Changes of telomere length with aging


Dr Kaiyo Takubo MD PhD, Research Team for Geriatric Diseases, Tokyo Metropolitan Institute of Gerontology, Tokyo 173-0015, Japan. Email:


We reviewed our methodology and results of telomere measurements, with reference to telomere length and aging. Human tissues always showed telomere shortening with age, except for the brain and myocardium. Yearly rates of telomere length reduction in various tissues were mostly within the range 20–60 bp, and thus compatible with that expected from only one round of mitosis. It was suggested that when telomeres were found to be longer in any specific organ in a given individual, then the other organs in that individual would also have longer telomeres. Using the quantitative fluorescence in situ hybridization (Q-FISH) method for telomere measurement, we were able to measure the telomere lengths of various cell types within tissues. Here we summarize the results obtained for various cell types in the stomach, tongue and breast. Our Q-FISH method using our original software program “Tissue Telo” is excellent for measuring telomere lengths using tissue sections and PNA probes. Geriatr Gerontol Int 2010; 10 (Suppl. 1): S197–S206.


Normal diploid cells from humans show a limited capacity for proliferation in culture,1 and this finite replicative lifespan has frequently been used as a model of human aging in mitotic tissues and organs. This phenomenon is considered to be associated with the reduction in telomere length as an indicator of the number of cell divisions undergone. There is mounting evidence to suggest that when the telomere shortens to a critical length, cell senescence is triggered. The human telomere is a simple repeating sequence of six bases, TTAGGG, located at the ends of chromosomes.2 It is thought that telomeres have multiple roles, including protection against degeneration, reconstruction, fusion and loss,3 as well as contributing to pairing of homologous chromosomes.4 The end-to-end chromosome fusions observed in some tumors could play a role in genetic instability associated with tumorigenesis, and possibly result from telomere loss.5 Telomeric repeats of DNA sequences at chromosome ends are shortened by 33–120 bp with each cell division in human fibroblasts6 and lymphocytes7in vitro, but the question of telomere shortening with aging in many other cell types, and human tissues and organs in vivo remains unclear. We have measured telomere lengths in almost all human tissues by Southern blot analysis, and determined annual telomere loss rates for several of them using the “tissue quantitative fluorescence in situ hybridization (Q-FISH)” method.8 Here we report the methodology and results of our telomere measurements, and discuss the relationship between telomere length and aging.

Telomere shortening in human tissues

We have reported telomere shortening with aging in mucosae of the esophagus,9 stomach10 and colon,11 the liver,8 and other sites (Table 1). Neurons and myocardium are exceptional in that they are relatively static with respect to cell turnover,12 but there have been no reports regarding their telomere status. Furthermore, relationships among telomere lengths in different organs or tissues in the same individual have not been examined. Because telomere lengths in human tissues, as measured by Southern blot analysis, show very large standard deviations among individuals,8–11,13,14 attention was concentrated on different tissues in many subjects.

Table 1.  Yearly reduction rates of telomere length in human tissues
Organs and tissuesYearly reduction rate (bp/year)Sample numberAge range (years)Reported year and author
  1. Approximate number; Our Study. bp, base pair; m, month old; NSD, no significant difference.

