We show here that Cdt1 is down-regulated at both the protein and RNA level when primary human fibroblasts exit the cell cycle into G0, and its expression is induced as cells re-enter the
Trang 13 Cdt1 and geminin are down-regulated upon cell cycle exit and
are over-expressed in cancer-derived cell lines
Georgia Xouri1, Zoi Lygerou1, Hideo Nishitani2, Vassilis Pachnis3, Paul Nurse4,* and Stavros Taraviras5
1
Laboratory of General Biology, Medical School, University of Patras, Rio, Patras, Greece;2Department of Molecular Biology, Graduate School of Medical Science, Kyushu University, Fukuoka, Japan;3Division of Molecular Neurobiology, National Institute for Medical Research, London, UK;4Cell Cycle Laboratory, Cancer Research UK, London Laboratories, London, UK;5Laboratory of Pharmacology, Medical School, University of Patras, Rio, Patras, Greece
Licensing origins for replication upon completion of mitosis
ensures genomic stability in cycling cells Cdt1 was recently
discovered as an essential licensing factor, which is inhibited
by geminin Over-expression of Cdt1 was shown to
predis-pose cells for malignant transformation We show here that
Cdt1 is down-regulated at both the protein and RNA level
when primary human fibroblasts exit the cell cycle into G0,
and its expression is induced as cells re-enter the cell cycle,
prior to S phase onset Cdt1’s inhibitor, geminin, is similarly
down-regulated upon cell cycle exit at both the protein and
RNA level, and geminin protein accumulates with a 3–6 h
delay over Cdt1, following serum re-addition Similarly,
mouse NIH3T3 cells down-regulate Cdt1 and geminin
mRNA and protein when serum starved Our data suggest a
transcriptional control over Cdt1 and geminin at the
trans-ition from quiescence to proliferation In situ hybridization
and immunohistochemistry localize Cdt1 as well as geminin
to the proliferative compartment of the developing mouse gut epithelium Cdt1 and geminin levels were compared in primary cells vs cancer-derived human cell lines We show that Cdt1 is consistently over-expressed in cancer cell lines at both the protein and RNA level, and that the Cdt1 protein accumulates to higher levels in individual cancer cells Geminin is similarly over-expressed in the majority of cancer cell lines tested The relative ratios of Cdt1 and geminin differ significantly amongst cell lines Our data establish that Cdt1 and geminin are regulated at cell cycle exit, and suggest that the mechanisms controlling Cdt1 and geminin levels may be altered in cancer cells
Keywords: cancer; Cdt1; G0; geminin; licensing
Genomic stability is maintained in proliferating cells
through control mechanisms which ensure that the cell’s
genetic content is duplicated entirely and only once in each
cell cycle, and is correctly partitioned to the two daughter
cells during mitosis [1] In eukaryotes, replication starts from
multiple origins along each chromosome during S phase,
and re-firing of the same origins is inhibited until mitosis is
completed This is achieved through the replication licensing
system [2], a regulatory system conserved in evolution from
yeast to humans, which licenses each origin for a single
round of DNA replication This license is lost upon origin
firing, and is reestablished only upon completion of mitosis,
thereby preventing over-replication of the genome
Recent studies, mostly using yeasts and a Xenopus laevis
in vitrolicensing system, have permitted an understanding
of the licensing process at the molecular level (reviewed
in [3–5]) A multisubunit complex is formed on origins of DNA replication upon completion of mitosis, by the stepwise association of licensing factors to origin sequences, which confers to each origin the license to replicate Origins are recognized by the six-subunit origin recognition complex (ORC), which, at least in lower eukaryotes, remains chromatin associated throughout the cell cycle Temporal regulation of origin licensing is achieved through the action of two loading factors, Cdc6/18 and Cdt1, which are required for the chromatin association of the six subunit mini chromosome main-tenance (MCM) complex The MCM complex is believed
to function as the replicative helicase [6], and its chromatin association confers to origins the license to replicate
Cdt1 was recently identified as a factor essential for the chromatin loading of MCM proteins upon comple-tion of mitosis in both lower and higher eukaryotes [7–11] The identification of the human homologue of Cdt1 permitted its analysis during the cell cycle in cultured human cells [12–14] Cdt1 is tightly regulated so that its protein accumulates only in G1, when licensing is legitimate This regulation is mediated mostly by targeted proteolysis of Cdt1 from S phase to mitosis, rather than
by transcriptional controls [13] In addition, Cdt1 binds strongly to and is inhibited by geminin [12,15] Geminin, originally identified in Xenopus as an inhibitor of
Correspondence to S Taraviras, Laboratory of Pharmacology,
Medical School, University of Patras, 26500, Rio, Patras, Greece.
Fax: +30 2610 994720
2 , Tel.: +30 2610 997638,
E-mail: taraviras@med.upatras.gr and Z Lygerou, Laboratory of
Biology, Medical School, University of Patras, 26500, Rio, Patras,
Greece Fax: +30 2610 991769, Tel.: +30 2610 997621,
E-mail: lygerou@med.upatras.gr
Abbreviations: HFF, human foreskin fibroblasts; MCM, mini
chro-mosome maintenance; ORC, origin recognition complex.
*Present address: The Rockefeller University, NY, USA.
