1. SHORT REPORT
Ubiquitin/proteasome-mediated degradation of p19INK4d
determines its
periodic expression during the cell cycle
Minna Thullberg1
, Jiri Bartek1
and Jiri Lukas*,1
1
Danish Cancer Society, Institute of Cancer Biology, Strandboulevarden 49, DK-2100 Copenhagen é, Denmark
Assembly and activity of the proto-oncogenic cyclin D/
CDK4(6) complexes, the major driving force of G1 phase
progression, is negatively regulated by a family of INK4
CDK inhibitors p16INK4a
, p15INK4b
, p18INK4c
, and p19INK4d
.
Expression of the INK4 family members is controlled at
the transcriptional level, through di€erential response to
environmental and intracellular signals such as cytokines,
oncogenic overload, or cellular senescence. Here we show
that the periodic oscillation of the p19INK4d
protein during
the cell cycle is determined by the ubiquitin/proteasome-
dependent mechanism, allowing the protein abundance to
follow the changes in its mRNA expression. Within the
INK4 family, this regulatory mode appears restricted to
p19INK4d
whose ubiquitination was dependent on the
integrity of lysine 62, and binding to CDK4. These
results highlight unexpected di€erences among the INK4
inhibitors, and suggest how p19INK4d
may help regulate
the rate of cyclin D/CDK4(6) complex formation, and
thereby timely progression through the mammalian cell
division cycle. Oncogene (2000) 19, 2870 ± 2876.
Keywords: INK4 family; p19INK4d
; cyclin D; CDK4;
ubiquitin/proteasome-dependent degradation; retino-
blastoma protein
Progression through the G1 phase of the cell cycle, and
initiation of DNA replication require neutralization of
the growth-restraining capacity of the retinoblastoma
tumor suppressor (pRb), whose unphosphorylated
form halts the G1/S transition by sequestering the cell
cycle-promoting transcription factors such as E2F.
Phosphorylation of pRb by cyclin-dependent kinases
(CDKs) triggers a wave of E2F-dependent synthesis of
positive regulators supporting completion of G1 and
initiation of DNA synthesis. This event was proposed
as a molecular basis of the passage through the
Restriction point, which integrates external signals
with the cell cycle machinery and determines the cell's
commitment to replicate the genome and complete one
round of the cell cycle (reviewed in Bartek et al., 1996).
Thus, the molecular network regulating pRb phos-
phorylation appears to underlie a mechanism which
prevents unscheduled proliferation.
Initiation of the pRb phosphorylation in mid-to-late
G1 phase is critically dependent on the accumulation
of the D-type cyclins and their assembly with CDK4 or
CDK6 catalytic subunits (reviewed in Sherr, 1996). The
activity of these holoenzymes is further regulated by
phosphorylation of the CDK, proteolysis of the cyclin
subunit, and by speci®c inhibitory proteins called
cyclin-dependent kinase inhibitors (CKIs) (reviewed in
Lees, 1995; Sherr and Roberts, 1999). Four members
of the CDK4(6)-speci®c family of CKIs have been
identi®ed to date. The founder of this so-called INK4
family, p16INK4a
, has been ®rmly established as a tumor
suppressor, being frequently lost or altered in many
types of human tumors (reviewed in Serrano, 1997;
Ruas and Peters, 1998). The other INK4 inhibitors
were named according to their mass and order of
discovery as p15INK4b
, p18INK4c
and p19INK4d
, respectively
(Hannon and Beach, 1994; Guan et al., 1994; Hirai et
al., 1995; Chan et al., 1995). All four INK4 proteins
share the ability to inhibit the kinase activity of
CDK4(6) by preventing their interaction with the D-
type cyclins. Consequently, overexpression of several
INK4 proteins arrests the cells in G1 phase, in a
strictly pRb-dependent manner (Guan et al., 1994;
Medema et al., 1995; Koh et al., 1995; Lukas et al.,
1995c). Despite the apparent redundancy in their
biochemical function(s) (Hirai et al., 1995; WoÈ l€ and
Naumann, 1999, and reviewed in Roussel, 1999), the
INK4 members di€er in several parameters such as
timing and tissue-speci®c patterns of expression, and
inducibility by diverse external or internal stimuli.
