Oncogene (2001) 20, 3541 ± 3552
ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00
www.nature.com/onc
Human ®broblast replicative senescence can occur in the absence of
extensive cell division and short telomeres
June Munro1, Karen Steeghs1, Vivienne Morrison1, Hazel Ireland1 and E Kenneth Parkinson*,1
1
The Beatson Institute for Cancer Research, CRC Beatson Laboratories, Garscube Estate, Switchback Road., Bearsden, Glasgow,
G61 1BD Scotland, UK
Ectopic expression of telomerase blocks both telomeric
attrition and senescence, suggesting that telomeric
attrition is a mitotic counting mechanism that culminates
in replicative senescence. By holding human ®broblast
cultures con¯uent for up to 12 weeks at a time, we
con®rmed previous observations and showed that
telomeric attrition requires cell division and also, that
senescence occurs at a constant average telomere length,
not at a constant time point. However, on resuming cell
division, these long-term con¯uent (LTC) cultures
completed 15 ± 25 fewer mean population doublings
(MPDs) than the controls prior to senescence. These
lost divisions were mainly accounted for by slow cell
turnover of the LTC cultures and by permanent cell cycle
exit of 94% of the LTC cells, which resulted in many
cell divisions being unmeasured by the MPD method. In
the LTC cultures, p27KIP1 accumulated and pRb became
under-phosphorylated and under-expressed. Also, coincident with permanent cell cycle exit and before 1 MPD
was completed, the LTC cultures upregulated the cell
cycle inhibitors p21WAF and p16INK4A but not p14ARF and
developed other markers of senescence. We then tested
the relationship between cell cycle re-entry and the cell
cycle-inhibitory proteins following subculture of the LTC
cultures. In these cultures, the downregulation of p27KIP1
and the phosphorylation of pRb preceded the complete
resumption of normal proliferation rate, which was
accompanied by the down-regulation of p16INK4A. Our
results show that most normal human ®broblasts can
accumulate p16INK4A, p21WAF and p27KIP1 and senesce by
cell division-independent mechanism(s). Furthermore, this
form of senescence likely requires p16INK4A and perhaps
p27KIP1. Oncogene (2001) 20, 3541 ± 3552.
Keywords: telomerase; telomeres; senescence; cell division; cell cycle inhibitors
Introduction
Human somatic cells are known to have a limited
proliferative lifespan that culminates in the process of
*Correspondence: EK Parkinson
Received 1 February 2001; revised 7 March 2001; accepted 14
March 2001
replicative senescence (Hay¯ick, 1965). Most previous
studies favour the view that senescence is dependent on
the cumulative number of cell divisions rather than
calendar time or metabolic time and hence that
replicative senescence is dependent on a mitotic
counting mechanism, which can apparently remember
the replicative age of the cells when division is halted
for a time (see Cristofolo and Pignolo, 1993 for a
review). One candidate for such a mechanism is the
process of telomeric attrition that takes place when
telomerase-de®cient cells age both in vitro and in vivo
(see Harley, 1991 for a review) and in support of this
hypothesis it has been shown that the extension of
telomere length can delay, or even prevent, the process
of replicative senescence, in all strains of human
®broblasts tested so far (Wright et al., 1996; Bodnar
et al., 1998; Vaziri and Benchimol, 1998; W Wright
2001, personal communication).
Telomere sequences may be lost from human
chromosomes as a result of the end replication
problem (Olovnikov, 1973), exonuclease degradation
(Makarov et al., 1997) or oxidative damage (von
Zglinicki et al., 1995) and whilst it is obvious that the
®rst mechanism would be dependent on cell division,
it is not so certain whether the other mechanisms
would be. Furthermore, recent evidence does not
favour the end replication problem as the major cause
of telomere loss in cultured human ®broblasts (von
Zglinicki et al., 1995; Makarov et al., 1997). Therefore, despite extensive correlative evidence (Harley,
1991; Allsopp et al., 1992), the dependence of
telomeric attrition on cell division has only been
directly tested in two previous studies.
Allsopp et al. (1995) found an insigni®cant level of
telomere shortening and Sitte et al. (1998) no
detectable shortening, when mid-lifespan ®broblasts
were held con¯uent for a period of 7 and 10 weeks
respectively, but these are shorter periods than those
required for the senescence of newly explanted human
®broblast cultures.
To address this question we kept telomerase-negative
human ®broblasts replicatively quiescent, but metabolically active, for the length of time it normally takes
them to senesce. We chose to quiesce the cells by
keeping them con¯uent in high levels of serum, rather
than serum-starved at low density, to ensure that the
metabolic rate of the quiescent test cells was similar to
Accumulation of p16
3542
INK4A
can be cell division-independent
J Munro et al
a
b
c
d
Oncogene
Accumulation of p16
J Munro et al
that of the dividing controls (see Sitte et al., 1998 and
references therein).
The results con®rmed and extended previous reports
that telomeric attrition is strongly dependent on cell
division, both in vitro (Allsopp et al., 1995; Sitte et al.,
1998) and in vivo (Allsopp et al., 1995). However, the
vast majority of the cells were very susceptible to the
induction of senescence by long-term con¯uence
(LTC). This form of senescence was not associated
with extensive cell division or telomeric attrition and
challenges the idea that replicative senescence is solely
dependent on cumulative cell divisions and telomeric
attrition.
Results
Telomeres do not shorten in LTC cultures of
human fibroblasts
The LTC cultures were shown to be very quiescent in
that only 10 ± 15% of the cells cycled in any 48 h
period following a change of medium and less than
half of the population cycled in 1 week. The telomere
length of the quiescent LTC human ®broblast cultures
was followed from 0 ± 10 weeks (Figure 1a,b) and no
signi®cant attrition took place during this time,
INK4A
can be cell division-independent
con®rming the results of Sitte et al. (1998), who used
WI 38 cells. Telomere length was also followed for
longer periods of time following successive periods of
expansion and LTC (Figure 1c) and in this case, the
total length of time in vitro corresponded to the time
normally required for the senescence of the HFF
®broblast strain (40 weeks). It can be seen from these
results that the telomeres do not shorten appreciably
during this time, even though the control cultures
shortened their telomeres from 10 to 3 kb (Figure
1c,d) and were approaching senescence as determined
by BrdU labelling index and senescence-associated bgalactosidase activity (SA-b-galactosidase). In both
the control cultures and the LTC cultures, telomere
length correlated closely with the number of MPDs
remaining (Figure 1e), rather than calendar time
(Figure 1d). Furthermore, the limited amount of
telomere shortening that did occur was closely
correlated with the number of cell divisions completed, once the appropriate corrections were made
for cryptic cell turnover in the LTC cultures and the
loss of many clonogenic cells during the period of
LTC (see Table 1 and below). The results, therefore,
support the idea that telomeric attrition is related to
the total number of cell divisions completed, rather
than calendar time, metabolic time or total in vitro
time.