(Received 19 April 2004, revised 17 June 2004, accepted 28 June 2004)
Trang 2licensing specifically degraded at the end of mitosis [16],
is believed to act through binding to Cdt1 [12,15] Cdt1
and geminin are, however, hardly coexpressed during the
cell cycle in cultured human cells [13], raising the
question of when and how geminin exerts its function
In contrast to other cell cycle inhibitors, geminin was
shown to be a marker of proliferating cells [17]
Cells cease to proliferate and exit the cell cycle (G0 phase)
in response to growth arrest and differentiation signals or
when deprived of growth factors The vast majority of cells
in multicellular organisms exist in out of cycle states, either
temporarily resting in G0, from which they can respond
to stimuli for cell cycle re-entry or differentiation, or in
terminally differentiated or arrested (senescent) states
Defects in the mechanisms that ensure the timely
prolifer-ation of human cells are key events in the development of
neoplasia
Cells exit the cell cycle from the G1 phase and
previous work has shown that licensing is lost when G1
cells exit to G0 Nuclei isolated from early G0 cells fail
to replicate in a Xenopus in vitro system, similar to G2
nuclei [18–21] ORC2-5 proteins persist on G0 chromatin,
but MCM proteins and Cdc6/18 rapidly dissociate from
chromatin and are gradually lost from G0 cells [20–22]
When cells re-enter the cell cycle, expression of Cdc6/18
and MCM proteins is induced [22,23] Cdc6/18, Orc1
and several of the MCM proteins have been shown to be
transciptionally regulated by E2F at the transition from
quiescence to proliferation [24–28] MCM proteins
have been proposed as sensitive proliferation markers
for the detection of premalignant and malignant states
[29–31]
In this study we examine whether Cdt1, a factor essential
for licensing across evolution and tightly controlled during
the cell cycle, is negatively regulated in quiescent cells We
studied Cdt1 and its inhibitor geminin at the transition from
quiescence to proliferation in cultured primary human cells
and NIH3T3 cells and we compared their expression
patterns in tissue sections and their expression levels in
primary and normal diploid vs cancer cell lines Our data
show a correlation of Cdt1 and geminin expression levels
with cell proliferation
Materials and methods
Cell culture
Human foreskin fibroblast (HFF), HeLa, MDAMB231,
MCF7, HT1080, U2OS, MRC5 and LNCAP cells were
grown in DMEM/high glucose medium with 10% (v/v)
fetal bovine serum NIH3T3 cells were grown in DMEM/
high glucose medium with 10% (v/v) calf serum Most cell
lines used were provided by the Cancer Research UK cell
line facility For serum starvation, HFF or NIH3T3 cells
were incubated in the presence of 0.1% (v/v) serum for 48 h
Cells were then induced to re-enter the cell cycle by addition
of 20% (v/v) serum For contact inhibition, NIH3T3 cells
were cultured in the presence of 10% (v/v) calf serum until
confluent (day 0) and then for the indicated number of days
following confluency To induce cell cycle re-entry following
contact inhibition, 4 days following confluency cells were
split 1 : 10
Plasmids Mouse Cdt1 and geminin full-length cDNAs were compiled by combining EST entries in the nucleotide databases Based on the deduced sequences, specific oligonucleotides were designed for PCR cloning the full-length open reading frames of mouse Cdt1 and mouse geminin into BamHI/HindIII and EcoRI/BamHI sites of pBluescript KS, respectively These were used to generate specific probes for Northern hybridization on total RNA extracted from NIH3T3 cells and mouse in situ hybrid-ization
Antibodies, Western blotting, immunofluorescence Antibodies against hCdt1 were described previously [13] Affinity purified anti-hCdt1 Ig was used for all experi-ments Anti(h-geminin) serum raised in rabbits against the C-terminal 94 amino acids of human geminin (expressed as a 6· His fusion protein in Escherichia coli and purified on an Ni-column) was affinity purified using the same recombinant fragment These affinity purified antibodies raised against geminin will be referred to hereafter as anti-Gem2 A major band with the expected apparent molecular mass for h-geminin (around 30 kDa) was detected by Western blotting using anti-Gem2 on HeLa total cell extracts RNAi directed against human geminin resulted in complete disappearance of this band (data not shown), verifying that it indeed corresponds to human geminin
For Western blotting, total cell lysates were prepared by lysing cell pellets directly in SDS/PAGE loading buffer and boiling Antibodies were used at the following dilutions: anti-hCdt1, 1 : 500; anti(h-geminin) (Santa Cruz), 1 : 500; anti-Gem2, 1 : 2000, anti-hCdc6/18 (Upstate Biotechno-logy), 1 : 1000; anti-cyclin A (Upstate BiotechnoBiotechno-logy),
1 : 2000, and anti(a-tubulin) (Sigma), 1 : 10 000
Immunofluorescence on HFF and HeLa cells, using affinity purified Cdt1 Ig (1 : 200 dilution), or anti-Gem2 Ig (1 : 1000) was carried out as previously described [13] Unrelated rabbit IgG or pre-immune serum was used
as a negative control
For BrdU staining, cells were incubated for 30 min in the presence of 20 lM BrdU (Sigma) added directly to the culture medium prior to collection Cells were then washed twice with ice-cold NaCl/Pi, fixed in 3.8% (v/v) formaldehyde for 10 min, washed twice with NaCl/Pi, and permeabilized with 0.3% (v/v) Triton X-100 in NaCl/Pi After washing cells three times with NaCl/Pi and once with double distilled H2O, DNA was denatured
by incubation in 2M HCl for 1 h at room temperature Cells were then washed for 5 min in 0.1M Tris/HCl,
pH 8.8, to neutralize the pH, and three times with NaCl/
Pi containing 0.1% (v/v) Tween Cells were treated with blocking buffer [3% (w/v) bovine serum albumin/10% (v/v) goat serum in NaCl/Pi] for 30 min and incubated with anti-BrdU (Sigma B2531, 1 : 150) in blocking buffer, overnight in a wet chamber Cells were washed
