Stochastic Variation in Telomere Shortening Rate Causes Heterogeneity of Human Fibroblast Replicative Life Span

Abstract

The replicative life span of human fibroblasts is heterogeneous, with a fraction of cells senescing at every population doubling. To find out whether this heterogeneity is due to premature senescence, i.e. driven by a nontelomeric mechanism, fibroblasts with a senescent phenotype were isolated from growing cultures and clones by flow cytometry. These senescent cells had shorter telomeres than their cycling counterparts at all population doubling levels and both in mass cultures and in individual subclones, indicating heterogeneity in the rate of telomere shortening. Ectopic expression of telomerase stabilized telomere length in the majority of cells and rescued them from early senescence, suggesting a causal role of telomere shortening. Under standard cell culture conditions, there was a minor fraction of cells that showed a senescent phenotype and short telomeres despite active telomerase. This fraction increased under chronic mild oxidative stress, which is known to accelerate telomere shortening. It is possible that even high telomerase activity cannot fully compensate for telomere shortening in all cells. The data show that heterogeneity of the human fibroblast replicative life span can be caused by significant stochastic cell-to-cell variation in telomere shortening. The replicative life span of human fibroblasts is heterogeneous, with a fraction of cells senescing at every population doubling. To find out whether this heterogeneity is due to premature senescence, i.e. driven by a nontelomeric mechanism, fibroblasts with a senescent phenotype were isolated from growing cultures and clones by flow cytometry. These senescent cells had shorter telomeres than their cycling counterparts at all population doubling levels and both in mass cultures and in individual subclones, indicating heterogeneity in the rate of telomere shortening. Ectopic expression of telomerase stabilized telomere length in the majority of cells and rescued them from early senescence, suggesting a causal role of telomere shortening. Under standard cell culture conditions, there was a minor fraction of cells that showed a senescent phenotype and short telomeres despite active telomerase. This fraction increased under chronic mild oxidative stress, which is known to accelerate telomere shortening. It is possible that even high telomerase activity cannot fully compensate for telomere shortening in all cells. The data show that heterogeneity of the human fibroblast replicative life span can be caused by significant stochastic cell-to-cell variation in telomere shortening. Heterogeneity is a hallmark of aging. Even genetically identical organisms in a controlled homogeneous environment age at different rates and have vastly different life spans. This is also true for aging of human somatic cells in vitro (1Kirkwood T.B.L. Finch C.E. Nature. 2002; 419: 794-795Crossref PubMed Scopus (68) Google Scholar). Replicative life spans differ widely between subclones (2Smith J.R. Whitney R.G. Science. 1980; 207: 82-84Crossref PubMed Scopus (233) Google Scholar), and there is an ever increasing number of senescent cells in dividing cultures (3Kill I.R. Faragher R.G. Lawrence K. Shall S. J. Cell Sci. 1994; 107: 571-579PubMed Google Scholar). Senescence of human fibroblasts is telomere-driven, i.e. induced by the uncapping of telomeres (4Blackburn E. Cell. 2001; 106: 661-673Abstract Full Text Full Text PDF PubMed Scopus (1765) Google Scholar). However, it is often assumed that only the life span of the longest surviving clone(s) is governed by telomere shortening, whereas termination of growth in the early senescing clones might be premature, i.e. caused by stress via telomere-independent mechanisms (5Wright W.E. Shay J.W. Nat. Biotechnol. 