Thus, only p18INK4c
and p19INK4d
are expressed in utero,
while the postnatal gradual appearance of p16INK4a
and
p15INK4b
has been implicated in triggering and/or
maintenance of replicative senescence (Zindy et al.,
1997; Serrano et al., 1997). p15INK4b
seems to be unique
as a mediator of the growth-restraining e€ects of
negative cytokines such as TGF-b (Hannon and Beach,
1994). Recently, rapid elevation of p16INK4a
was
observed in human skin grafts exposed to UV light
and suggested to be a part of the UV-induced DNA
damage checkpoint response (Pavey et al., 1999). One
feature of p19INK4d
which appears distinct from the
remaining INK4 CKIs is the pronounced periodic
accumulation of its protein during the cell cycle: In
synchronized mouse macrophages the protein synthesis
of p19INK4d
was low in G0/G1, and peaked in S phase
(Hirai et al., 1995). Although the mRNA of p18INK4c
was also reported to be induced by mitogens, only the
kinetics of the p19INK4d
protein accumulation seemed to
correlate tightly with its mRNA expression (Hirai et
al., 1995).
The cyclin D/CDK4(6) complexes perform a dual
role in cell cycle regulation, via their speci®c pRb-
directed kinase activity, and as a reservoir for another
class of CKIs, represented by the p21CIP1
and p27KIP1
Oncogene (2000) 19, 2870 ± 2876
ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00
www.nature.com/onc
*Correspondence: J Lukas
Received 20 December 1999; revised 10 March 2000; accepted 15
March 2000

2. proteins (Reynisdottir et al., 1995; Sherr and Roberts,
1999). These CKIs inhibit a broader range of CDKs
including CDK2, an activity operating downstream of
CDK4 and CDK6, which appears necessary and
sucient for the initiation of DNA replication (Lukas
et al., 1997). Both the kinase activity and the
stoichiometric `titration' e€ects of cyclin D/CDK4(6)
complexes help orchestrate normal somatic cell cycles
(Jiang et al., 1998; Lukas et al., 1999; McConnell et al.,
1999) and, when overexpressed, may contribute to
deregulated proliferation of cancer cells (Haas et al.,
1997, and reviewed in Sherr, 1996).
Periodic expression of an INK4 inhibitor could
signi®cantly a€ect the extent of productive formation
of the cyclin D/CDK4(6) complexes, and thus help
ensure the physiological length of G1. In this paper, we
explored the oscillation of the p19INK4d
protein in
human normal diploid cells and tumor-derived cell
lines, and report on a previously unrecognized
regulatory mode of this CKI, mediated by its
proteolysis through the ubiquitin/proteasome-depen-
dent pathway. To our knowledge, this is the ®rst
evidence for ubiquitin-mediated degradation of any
INK4 inhibitor, a ®nding with potentially important
implications for current concepts of the somatic cell
cycle control and oncogenesis.
As the dynamics of the p19INK4d
protein abundance
during the cell cycle has so far been studied
exclusively in murine models, we ®rst examined
whether its periodicity is conserved also in human
cells. In human diploid ®broblasts, synchronized by
contact inhibition and subsequent re-plating to lower
density, the amount of the p19INK4d
protein remained
low-to-undetectable in starved cells and throughout
the whole G1 period, but became progressively
accumulated during S and G2/M phases (Figure 1a).
The accumulation of the p19INK4d
protein signi®cantly
lagged behind the D-type cyclins, the expression of
which peaks during the G1 phase, and correlated well
with the appearance of cyclin A, an established
regulator of S phase progression and G2/M transition.
The protein levels of the p19INK4d
speci®c interactors,
CDK4 and CDK6, remained constant throughout the
cell cycle (Figure 1a).
The next issue we addressed was whether or not the
p19INK4d
protein oscillates also in continuously prolif-
erating cells. We synchronized exponentially growing
human ML-1 cells by centrifugal elutriation, and
immunoblotting analysis of the individual cell fractions
revealed a characteristic pattern of the p19INK4d
protein
accumulation similar to ®broblasts released from
quiescence. These results con®rmed that similarly to
normal ®broblasts (see Figure 1a), also in tumor-
derived ML-1 cells, the p19INK4d
protein was low in G1
cells and gradually accumulated as cells passed the G1/
S transition (Figure 1b). Densitometric analysis
revealed 18-fold increase of the p19 protein in G2
fractions compared to the lowest G1 amount detected
in fraction 4. On the contrary, analysis of the matched
elutriated fractions showed that the protein level of
p18INK4c
¯uctuated only moderately with maximum
twofold di€erences between G1 and S/G2 fractions
(Figure 1b). These data con®rmed and extended
previous observations in mouse cells (Hirai et al.,
1995), and showed that in both normal and tumor-
derived cells of human origin, p19INK4d
protein levels
dramatically oscillate, with a nadir in G0/G1, and a
peak in S/G2 phases of the cell cycle.