in NaCl/Picontaining 0.1% (v/v) Tween three times and incubated with an Alexa 488 conjugated goat anti-mouse secondary antibody (Molecular Probes) After washing, DNA was stained briefly with Hoechst 33258
Trang 3Quantitation of protein levels in cancer cell lines and
primary cells
To calculate the number of hCdt1 and h-geminin molecules
present per HeLa cell, full-length hCdt1 (HisT7Cdt1)
and full-length h-geminin (Hisgeminin) were expressed as
His-tagged proteins in E coli (using vectors pET28a-Cdt1
and pQE-geminin), purified on Ni-agarose, and protein
amounts of each full-length protein were quantified by
comparison with increasing amounts of bovine serum
albumin on an SDS/PAGE stained with Coomassie brilliant
blue Increasing amounts of each recombinant protein were
then loaded on an SDS/PAGE alongside a total cell extract
corresponding to 1.5· 105asynchronously growing HeLa
cells and immunoblotted using anti-Cdt1 and anti-geminin
specific antibodies Comparison of the Western blot signals
showed that approximately 0.4 ng of Cdt1 protein and
0.2 ng of geminin protein are present in 1.5· 105 HeLa
cells Because the molecular mass of Cdt1 is nearly twice that
of geminin, it was calculated that about 30 000 molecules of
each protein are present on average in each HeLa cell It
should be noted, however, that Cdt1 is present in cells that
are in G1 while geminin is present in cells in S to M phases
In order to quantify the amount of Cdt1 detected by
immunofluorescence in individual HFF and HeLa cells,
indirect immunofluorescence was carried out as described
above and signal intensity was quantified using IPLAB
software (Scanalytics Inc., Fairfax, VA, USA)
fluorescence intensity for 25 high-power fields (40·
magni-fication) for HFF cells and 25 high-power fields for HeLa
cells, from three independent experiments, was quantified by
defining the respective area as a region of interest and after
applying background correction Western blots were
quan-tified using theQUANTIFY ONE
Northern blot analysis and semiquantitative RT-PCR
analysis
Total cell RNA was prepared by the TrizolTM method
7(Invitrogen) and 10 lg of total RNA per sample was used
for Northern blot analysis Northern blot analysis was as
described [13,32] Probes specific for the mouse Cdt1 and
geminin cDNAs, generated by random priming, were used
for NIH3T3 cells (see above) while probes specific for the
human geminin gene were generated by random priming
using the complete open reading frame of the human
geminin cDNA A probe directed against the actin mRNA
served as a loading control A blot containing total mRNA
from human tumor and normal samples was purchased
from ResGen (Invitrogen Corporation) Northern blots
were quantified using theQUANTIFY ONEprogram (BioRad)
Semiquantitative RT-PCR analysis was performed to
examine hCdt1 and h-geminin mRNA levels in HFF cells
Total RNA was isolated from cycling, serum deprived and
re-stimulated HFF cells using the Trizol method
(Invitro-gen) Reverse transcription was performed using 1–5 lg
total RNA and random primers according to the
manufac-turer’s protocol (Superscript; Invitrogen) cDNA was
amplified by PCR using specific sets of primers for hCdt1,
h-geminin and h-actin Primers used were: 5¢-AAGGATC
CCGCCTACCAGCGCTTCC-3¢ and 5¢-CCAAGCTTGA
AGGTGGGGACACTG-3¢ for hCdt1 (288 nucleotide
product); 5¢-CTTCTGTCTTCACCATCTACA-3¢ and 5¢-AGTGGAGGTAAACTTCGGCAG-3¢ for h-geminin (710 nucleotide product) and 5¢-CACCTTCTACAATG AGCTGC-3¢ and 5¢-AGGCAGCTCGTAGCTCTTCT-3¢ for h-actin (437 nucleotide product) PCR was performed under the following conditions: denaturation at 94C for
45 s, annealing for 30 s at 65C for hCdt1 and 62 C for h-geminin and h-actin, extension at 72C for 1 min; 26 cycles were used for the amplification of hCdt1 and h-geminin cDNAs and 20 cycles for the amplification of h-actin cDNA The number of cycles was adjusted to ensure that the reaction was in the linear range PCR products were analyzed by agarose gel electrophoresis Two PCRs with twofold dilution of cDNA were performed for each sample,
to show linearity in detection
In situ hybridization and immunohistochemistry Non-radioactive in situ hybridization was performed on fresh-frozen sections of E17 mouse embryos
obtained from timed pregnancies of outbred (Parkes) mice All animal work was performed according to the Home Office (UK) and local (NIMR-MRC) Ethical Commitee guidelines Frozen sections were postfixed for 10 min at room temperature with 4% (v/v) paraformaldehyde Sub-sequently the slides were pretreated with 0.25% (w/v) acetic anydride for 10 min and hybridization was carried out in a 5· NaCl/Cit humidified chamber overnight at 65 C The slides were then washed at high stringency (0.2· NaCl/Cit at
65C) for 1 h and transferred to 0.2· NaCl/Cit at room temperature for 5 min Slides were blocked for 2 h at room temparature, with 10% (v/v) sheep serum in 0.1MTris/HCl
pH 7.5/0.15M NaCl and incubated overnight at 4C with anti-DIG Ig (1 : 5000 dilution, Roche) in 0.1M Tris/HCl pH 7.5/0.15MNaCl Slides were rinsed in 0.1M Tris/HCl pH 7.5/0.15M NaCl, equilibrated in 0.1M Tris/ HCl pH 9.5/0.1M NaCl/50 mM MgCl2 and incubated with 262.5 lgÆmL)1 Nitro Blue tetrazolium
and 175 lgÆmL)1 5-bromo-4-chloroindol-2-yl phosphate
10(Roche) in 0.1M Tris/HCl pH 9.5/0.1M NaCl/50 mM MgCl2for 3–6 h [33] Antisense riboprobes were generated using the full-length open reading frames of mouse Cdt1 and geminin as templates and the T3 and T7 polymerase, respectively, while sense probes served as negative controls and were generated using T7 and T3 polymerase
E17 dpc mouse embryos used for immunohistochem-istry experiments were fixed overnight with 4% (v/v) paraformaldehyde, transferred to a 30% (v/v) sucrose solution in NaCl/Pi for 24 h and embedded in OCTTM
11compound (BDH) Immunohistochemistry was performed
on consecutive fresh-frozen sections that were postfixed in 4% (v/v) paraformaldehyde in NaCl/Pi, washed with NaCl/Piand permeabilized with 0.3% (v/v) Triton X-100
in NaCl/Pi Horseradish peroxidase activity was quenched