2002; 20: 682-688Crossref PubMed Scopus (298) Google Scholar). In fact, a senescent phenotype can be generated in response to a variety of stresses that are telomere-independent and inhibit cell growth, for instance, in response to generalized DNA damage (6Gorbunova V. Seluanov A. Pereira-Smith O.M. J. Biol. Chem. 2002; 277: 38540-38549Abstract Full Text Full Text PDF PubMed Scopus (195) Google Scholar), oncogene activation (7Wei W. Hemmer R.M. Sedivy J.M. Mol. Cell. Biol. 2001; 21: 6748-6757Crossref PubMed Scopus (190) Google Scholar), histone acetylation (8Langley E. Pearson M. Faretta M. Bauer U.M. Frye R.A. Minucci S. Pelicci G. Kouzarides T. EMBO J. 2002; 21: 2383-2396Crossref PubMed Scopus (755) Google Scholar), or by induction of the p16 pathway through so far not well characterized mechanisms (9Itahana K. Zou Y. Itahana Y. Martinez J.L. Beausejour C. Jacobs J.J.L. Van Lohuizen M. Band V. Campisi J. Dimri G.P. Mol. Cell. Biol. 2003; 23: 389-401Crossref PubMed Scopus (352) Google Scholar). An alternative to this idea of premature senescence is the proposal that stochastic telomere uncapping can cause heterogeneity in replicative life span (10Blackburn E.H. Nature. 2000; 408: 53-56Crossref PubMed Scopus (1099) Google Scholar). The probability of telomere uncapping is dependent on a number of factors, including the presence of active telomerase, the integrity of the single-stranded telomeric G-rich overhang and of the higher order structure of telomeres, the functionality of a number of telomere-binding proteins, and telomere length (for review see Ref. 4Blackburn E. Cell. 2001; 106: 661-673Abstract Full Text Full Text PDF PubMed Scopus (1765) Google Scholar). Finally, there is evidence for a substantial interaction between stress and specifically mild oxidative stress and telomeres in cell replicative senescence. Stress accelerates telomere shortening because of a telomere-specific single strand break repair deficiency (11von Zglinicki T. Trends Biochem. Sci. 2002; 27: 339-344Abstract Full Text Full Text PDF PubMed Scopus (1837) Google Scholar). For most human cells, standard cell culture conditions exert mild oxidative stress (12von Zglinicki T. Serra V. Lorenz M. Saretzki G. Lenzen-Grossimlighaus R. Gessner R. Risch A. Steinhagen-Thiessen E. Lab. Invest. 2000; 80: 1739-1747Crossref PubMed Scopus (264) Google Scholar). Thus, heterogeneity of human cell replicative senescence might be due to either heterogeneous shortening of telomeres in a stress-dependent fashion, to stochastic length-independent uncapping of telomeres, or to premature induction of senescence completely independent of telomeres. To decide between these alternatives would clarify the nature of cellular senescence and will help to resolve the issue as to whether it occurs in vivo as an essential part of the aging process or not (13Campisi J. Exp. Gerontol. 2003; 38: 5-12Crossref PubMed Scopus (173) Google Scholar). To resolve the role of telomeres for the heterogeneity of replicative senescence, we sorted “prematurely” senescent human fibroblasts from growing cells by fluorescence-activated cell sorting (FACS) 1The abbreviations used are: FACS, fluorescence-activated cell sorter; BrdUrd, 2-bromo-5-deoxyuridine; EYFP, enhanced yellow fluorescent protein; FL1, fluorescence channel 1; FISH, fluorescent in situ hybridization; FSC, forward scatter channel; GAPDH, glutaraldehyde-3-phosphate dehydrogenase: hTERT, human telomerase reverse transcriptase; PD, population doubling; PBS, phosphate-buffered saline; ELISA, enzyme-linked immunosorbent assay. and measured telomere length separately in both populations. The first data revealed that telomeres in sorted prematurely senescent MRC5 fibroblasts were, on average, as short as those at the end of the replicative life span of the cultures (14von Zglinicki T. Petrie J. Kirkwood T.B.L. Nat. Biotechnol. 2003; 21: 229-230Crossref PubMed Scopus (66) Google Scholar). Using clonal populations we show now that telomeres shorten faster in those sublineages that senesce early. This faster telomere shortening causes senescence, because the majority of cells can be rescued from early senescence by the expression of telomerase. However, a small fraction of cells still arrest with short telomeres despite active telomerase. This fraction increases under oxidative stress, which accelerates telomere shortening. We conclude that stochastic, stress-induced cell-to-cell variation of telomere shortening is the main cause of the intrinsic heterogeneity in the division potential of human fibroblasts. Cell Culture—Human embryonic lung fibroblasts MRC5 were obtained from the ECACC. Human foreskin fibroblasts BJ originated from the laboratory of J. R. Smith (Houston, TX). The cells were grown in Dulbecco's modified Eagle's medium (Sigma) plus 10% fetal calf serum (Sigma) under either normoxia (95% air, 5% CO2) or 40% normobaric hyperoxia using a 3-gas cell culture incubator (Zapf Instruments, Sarstedt, Germany). Retroviral particles were generated upon transfection of the vector pLCP-hTERT (Clontech, Palo Alto, CA) containing the gene for the catalytic subunit of the human reverse transcriptase (hTERT) under control of the cytomegalovirus promoter into the packaging line ΦNX. MRC5 (PD 33) were infected with the retroviral supernatant, and cells were selected with 0.75 μg/ml for of was by G. S. K. Zglinicki T. 2002; PubMed Scopus Google and activity was as in the human cell line fibroblasts were grown in (Sigma) under an of 5% 5% and of a Cell by cell line was from a human fibroblast cell from embryonic lung W. Sedivy J.M. Science. 