively (Fig. 1). Donkey antibody was chosen to prevent cross-reactivity with human tissue.

Figure 1.

Images of different intrahepatic cell lineages stained with Q-FISH. The nuclei of all cell lineages were stained with DAPI (blue); telomeres were stained with a Cy5 PNA probe (pink); cytoplasm was stained indirectly with Cy3 (green) to distinguish (A) Kupffer cells (CD68-positive), (B) CD4+ lymphocytes (CD4-positive), (C) cholangiocytes (CK-19 positive), (D) hepatocytes (Hepar-1–positive), (E) CD8+ lymphocytes (CD8-positive), and (F) stellate cells (smooth muscle actin–positive).

Fluorescent Microscopy Image Acquisition.

Image acquisition was performed with an Olympus ScanR high content screening fluorescence microscope with a charge-coupled device camera (Hamamatsu) equipped with a triple bandpass Semrock filter at DAPI-1160A/Cy3-4040B/Cy5-4040A. The starting Z position for autofocus was set and a Z stack of images at six levels separated by 1.67 μm were acquired at ×60 oil magnification, using a Plan Apo N 1.42 lens, through each section and compressed to give a Z composite projection. Six levels for image acquisition were chosen to detect the maximum number of telomeres per nucleus. Ultraviolet light source exposure time and illumination intensity were optimized to prevent saturation for each fluorochrome; these settings were maintained to permit valid direct comparison. Adopting this approach maximized detection, separation, and resolution of telomeres within the cell lineage of interest. Images were acquired using Cy5 Z maximum fluorescence intensity and DAPI Z maximum fluorescence intensity, as these provided optimum detection and resolution of telomeres and nuclei, respectively. Maximum telomere pixel intensity for each telomere in each cell was analyzed with ScanR analysis software. A control sample, to monitor reproducibility and uniformity, had similar mean maximum Cy5 intensity in every run, confirming reproducibility.

Three hundred images per biopsy were collected for hepatocytes and Kupffer cells. Given the relative paucity of cholangiocytes, lymphocytes, and stellate cells, 495 images per biopsy were acquired. Images were obtained over the entire length of liver samples to reduce anomalies. Location-dependent heterogeneity is described for cholangiocytes and hepatocytes, so analysis was restricted to cholangiocytes lining interlobular or larger intrahepatic bile ducts. Comparison of hepatocyte telomere length in periportal and centrilobular zones was similar (Supporting Fig. 1).

Image Analysis.

Definitions were established for each cell lineage to enable accurate automated detection of nuclei and telomeres, including initial optimization of image processing with background correction filters set at 300 for DAPI, 200 for Cy3, and 2 for Cy5 to prevent artifact. Settings were maintained for analysis of data for each cell lineage.

Cells with a Haywood circularity factor of 1 (perfect circle) to 1.5 were identified for analysis, excluding clumped nuclei and those cut tangentially. Using DAPI Z maximum fluorescence intensity and threshold object recognition, outlines of individual nuclei within this population were identified via application of maximum and minimum thresholds and a watershed algorithm, enabling clear separation of neighboring nuclei along the intensity saddle between local maxima; these settings were uniform for each cell lineage. The software ignored partial nuclei at the borders of each image.

Cy3 staining was used to identify further the nuclei of cell lineages of interest. Using the ScanR analysis software, a mask was incorporated that defined Cy3-positive cells as those where cytoplasmic staining overlapped −5 pixels inside the DAPI-stained nucleus and extended +15 pixels outside the nucleus. Data were displayed in histograms, so that cells were gated on the basis of intensity of Cy3 staining, preventing incorrect inclusion of Cy3-negative cells adjacent to Cy3- positive cells. This gate was maintained for each cell lineage.

To identify telomeres in Cy3-positive cell nuclei, Cy5 Z maximum fluorescence intensity was used with application of threshold object recognition and a watershed algorithm to outline individual telomeres, enabling maximum separation. These were depicted in histograms. Data obtained for Cy3-positive nuclei and the subobject of individual telomeres in each nucleus were exported into Microsoft Excel spreadsheets.