3543
e
Figure 1 Telomere length correlates with the number of MPDs remaining, not with calendar time (a) Southern blot of the RsaI/
HinfI digested genomic DNA of HFF cells held con¯uent for 0 ± 10 (LTC) weeks probed with a biotinylated 51mer telomere probe
(see Materials and methods). (b) A plot of the average telomere length derived from the Southern (a) showing that the decrease in
telomere length is minimal over this period. (c) Southern blot of the TRFs of control HFF cultures and HFF cultures subjected to
successive periods of LTC and subculture for a total of 242 days after an initial period of 35 days of standard cell culture
conditions. To test whether the mass culture was senescent, the cells were harvested after 17 days of standard cell culture conditions
following each period of LTC, stained for SA-b-GAL and assayed for BrdU incorporation over 48 h. It can be seen that the per
cent of cells incorporating BrdU declines with time in the control cultures, and the per cent expressing SA-b-GAL increases, whilst
only minimal changes occur in the cultures subjected to repeated LTC. (d) The average telomere length of the cultures plotted
against calendar time. This shows that whilst there is clear telomeric attrition of the control cultures (circles), the cultures that were
restricted in the numbers of times that they could divide by successive periods of con¯uence (triangles) showed very little telomere
shortening. The 3.3 kb value of the completely senescent control cultures was derived from a separate Southern blot using the `58
MPD remaining' cells as a standard. (e) The same telomere data as in (d) now plotted against the number of population doublings
remaining for each culture. The LTC telomere lengths (triangles) are within the 95% con®dence limit of the regression analysis of
the control telomere lengths, implying that they may be ®tted to the same straight line
Oncogene
Accumulation of p16
INK4A
can be cell division-independent
J Munro et al
3544
The LTC cultures can still undergo replicative senescence
The quiescent LTC cultures were not senescent, since
they had around 32 MPDs left when they were ®nally
released from quiescence (Figure 1c,e) and at the end
of their replicative lifespan, around 12 weeks later, the
telomere length had shortened to the same length as
control senescent cultures that were allowed to
proliferate under standard culture conditions with
regular sub-culturing (data not shown). Furthermore,
these same cultures became SA-b-galactosidase positive, upregulated the senescence markers IGFBP3
(insulin-like growth factor binding protein 3) and
MMP-1 (matrix metalloproteinase 1) (data not shown)
and less than 20% incorporated BrdU in 48 h.
Interestingly, neither the control, nor the LTC cultures
upregulated p16INK4A upon reaching senescence. This
has been noticed before, when dermal ®broblasts
(Shelton et al., 1999, W Wright, 2001, personal
communication), rather than lung ®broblasts (Hara
et al., 1996; Alcorta et al., 1996) are used; the reason
for this is unclear. Nevertheless, our results further
support the telomeric attrition hypothesis of replicative senescence and illustrate that the long periods of
con¯uence do not cancel the programme of replicative
senescence in these cells or select for variant cells that
had an extended replicative lifespan (see Table 1 and
below).
The apparent replicative lifespan of the LTC fibroblasts
is reduced
The LTC quiescent cells would be expected to have a
total replicative lifespan of around 70 MPD based on
the lifespan of the control cultures. However, the total
lifespan of the culture that had been held repeatedly
con¯uent for 242 days was only 50 MPD (Table 1). As
cells may be constantly `exfoliated' from the surface of
long term con¯uent cultures, or otherwise not revealed
by an increase in cell number, we measured the
labelling index of our cultures over a 7 day period in
order to establish the number of extra MPDs that
would be predicted to take place over the test period.
We found that the BrdU labelling index was not
constant throughout the 10 week period of con¯uence.
The cultures took 15 days to complete 1 MPD during
Table 1
The relative contributions of cryptic cell turnover and permanent cell cycle exit to the loss of total replicative lifespan, following
various periods of long-term con¯uence in human ®broblast cultures
Total time of repeated
LTC (days)b
81
133
190
228
242
Control
a
the ®rst 3 weeks of the test period but 44 days/MPD in
the remaining 7 weeks. These estimates translate into a
total of 2.9 ± 9.6 extra MPD completed by the cultures
held con¯uent for 81 ± 242 days (Table 1) but this does
not necessarily put them within the range of variation
of control cultures (52 ± 61 MPD versus 70 MPD;
Table 1), even though control cultures can vary in
lifespan by up to 10 MPD (Holliday, 1996; Goldstein
and Singal, 1974). We also noticed by routine
observation of our LTC cultures, that when they were
disaggregated and replated, although the viability was
high, the cultures were slow to complete 1 MPD. Even
48 h after subculturing, the BrdU labelling index was
low, many cells appeared ¯at and approximately ®ve
times as many cells as expected, stained weakly positive
for SA-b-galactosidase (Figure 2). We therefore
investigated whether some of the clonogenic cells of
the LTC HFF culture had failed to re-enter the cell
cycle following the period of LTC. If this number was
signi®cant, then the number of extra divisions needed
to repopulate the culture by the remaining clonogenic
cells could account for the remaining lost MPDs.
Figure 3 shows that this is indeed the case and that
only 6% of the clonogenic HFF remained after a 10
week period of LTC, when compared with recently
con¯uent MPD-matched control cultures. Such a
reduction in cloning eciency would result in four
extra population doublings being required to repopulate the HFF culture after each period of LTC.
However, we went on to test whether the cells
recovered from the ®rst period of LTC could
regenerate cells that were sensitive to LTC-induced
senescence or were a distinct ®broblast population.
These results showed that a culture that had been
subject to 81 days of LTC and then cultured through 3
MPD was now completely resistant to LTC-induced
senescence (Figure 4). So these cultures and all LTC
cultures harvested subsequently, would be expected to
have lost an extra 4 MPD (Table 1). When this
additional correction was made to the total replicative
lifespan completed by the experimental LTC groups,
most cultures were now within the expected range of
variation of the controls (56 ± 65 versus 70 MPD; Table
1, see Holliday, 1996 and also above). There was no
obvious trend towards a longer replicative lifespan with
each successive period of LTC (Table 1).