 1. Peripheral blood334720–851990 Hastie et al.
 2. Epidermal cells19.8210–921991 Lindsey et al.
 3. Fibroblasts15430–931992 Allsopp et al.
 4. Peripheral lymphocytes411400–1071993 Vaziri et al.
 5. Peripheral lymphocytes (twins)311232–951994 Slagboom et al.
 6. Endothelium47–147133.5–1021995 Chang et al.
 7. Large and small intestine42530–891996 Hiyama et al.
 8. Esophageal mucosa601770–1021999 Takubo et al.
 9. Endothelium28511 m–801999 Okuda et al.
10. Arterial mediastinum25511 m–801999 Okuda et al.
11. Liver55940–1012000 Takubo et al.
12. Colonic mucosa591290–972000 Nakamura et al.
13. Gastric mucosa47380–992000 Furugori et al.
14. CD4+T Cells351210–942000 Son et al.
15. CD8+T Cells261210–942000 Son et al.
16. CD19+B Cells191210–942000 Son et al.
17. Renal cortex29240–882000 Melk et al.
18. Renal medulla9 & 13200–882000 Melk et al.
19. Liver1202317–812000 Aikata et al.
20. Renal cortex461370–1012002 Takubo et al.
21. Liver601910–1042002 Takubo et al.
22. Cerebral cortexNSD1370–1042002 Takubo et al.
23. Cardiac muscleNSD1680–1042002 Takubo et al.
24. Spleen29300–1022002 Takubo et al.
25. Thyroid90440–982002 Kammori et al.
26. Parathyroid gland94190–832002 Kammori et al.
27. Epidermal tissue36520–1012002 Nakamura et al.
28. Lingual mucosa30480–1012002 Nakamura et al.
29. Dental pulp7210016–702003 Takasaki et al.
30. Pancreas36690–1002006 Ishii et al.
31. Epidermal tissue9100 2006 Sugimoto et al.
32. Dermal tissue1160 2006 Sugimoto et al.
33. Cerebral cortexNSD720–1002007 Nakamura et al.
34. Cerebral white matterNSD720–1002007 Nakamura et al.

Telomere lengths in newborns and centenarians

For neonates, mean telomere lengths for cerebral cortex, myocardium, liver and renal cortex were 13.1 ± 1.1, 12.6 ± 1.1, 13.7 ± 2.5 and 13.7 ± 2.2 kbp, respectively.8 For subjects less than 10 years-of-age, including neonates, the respective values were 13.1 ± 1.8, 13.0 ± 1.0, 13.6 ± 2.3, 13.6 ± 2.0 and 12.8 ± 1.0 kbp. For centenarians, the respective values were 13.1 ± 2.3, 11.3 ± 1.6, 8.7 ± 1.4 and 11.8 kbp. In centenarians, mean telomere lengths in these four tissues were not shorter than 6 kbp, which was considered in the telomere hypothesis to be the telomere length of mortality stage l. However, telomere lengths were very heterogeneous, and the ratio of telomere lengths within 6 kbp increased in every tissue with aging.

Correlations of telomere length among four organs in individuals

Examinations of correlations in telomere length among organs in individuals showed a number of significant correlations, such as cerebral cortex versus myocardium; cerebral cortex versus liver; cerebral cortex versus renal cortex; myocardium versus liver; myocardium versus renal cortex; and liver versus renal cortex.8 Therefore, it was suggested that when longer telomeres are shown in any particular organ in a given individual, the other organs will also have longer telomeres.

Annual telomere reduction rate and tissue renewal times

We reviewed telomere reduction rates in human cells, tissues and organs (Table 1).7–11,13–24 These values were mostly within the range 20–60 bp per year. In an autoradiographic study in mice, cellular renewal times of hepatocytes, renal tubular and glomerular cells, and splenic tissue were found to be 460–480, 140–190 and 1–21 days, respectively,25 whereas very rapid renewal was found to occur in epithelial cells of the gastrointestinal mucosae, showing a turnover of within 7 days.12,25 Previous data we had obtained for annual telomere reduction in gastrointestinal mucosae9–11 showed values lower than 70 bp. Unlike mouse tissues and organs, renewal times of human tissues have been examined for only some types of lymphocytes.26 Data available for lymphocyte turnover times have shown considerably large differences between humans and mice.26,27 There are apparently no marked differences among telomere length reduction rates for the liver and renal cortex, and those we reported previously for gastrointestinal mucosae, despite what must be very large differences in renewal times, if the renewal times in these human and mouse tissues are similar. Therefore, it can be said that annual telomere length reduction rates are not directly related to tissue turnover. In digestive tract mucosae, after mitosis, one of two daughter cells migrates up from the proliferative zone to the surface layer and then desquamates into the gastrointestinal lumen within 7 days.12,25 Hepatocytes in the portal space gradually stream toward the terminal hepatic vein, where they are probably eliminated by apoptosis; this process takes approximately 201 days.28 Renal cortical cells have never been reported to undergo cellular streaming. Renewal times determined using autoradiography, and patterns of cellular streaming in the mucosal epithelia and liver indicate that early differentiated or transit-amplifying cells need to undergo very many mitotic divisions per year. It is likely that senescent epithelial cells in digestive tract mucosae disappear into the digestive tract lumina along with young epithelial cells, and that senescent hepatocytes in the liver stream disappear near central veins of the liver along with young hepatocytes. Therefore, few or no senescent cells might remain in vivo.