Telomere area was measured as the mean telomere area in pixels2 per nucleus. Nuclear area and nuclear density for each cell line were measured by the mean DAPI staining pixels2 for each nucleus and mean DAPI Z maximum fluorescence intensity, respectively.

Statistical Analysis.

Statistical analysis was performed on GRAPH PAD Prism5 software (Graph Pad, San Diego, CA) and SPSS using a linear regression test as data demonstrated a normal distribution (Supporting Fig. 2). A P value of ≤0.05 was considered significant. Analysis of variance and Dunn's multiple comparison tests were used to compare cell lineages.



Five hyperoxalosis liver explants and sixteen age-matched time-zero livers were compared to determine whether organ acquisition, storage, or cold/warm ischemia influenced telomere length. Explanted hyperoxalosis livers were chosen because they had normal histology and were processed immediately with the shortest possible ischemia time. There was no significant difference in telomere length or telomere number measured by Q-FISH for any cell lineage (Fig. 2). There was no evidence that cold or warm ischemia influenced telomere length.

Figure 2.

Validation of the use of time zero liver biopsies as normal liver tissue. Liver tissue from five patients with hyperoxalosis (without cold or warm ischemia) was compared with liver tissue from 16 age- and sex-matched time-zero liver biopsies using Q-FISH. Telomere length measured as mean telomere Cy5 intensity was similar in each intrahepatic cell lineage in hyperoxalosis and time-zero biopsy.

Q-FISH was validated against real-time PCR (the gold standard measure of telomere length) (Fig. 3). Fresh tissue from eight patients with liver disease obtained at liver resection or transplantation was analyzed in tandem using Q-FISH and real-time PCR. Cy5 staining of telomeres using Q-FISH (Fig. 1) provided data on telomere number, area, and intensity. Telomere intensity for all nuclei in liver tissue (not separated by cell lineage) correlated most closely with telomere length in liver homogenate measured by real-time PCR (R2 = 0.659, P = 0.015). Mean Cy5 intensity was therefore used in subsequent experiments as a measure of telomere length.

Figure 3.

Comparison of telomere length by Q-FISH method and real-time PCR. (A) Mean intensity Cy5 telomere staining for hepatocytes alone. (B) Mean intensity Cy5 telomere staining for all intrahepatic cell lineages. (C) Telomere area per nucleus for all cell lineages. (D) Number of telomeres per nucleus for all cell lineages.

There was no correlation between hepatocyte telomere length analyzed by Q-FISH and whole liver tissue homogenate analyzed by real-time PCR (R2 = 0.010, P = 0.800). There was, however, close correlation between whole liver telomere length analyzed by Q-FISH and real-time PCR (R2 = 0.659, P = 0.015), suggesting not all intrahepatic cell lineages have similar telomere length. Studies of liver aging or senescence using tissue homogenates may be misleading.

Telomere Length.

Telomere length was analyzed in large numbers of cells (Table 2, Fig. 4) from each intrahepatic cell lineage. A relation between increasing age and reduced telomere length was detected in just two cell lineages, both sinusoidal: hepatic stellate cells (R2 = 0.2613, P = <0.0001) and Kupffer cells (R2 = 0.1039, P = 0.0054). In contrast, there was no relation between telomere length and age in hepatocytes (R2 = 0.03756, P = 0.1004), cholangiocytes (R2 = 0.01164, P = 0.3637), or either T cell subset (R2 = 0.02724, P = 0.1629 [CD4 lymphocytes] and R2 = 0.05092, P = 0.0954 [CD8 lymphocytes]).

Table 2. Telomere and Nuclear Staining with Cy5 and DAPI, Respectively, in Normal Liver, According to Cell Type
Cell LineageNo. of Cells AnalyzedNuclear Area in Pixels2Nuclear Staining IntensityNo. of Telomeres Detected per NucleusTelomere Area (Pixels2) Detected per Nucleus
  • Results are expressed as the mean (SD).