Total culture
time (days)a
Apparent replicative
lifespan (MPD)
MPD lost by cell cycle
exit during LTC
MPD lost by cryptic
cell turnover
Corrected lifespan
(MPD)
116
168
225
263
279
281
53.8
47.2
45.3
51.7
50.2
70.0
4.0
4.0
4.0
4.0
4.0
n/a
2.9
5.1
7.4
9.3
9.6
n/a
60.8
56.3
56.7
65.0
63.8
70.0
Data includes the period of low density culture (7 ± 10 days) between each period of LTC. bData includes the HFF proliferation prior to the
commencement of the experiment (35 days)
Oncogene
Accumulation of p16
J Munro et al
INK4A
can be cell division-independent
3545
Figure 2 The morphology and SA-b galactosidase activity of LTC HFF cultures 48 h after release from con¯uence. (a) Immortal
human ®broblast line GM847 (negative control). (b) Control HFF (MPD level 26). (c) LTC HFF 48 h after release from 10 weeks
of con¯uence. The arrows point to cells faintly stained for SA-b galactosidase, the lower one of which is very ¯at. (d) Senescent
HFF (MPD level 68; positive control). Bar=10 mm
Figure 3 Loss of clonogenicity in LTC HFF cultures. (a)
Control HFF culture (24 h con¯uent, MPD level 26; positive
control) 500 cells/plate. (b) 10 week LTC HFF culture 5000 cells/
plate. (c) 10 week LTC HFF culture 500 cells/plate. (d) Senescent
HFF culture (MPD level 68; negative control) 5000 cells/plate.
No more colonies formed in (c) and (d) when the plates were left
for up to 8 weeks
These results, together with the small number of
divisions accounted for by slow turnover of the
con¯uent cultures, largely explained the lost MPD
from the LTC mass cultures. However, the few
remaining lost divisions could be accounted for by
the breakage and rapid shortening of the telomeres as
previously reported for WI 38 LTC cultures (Sitte et
al., 1998).
Collectively, our results were in agreement with
previous studies that telomere length is strongly
correlated with the remaining lifespan of mass
human ®broblast cultures (Allsopp et al., 1992)
and that telomeric attrition requires cell division
Figure 4 One period of LTC selects for HFF cells that are
resistant to LTC-induced senescence. The HFF cells were allowed
to remain con¯uent for 81 days and then cultured through 3
MPD. These cells were then allowed to become con¯uent again
and assayed immediately or after 10 weeks for cloning eciency
(see Figure 3 and Materials and methods). The solid histogram
bar represents the cloning eciency in the LTC-pre-treated group
and the open bar the cloning eciency in the control, which were
calendar time-matched cultures not previously subjected to LTC.
The values given are cloning eciencies, relative to the respective
cloning eciencies at zero time
(Allsopp et al., 1995; Sitte et al., 1998). However,
they also suggested a telomeric attrition-independent
mechanism of senescence, so we investigated this
further.
Oncogene
Accumulation of p16
INK4A
can be cell division-independent
J Munro et al
3546
The loss of clonogenicity in LTC HFF cultures is a
permanent cell cycle arrest that requires cell cycle exit
and perhaps p16INK4A, p21WAF and p27KIP1
When the HFF cells were released from the 10 weeks
of LTC more than 95% of them attached to the culture
dish and there was no subsequent evidence of cell
death. Furthermore, 10% of the culture consisted of
small cells that incorporated BrdU within 4 h. The
remaining fraction consisted of large cells that did not
incorporate any BrdU in 48 h. Although we have not
measured phenomena such as apoptosis directly, we
did not notice any evidence of nuclear fragmentation in
the BrdU experiments where the cells were counterstained with propidium iodide, nor did we observe any
cellular blebbing, cell detachment, or a reduction in the
number of cells per plate. These results strongly
suggested that most LTC cells, which have a G0/G1
DNA content on entering the con¯uent state (Dietrich
et al., 1997, data not shown), fail to re-enter the cell
cycle following the period of LTC.
The cell cycle inhibitors p21WAF and p16INK4A
accumulated in the LTC cultures (Figure 5a,b) in a
similar fashion to that of senescent cultures (Alcorta et
al., 1996). The p21WAF protein accumulated to levels
approximately fourfold higher than in the zero time
con¯uent controls and then declined (Figure 5b),
whereas p16INK4A protein accumulated rather later, from
4 weeks, and was sustained at approximately 15 ± 16
times the level found in the proliferating and zero time
controls (Figure 5a), even when p21WAF had begun to
decline (Figure 5b). These observations show that cells
Figure 5 Induction of a senescence-like programme in LTC HFF cultures. (a) The p16INK4A levels rise sharply between weeks 4
and 8 where they are 15-fold higher than at zero time and are maintained at week 10. The two lanes on the far right are SCC-13
(negative control) and SVHFK (positive control). (b) The p21WAF levels rise steadily between weeks 2 and 8 post-con¯uence
reaching levels of fourfold those at zero time. They then decline to threefold the zero time value by week 10. The lanes on the far
right are controls. HaCat (p21WAF low); HFF 20 h after 4 Gy of g irradiation (positive control); HFF actively growing control cells.
(c) The p27KIP1 levels rise fourfold within 2 weeks of the HFF cells becoming con¯uent but do not increase further for the 10 weeks
of the experimental period. The controls are HeLa cells (Positive control; control lane 1) and BICR31 cells (Low control; control
lane 2). (d) The p53 levels are lower than in the actively growing HFF cells (lane 6) and remained fairly constant throughout the 10
week period. The SaOS-2 cells (control lane 1) and the SVHFK cells (control lane 2) were used as negative and positive controls
respectively. (e) The p14ARF levels were undetectable at all time points and in the actively proliferating HFF controls (lane 6) or
irradiated HFF (lane 8) but were clearly detected in the SaOS-2 positive control (control lane 1). The extremely low levels of p14ARF
are typical of normal human cells with intact pRb and p53 genes. (f) The cyclin B1 protein was only 6% of the level of the actively
proliferating control HFF (lane 6) by 2 weeks postcon¯uence and declined progressively until it was only 0.5% of the control's at
week 10. It was also reduced by irradiation (control lane 1). The SaOS-2 (control lane 2) served as a positive control. (g) The MMP1 protein accumulated to around 2 ± 3-fold the level of the zero time levels by weeks 8 ± 10 post-con¯uence. These levels were also
higher than in the immortal cell lines SCC-13 and SVHFK (control lanes 1 and 2). The apparent reduction in MMP-1 levels at week
2 is accounted for by loading. (h) The IGFBP3 levels began to rise by week 2 post-con¯uence (6.5-fold) rose steadily to week 8 (12fold) and then sharply by week 10 to 40-fold the zero time level. The 10 week level was also 10-fold higher than the proliferating
control HFF cells (control lane 3) and their g irradiated counterparts (control lane 2)
Oncogene
Accumulation of p16
J Munro et al
do not need to divide extensively to accumulate the cell
cycle inhibitors p16INK4A and p21WAF. In addition,
p27KIP1 accumulated to fourfold the levels at zero time
after 2 weeks of con¯uence as reported previously for
con¯uent ®broblast cultures (Polyak et al., 1994) but
did not accumulate further during the 10 weeks of LTC
(Figure 5c).