Yearly telomere reduction rates of less than 100 bp are comparable with those resulting from only one round of mitosis in fibroblasts6 and lymphocytes7in vitro. Although there is a possibility that the extent of telomere shortening per mitotic division in vivo is much less than that in vitro, it would seem that stem cells in human tissues and organs require a mechanism for telomere elongation. It has been reported that telomere reduction rates per cell division differ significantly among human cell types in vitro.17 However, no data have been reported for the reduction rate per cell division in vivo. Organs with a high cell turnover, such as skin, lymphoid tissue and the gastrointestinal tract mucosal epithelium, would be most affected by an accelerated loss of telomere repeats. Indeed, telomere maintenance has been reported for stem cells29 and gastrointestinal mucosal epithelial cells,15,30 and lymphocytes31 are positive for telomerase activity as determined by telomere repeat amplification protocol assay.32 The differences in in vitro telomere reduction rates per cell division and tissue stem cell telomerase activity, and the present data indicating that there are no significantly large differences in annual reduction rates for rapidly renewing tissues such as gastrointestinal epithelia, suggest that the latter might show less telomere reduction per cell division as a result of simultaneous telomere elongation by telomerase, whereas slowly renewing tissues, such as liver or kidney, might show comparatively more telomere reduction per division.

Telomere lengths of postmitotic cells

In cerebral tissue and myocardium, telomere lengths in individuals were significantly correlated, and it was considered that they retained their original telomere lengths. Our recent study of telomere lengths in the cerebral gray and white matter, obtained from autopsied individuals, suggested that telomere lengths are well maintained under postmitotic conditions and that longer telomeres are associated with longer lifespan.23

Relative telomere lengths measured by Q-FISH using human tissue sections

Recently, the Q-FISH method has been used to measure telomere lengths in histological sections. However, there have been many differences among the Q-FISH methods reported by different authors, including the internal controls (Table 2), probes and PNA. Some studies compared telomere lengths of different cell types on the basis of visual assessment, and the results obtained were not considered to be reproducible. Some studies have used “stromal cells”, including fibroblasts, lymphocytes and endothelial cells. Although fibroblasts showed relatively constant lengths in humans of different ages,33 we also considered that any method using a mixture of stromal cells as a control would not yield reproducible results. We considered that more than 100 cells of each cell type should be subjected to Q-FISH, but many studies have assessed only a small number of each cell type. The methods we have used are described in the next section “Q-FISH and image analyses for telomere measurement.”

Table 2.  Methodology in telomere length measurement by tissue Q-FISH
Year reported and authorsSitenCells analyzedInternal control for quantification of telomere signal 
  1. Our studies. –, not mentioned. DAPI, 4′,6-diamidino-2-phenylindole dihydrochloride.