  • *

    The number of telomeres detected in hepatocytes and cholangiocytes was larger than in other intrahepatic cell types.

Hepatocytes40,000 (9123)4257 (525)2019 (569)15.4 (3.0)*11.1 (1.3)
Cholangiocytes5000 (988)2961 (408)2561 (638)15.3 (3.3)*13.4 (3.0)
CD4+ lymphocytes686 (162)2911 (629)2339 (640)13.2 (2.9)11.2 (1.5)
CD8+ lymphocytes453 (139)2729 (500)2148 (606)13.7 (2.8)11.5 (1.7)
Hepatic stellate cells942 (157)2959 (689)2017 (641)12.0 (3.0)11.2 (1.6)
Kupffer cells30,000 (6893)3893 (538)1681 (463)13.5 (2.8)10.7 (1.4)
Figure 4.

Age and telomere length (measured as the mean intensity of Cy5 telomere staining) in healthy liver according to intrahepatic cellular lineage.

There was a striking difference between cholangiocyte telomere length, which was longer, when compared with other intrahepatic cell lineages (Dunn's multiple comparison test for the difference in rank sum for hepatocytes versus cholangiocytes was −235; for cholangiocytes versus Kupffer cells, −218; for cholangiocytes versus stellate cells, −169; and for cholangiocytes versus CD8 and CD4 lymphocytes, −245 and −226, respectively). All P values were <0.05 (Fig. 5).

Figure 5.

Variation in (A) median telomere number per nucleus, (B) mean nuclear area per nucleus, and (C) mean telomere length (Cy5 staining intensity) for each intrahepatic cell lineage.

Telomere Area.

Telomere area did not correlate with age in any cell lineage (Supporting Fig. 3). Mean telomere area per nucleus was higher in cholangiocytes compared with all other cell lineages (P < 0.05 using Dunn's multiple comparison test), probably reflecting longer cholangiocytes telomeres (see Table 2).

Telomere Number.

The number of telomeres detected per nucleus was higher in hepatocytes and cholangiocytes compared with other cell lineages (mean 15.4 [SD 3.3] and mean 15.3 [SD 3.3], respectively) using Dunn's multiple comparison test (Fig. 5, Table 2). This finding may reflect the difficulty in detection of telomeres in other cell lineages because of morphology and size; nuclei in hepatocytes and cholangiocytes were larger than in other lineages (Fig. 5, Supporting Fig. 4). There was no relation between age and the number of telomeres detected per cell for any intrahepatic cell lineage (Fig. 6).

Figure 6.

Variation in the number of telomeres per nucleus for each intrahepatic cell lineage with increasing age of liver.

Nuclear Area.

There was no relation between nuclear area and age for any lineage within healthy liver (Supporting Fig. 4). Nuclear area was greater in hepatocytes and Kupffer cells in comparison to other lineages (Table 2, Fig. 6).

Nuclear Intensity.

There was no relation between the intensity of nuclear staining with DAPI and age for any cell lineage (Supporting Fig. 5). However, nuclear intensity was lower in Kupffer cells than in other lineages (P < 0.05) but was higher in cholangiocytes compared with all other lineages (P < 0.05) (Table 2).


Since the discovery of telomeres, only two studies have addressed age-related changes in telomere length in “healthy” liver.25, 26 However, both were limited by small sample size, limited age range, the use of liver tissue derived from individuals with an increased risk of senescence (patients with malignancy) and unknown levels of steatosis, which causes premature hepatocyte telomere shortening and impaired regeneration in nonalcoholic fatty liver disease.31 Additionally, both studies analyzed telomeres in whole liver homogenates and assumed that findings were representative of hepatocytes. This assumption is flawed, however, because only 64% of cells in liver tissue are hepatocytes,27 and in our study there was no correlation between whole liver telomere length measured by real-time PCR and hepatocyte telomere length assessed by Q-FISH. This and the utilization of archived paraffin-embedded material emphasizes the advantages of Q-FISH. The ability to include only cells that meet tight definitions excludes cells with unusual morphology, and the large numbers of cells available for analysis increases methodological robustness.