The p53 (Figure 5d) and p14ARF (Figure 5e) proteins
did not increase in the LTC cultures, which is similar
to results previously reported for senescent human
®broblasts (Vaziri et al., 1997; Robles and Adami,
1998), as is the decline in cyclin B1 levels (Figure 5f,
see Stein and Dulic, 1995). We also tested for the
accumulation of proteins that have previously been
linked to the senescent state. The human interstitial
collagenase (MMP-1) was present at high levels as the
cells approached con¯uence but accumulated approximately threefold during the con¯uence period (Figure
5g) and IGFBP3 accumulated up to 40-fold after 10
weeks of con¯uence (Figure 5h). The data taken
together suggests that the LTC cultures enter an
irreversible G1 arrest state that is very similar, though
not identical, to replicative senescence. LTC-induced
senescence is distinct from the G1 arrest induced by
DNA double-strand breaks (Robles and Adami, 1998)
or by the overexpression of certain activated oncogenes
(Serrano et al., 1997; Palmero et al., 1998), that appear
to mediate their e€ects via the induction of p14ARF and/
or the stabilization of p53.
Fibroblast cultures that harbour SV40 T antigen did
not exit the cell cycle under conditions of LTC and did
not undergo a reduction in cloning eciency on
transfer, suggesting that cell cycle exit was an
important prerequisite for LTC-induced senescence
(data not shown).
HFF cells recovering from LTC-induced senescence
down-regulate p16INK4A and p27KIP1
To test the relationship between cell cycle re-entry and
the cell cycle-inhibitory proteins, we monitored the
proliferation rate of the HFF cells and the expression
of the cell cycle inhibitory proteins when the LTC
cultures were subcultured. We were able to show that
p27KIP1 was downregulated and pRb became phosphorylated well in advance of the proliferation rate of
the cells returning to normal as assessed by both MPD
completed per week and by the BrdU-labelling index
(Figure 6a,b). However, the cells began to proliferate
normally once p16INK4A levels had also returned to
normal (Figure 6c). P21WAF on the other hand, was not
down-regulated on returning the HFF cells to a
subcon¯uent state (data not shown). These results
suggest that p16INK4A and possibly p27KIP1, but not
necessarily p21WAF, are co-operative e€ectors of LTCinduced senescence.
We also tested whether p16INK4A was upregulated in
LTC cultures prepared from HFF cultures, previously
subjected to one round of LTC (see Figure 4). P16INK4A
still accumulated (data not shown), indicating that the
ability to downregulate p16INK4A upon subculture,
INK4A
can be cell division-independent
3547
pRb
p27
p16
ERK
Figure 6 A reduction in p16INK4A levels is associated with the
recovery of normal rates of multiplication of LTC cultures. The
®gure shows Western blots of the di€erent cell cycle proteins as
the LTC cultures recover and begin to proliferate normally. The
p27KIP1/ERK2 and p16INK4A/ERK2 ratios are normalized to the
ratios of the zero time LTC cultures. The return of p27KIP1 levels
to normal levels correlates with a return to near normal pRb
phosphorylation but normal proliferation rates are only restored
once p16INK4A declines to below the level found in con¯uent
cultures at zero time. (a) pRb, ppRb indicates the phosphorylated
forms of the protein. (b) p27KIP1. (c) p16INK4A. (d) ERK2
rather than its lack of upregulation, may be more
important for the resistance to LTC. We also examined
LTC cultures prepared from near-senescent HFFs,
which cannot recover from LTC at all. These LTC
cultures still elevate p16INK4A and do not downregulate
it on subculture, consistent with the above hypothesis.
LTC-induced senescence is not blocked by the
ectopic expression of telomerase
To further understand the nature of LTC-induced
senescence we reconstituted telomerase activity in the
telomerase negative ®broblast strain HFF by means of
an amphotropic retrovirus (Vaziri and Benchimol,
1998). We chose to work with infected pools of cells
rather than clones because we had already established
that 6% of HFF cells were very resistant to LTCinduced senescence (see above). We established that
12/12 G418-resistant HFF colonies that received the
hTERT virus were strongly telomerase positive and
that 13/13 control colonies were not. Furthermore, we
established that the pool of G418-selected hTERTexpressing cells did have an extended proliferative
lifespan, as reported for other human ®broblasts
strains (Bodnar et al., 1998; Vaziri and Benchimol,
1998; Kiyono et al., 1998). Despite this, the ectopic
Oncogene
Accumulation of p16
3548
INK4A
can be cell division-independent
J Munro et al
expression of hTERT in the LTC cultures failed to
block senescence (Figure 7a,b), arguing that LTC does
not mediate its e€ects via telomeric attrition.
The ectopic expression of hTERT in the LTC
cultures also failed to block the accumulation of
p16INK4A (Figure 7c) and p27KIP1 (data not shown).
Furthermore, the LTC cultures expressing hTERT still
showed underphosphorylated pRb (data not shown).
Although it is formally possible that the prematurely
senescent LTC cells each have one critically short
telomere, ectopic hTERT expression would be expected
to correct this (Bodnar et al., 1998; Vaziri and
Benchimol, 1998). We also tested the average telomere
length in HFF/TERT cells and HFF/NEO cells and
the results are shown in Figure 7d. It can be seen from
these data that the telomere lengths of the HFF/TERT
cells (13.1 kb) are slightly higher than the HFF/NEO
cells (11 kb), and indistinguishable from the early
passage HFF cells (14 kb), consistent with hTERT
maintaining telomere length.
Discussion
It has long been known that human ®broblasts possess
a limited replicative lifespan in vitro and telomeric
attrition has been proposed as the mitotic counting
mechanism that underpins this limit (Harley, 1991).
Several lines of evidence now suggest that this
hypothesis may be correct, at least for dividing human
®broblasts. The ectopic expression of telomerase in
human ®broblasts can prevent both telomeric attrition
Figure 7 The reconstitution of telomerase in HFF does not antagonize LTC-induced senescence or the accumulation of p16INK4A.