by a 10 min incubation in 10% methanol/10% H2O2 (v/v) Sections were then blocked in 3% (w/v) bovine serum albumin, 10% (v/v) goat serum in NaCl/Pifor 2 h and incubated overnight at 4C with hCdt1 or anti-geminin Ig (Santa Cruz) at 1 : 100 dilution in blocking buffer Secondary antibodies, anti-rabbit or anti-goat horseradish peroxidase conjugated (Roche), were used Incubation with the pre-immune serum or secondary
Trang 4antibody only was used to determine the specificity of the
primary antibodies used
Results
Cdt1 protein levels are low in quiescent cells
Our previous work showed that Cdt1, a DNA licensing
factor, is tightly controlled by proteolysis during the cell cycle
in human cells, accumulating only during the G1 phase,
when licensing is legitimate [13] When cells exit the cell cycle,
licensing in lost, and is established again as cells prepare to
reenter the cell cycle While a previous study did not detect a
significant down-regulation of Cdt1 in serum deprived cells
[14], a different study showed that cells which express higher
levels of Cdt1 in G0 exhibit a quicker entry into S phase and
are predisposed for malignant transformation [34]
We therefore investigated whether Cdt1 is
down-regula-ted upon cell cycle exit To this effect, HFF cells were
deprived of serum for 48 h to induce cell cycle exit Serum
was then re-added and samples taken as cells progressed
into the cell cycle Total Cdt1 levels were measured by
Western blotting (Fig 1A) Cyclin A and Cdc6 served as
controls of proteins previously shown to be down-regulated
during G0, while tubulin served as a loading control Cdt1 is
markedly down-regulated upon cell cycle exit and then
quickly re-accumulates as cells re-enter the cell cycle
Antibodies against hCdt1 were used to assess by
immu-nofluorescence the percentage of cells expressing Cdt1 in
asynchronously growing HFF cells, and at different time
points during the transition from quiescence to proliferation
(Fig 2A, left panels, immunofluorescence images; Fig 2B,
quantitation) The percent of BrdU positive cells at each
Fig 1 Cdt1 protein expression in human fibroblasts (HFF) at the
transition from quiescence (G0) to proliferation Total cell extracts from
HFF cells were analyzed by Western blotting using Cdt1, cyclin A,
Cdc6 and tubulin specific antibodies Lane 1, proliferating HFF cells;
lane 2, HFF cells deprived of serum for 48 h; lanes 3–8, serum deprived
HFF cells induced to re-enter the cell cycle by addition of serum and
collected at 6, 12, 15, 18, 21 and 24 h, respectively The band
corres-ponding to Cdt1 has been marked by an arrow, while a cross-reacting
band running above Cdt1 is indicated by an asterisk The position of
migration of the 66-kDa molecular mass marker band is indicated at
the left of the Cdt1 blot.
A
Pr
0 h
6 h
9 h
12 h
15 h
18 h
21 h
24 h Anti-Cdt1 DAPI Anti-geminin DAPI
B
Fig 2 Cdt1 and geminin protein levels and localization at the transition from quiescence to proliferation Proliferating HFF cells (Pr), HFF cells deprived of serum for 48 h (0 h), or serum deprived cells induced to re-enter the cell cycle by serum re-addition for 6, 9, 12, 15, 18, 21 and 24 h were processed for indirect immunofluorescence using anti-Cdt1 and anti-geminin (anti-Gem2) Ig (A) Microscopy images (recorded with identical exposure settings for all time points) (B) Percentage of cells showing staining for Cdt1 (white bars), BrdU (black bars)
(hatched bars) in each time point Over 200 cells were measured for each time point In order to assess cell cycle progression, cells were incubated with BrdU 30 min prior to fixation, and processed for BrdU staining as described in Materials and methods Percent of BrdU positive cells in each time point was scored.
Trang 5time-point is also shown in Fig 2B for comparison Cdt1 is
detected in the nucleus of approximately one-third of cells
from an asynchronous population but its levels are
mark-edly decreased in cells cultured in the absence of serum Cells
staining for Cdt1 appear around 12 h following serum
re-addition, several hours before the peak of cells in S phase
(21–24 h) We wished to compare the behavior of Cdt1 to
that of its inhibitory molecule, geminin, during cell cycle exit
and re-entry To that effect, we assessed the levels of the
geminin protein by immunofluorescence, in proliferating
HFF cells, upon serum withdrawal and upon serum
re-addition, in parallel to Cdt1 immunofluorescence detection
described above (Fig 2A, right panels, immunofluorescence
images; Fig 2B quantitation) Geminin was detected in the
nucleus of around one-quarter of asynchronously growing
HFF cells Upon serum withdrawal geminin levels were
markedly decreased, similar to Cdt1 Geminin staining first
re-appeared in a small number of cells around 15 h
following serum readdition, and peaked at 21–24 h together
with the peak of cells in S phase, as judged by the percentage
of BrdU positive cells Geminin appeared in the cell
nucleus significantly later upon serum re-addition than
Cdt1 (a 3–6 h delay) and its accumulation paralleled the
accumulation of BrdU positive cells
Our data suggest that both Cdt1 and geminin are
down-regulated in quiescent HFF cells When cells re-enter the cell
cycle, Cdt1 is expressed first, as cells prepare for a new
round of S phase, while geminin accumulates as cells enter
S phase
Geminin, an inhibitor of Cdt1 is severely down-regulated
upon cell cycle exit
Cdt1 is negatively regulated by geminin [12,15], and, during
the cell cycle, geminin accumulates in S phase and G2, when
Cdt1 levels are low, while Cdt1 accumulates in G1, when
geminin is undetectable [12,13,16] Our immunofluorescence findings, showing that upon serum starvation of human fibroblasts geminin is down-regulated similar to Cdt1, were somewhat surprising, as geminin might have been expected
to be up-regulated upon cell cycle exit We therefore wished