2002Meeker AK et al.Prostate176–23DAPI signal intensityAm J Pathol160: 1259–68
2002O'Sullivan JN et al.Colon34Telomere signal intensity of stromal cellsNat Genet32: 280–4
2002van Heek NT et al.Pancreas410DAPI signal intensityAm J Pathol161: 1541–7
2002Meeker AK et al.Prostate65–20DAPI signal intensityCancer Res62: 6405–9
2003Ferlicot S et al.Liver, kidney, thyroid30100–200NoneJ Pathol200: 661–6
2003Kuniyasu H et al.Stomach219100Mean nuclear area, TMK-1 gastric cancer cellsOncology65: 275–82
2003Vukovic B et al.Prostate1520Centromere signal intensityOncogene22: 1978–87
2004Meeker AK et al.Breast610–20DAPI signal intensityAm J Pathol164: 925–35
2004O'Sullivan JN et al.Colon, esophagus3540Telomere signal intensity of stromal cellsCytometry A58: 20–31
2004Nakajima T et al.Liver3015–20Pixels of DAPI, non-lymphocyte stromal cellsBiochem Biophys Res Commun325: 1131–5
2006Nakajima T et al.Liver4415–20Pixels of DAPI, lymphocyteLiver Int26: 23–31
2006Tabori U et al.Glioma825Centromere signal intensityNeoplasia 8: 136–42
2006O'Sullivan JN et al.Colon22350Telomere signal intensity of stromal cellsCancer Epidemiol Biomarkers Prev15: 573–7
2006Hansel DE et al.Biliary tract512–26DAPI signal intensityMod Pathol 19: 772–9
2006Finley JC et al.Esophagus8150Telomere signal intensity of stromal cells,Cancer Epidemiol Biomarkers Prev15: 1451–7
2006Maida Y et al.Cervix, endometrium65DAPI signal intensityJ Pathol210: 214–23
2006Kammori M et al.Esophagus15average 86Centromere signal intensityOncology71: 430–6
2007Aida J et al.Stomach11200Centromere signal intensityHum Pathol38: 1192–200
2007Joshua AM et al.Prostate6850Centromere signal intensityNeoplasia9: 81–9
2007Rashid-Kolvear F et al.Breast18minimum 20Chromosome 17q signal intensityNeoplasia9: 265–70
2007Perrem K et al.Skin152minimum 10DAPI signal intensityHum Pathol38: 351–8
2007Morton MJ et al.Bladder3425Normal urotheliumClin Cancer Res13: 6232–6
2007Bechan GI et al.Langerhans cell histiocytosis414–92DAPI signal intensityBr J Haematol140: 420–8
2008Hashimoto Y et al.Pancreas1710–25DAPI signal intensityJ Gastrointest Surg12: 17–29
2008Aida J et al.Tongue21minimum 660Centromere signal intensity, cell block-section of a cultured cell line, TIG-1Exp Gerontol43: 833–9
2008Kurabayashi et al.Breast30minimum 373Centromere signal intensityHum Pathol39: 1647–55
2009Zheng YL et al.Esophagus4730NoneCancer Res69: 1604–14
2009Shiraishi et al.Esophagus11200–500Centromere signal intensityScand J Gastroenterol (in press)
2004Montgomery E et al.Sarcoma18Visual assessmentAm J Pathol164: 1523–9
2004Meeker AK et al.Bladder, cervix, colon, esophagus, oral cavity25Visual assessmentClin Cancer Res10: 3317–26
2005Lantuejoul S et al.Bronchus51Scored as 0, no staining; 1, <10% of nuclei; 2, 20–60%; 3, 70% and moreClin Cancer Res11: 2074–82
2006Patton KT et al.Leiomyoma3Visual assessmentMod Pathol19: 130–40
2007Kawai T et al.Lung6110 or moreThe number of telomere signals per nucleusAm J Clin Pathol127: 254–62
2008Akbay et al.Endometrium29Visual assessmentAm J Pathol173: 536–44

We measured telomere lengths of various cell types in human tissues and organs, and decided their relative sizes (Table 3). Furthermore, we plan to apply our telomere measurement techniques to normal, dysplastic and cancerous tissues in order to better understand the role of telomeres in intraepithelial cancer, and for diagnosis of intraepithelial lesions.

Table 3.  Relative telomere length of different cell types in human tissues measured by tissue Q-FISH using our original software “Tissue Telo”
Tissues (authors)Telomere length of cell types
Lingual mucosa (Aida et al. 2008)Fibroblasts > basal cells > parabasal cells > prickle cells
Esophageal mucosa (Kammori et al. 2006)Basal cells > fibroblasts > prickle cells
Gastric mucosa (Aida et al. 2007)Fibroblasts > chief cells, parietal cells > neck cells > foveolar cells
Mammary gland (Kurabayashi et al. 2008)Myoepithelial cells > fibroblasts > luminal epithelial cells
Transformed (cancer associated) fibroblasts > non-transformed fibroblasts

Gastric mucosa

Through the measurement of cell types in the gastric mucosa, we were able to show that progressively shorter telomere lengths were observed in turn from mucosal fibroblasts to fundic gland cells, to glandular neck cells and to surface foveolar epithelial cells. The telomeres of intestinal metaplastic cells appeared to be longer than those of glandular neck and foveolar cells. The telomeres of adenocarcinoma cells were not always shorter than those of non-cancerous epithelial cells.