(a) The loss of cloning eciency after 10 weeks LTC is not antagonized by hTERT expression. Upper panel left HFF/NEO 2 days
con¯uent (zero time); Upper panel right HFF/NEO after 10 weeks of LTC; Lower panel left HFF/TERT 2 days con¯uent (zero
time); Lower panel right HFF/TERT after 10 weeks of LTC. No further colonies formed even when the plates were left for up to 8
weeks (see also Figure 3). (b) The colony counts from two of the above experiments relative to the zero time controls. The values are
the means of four experiments+standard deviation. The LTC values were expressed as a percentage of the zero time controls. (c)
Telomerase reconstitution does not block the induction of p16INK4A by LTC. Top panel, p16INK4A Western blot of HFF/NEO
cultures after 2 days of con¯uence (zero time; HFF NEO CONF.) or in exponential phase (HFF NEO EXP.) and the same groups
for HFF/TERT (HFF TERT CONF. and HFF TERT EXP.). There is a large induction of p16INK4A protein in both HFF/NEO
(HFF NEO 10 WEEKS) and an equal accumulation in HFF/TERT (HFF TERT 10 WEEKS) after 10 weeks of LTC. Controls
included HFF late pass controls (HFF.21), the same cells irradiated with 4 Gy (HFF.21 4 Gy), the p16INK4A deleted line BICR31
and the p16INK4A over-expressing line SVHFK. Bottom panel, the same blot re-probed with an antibody to ERK2. (d) Telomere
Southern blot showing that the average telomere length of HFF/TERT is very similar to those of HFF/NEO and does not exceed
the length of early pass HFF. Lane 1 HFF NEO, lane 2 HFF/TERT, lane 3 HFF/MPD 12
Oncogene
Accumulation of p16
J Munro et al
and replicative senescence (Bodnar et al., 1998) without
transforming the cells (Morales et al., 1999; Jiang et al.,
1999). Furthermore, the targeted disruption of the
telomerase RNA gene in mouse cells (Lee et al., 1998),
or the disruption of telomere caps (van Steensel et al.,
1998; Karlseder et al., 1999), create dicentric chromosomes and trigger proliferation arrest or apoptosis.
However, for telomeric attrition to be considered as a
mitotic counting mechanism, it should be demonstrated
that it is dependent on cell division rather than
calendar time (see Cristofolo and Pignolo, 1993). Here,
we con®rm and extend previously published work on
other ®broblast strains (Allsopp et al., 1995; Sitte et al.,
1998) and show that telomeric attrition is strongly
dependent on extensive cell division and that even over
a period of nearly 10 months of intermittent con¯uence, very little telomeric attrition takes place.
Virtually all telomere loss could be accounted for by
cell division in the mass HFF ®broblast cultures and all
telomere loss was linked with a decrease in replicative
lifespan, as predicted by earlier studies (Allsopp et al.,
1992). These results support the candidacy of telomeric
attrition as a mitotic counting mechanism in human
®broblasts (Harley, 1991).
During the course of the above experiments, we
noticed, as others had (Goldstein and Singal, 1974;
Sitte et al., 1998), that when human ®broblasts were
held con¯uent for long periods in the presence of high
concentrations of serum growth factors, the remaining
proliferative lifespan was reduced by around 0.5 MPD
per week of con¯uence. Furthermore, such cultures are
slow to recover following disaggregation and replating
(Goldstein and Singal, 1974, this study), suggesting
that a signi®cantly large subset of cells were undergoing premature senescence. The data presented in
Figures 2, 3 and 5 clearly show that this is the case and
that the mass population is regenerated by a very small
subset of LTC-resistant HFF cells (Figure 4) whose
replicative senescence is totally dependent on telomeric
attrition and extensive cell division, thus explaining
earlier notions that the replicative senescence of mass
cultures is tightly linked to the number of cell divisions.
It has been shown that many cells in human ®broblast
cultures can senesce prematurely and that this is
unrelated to their proliferative history (Smith and
Whitney, 1980) but it is unclear whether this stochastic
senescence and LTC-induced senescence are mechanistically linked. It has, in fact, been proposed that
stochastic senescence can be explained by abrupt
telomere loss (Rubelj and Vondracek, 1999) but this
idea has not yet been experimentally tested.
Our data suggests that some clones of ®broblasts
may be resistant to LTC-induced senescence and it
remains to be seen whether all ®broblast strains behave
in the same way as HFFs. It is well documented that
there is clonal heterogeneity within human ®broblast
populations and that embryonic ®broblasts have
di€erent properties from adult ®broblasts (Schor and
Schor, 1987). The molecular pro®le of senescent human
®broblasts can also vary between di€erent ®broblast
strains (Hara et al., 1996; Alcorta et al., 1996; Morales
INK4A
can be cell division-independent
et al., 1999; Shelton et al., 1999). One could speculate
that, as some ®broblasts are resistant to LTC-induced
senescence, senescence by this mechanism may be
associated with a later stage of development or
di€erentiation. This clearly deserves further investigation.
The pattern of cell cycle arrest provoked by LTC
was very similar to the G1 arrest seen in replicatively
senescent human ®broblasts (Alcorta et al., 1996;
Vaziri et al., 1997) in that there was no increase in
either p14ARF (Robles and Adami, 1998) or p53 protein
(Vaziri et al., 1997). These results also distinguish the
LTC cell cycle arrest from that seen by the overexpression of activated oncogenes (Serrano et al., 1997;
Palmero et al., 1998) and from mouse ®broblast
senescence (Kamijo et al., 1999), neither of which are
associated with telomeric attrition or extensive cell
division. No elevation of p27KIP1 transcript (Wong and
Riabowol, 1996) or protein (McConnell et al., 1998)
has been observed in senescent human ®broblasts that
have reached the Hay¯ick limit. Furthermore, the
elevated levels of p27KIP1 protein in the LTC cultures
are fully consistent with those previously reported for
quiescent cells that re-enter the cell cycle when returned
to the subcon¯uent state (Polyak et al., 1994).
However, an elevation of p27KIP1 in concert with
high levels of p16INK4A was recently reported in
senescent thyroid epithelial cells (Jones et al., 2000).
As in the study of Jones et al. (2000), we were also
able to show that LTC-induced senescence, which
occurs in the absence of cell division and detectable
telomeric attrition, could not be blocked in HFF
cells by ectopically expressing telomerase, arguing
against a telomere-based mechanism for LTC-induced
senescence.