to examine this point more carefully The low levels of geminin protein and mRNA present in HFF cells however (see below) hampered a detailed analysis in these cells We therefore turned to mouse NIH3T3 cells, which express geminin to levels similar to HeLa cells (Fig 3, left: compare lanes 1 and 2), but can be induced to exit the cell cycle by serum withdrawal or contact inhibition
Figure 3 shows the levels of Cdt1 and geminin in NIH3T3 cells, which are induced to exit the cell cycle either
by serum deprivation or contact inhibition Cdc6/18 and cyclin A protein levels serve as controls for proteins previously shown to be down-regulated upon cell cycle exit, while tubulin serves as a loading control As shown above for HFF cells, Cdt1 protein levels are significantly reduced
in serum starved NIH3T3 cells and re-accumulate upon addition of serum Cdt1 protein levels are much less affected
by contact inhibition (still present 4 days following conflu-ency) Geminin is severely down-regulated by both serum deprivation and contact inhibition, similar to cyclin A and more dramatically than Cdt1 For example, geminin protein levels are undetectable upon serum starvation and are already reduced from the first day following confluency, when Cdt1 levels are still unaffected
We conclude that geminin is dramatically down-regulated
in NIH3T3 cells in G0, consistent with our findings with HFF human cells
Cdt1 and geminin mRNAs are down-regulated in G0 During the cell cycle, Cdt1 and geminin mRNA levels are mostly stable and protein levels are primary controlled by
Fig 3 Expression of Cdt1 and geminin proteins in quiescent and proliferating NIH3T3 cells NIH3T3 cells were induced to exit the cell cycle by serum starvation (–S) or contact inhibition (Ci) Total cell extracts at the conditions indicated below were prepared and Western blot analysis was performed using antibodies that recognize specifically Cdt1, geminin (Santa Cruz), cyclin A, Cdc6 and tubulin proteins Lane 1, proliferating HeLa cells; lane 2, proliferating NIH3T3 cells; lane 3, serum deprived NIH3T3 cells (cultured for 48 h in low serum); lane 4, NIH3T3 cells induced to re-enter the cell cycle upon serum addition for 6 h; lanes 5–8, contact inhibited NIH3T3 cells, 2, 3 and 4 days following confluency; lane 9, NIH3T3 cells induced to re-enter cell cycle by splitting 1 : 10, 4 days after confluency The arrow on the cyclin A blot indicates the band corresponding to cyclin A while the asterisk marks a cross-reacting band Mouse Cdt1 migrates slower than human Cdt1.
Trang 6proteolysis [13] Given the down-regulation of both proteins
upon serum deprivation, we wished to examine their
respective mRNA levels
In Fig 4A, the mRNA levels of Cdt1, geminin and actin
(as loading control) are shown in serum deprived NIH3T3
cells, and 6 and 12 h following serum re-addition, and
compared with mRNA levels in proliferating cells Both
Cdt1 and geminin mRNAs are down-regulated upon
serum deprivation and re-accumulate as cells prepare for
S phase Densitometry scanning and data normalization
against the actin control shows that geminin mRNA is at
background levels in serum deprived NIH3T3 cells and at
6 h following serum re-addition, while at 12 h it has
returned to the level detected in proliferating cells Cdt1
mRNA levels are reduced twofold and similarly return to
the levels detected in proliferating cells by 12 h following
serum readdition
We then examined how quickly upon serum deprivation
Cdt1 and geminin mRNA levels are reduced (Fig 4B)
Densitometry analysis showed that geminin mRNA levels
appear significantly reduced already at early time points
(12 h minus serum, twofold reduction) and are further
reduced when cells are cultured longer in the absence of
serum (reaching a sevenfold reduction at 40 h minus serum)
Cdt1 mRNA levels show a twofold reduction 24 h
follow-ing serum deprivation and remain at approximately the
same level to the end of the time course
In order to reproduce our findings also in human HFF cells, and given that Cdt1 and geminin mRNAs were hardly detectable in these cells by Northern blotting (see below) we employed reverse-transcription followed by semiquantita-tive PCR amplification (RT-PCR) of human Cdt1 and geminin mRNAs in cycling, serum deprived and re-stimu-lated HFF cells (Fig 4C) Consistent with our finding with NIH3T3 cells, geminin mRNA levels were dramatically down-regulated in serum starved HFF cells and geminin mRNA accumulated again around 18 h following serum re-addition Cdt1 mRNA levels were also decreased in serum starved HFF cells and re-accumulated from 12 h following serum re-addition Similar to our findings with NIH3T3 cells, mRNA fluctuations upon serum withdrawal
in HFF cells appeared less dramatic for Cdt1 than geminin
We conclude that Cdt1 and geminin mRNA levels are reduced in quiescent cells, suggesting that in contrast to their regulation during the cell cycle [13], upon exit and entry to the cell cycle both genes are controlled, at least in part, transcriptionally
Cdt1 and geminin are highly expressed in proliferating cellsin vivo
The experiments with cultured cells have suggested that Cdt1 and geminin mRNA and protein are down-regulated upon cell cycle exit and are progressively up-regulated upon
Fig 4 Transcriptional control of Cdt1 and
geminin in quiescent cells (A) Total cell RNA
prepared from proliferating NIH3T3 cells
(lane 1), from cells deprived of serum for 48 h
(lane 4) or from cells first serum deprived for
48 h and then cultured for 6 or 12 h in the
presence of 20% (v/v) serum (lanes 3 and 2,
respectively) was subjected to Northern blot
analysis using a probe specific for human Cdt1
(upper), human geminin (middle) or actin as a
loading control (lower) (B) Northern blotting
analysis was performed using total cell RNA
prepared from NIH3T3 cells that were grown
in the presence of serum (lane 1), in the
ab-sence of serum for 12, 24, 32, 40 and 48 h
(lanes 2–6) or for 18 h after re-addition of
serum to cells which were previously serum
deprived for 48 h (lane 7) and hybridized using
specific probes for Cdt1, geminin and actin.