Southern blot analysis has shown that the intestinal metaplastic gastric mucosa has shorter telomeres than non-metaplastic gastric mucosa.10 In contrast, however, our Q-FISH data showed that intestinal metaplastic cells had longer telomeres than normal epithelial cells.34 We considered two possible explanations for this result. First, in our previous study, we had obtained gastric specimens from apparently atrophic mucosa, whereas in the subsequent study we measured telomeres of intestinal metaplastic cells intermixed with normal fundic glands. In the two cases of atrophic gastric mucosa, the telomere : centromere ratio (TCR) of intestinal metaplastic cells appeared smaller than that in eight cases of mild intestinal metaplasia. In four of these eight cases, intestinal metaplastic cells appeared to have longer telomeres than the fundic gland cells. The intestinal metaplastic cells located in the fundic gland mucosa might have telomeres that are elongated as a result of human telomerase reverse transcriptase (h-TERT) activity. As atrophy of the gastric mucosa progresses, epithelial telomeres might shorten. The second explanation is that we evaluated telomere length using Southern blotting, which takes the peak value as the telomere length for any given tissue. In contrast, our Q-FISH method yielded the mean TCR. Because the telomere (TCR) distributions are not symmetrical and longer telomeres tend to predominate, the mean telomere value tends to be larger than the peak value.

In cases showing Helicobacter pylori infection, the inflammatory process leads to the development of atrophic gastritis with apoptosis of gastric fundic and pyloric gland cells, followed by intestinal metaplasia, dysplasia and, ultimately, early and advanced cancer.35 Furthermore, H. pylori has the proven ability to trigger gene mutation and cancer progression.36 Our results suggest that telomere shortening provides evidence of apoptosis progression in infected fundic glands35 and support those of previous studies showing telomere shortening in infected gastric mucosae.37 In contrast, the fact that telomeres were elongated in infected intestinal metaplastic cells suggests possible telomerase activation in infected metaplastic mucosae.38

Cancerous tissues have been shown to have relatively shorter telomeres than normal mucosae, irrespective of histological type.10,39 Furthermore, non-cancerous gastric mucosae from patients with cancer are reported to have shorter telomeres than normal mucosae from healthy individuals, and four of the eight cases we examined showed longer telomeres in cancerous tissues than in non-cancerous mucosae.37 This finding is contrary to that of previous studies, perhaps because of the fact that all the non-cancerous mucosal specimens we used for comparison were obtained from cancer-bearing stomachs.

Lingual mucosa and epithelial tissue stem cells

In previous studies by Kang et al.40 and our group,41 the Southern blotting technique was used, and Meeker et al.42 used Q-FISH to estimate telomere lengths in oral mucosa. The latter workers showed differences in telomere length among normal, dysplastic and cancerous tissues. However, they did not observe differences in telomere length among various cell types within the epithelium. We applied the tissue Q-FISH method to estimate telomere lengths of individual cell types, using TCR.33 As our methodology made it easy to extract data from very many cells of different types within the mucosa, statistically relevant results were generated.

Age-related telomere length reduction in human tissue is thought to be caused by telomere shortening and/or reduction in the number of stem cells.43,44 Our previous study showed that the number of cells with longer telomeres (NTCR, mentioned later) in younger individuals was about 15-fold that in adults. Therefore, the largest regression rate in the basal cell layer suggests a probable reduction of tissue stem cell number in this layer. This stem cell reduction means that cells with the longest telomeres undergo numerical reduction with age, leading to a reduction in the variation of telomere length in each cell type. This indicates that shorter telomeres accumulate in adults, supporting the in vitro findings of Kang et al.45

Our investigation of the correlation of telomere length between cell types in the lingual mucosa showed a close correlation among different epithelial cell groups. Tissue stem cells might be located in the basal layer of the lingual epithelium, and cells undergoing transient amplification, derived from stem cells by asymmetric cell division, might be present in the parabasal layer of the stratified squamous cell epithelium.46 Cells undergoing transient amplification, Ki-67-positive proliferating cells, undergo repeated cell division to produce differentiated prickle or keratinized epithelial cells, and thus might show a reduction of telomere length in the parabasal layer.47 Therefore, basal cells might affect the telomere length of parabasal and prickle cells.