In both our study and that of Jones et al. (2000), it is
uncertain whether the presence, let alone the accumulation, of the cell cycle inhibitors, is instrumental in the
permanent cell cycle arrest. However, SV40 T antigen
antagonized both cell cycle exit and senescence in LTC
cultures and it is documented that, compared to wildtype cells, oligodendrocyte precursor cells from p27KIP1de®cient mice exit the cell cycle less eciently (Durand
et al., 1998) and mouse embryo ®broblasts from
p16INK4A-de®cient mice proliferate to a higher saturation
density at con¯uence (Serrano et al., 1996). Furthermore, the ectopic expression of these cell cycle
inhibitors in human ®broblasts induces a cell cycle
arrest reminiscent of replicative senescence (McConnell
et al., 1998; Collardo et al., 2000). One possible
explanation for LTC-induced senescence may be
connected to the recent observation that inhibition of
PI3 kinase in human ®broblasts elicits senescence
(Tresini et al., 1998) by a p27KIP1-dependent mechanism
(Collardo et al., 2000). Ras-mediated PI3 kinase
activity is thought to be responsible for the degradation of p27KIP1 that occurs in late G1 as quiescent cells
re-enter the cell cycle (Takuwa and Takuwa, 1997) and
low PI3 kinase activity in the LTC cultures could
contribute to the accumulation of p27KIP1. This
hypothesis would also be consistent with the observa-
3549
Oncogene
Accumulation of p16
3550
INK4A
can be cell division-independent
J Munro et al
tion that ectopic expression of mutant Ras, in certain
settings, can overcome the cell cycle block imposed by
p27KIP1 and extend proliferative lifespan in thyroid
epithelial cells (Jones et al., 2000). However, the
inhibition of PI3 kinase is unlikely to explain the
accumulation of p16INK4A that is seen in both our LTC
cultures and in senescent thyroid epithelial cells (Jones
et al., 2000), since the induction of senescence with PI3
kinase inhibitors is associated with reduced p16INK4A
levels (Collardo et al., 2000).
Regardless of the mechanism of LTC-induced
senescence, the accumulation of the cell cycle inhibitors, without intervening cell division or telomeric
attrition, challenges the idea that the quantal increase
of these inhibitors with each successive cell division,
can act as a reliable mitotic counting mechanism within
the context of replicative senescence (Hara et al., 1996;
Mazars and Jat, 1997; Jones et al., 2000, see also
Durand et al., 1997). The results also question further
the assertion that cell division and telomeric attrition is
necessarily required for a replicative senescence like cell
cycle arrest (Harley, 1991; Cristofolo and Pignolo,
1993), even in the absence of an over-expressed
activated oncogene (Serrano et al., 1997).
Although human ®broblasts never reach the kind of
densities in vivo that the LTC ®broblast cultures
experience in vitro, other quiescent tissues in vivo, such
as liver, do. Furthermore, the observed accumulation
of p16INK4A protein in the non-dividing LTC ®broblast
cultures and its tight association with the phenomenon,
o€ers a possible explanation for the observed increase
in p16INK4A transcript in non-dividing mouse tissues in
vivo with increasing donor age (Zindy et al., 1997).
These mouse tissues, like LTC cultures, would also be
expected to be composed of cells with long telomeres
(Kipling and Cooke, 1990) which would remain long
throughout the life of the animal (Allsopp et al., 1995).
The LTC cultures o€er a possible system for
investigating telomere-independent senescence mechanisms in human cells.
Materials and methods
Cell culture
The human foetal ®broblast strain HFF was derived by
collagenase digestion of 16-week-old foetal skin (Burns et al.,
1993) and was passaged every week at a density of 26104
cells/sq cm in Dulbecco's Modi®ed Eagles Medium containing 20% vol/vol foetal bovine serum (FBS). The cells were
98% negative for desmin showing them to be undi€erentiated
®broblasts and not muscle satellite cells. The cells were also
negative for telomerase activity (see below). The cells took
approximately 40 weeks to senesce, as assessed by the
presence of the SA-b-galactosidase (Dimri et al., 1995) in
more than 50% of the cells and a labelling index of less than
5% when the cells were incubated with BrdU for 48 h. The
cell inputs (No) and cell yields (N) were recorded at each
passage and the population doublings calculated from the
formula PDL=3.32 (log10 N ± log10 N0). After 10 population
doublings some cultures were allowed to remain con¯uent,
Oncogene
whilst parallel cultures were split once a week. The con¯uent
cultures could be left con¯uent for up to 10 weeks after which
time they were disaggregated with collagenase (Sigma type IA
for 15 min in DMEM) followed by 0.05% trypsin (Worthington; Lorne Laboratories Limited, Reading, UK) and 0.01%
EDTA (BDH Limited, Poole, UK), counted, re-plated and
allowed to become con¯uent once more. Failure to replate at
intervals, resulted in the peeling of some cultures from the
dish and their consequent loss. The cultures were labelled
with BrdU for 2 ± 7 days every 2 weeks and the cell cycle time
of the con¯uent cells was estimated from the labelling indices.
From this, the number of extra cell divisions due to cell
turnover in the con¯uent monolayer could be calculated. At
the same time points cell samples were recovered for telomere
length measurement and Western analysis. Further samples
were recovered after repeated periods of con¯uence at the
times indicated in the ®gures.
Colony forming efficiency
HFF cells were disaggregated as indicated above using
collagenase and trypsin and were plated at 500 or 5000 cells
per 9 cm dish together with 56104 lethally irradiated Swiss
3T3 cells to act as a feeder layer. The plates were medium
changed once weekly and ®xed with 10% formalin after 4
weeks. The colonies were then stained with 20% Giemsa
stain, washed in water and air dried. Only colonies of 50 cells
or more were scored. Some plates were left for 8 weeks to
check for exceptionally slow growing colonies.
Retrovirus production and ectopic expression of hTERT
The retroviruses were produced by transfecting the amphotropic packaging line PA317 with pBabeNEO or pBABest2
(Vaziri and Benchimol, 1998) and selecting with G418. When
the cells were 80% con¯uent they were fed with medium
lacking G418 and the supernatants recovered ®ltered and
frozen 24 h later. The HFF cells were plated at 46105 per
5 cm dish and infected 24 h later with the viral supernatants
containing 4 ± 8 mg/ml of polybrene. Twenty-four hours later
the cells were disaggregated and seeded at 36105 per 9 cm
plate and selected 48 h later with 200 mg/ml of G418 for 10
days. Pooled G418-resistant colonies were used in all
experiments to avoid variation due to clonal heterogeneity
of the normal HFF population.
Telomerase activity was assayed by the telomere repeat
ampli®cation protocol (TRAP) using the TRAPEZE Telomerase Detection Kit (Intergen Company, Oxford, UK)
according to manufacturer's instructions. 0.2 mg protein was
analysed for each sample. Products were resolved on a 10%
non denaturing polyacrylamide gel and products visualized
by staining the gel with SYBR Green (Molecular Probes
Europe BV, Leiden, The Netherlands).