(C) Total RNA extracted from HFF cells was
subjected to reverse transcription and PCR
amplification with oligonucleotides specific for
the hCdt1 and h-geminin cDNAs PCR with
oligonucleotides specific for actin served as a
loading control Twofold dilutions of starting
cDNA were used to show linearity (data not
shown) Lane 1, proliferating HFF cells; lane
2, HFF cells deprived of serum for 48 h; lanes
3–7, serum deprived HFF cells upon serum
readdition for 6–24 h.
Trang 7re-entry to the cell cycle A recent report showed that
geminin protein levels are positively correlated with cell
proliferation [17] We investigated the in vivo expression of
Cdt1 and compared it to that of geminin in the developing
mouse gut epithelium, a tissue in which a proliferating and a
differentiating zone can be distinguished histologically Gut
epithelium differentiation is initiated after E14dpc in mouse
embryos and continues postnatally We determined Cdt1
and geminin mRNA and protein expression using in situ
hybridization and immunohistochemistry, respectively, on
sections from the gastrointestinal tract of an E17dpc mouse
embryo At this stage of development, the gut epithelium is
organized into villi, which are separated at their bases by a
proliferating compartment known as the intervillus
epithe-lium
Cdt1 mRNA is mainly localized at the bases of the
developing intestinal villi (the intervillus epithelium), where
the proliferating cells of the intestinal epithelium are
localized (Fig 5A) Geminin mRNA has a distribution
similar to Cdt1 in the small and large intestine
Immuno-histochemistry using antibodies specific for Cdt1 and
geminin (Fig 5B) showed that both proteins are detected
in the proliferating cell layer of the developing gut
epithe-lium in a similar expression pattern
Therefore, Cdt1 and geminin mRNA and protein reveal a similar distribution along the gastrointestinal tract, localiz-ing mainly in the proliferatlocaliz-ing zone of the gut epithelium Cdt1 and geminin are over-expressed in cancer cells Given the correlation we observed between Cdt1 and geminin mRNA and protein levels and proliferation, we wished to examine the expression levels of these two genes in different tumor cell lines and compare them to primary cells Cellular lysates were prepared from human foreskin fibro-blasts, a primary cell line, and the tumorigenic cell lines Saos, MDAMB231, MCF7, HeLa and LNcap Western blot analysis using anti-Cdt1 Ig showed that Cdt1 protein is detected at much lower levels in the primary HFF cells compared with all the tumorigenic cell lines tested (Fig 6A) Western blotting with commercial antibodies against gem-inin showed that gemgem-inin protein levels are also increased in cancer cell lines, while different cancer cell lines appear to over-express geminin to varying degrees In order to more carefully compare the relative levels of Cdt1 and geminin in different primary and cancer cell lines, we utilized a more sensitive antibodies against geminin (anti-Gem2) and inclu-ded primary endothelial cells (Huvec), the normal diploid human cell line MRC5 and two more cancer cell lines (HT1080 and U2OS), in addition to HFF, HeLa and MCF7 analyzed in Fig 6A As shown in Fig 6B, cancer cell lines appear to consistently over-express Cdt1 in comparison with primary and normal diploid cell lines (for lanes 5–8, a higher exposure is shown for the Cdt1 blot, as evident by the intensity in lanes 2 and 8, which both correspond to HeLa cell, to permit detection of Cdt1 in the normal diploid MRC5 cells) Quantitation of the blots in Fig 6A,B showed that the majority of cancer cell lines express Cdt1 over 10-fold more than primary cell lines Geminin is hardly detectable in the primary cell lines, and accumulates to higher levels in the majority of cancer cell lines tested It is noteworthy, however, that geminin levels vary significantly amongst the cancer cell lines tested (compare, for example, HeLa, lane 2, to MCF7, lane 3), suggesting that the relative amount of Cdt1 and its inhibitor geminin may differ in different cell lines
In order to further investigate this point, we estimated how many molecules of Cdt1 and geminin are present on average per cell in an asynchronous population of HeLa cells To this end, known amounts of recombinant full-length Cdt1 (HisT7-Cdt1) and recombinant full-length geminin (His-geminin) were loaded on an SDS/PAGE gel alongside total cell extract from 1.5· 105 asynchronously growing HeLa cells, and Western blotted with Cdt1 and anti-geminin Ig (Fig 6C) We estimate that approximately 30 000 molecules of Cdt1 and an equal number of geminin molecules are present on average per HeLa cell (Materials and methods), suggesting that rather similar levels of Cdt1 and its inhibitor are produced in this cell line, though at different cell cycle stages [13] In contrast, we calculate a ratio
of Cdt1 to geminin of approximately 10 : 1 for MCF7 cells The difference in Cdt1 protein levels which is consistently observed between primary and cancer cell lines tested could have been due to the larger in cycle pool of the cancer cell lines In order to address this, we used immunofluorescence
to assess whether Cdt1 is over-expressed in individual cells
Fig 5 Cdt1 and geminin are expressed in proliferating cells of the
gastrointestinal tract In situ hybridization (·5, ·10)
immunohistochemical (·5) (B) analysis of Cdt1 (left) and geminin
(right) expression on frozen section of an E17 dpc mouse embryo Cdt1
and geminin mRNA and protein expression show a similar expression
profile Consecutive sections of the gastroinestinal tract are shown.