It was noteworthy that NTCR of fibroblasts in the lingual mucosa showed larger variations than those of the epithelial cells in all age groups. Furthermore, in adults, fibroblasts had larger NTCR than epithelial cells. It has been reported that telomere length measurements do not show a clear correlation with tissue renewal times in vivo, but rather are characteristic for individuals and are correlated with those in other organs.48 Therefore, fibroblasts might retain their original telomere length to a greater degree than epithelial cells, supporting the findings for human oral fibroblasts.49

The telomere length patterns differed between neonates and adults. Telomeres of fibroblasts were relatively shorter than those of epithelial cells in neonates less than 1 month-old, but were longer than those of epithelial cells in individuals over 1 month-old. Our results appear to support the accelerated telomere shortening in normal oral keratinocytes,45 esophageal mucosa9 or whole blood leukocytes44 reported for very young individuals, indicating the possibility of accelerated telomere shortening in oral epithelial cells during development, when massive cell proliferation occurs. Our findings indicated the possible presence of tissue stem cells and/or progenitor cells in the basal layer.


Prior studies of telomere length in breast cancers using Southern blotting or slot-blot analysis have shown telomere shortening in breast cancer tissues, but the data regarding correlation with clinical factors such as histological grade and prognosis have not always been concordant.50–52 Although Odagiri et al.50 reported that grade 3 tumors had shorter telomeres and that telomere length was not correlated with prognosis, Rogalla et al.51 found no correlation between telomere length and histological grade. In contrast, Fordyce et al. reported that telomere length was correlated with both stage and prognosis.52 These disagreements might be partly explained by the limitations of Southern blotting and slot-blot analysis, which are unable to estimate the telomere lengths of individual cells and provide only information on telomere lengths for a mixed cell population, including inflammatory cells and other stromal cells. Using our Q-FISH method, we analyzed 30 surgical specimens of invasive breast carcinoma from women aged 34–91 years and estimated telomere lengths as the telomere-to-centromere ratio in the five different cell types comprising breast tissue in order to clarify telomere length variations within and between individuals.53 We obtained four novel findings: (i) in corresponding normal tissues, telomere length decreased in the order myoepithelial cells > normal-appearing fibroblasts > luminal epithelial cells, and telomere lengths were characteristic in these three cell types within each individual; (ii) as expected, cancer cells had significantly shorter telomeres than myoepithelial cells and normal-appearing fibroblasts, but there was no significant difference in telomere length between luminal cells and cancer cells; (iii) fibroblasts adjacent to cancer had longer telomeres than normal-appearing fibroblasts distant from cancer; and (iv) telomere lengths of cancer cells were not related to any clinocopathological factors.

Ongoing studies by our group

We are now measuring telomere lengths of different cell types in both normal and dysplastic tissues of the thyroid, parathyroid, liver, esophagus, colon, brain, pancreas and skin.

Methods of telomere measurement

Southern blotting

Subjects, tissues and DNA samples

When we used tissues from autopsied subjects, all the cadavers examined were kept refrigerated at 3°C until autopsy. We had found previously that no DNA degradation was observed in most tissues from autopsied subjects kept at 3°C for more than 12 h.

All samples were collected in the autopsy room, frozen rapidly in liquid nitrogen and stored at −80°C until use.

Histological examination by pathologists

The tissues adjacent to all areas we used were histologically examined by specialists in anatomical and surgical pathology, and areas exhibiting marked retention of other tissues, marked accumulation of inflammatory cells and/or autolysis were excluded.