Determination of labelling index by BrdU incorporation
HFF cells were plated at 36103 cells/sq cm in chamber slides
(Labtek, Nalge Nunc International Inc., Naperville, USA) for
24 h prior to labelling for the desired intervals with 10 mM
BrdU (Sigma Chemical Co., Poole, UK). The cells were
washed in phosphate-bu€ered saline (PBS), ®xed in 1 : 1
methanol/acetone for 10 min, washed in PBS again for
10 min and permeabilized for 10 min in 0.025% Tween 20
in PBS. The cells were then denatured with 1 M HCl at 608C
for 5 min. After blocking in 10% FBS, the BrdU was
visualized by incubation with an anti BrdU antibody (Harlan
Sera-Lab. Ltd., Loughborough, UK) at 48C overnight and an
Accumulation of p16
J Munro et al
FITC-conjugated anti-rat IgG (Sigma Chemical Co., Poole,
UK) at a dilution of 1 : 125 for 60 min in the dark. Before
and after each antibody incubation, the cells were washed in
10% FBS/1xKRH/Tween 80. The nuclei were counterstained
with propidium iodide (1 ng/ml) immediately prior to
mounting using Vectashield (Vector Laboratories, Peterborough, UK) and visualization on the ¯uorescent microscope.
Controls included cells that did not receive BrdU, exponentially proliferating HFF and 3T3 cells (positive controls) that
did receive BrdU and the same cells after 48 h in 0.5% FBS
(negative controls) that also received BrdU.
Telomere length measurement
Genomic DNA was extracted using the QIAamp tissue kit
(Qiagen, Crawley, West Sussex, UK). For telomere length
comparisons, 1 or 10 mg of genomic DNA was digested with
RsaI and HinfI restriction enzymes, separated on a 0.6% w/v
agarose gel, and transferred to HybondN+ membrane
(Amersham International, Amersham, UK) in 0.4 N NaOH.
The blot was hybridized with a biotinylated 51-mer telomere
probe (Teloquant kit, Pharmingen, Becton Dickinson, Cowley, Oxford, UK) and subsequently washed according to the
manufacturer's instructions. Non-speci®c binding sites were
blocked in 46SSC/0.05% Tween 20/16 blocking solution
(Boehringer Mannheim, Roche, Lewes, East Sussex, UK) for
1 h at room temperature, followed by incubation with
100 ng/ml streptavidin (Teloquant kit, Pharmingen, Becton
Dickinson, UK) in the same blocking solution. The
membrane was washed for 15 min with 46SSC/0.05%
Tween 20 and then developed with enhanced chemiluminescence substrate (ECL).
The telomere lengths were estimated according to the
instructions provided by the Teloquant kit. Brie¯y, the signal
from each lane was scanned by using the Quantity One
package (PDI Inc., Huntingdon Station, NY, USA) and the
area divided into a grid of 25 boxes. The density reading
from each individual box was then incorporated into the
formula Telomere Length=S (ODi6Li)/S (ODi), where
(ODi)=the density of the box and Li=the size in bp of a
band positioned at the centre of the box, relative to the
molecular weight markers. The straight lines ®tted by linear
regression using the Microsoft Excel statistics programme
(Microsoft, UK). The regression constant was accurate to
within the 95% con®dence limit, unless otherwise stated.
Western blotting and antibodies
Cell pellets (1 ± 26106 cells/pellet) were lysed for 30 min in
bu€er containing 20 mM HEPES [pH 7.9], 5 mM EDTA,
INK4A
can be cell division-independent
10 mM EGTA, 5 mM NaF, 0.1 mg/ml okadaic acid, 10%
glycerol, 1 mM dithiothreitol (DTT), containing 0.4 M KCl,
0.4% Triton X-100, and protease inhibitors as follows: 5 mg/
ml each of aprotinin and pepstatin A, 1 mM benzamidine,
50 mg/ml phenylmethylsulphonyl ¯uoride. The extracts were
cleared by centrifugation and the supernatants stored at
7708C. 40, 100 or 200 mg of protein was subjected to
electrophoresis on 7.5% (pRb), 10% (p53, cyclin B1 or
MMP-1), 12.5% (p27KIP1 or IGFBP3), 17% (p14ARF, p16INK4A
or p21WAF-1), Tris-glycine SDS polyacrylamide gels. After
semi-dry blotting onto Immobilon-P ®lters (Millipore, UK),
non-speci®c binding sites were blocked by incubating the
membrane in Tris-bu€ered saline (TBS), 5% non fat dried
milk. Primary antibody incubations were carried out overnight at 48C in TBS-5% milk with antibodies against the
following human proteins: p14ARF (see Stott et al., 1998),
p16INK4A (C-20; Santa Cruz Biotechnology, California, USA),
p53 (DO-1; Santa Cruz), p21WAF-1 (Transduction Laboratories, Becton Dickinson, UK), p27 (Kip-1; Transduction
Laboratories), ERK2 (Transduction Laboratories), cyclin B1
(H-20; Santa Cruz), MMP-1 (Triple Point Biologics, Santa
Cruz, California, USA) and IGFBP3 (C-19; Santa Cruz).
After being washed, membranes were incubated with the
appropriate horseradish peroxidase-conjugated secondary
antibodies (Amersham Pharmacia Biotech, Amersham, UK)
and then developed with ECLTM Western blotting reagents
(Amersham Pharmacia Biotech). Negative controls were cell
lines lacking the INK4A locus (p14ARF and p16INK4A), SaOS-2
osteosarcoma cells (p53 and pRb) and HaCaT (p21WAF-1 low).
Positive controls were SaOS-2 cells (p14ARF), SV61HFK
(p16INK4A), Hela cells (p27KIP1) and human diploid ®broblasts
20 h post-irradiation with 4 Gy of g rays (p53 and p21WAF-1).
The ERK2 antibody was used as a loading control
throughout as it has been shown not to vary when ®broblasts
undergo senescence (Alcorta et al., 1996).
3551
Acknowledgements
The authors are very grateful to H Vaziri for the generous
gift of the pBabest2 retroviral construct, to Karen Vousden
and Gordon Peters for gifts of p14ARF antibodies, to the
Cancer Research Campaign and the European Molecular
Biology Organisation for ®nancial support (K Steeghs), to
Keith Vass for help with statistical analysis and to John
Wyke for critical reading of the manuscript.
References
Alcorta DA, Xiong Y, Phelps D, Hannon G, Beach D and
Barrett JC. (1996). Proc. Natl. Acad. Sci. USA, 93,
13742 ± 13747.
Allsopp RC, Vaziri H, Patterson C, Goldstein S, Younglai
EV, Futcher AB, Greider CW and Harley CB. (1992).
Proc. Natl. Acad. Sci. USA, 89, 10114 ± 10118.