Trang 8of a tumor cell line population in comparison with primary
cells (Fig 7) The number of HeLa cells staining positive
for Cdt1 was higher (approximately 50% of HeLa cells
in comparison with 35% of HFF cells), consistent with a
higher percentage of HeLa cells actively cycling In addition
to that however, the staining observed in individual HeLa
cells was higher than the staining observed in individual
HFF cells (Fig 7A) Quantitation of 25 high-power fields
each for HFF and HeLa immunostainings shows that a
range of expression levels are observed in both HFF and
HeLa cells, as expected for a protein whose levels fluctuate
during the cell cycle, individual HeLa cells, however, express
on average over twofold higher levels of Cdt1 than HFF
cells (Fig 7B) These data show that Cdt1 is expressed to higher levels in individual cycling cancer cells in comparison with cycling primary cells
To address whether Cdt1 and geminin up-regulation in cancer cell lines would also occur at the mRNA level, we used Northern blot analysis with total cell RNA extracted from the primary HFF cells and different tumor cell lines Similar to protein levels, both Cdt1 and geminin mRNA was markedly increased in all the tumor cell lines tested compared with the primary HFF cells (Fig 8A) We employed densitometry in order to compare the levels of hCdt1 and h-geminin mRNA amongst the different cancer cell lines tested When compared with HeLa cells,
Fig 6 Cdt1 and geminin are highly expressed in cancer cells (A,B) Western blotting analysis was used to determine the expression of Cdt1, geminin and tubulin as a loading control in cellular extracts from different human cell lines (A) Lanes 1–6, HFF, Saos, MDAMB231, MCF7, HeLa and LNcap, respectively (B) Lanes 1–8, HFF, HeLa, MCF7, Huvec, HT1080, U20S, MRC5 and HeLa, respectively A commercial anti-geminin Ig (Santa Cruz) was used for (A), while anti-Gem2, which shows a higher sensitivity, was used for (B) For Cdt1, a higher exposure is shown for lanes 5–8 in respect to lanes 1–4, to allow visualization of Cdt1 in MRC5 cells For lanes 1–4, a higher exposure of the geminin blot (Gem long) is also shown at the bottom of the panel, to permit visualization of geminin in the primary HFF cells (C) Estimation of the number of Cdt1 and geminin molecules present in HeLa cells Western blotting of total cell extract from 1.5 · 10 5 asynchronously growing HeLa cells (marked HeLa) was run alongside known amounts of recombinant full-length Cdt1 and geminin (HisT7-geminin and His-geminin, amount of recombinant protein run in each lane in ng is indicated) and immunoblotted with anti-Cdt1 and anti-geminin Ig See Materials and methods for calculations.
Trang 9MDAMB231 cells showed somewhat decreased levels for
both hCdt1 and h-geminin (approximately threefold
reduc-tion) while MCF7 cells showed increased levels of hCdt1
mRNA (twofold) and decreased levels of h-geminin mRNA
(threefold reduction), supportive of our findings concerning
hCdt1 and h-geminin protein levels in this cell line
Therefore, while all cancer cell lines tested express much
higher levels of hCdt1 and h-geminin mRNAs when
compared with primary cells, cancer cell lines may differ
in the levels of overexpression of these mRNAs
In Fig 8B, a Northern blot containing total RNA
extracted from tumorigenic and matched normal specimens
(kidney cancer, liver cancer and lung cancer and respective
normal specimens) was hybridized with an h-geminin
specific probe and actin as a loading control Densitometry
analysis shows that h-geminin mRNA is increased in the
tumor specimens tested (approximately twofold) in
com-parison with the matched normal specimens, consistent with
our findings with cancer cell lines
Discussion
In cycling cells, Cdt1 is specifically expressed during the G1
phase of the cell cycle and is believed to act together with
Cdc6 to load the MCM protein complex onto chromatin,
thereby licensing DNA for a further round of DNA replication When G1 cells exit the cell cycle and enter quiescence, licensing is gradually lost We show here that Cdt1 is down-regulated when cells are induced to transit from G1 into G0 by serum deprivation Cdt1 protein levels are low in serum deprived human primary fibroblast HFF cells and NIH3T3 cells Cdt1 appears again as cells are induced to re-enter the cell cycle upon serum re-addition, before a new round of S phase is initiated, consistent with a requirement for Cdt1 in re-licensing G0 chromatin for a new round of DNA replication Cdt1 protein levels are not appreciably affected early upon contact inhibition, suggest-ing that the degree of down-regulation of Cdt1 may vary depending on how cells have entered the quiescent state
Fig 7 Quantitative immunofluorescence to compare Cdt1 protein levels
in individual HFF and HeLa cells Triplicates of HFF and HeLa cells
grown on coverslips were subjected to immunofluorescence with
anti-Cdt1 Ig (A) Microscopy images recorded under identical conditions
are shown (B) A scatter plot of the expression values of Cdt1 in 25
different high-power fields each for HFF and HeLa cells is shown.
Relative expression values were measured with IPLAB software in
arbitrary units.
Fig 8 Cdt1 and geminin mRNA levels in cancer cell lines and tumours.
16 (A) Northern blotting analysis was performed to determine mRNA levels of hCdt1, h-geminin and actin in different human cell lines Lanes 1–5, HeLa, MDAMB231, LNcap, MCF7 and HFF cells, respectively (B) Northern blot bearing total mRNA from human kidney, liver and lung tumors and corresponding normal specimens was hybridized with a geminin specific probe, and an actin specific probe as loading control.