Examination of DNA degradation

Genomic DNA was prepared from each sample by treatment with proteinase K and sodium dodecyl sulfate (SDS), followed by repeated phenol-chloroform extraction. In preliminary experiments with genomic DNA from all samples, pulse-field gel electrophoresis54 was carried out in 1.0% (w/v) agarose (BMA Products, Rockland, ME, USA) gels, using the Genofield system (ATTO, Tokyo, Japan), a biased sinusoidal field gel electrophoresis system, to examine post-mortem autolytic changes. The genomic DNA was subjected to electrophoresis at room temperature for 7 h in 0.5 × Tris-boric acid-EDTA (TBE) buffer using a 2.5 V/cm direct current and a 9.8 V/cm alternating current. The polarity alternation frequencies were 0.07 Hz (initial) and 0.3 Hz (final) with a logarithmic gradient. Only DNA of more than 100 kbp was used for experiments.

Measurement of terminal restriction fragment length by Southern blotting

We usually used the standard method to measure telomere terminal restriction fragment (TRF) length. A 5-mg sample of DNA was digested with the restriction enzyme HinfI (Boehringer Mannheim Biochemica, Germany) and complete cleavage by the enzyme was confirmed by electrophoresis of DNA digests on 0.8% (w/v) agarose gels overnight at 20 V AC. Fractionated DNA fragments were transferred to nylon membranes (Hybond-N+; Amersham, Buckinghamshire, UK) by an alkaline transfer technique that used capillary blotting, followed by hybridization for 12 h at 50°C in an appropriate solution (6 × SSPE [1 ×; 0.15 M NaCl, 10 mmol/L sodium phosphate, 1 mmol/L EDTA, pH 7.4], 1% [w/v] SDS) with a (TTAGGG)4 probe labeled by [g-32P] adenosine triphosphate (Amersham) at the 5′ end with T4 polynucleotide kinase (Toyobo, Japan). Membranes were washed in 2 × SSC (1 ×; NaCl 17.55 g/L, sodium citrate 8.82 g/L) at room temperature and then in 6 × SSC, 0.1% SDS at 50°C for 15 min while being shaken. After drying the membranes with filter paper, Fuji Imaging Plates (Fuji Photo Film, Tokyo, Japan) were exposed to them for 20 min at room temperature. A BAS-2500 Mac image analyzer (Fuji Photo Film) and the programs Image Reader (version 1.1, Fuji Photo Film) and Mac Bas (version 2.4, Fuji Photo Film) were used for the analysis. We estimated mean TRF length at the peak position of the hybridization signal in each lane; TRF lengths were determined by comparing the positions of maximum radioactivity in each lane with those of molecular size markers, as described previously.8–11,13–15,48

Telomere lengths in tissues containing multiple cell types

The presence of multiple cell types is always problematic in studies measuring telomere lengths in human tissues or organs,8,11 unlike the case with cultured cell lines. The cerebral cortex contains neurons and glia, and the myocardium contains a few endothelial cells and fibroblasts. Although hepatocytes only account for about 79% of hepatic tissue volume,55 our previous study showed no significant difference in telomere lengths between isolated hepatocytes and whole liver tissue.8 The renal cortex contains cells from renal tubules, glomeruli and other components. The spleen consists of lymphocytes, endothelial cells, reticular cells and some other cells. Reduction of telomere length might be expected because glia and endothelial cells coexist with nervous tissue in the cerebral cortex, and fibroblasts and endothelial cells coexist in the myocardium and can undergo mitoses; but clearly this cannot exert a major influence at the tissue level. Furthermore, by quantitative fluorescence and in situ hybridization, telomere lengths of specific chromosomes have been shown to differ among cells of the same type, some being long and others short.56,57 Our results suggest that individuals have characteristic telomere lengths that correlate across most organs and tissues.

However, in order to clarify telomere metabolism, it is necessary to examine individual cell types in tissues and organs, and for this purpose we have independently developed tissue Q-FISH methods using tissue sections and PNA probes.

Q-FISH and image analyses for telomere measurement

Tissue processing

Tissues were fixed for 6 h in 10% buffered formalin and then subjected to standard tissue processing and paraffin embedding. Tissues were sliced into sections 3–4 µm thick for histological and immunohistochemical examinations.