Allsopp RC, Chang E, Kashe®-Aazam M, Rogaev EI,
Piatyszak MA, Shay JW and Harley CB. (1995). Exp.
Cell Res., 220, 194 ± 200.
Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu C-P,
Morin GB, Harley CB, Shay JW, Lichsteiner S and Wright
WE. (1998). Science, 279, 349 ± 352.
Burns JE, Baird MC, Clark LJ, Burns P, Edington K,
Chapman C, Mitchell R, Robertson G, Soutar D and
Parkinson EK. (1993). Br. J. Cancer, 67, 1274 ± 1284.
Collardo M, Medema RH, Garcia-Cao I, Dubuisson MLN,
Barradas M, Glassford J, Rivas C, Burgering BMT,
Serrano M and Lam EW-F. (2000). J. Biol. Chem., 275,
21960 ± 21968.
Cristofolo VJ and Pignolo RJ. (1993). Physiol. Rev., 73,
617 ± 638.
Dietrich C, Wallenfang K, Oesch F and Wieser R. (1997).
Exp. Cell Res., 232, 72 ± 78.
Oncogene
Accumulation of p16
3552
Oncogene
INK4A
can be cell division-independent
J Munro et al
Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C,
Medrano EE, Linskens M, Rubelj I, Pereira-Smith O,
Peacocke M and Campisi J. (1995). Proc. Natl. Acad. Sci.
USA, 92, 9363 ± 9367.
Durand B, Gao F-B and Ra€ M. (1997). EMBO J., 16, 306 ±
317.
Durand B, Fero ML, Roberts JM and Ra€ MC. (1998). Curr.
Biol., 8, 431 ± 440.
Goldstein S and Singal GP. (1974). Exp. Cell Res., 88, 359 ±
364.
Hara E, Smith R, Parry D, Tahara H, Stone S and Peters G.
(1996). Mol. Cell Biol., 16, 859 ± 867.
Harley CB. (1991). Mutat. Res., 256, 271 ± 282.
Hay¯ick L. (1965). Exp. Cell Res., 37, 614 ± 636.
Holliday R. (1996). BioEssays, 18, 3 ± 5.
Jiang X-R, Jimenez G, Chang E, Frolkis M, Kusler B, Sage
M, Beeche M, Bodnar AG, Wahl GM, Tlsty TD and Chiu
C-P. (1999). Nature Genet., 21, 111 ± 114.
Jones CJ, Kipling D, Morris M, Hepburn P, Skinner J,
Bounacer A, Wyllie FS, Ivan M, Bartek J, WynfordThomas and Bond JA. (2000). Mol. Cell Biol., 20, 5690 ±
5699.
Kamijo T, van de Kamp E, Chong MJ, Zindy F, Diehl JA,
Sherr CJ and McKinnon PJ. (1999). Cancer Res., 59,
2464 ± 2469.
Karlseder J, Broccoli D, Dal Y, Hardy S and de Lange T.
(1999). Science, 283, 1321 ± 1324.
Kipling D and Cooke HJ. (1990). Nature, 347, 400 ± 402.
Kiyono T, Foster SA, Koop JI, McDougal JK, Galloway
DA and Klingelhutz AJ. (1998). Nature, 396, 84 ± 88.
Lee HW, Blasco MA, Gottlieb GJ, Horner II JW, Greider
CW and DePinho RA. (1998). Nature, 392, 569 ± 574.
Makarov VL, Hirose Y and Langmore JP. (1997). Cell, 88,
657 ± 666.
Mazars GR and Jat PS. (1997). Proc. Natl. Acad. Sci. USA,
94, 151 ± 156.
McConnell BB, Starborg M, Brookes S and Peters G. (1998).
Curr. Biol., 8, 351 ± 354.
Morales CP, Holt SE, Ouellette M, Kaur KJ, Yan Y, Wilson
KS, White MA, Wright WE and Shay JW. (1999). Nature
Genet., 21, 115 ± 118.
Olovnikov AM. (1973). J. Theor. Biol., 41, 181 ± 190.
Palmero I, Pantoja C and Serrano M. (1998). Nature, 395,
125 ± 126.
Polyak K, Kato J-Y, Solomon MJ, Sherr CJ, Massague J,
Roberts JM and Ko€ A. (1994). Genes Dev., 8, 9 ± 22.
Robles SJ and Adami GR. (1998). Oncogene, 16, 1113 ± 1123.
Rubelj I and Vondracek Z. (1999). J. Theor. Biol., 197, 425 ±
438.
Serrano M, Lee HW, Chin L, Cordon-Cardo C, Beach D and
DePinho RA. (1996). Cell, 85, 27 ± 37.
Serrano M, Lin AW, McCurrach ME, Beach D and Lowe
SW. (1997). Cell, 88, 593 ± 602.
Schor SL and Schor AM. (1987). BioEssays, 7, 200 ± 204.
Shelton D, Chang E, Whittier PS, Choi D and Funk WD.
(1999). Curr. Biol., 9, 939 ± 945.
Sitte N, Saretzki G and von Zglinicki T. (1998). Free Rad.
Biol. Med., 24, 885 ± 893.
Smith JR and Whitney RG. (1980). Science, 207, 82 ± 84.
van Steensel B, Smogorzewska A and de Lange T. (1998).
Cell, 92, 401 ± 413.
Stein GH and Dulic V. (1995). BioEssays, 17, 537 ± 543.
Stott FJ, Bates S, James MJ, McConnell BB, Starborg M,
Brookes S, Palmero I, Ryan K, Hara E, Vousden K and
Peters G. (1998). EMBO J., 17, 5001 ± 5014.
Takuwa N and Takuwa Y. (1997). Mol. Cell Biol., 17, 5348 ±
5358.
Tresini M, Mawal-Dewan M, Cristofolo VJ and Sell C.
(1998). Cancer Res., 58, 1 ± 4.
Vaziri H and Benchimol S. (1998). Curr. Biol., 8, 279 ± 282.
Vaziri H, West MD, Allsopp RC, Davison TS, Wu Y-S,
Arrowsmith CH, Poirier GG and Benchimol S. (1997).
EMBO J., 16, 6018 ± 6033.
Wong H and Riabowol K. (1996). Exp. Geront., 31, 311 ±
325.
Wright WE, Brasiskyte D, Piatyszek MA and Shay JW.
(1996). EMBO J., 15, 1734 ± 1741.
von Zglinicki T, Saretzki G, Docke W and Lotze C. (1995).
Exp. Cell Res., 220, 186 ± 193.
Zindy F, Quelle DE, Roussel MF and Sherr CJ. (1997).
Oncogene, 15, 203 ± 211.