Obtaining normal healthy liver tissue for research over a broad age range is challenging. Tissue from hepatic resections for malignancy, distant to the tumor with normal macroscopic and microscopic appearances, demonstrate shortened telomeres25-27, 28-30 and is only available over a narrow age range. Our subjects were highly selected for normality and may represent unusually healthy liver. Comparison with age-matched hyperoxalosis normal liver tissue, often used in domino liver transplantation,32-38 vindicated this approach, as there was no discernible difference in telomere length between the groups.

Different intrahepatic lineages in healthy liver aged at different rates. Age-related telomere shortening was restricted to Kupffer cells and hepatic stellate cells. Maintained telomere length with increased age in cholangiocytes and hepatocytes (in contrast to previous studies25, 26) may reflect low turnover of these populations, thus preserving regenerative capacity.

The preservation of hepatocyte telomere length with age contrasts with observations of reduced regenerative capacity with increasing age and clinical experience that older individuals are more susceptible to liver injury. Two factors may explain this anomaly. First, great lengths were undertaken to identify normal liver so that excellent donor liver function at 1 year and exclusion of concomitant senescence-related disease, steatosis and graft injury were defined entry criteria and less than 8% of available donor livers were studied. The study group was healthy and normal (unlike previous studies), but not necessarily typical of the everyday. Second, liver function is not related to hepatocytes alone but to all intrahepatic cells and the finding that sinusoidal cells showed age-related telomere shortening may be an important observation in relation to age-related liver function.

The unexpected and profound difference between interlobular/large bile duct cholangiocyte telomere length and that of the other intrahepatic cell lineages may reflect greater potential for multiple cell cycles in cholangiocytes, and this population may include progenitor cells, which drive bile duct proliferation in response to hepatocyte injury as described by Theise et al.39

More intrahepatic lymphocytes were detected than expected in normal liver and may represent the response to handling of the liver during harvesting and implantation.

The reason behind the reduction in sinusoidal cell telomere length with age (in Kupffer cells and hepatic stellate cells) was beyond the scope of this study. Sinusoidal cells have a different origin, namely bone marrow,40, 41 and are subject to constant immune stimulation through contact with portal blood. Others have demonstrated that hepatocyte telomeres shorten in cirrhotic liver but that hepatic stellate cells and lymphocytes in regions of liver fibrosis have longer telomeres.24 These studies have only looked at small numbers of each hepatic cell lineage and may not be representative, particularly given the heterogeneity seen in liver tissue. They may also reflect the recruitment of cells with longer telomeres to the injured liver from bone marrow. In chimeric mice, hepatic stellate cells originate from hematopoetic bone marrow stem cells, particularly following hepatic injury.42 Finally, telomere shortening in sinusoidal cells may reflect a reduction in hepatic blood flow, which is especially marked after the age of 50.43, 44 It has been suggested that reduced hepatic flow alone could explain delayed hepatic regeneration after injury.45

Kupffer cells populating the liver originate from bone marrow in both mice and humans after bone marrow transplant46 and constitute a large intrahepatic population. Quantitative study of monocyte production in bone marrow and transit through the circulation showed that in the normal steady state, over 50% of monocytes leaving the circulation become Kupffer cells. Considering the Kupffer cells as kinetically homogeneous, gives a mean turnover time of the total population of Kupffer cells of 21 days.47 Further studies in normal rat liver revealed an age-related decline in antigen presentation ability by Kupffer cells,48 and these cells in mice have been shown to have a stimulatory role in liver regeneration.49

In conclusion, we developed a robust, high-volume Q-FISH method for analysis of telomeres in different hepatic cell lineages that highlighted the pitfall of using liver homogenates in the study of aging and senescence. Furthermore, we demonstrated very long telomeres in cholangiocytes in normal liver over a wide age range and age-related telomere attrition restricted to sinusoidal cells. Understanding the normal process of aging in the liver is important in many aspects of hepatology from pharmacology to selection of older donors, and the findings encourage careful selection of older liver donors.