Telomere fluorescence analysis

After the samples had been embedded in paraffin and cut into 5-µm sections, the slides were immersed in O-xylene for 7 min at room temperature (RT), dipped in 100% ethanol, and air-dried. They were then dipped in 0.2 mol/L HCl for 20 min at RT, followed by rinsing in H2O for 10 min at RT, incubation in 1 mol/L NaSCN for 30 min at 80°C, and rinsing again in H2O. The slides were then incubated in pepsin (1 mg/mL acid H2O) for 15 min at 37°C, rinsed in H2O, dipped in 100% ethanol and air-dried. The slides were then treated with 0.5 µg/mL of ribonuclease A for 10 min at 37°C, rinsed with phosphate-buffered saline, dehydrated through an ethanol series (70%, 90%, 100%), and dried.

We hybridized the tissue sections with 30 mol/L of the PNA telomere probe (telo C Cy3 probe: 5′-CCCTAACCCTAACCCTAA-3′V; Fasmac, Atsugi City, Kanagawa, Japan) conjugated to Cy3 (final concentration 0.32 µmol/L) and the PNA centromere probe (Cenp1 probe: 5′-CTTCGTTGGAAACGGGGT-3′; Fasmac) conjugated to fluorescein isothiocyanate (FITC; final concentration 0.12 µmol/L) for 3 min at 80°C, and the slides were incubated for 1 h at RT. We then washed the slides four times in 70% formamide buffer for 15 min each time, followed by four washes in TBST (0.1 mol/L Tris, 0.15 mol/L NaCl, 0.08% Tween 20) for 5 min each time. Finally, the slides were dehydrated through an ethanol series and air-dried. The nuclei were then stained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI; Molecular Probes, Eugene, OR, USA) and the slides were mounted with Vectashield (Vector Laboratories, Burlingame, CA, USA).

Q-FISH and image analysis of telomeres

Q-FISH digital images were captured by a charge-coupled device camera (ORCA-ER-1394; Hamamatsu Photonics KK, Hamamatsu, Japan) mounted on an epifluorescence microscope (80i; Nikon, Tokyo, Japan) equipped with a triple band-pass filter set for DAPI/FITC/Cy3 (part 61010, Chroma Technology, Rockingham, VT, USA) with an objective lens (Plan Fluor 40/0.75, Nikon). Microscope control and image acquisition were carried out using the Image-Pro Plus software package (version 5.0, Media Cybernetics, Silver Spring, MD, USA). The captured images were analyzed using our own original tissue analysis software package, Tissue Telo, which allows manual identification of nuclear regions from the composite color image: DAPI, blue channel; FITC, green; Cy3, red. Telomere and centromere signals were then determined as pixels showing the brightest intensities (top 5%) within each selected nuclear region. The measured signal intensities were corrected for background autofluorescence as determined from the mean of the pixels showing the lowest intensities (lowest 20%). The top 5% and lowest 20% thresholds have previously been shown to give consistent results.58 Because there is no guarantee that the entire nucleus is captured within a tissue section, the total corrected telomere signal (integrated optical density) for each nucleus is further normalized by the corresponding integrated optical density of the centromere. We have previously verified in a fluorescence flow study that the mean optical density of our centromere probe was constant among human blood cells from individuals of various age groups and cultured fibroblasts (TIG-1, 3 population doubling levels). In summary, our cell-normalized telomere length estimate for each nucleus is defined as the ratio of the detected telomere signal intensity to the centromere signal intensity (TCR). TCR from 100 to 300 cells were analyzed.

TCR normalized by cell block

As a control for variations in sample preparation, we also carried out Q-FISH on a cell block-section of a cultured cell line, TIG-1,59 that had been subcultured (34 PDL; population doubling level) and had a telomere length of 8.6 kbp measured by Southern blot analysis, placed on the same slides with the lingual tissue sections. Every TCR of cells was divided by the mean TCR for the cell block on the same slide to give the NTCR.


Human tissues always show telomere shortening with age, except for the brain and myocardium. Telomere lengths in human tissues show significant correlations within individuals. Our tissue Q-FISH method is excellent for measuring telomere lengths using tissue sections and PNA probes.

Conflicts of interest