Trang 10This would explain why a significant down-regulation of
Cdt1 in G0 was not previously detected [14] Cdc6, Cdt1’s
partner for DNA licensing, has been shown to be
down-regulated in G0 cells and induced upon cell cycle re-entry
[23] and to be under the transcriptional control of E2F [24–
26] We show here that, similar to Cdc6, Cdt1 mRNA levels
are reduced in G0 and re-accumulate at the G0 to cell cycle
transition, suggesting that, in this transition, Cdt1 may be
controlled transcriptionally The presence of predicted E2F
binding sites on the putative Cdt1 promoter (D Kougiou
and S Taraviras, unpublished observation) attests to a
possible E2F mediated regulation of Cdt1 Indeed E2F was
recently reported to regulate the transcription of Cdt1 [34]
In contrast, in cycling cells, Cdt1 appears to be controlled
mostly post-transcriptionally [13] The difference in Cdt1
regulation we observe between serum deprived and contact
inhibited NIH3T3 cells may indicate a regulation of Cdt1 by
growth factors present in the serum In that respect it is
interesting to note that consensus sites for myc and TCF-1
A are also present on the Cdt1 promoter In addition, we
detected slower migrating forms of Cdt1 in serum deprived
HFF cells upon longer exposure (data not shown),
suggesting that an additional control of Cdt1 at the
post-translational level may also operate in G0
Geminin is believed to act as a cell cycle inhibitor, with a
role to prevent untimely licensing by specifically binding to
Cdt1 and inhibiting its MCM loading function [12,15,16]
During the cell cycle, geminin is expressed in S and G2
phases, when licensing should be inhibited and when the
Cdt1 protein is undetectable [13,16] Based on these
findings, one might have expected geminin to be induced
when G1 cells exit the cell cycle into G0 In contrast,
however, we show here that geminin levels are extremely
low in quiescent cells, with a decrease even more
pro-nounced than that observed for Cdt1 For example, geminin
is already undetectable in NIH3T3 cells in early confluency,
when Cdt1 levels are not affected Geminin levels closely
mirror the levels of cyclin A in these experiments Geminin
mRNA levels also appear to be reduced early upon serum
withdrawal and more dramatically than Cdt1 mRNA levels
These findings are consistent with a recent publication
linking geminin expression to the proliferating cell [17] An
E2F binding site is also present on the predicted geminin
promoter (D Kougiou and S Taraviras, unpublished
observations) suggesting that Cdt1 and its inhibitor might
be under similar transcriptional regulatory mechanisms
Indeed, while this manuscript was under review, regulation
of geminin by E2F and RB was reported [34,35] Geminin
protein accumulates in HFF cells re-entering the cell cycle
from G0 3–6 h later than Cdt1, and as cells enter into
S phase (Fig 2) The accumulation of geminin may inhibit
further licensing and define the window of opportunity for
licensing at the transition from quiescence to proliferation
The distribution of Cdt1 and geminin in the gut
epithelium mirrors our findings with cultured cells We
show by in situ hybridization and immunohistochemistry
that both Cdt1 and geminin are expressed in the
prolifer-ative compartment of the developing mouse gut epithelium,
consistent with a down-regulation of these factors upon cell
cycle exit This is the first report of the distribution of Cdt1
in a mammalian tissue, while our findings are consistent
with a previous report on geminin’s localization [17]
Cancer cells have defects in the control mechanisms regulating cell cycle exit and therefore divide uncontrolla-bly In addition, cancer cells often exhibit genomic instability and are aneuploid A recent report showed that over-expression of Cdt1 can predispose cells to a malignant transformation [36], thereby identifying Cdt1 as a putative oncogene In addition, over-expression of Cdt1 together with Cdc6 has been shown to result in re-replication and genomic instability in both yeast and human cells [7,37]
We wished to examine whether Cdt1 is over-expressed in cancer cells in culture, in comparison with normal cycling cells We show that both Cdt1 protein and mRNA accumulate to much higher levels in cancer cells Immuno-fluorescence experiments showed that Cdt1 is present in higher levels in individual cancer cells, vs cycling primary fibroblasts Over-expression of Cdt1 in cancer cell lines can therefore not be solely accounted for by the larger in cycle fraction of cancer cells It would be interesting to investigate the mechanism that leads to over-accumulation
of Cdt1 in cancer cells and the functional significance of this over-expression for malignant transformation The presence of increased mRNA levels suggests that over-expression may be at least partly due to increased transcription or increased gene copy number Over-expres-sion of a stable form of geminin was recently shown to induce cell cycle arrest or apoptosis in human cell lines [17,38] However, endogenous geminin is over-expressed in the majority of cancer cells tested, in comparison with normal cells This is observed both in cancer cell lines and human tumors and is consistent with a recent report [17] The degree of over-expression of the geminin protein differs between cell lines, resulting in gross differences in relative amounts of Cdt1 and its inhibitor in different cell lines It would be interesting to investigate whether geminin and/or Cdt1 levels or the ratio of Cdt1 to its inhibitor geminin may show a correlation with the type, aggressive-ness or molecular pathology of a given tumor
Acknowledgements
We would like to thank A Pyriohou, D Kalatzis, M Iliou and V Roukos for assistance with experiments, M Ohtsubo and N Tsopanoglou for cell lines, the Bastiaens laboratory for help with quantitations and Profs G Maniatis, C Flordellis and A Athanassia-dou for their support and critical reading of the manuscript Our work
is supported by grants from the Association for International Cancer Research, University of Patras-Karatheodori Program, the Emperirikio Foundation, Human Frontiers Science Program, and the Japanese Ministry of Education, Culture, Sports, Science and Technology.
References
1 Nurse, P (1994) Ordering S phase and M phase in the cell cycle Cell 79, 547–550.
2 Blow, J.J & Laskey, R.A (1988) A role for the nuclear envelope in controlling DNA replication within the cell cycle Nature 332, 546– 548.
3 Blow, J.J & Hodgson, B (2002) Replication licensing – defining the proliferative state? Trends Cell Biol 12, 72–78.
4 Bell, S.P & Dutta, A (2002) DNA replication in eukaryotic cells Annu Rev Biochem 71, 333–374.
5 Nishitani, H & Lygerou, Z (2002) Control of DNA replication licensing in a cell cycle Genes Cells 7, 523–534.