277: PubMed Scopus Google Scholar), as W. A. Sedivy J.M. Cell. 2003; PubMed Scopus Google Scholar). the vector was modified by the of the An was between and to expression of the cells were gene by R.M. W. A. Sedivy J.M. Mol. Biol. 2003; Google Scholar). into of the gene was by and the of the as well as the of were by of from the was by cells senescence W. Sedivy J.M. Science. 277: PubMed Scopus Google Scholar), and cell were to was with hTERT, and telomerase activity was by assay. The cell line used in this showed DNA damage with or including expression of both and W. A. Sedivy J.M. Cell. 2003; PubMed Scopus Google Scholar). and cells were in Dulbecco's modified Eagle's medium plus 10% serum at and used for and sorting in a flow and cell Germany). of cells was measured in Cell was by forward scatter were used for of senescent fibroblasts were at higher of both and as with cultures The cells were sorted into and for at The sorting were to all cells or the of MRC5 or BJ cells with the or for and For sorting of cells, the were in with MRC5 under identical of cells revealed for and high and sorting was using in sorted cells were in at a of with Zglinicki T. R. Biol. 2000; PubMed Scopus Google Scholar). DNA was completely by at The were in a by The were to and with the telomeric which was with The were by To for of the were with the not was as R.M. 2002; PubMed Scopus Google with the DNA was from cells using DNA DNA was to in The single gene was as control using the and at and of the with an length of for the for both telomere and was using of in a of with and of DNA of conditions were at by of at and at with and was with at by of at and at and were for and for were in an in The were by a well for and for to for in fluorescence in of a DNA of known telomere length of the were in and on the were between telomere and and between standard was for using the DNA which the measured number into the of length is as the of for telomeres to that for The between of telomere length by and was measured in human cell including and different cell with was and were used for of telomerase activity in cells. The was with and of whereas in the of were identical activity of cells was measured as a cells were grown on with and with μg/ml number by μg/ml number For the sorted cells were grown for on with in and with The were in a cells were with for to The sorted cells were in in plus serum and at for in in plus serum for in the cells were with number of in of or control number DNA was with μg/ml for and the were in the with in and in with PBS, the cells were with in for were in μg/ml and for at The cells in were To cells with a senescent phenotype from growing fibroblast populations by the of DNA for telomere length we used senescence that be in cells. fibroblasts are than K. R. K. M. Sci. S. A. PubMed Scopus Google Scholar), and of cells with a senescent phenotype from fibroblast cultures by i.e. FSC, V. Seluanov A. Pereira-Smith J. Biol. Chem. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). fibroblasts as replicative senescence A. PubMed Scopus Google K. T. Zglinicki T. Exp. Gerontol. 2001; PubMed Scopus Google Scholar). as measured by in fluorescence channel is a of cellular M. T. A. Zglinicki T. J. 2000; PubMed Google 23: PubMed Scopus Google Scholar). In fact, senescent fibroblasts show higher for and than In growing populations of there is a fraction of cells with high levels of These cells are for a for cycling cells and often for a of DNA damage Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google Scholar), and most senescent cells. To cells by FACS, we the the to cells with a senescent phenotype the of all cells with the and whereas the including the of all cells with the and sorted cells all cells was used as of the sorted significant of small cells with sorting and of cells sorting fibroblasts were to fibroblasts and to high whereas cells showed growth a senescent The fraction of cells that for activity G.P. G. M. G. C. M. Pereira-Smith Sci. S. A. PubMed Scopus Google was higher in cells as with and than higher in with cells of of their replicative life of BJ fibroblasts not a into the increased the fraction of cells to whereas all cells were sorting We conclude that sorting cells with a senescent in sorted MRC5 and BJ were measured by sorting not telomere length in with sorting through a all cells However, telomere length in MRC5 fibroblasts to to those in senescent cells, because of a higher fraction of short telomeres and (14von Zglinicki T. Petrie J. Kirkwood T.B.L. Nat. Biotechnol. 2003; 21: 229-230Crossref PubMed Scopus (66) Google Scholar). shorter telomeres was obtained for BJ fibroblasts as with not which is in with in telomere shortening rates that to doubling for MRC5 and than for BJ M. Saretzki G. S. Zglinicki T. Biol. 2001; PubMed Scopus Google Scholar). we to a to telomere length to the of cells for senescent cells cannot be by fluorescent in situ and by flow and is in human in senescent with their because of and telomeric J. T. 2002; PubMed Scopus Google R.G. J. K. J. Cell Sci. 2001; Google Scholar). J. Nat. 2003; PubMed Scopus Google only the length of single Thus, we a R.M. 2002; PubMed Scopus Google to telomere in to the of a This and telomere length from as as cells and in a with data a of telomere It the in telomere length between and MRC5 fibroblasts as by These with increasing PD, because of a in telomere length in cells as Thus, human fibroblasts that a senescent phenotype early in mass culture have short telomeres. To whether these were caused by in telomere length or by cell-to-cell variation of the rates of telomere shortening, we MRC5 cells at of subclones, still at a between and In of these telomere length was shorter in cells was variation in telomere length that might be due to variation between the cells for between and cells were to those in mass culture and the in a that was than from senescence at the of between and telomeres was there is heterogeneity in the rate of telomere shortening. The in shortening rates are most because at of the will have through cell than the growing cells and dividing the of We conclude that variation of telomere shortening rate occurs the of a single cell and can cause early telomere in a of human fibroblasts. To whether telomere or causes early senescence in these cells, we MRC5 fibroblasts with hTERT, the catalytic subunit of human telomerase. activity telomere shortening and fibroblasts from senescence at the of the population M. M. Shay J.W. S. W.E. Science. PubMed Scopus Google Scholar). length and growth rate in a mass culture of MRC5 were stabilized for at transfection Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google and data not 5% of into the which of MRC5 and and this fraction all Thus, as in human fibroblast the majority of MRC5 is rescued from senescence by telomere This that telomere shortening causes most of the early senescence in We that these 5% of the would either be with telomeres as as in their cycling counterparts or would be However, had shorter telomeres than cells This was not due to telomerase activity in cells. the telomerase activity in cells under normoxia was even higher than in fibroblasts To these using an independent for senescence, we sorted human fibroblasts to their expression is an fibroblast in which an was in of the gene W. A. Sedivy J.M. Cell. 2003; PubMed Scopus Google Scholar). is an of A. V. Mol. Cell. Biol. PubMed Scopus Google Scholar). In senescence, DNA damage at of telomeres via of and Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google Scholar), to activation of of is an essential of the pathway that human cell into replicative senescence W. Sedivy J.M. Science. 277: PubMed Scopus Google A. T. EMBO J. 2002; 21: PubMed Scopus Google W. S. A. Sedivy J.M. EMBO 2003; PubMed Scopus Google Scholar). Thus, under growth conditions and in the of DNA is an early for human fibroblast replicative senescence. Cell populations with high and fluorescence were by flow cytometry. cells with high indicating high promoter had short telomeres the obtained by telomeres and replicative senescence in human fibroblasts via of DNA damage Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google A. T. Biol. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). is a of these To find out whether short telomeres in fibroblasts would be to the we in sorted cells with that in MRC5 fibroblasts In with data Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google Scholar), is in cultures of MRC5 and cells, a of senescent cells in these It is even in cells. both MRC5 and show high of cells suggesting that the short telomeres in both can the of DNA damage and are telomeres in a of cells despite high telomerase activity in the be due to with the of telomerase to telomeres in these cells. telomeres in these cells shorten so that even high telomerase activity would be to telomere To resolve this we under conditions that accelerate telomere shortening, i.e. under 40% normobaric This increased oxidative stress in MRC5 fibroblasts as by levels of and M. Saretzki G. S. Zglinicki T. Biol. 2001; PubMed Scopus Google Scholar). it increased the rate of telomere shortening and the replicative life span of MRC5 V. Zglinicki T. Lorenz M. Saretzki G. J. Biol. Chem. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). However, there is that it the cells under 40% hyperoxia showed a of growth and of cells in the not in to cells, not a and growth However, at of cells were now the that these cells had short telomeres despite telomerase activity for and cells We conclude that variation in the rate of telomere shortening, under conditions of chronic stress, is than can be for by telomerase. The replicative life span of human cells in culture is the and even the a fraction of cells to from the cell and to a senescent phenotype at every population doubling (3Kill I.R. Faragher R.G. Lawrence K. Shall S. J. Cell Sci. 1994; 107: 571-579PubMed Google Scholar). in culture are to a variety of stresses including oxidative stress, and these stresses can senescence prematurely in a premature premature senescence and replicative senescence not only are also induced via the the DNA damage response and of DNA damage and activation of and Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google A. T. Biol. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). Thus, the early senescence of human cells under standard culture conditions might be either telomere-independent and premature (5Wright W.E. Shay J.W. Nat. Biotechnol. 2002; 20: 682-688Crossref PubMed Scopus (298) Google K. Zou Y. Itahana Y. Martinez J.L. Beausejour C. Jacobs J.J.L. Van Lohuizen M. Band V. Campisi J. Dimri G.P. Mol. Cell. Biol. 2003; 23: 389-401Crossref PubMed Scopus (352) Google or caused by stochastic uncapping of telomeres in a telomere length-independent (10Blackburn E.H. Nature. 2000; 408: 53-56Crossref PubMed Scopus (1099) Google or might be due to heterogeneity in the rate of telomere shortening (11von Zglinicki T. Trends Biochem. Sci. 2002; 27: 339-344Abstract Full Text Full Text PDF PubMed Scopus (1837) Google Scholar). data show that cells that a senescent phenotype early have telomeres that are on as short as those in senescent populations. that this is due to faster telomere shortening in the subclones that senesce early. of telomere length by transfection most cells from early senescence and that the telomere shortening causes early senescence in the majority of cells. However, active telomerase not all cells from senescence. is a small fraction of cells that senesce despite active telomerase. These cells on average, short telomeres as well as containing which is a for DNA damage in response to telomeres Zglinicki T. Saretzki G. Nature. 2003; PubMed Scopus Google A. T. Biol. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). it is to that short telomeres might be to senescence even in a minor fraction of cells, it be that the evidence for that is so far only It is possible that senescence in that it occurs with short telomeres, is by stress, in to telomere shortening, might also that be for telomere-independent senescence. For instance, of p16 an telomere-independent to in human and at fibroblast C. A. M. Campisi J. EMBO J. 2003; PubMed Scopus Google Scholar). p16 is by oxidative stress is not completely (9Itahana K. Zou Y. Itahana Y. Martinez J.L. Beausejour C. Jacobs J.J.L. Van Lohuizen M. Band V. Campisi J. Dimri G.P. Mol. Cell. Biol. 2003; 23: 389-401Crossref PubMed Scopus (352) Google K. A. K. J. Biol. Chem. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). it was that of the single-stranded overhang than shortening of telomeres be the main of senescence R.A. Nat. 2003; PubMed Scopus Google Scholar). telomeric G-rich single-stranded a growth arrest with of senescence G. Zglinicki T. PubMed Scopus Google R. Sci. S. A. 2003; PubMed Scopus Google Scholar). expression of telomerase, which in is not to telomere shortening, to telomere uncapping even in human fibroblasts K. S. J.L. S. S. R.A. Cell. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar), and this be due to of telomeric Thus, data that telomere uncapping can in the fraction of cells that telomere shortening cannot out for stress-induced premature senescence or length-independent telomere uncapping in this minor fraction of cells. It also be that are not in with a in which cells with a senescent phenotype were sorted from fibroblast mass cultures using as sorting V. Seluanov A. Pereira-Smith J. Biol. Chem. 2003; Full Text Full Text PDF PubMed Scopus Google Scholar). Using flow FISH, these were not to in telomere length between the sorted fibroblast populations. this might well be due to of the flow to human fibroblasts with short telomeres and in between the sorted populations. of telomere length would not be by these and we obtained with completely independent sorting In data show that MRC5 fibroblasts in early senescence have on short telomeres and that of telomere shortening by expression also not cells from early senescence. Thus, stochastic variation of the telomere shortening rate even between cells from a single to be the cause of heterogeneity in the replicative life span for most MRC5 human fibroblasts. of oxidative stress and can the rate of telomere shortening (11von Zglinicki T. Trends Biochem. Sci. 2002; 27: 339-344Abstract Full Text Full Text PDF PubMed Scopus (1837) Google Zglinicki T. R. Biol. 2000; PubMed Scopus Google V. Zglinicki T. Lorenz M. Saretzki G. J. Biol. Chem. 2003; Full Text Full Text PDF PubMed Scopus Google G. Zglinicki T. Cell. 2003; PubMed Scopus Google J. J.R. Smith Cell. 2002; PubMed Scopus (173) Google Shay J.W. W.E. Cell. 2003; PubMed Google Scholar). is evidence for increased oxidative stress in the early senescent cells because which was used to these cells, is in a for cellular oxidative stress A. PubMed Scopus Google K. T. Zglinicki T. Exp. Gerontol. 2001; PubMed Scopus Google Scholar). Thus, oxidative stress is a possible cause for the stochastic variation in the telomere shortening This is by the obtained with fibroblasts. oxidative stress increases the fraction of cells with a senescent phenotype and short telomeres. The most for this is that even telomerase expression in high activity is not to compensate telomere shortening in all cells and that the fraction of cells with telomere shortening faster than can be by telomerase increases under This might help to human fibroblast with V. Zglinicki T. Lorenz M. Saretzki G. J. Biol. Chem. 2003; Full Text Full Text PDF PubMed Scopus Google were not by transfection under Shay J.W. W.E. Cell. 2003; PubMed Google Scholar). for the of variation in telomere length in an stochastic at the of a single telomere J. Nat. 2003; PubMed Scopus Google Scholar). the length of individual telomeres in MRC5 it was that most telomeres with increasing telomeres at of the telomeres with a than of the J. Nat. 2003; PubMed Scopus Google Scholar). that all telomeres these data high that at telomere cell might be that are to the of senescent cells in the These data are also in with of strand break in all MRC5 telomeres and the telomere shortening rates Zglinicki T. R. Biol. 2000; PubMed Scopus Google Saretzki G. Zglinicki T. Biol. PubMed Scopus Google Scholar). It is not whether the of telomeres with However, with the that the telomere length at senescence was widely different from to J. Nat. 2003; PubMed Scopus Google Scholar), the data are in with the idea that in the majority of it is not the shortening of telomeres that causes senescence the of stochastic on telomeres. with telomerase have that short telomeres can arrest cell growth C. Cell. 2001; 107: Full Text Full Text PDF PubMed Scopus Google Scholar). Thus, it is possible that the telomere shortening rates measured are by on individual telomeres and the probability of to In data that at the of a single cell telomeres are not senescence is induced by stochastic telomere In stress-induced senescence and replicative senescence are not there a in which telomere shortening is driven to a by stress stochastic shortening by intrinsic mechanisms the end an to the replicative life which might only be damage to telomeres is to the human fibroblast replicative life span is