The highly periodic accumulation of p19INK4d
but not
p18INK4c
protein was intriguing since the mRNA
expression of both genes ¯uctuated with a very similar
kinetics in a cell cycle dependent manner: Northern
blot analyses demonstrated that the steady-state levels
of both mRNAs were low in G1 and induced upon
entry into S phase (Hirai et al., 1995, and our
unpublished data). Our observation that only the
p19INK4d
protein levels closely followed the pattern of
its mRNA expression could be explained if p19INK4d
was an unstable protein. This would be, however,
Figure 1 Periodic accumulation of human p19INK4d
during the
cell cycle. (a) Skin diploid ®broblasts (BJ strain) were
synchronized by incubating the con¯uent cultures for 5 days
and subsequent trypsinization and re-plating into a fresh medium.
After lysis at the indicated timepoints, the SDS ± PAGE-separated
protein lysates (50 mg per lane) were immunoblotted with the
indicated antibodies: DCS 100 to p19INK4d
(Thullberg et al.,
2000), DCS 156 to CDK4 (Lukas et al., 1999), DCS 183 to CDK6
(Lukas et al., 1999), a mixture of antibodies to the D-type cyclins
described earlier (Lukas et al., 1994; Bartkova et al., 1998), and
rabbit antiserum to cyclin A (gift from M Pagano). Conditions
for Western blotting were described (Lukas et al., 1995a). The
diagram on the top shows the degree of synchronous progression
through the cell cycle as determined by ¯ow cytometry.
Exp.=control, asynchronously growing cells. (b) Exponentially
growing ML-1 cells were synchronized by centrifugal elutriation
as described (Lukas et al., 1995b). Cell cycle distribution of
individual fractions were determined by ¯ow cytometry (top), and
the corresponding lysates (50 mg per lane) were analysed by
SDS ± PAGE and immunoblotting with antibodies against
p19INK4d
(DCS 100 [Thullberg et al., 2000]) and p18INK4c
(DCS
118 [Thullberg et al., 2000]). Exp.=control, asynchronously
growing cells
Oncogene
Proteasome-dependent turnover of p19
INK4d
M Thullberg et al
2871

3. unexpected since it was repeatedly reported that the
other INK4 proteins such as p16INK4a
and p15INK4b
are
very stable, with half-lives over 8 h (Parry et al., 1995;
Shapiro et al., 1995; Sandhu et al., 1997). To measure
the protein turnover of endogenous p19INK4d
directly in
human cells and in parallel to p16INK4a
, we used the
immortalized keratinocyte cell line HaCat expressing
both of these INK4 proteins at the levels amenable to
metabolic labeling and thus allowing reliable estima-
tion of the protein half-lives of endogenous p19INK4d
and p16INK4a
. By metabolic pulse labeling with
[35
S]methionine and subsequent chase over a 4 h
period, we found that the p16INK4a
protein was stable
throughout the entire chase period, while p19INK4d
was
rapidly degraded with the protein half-life between 20
and 30 min (Figure 2a).
To directly compare the protein stability of p19INK4d
and p18INK4c
, the only two INK4 family members
showing signi®cant mRNA oscillation during the cell
cycle, we analysed their turnover in the human
osteosarcoma cell line U-2-OS, expressing comparable
levels of both of these inhibitors. Due to the lack of a
suitable antibody supporting quantitative immunopreci-
pitation of the total cellular pool of human p18INK4c
, we
assessed the protein turnover by inhibiting translation
with cycloheximide (CHX). To validate such approach,
we ®rst con®rmed that the protein half-lives of both
p19INK4d
and p16INK4a
in cells treated with CHX were
nearly identical (30 min and 6 h, respectively) to those
obtained through the [35
S]methionine pulse chase (Figure
2b, and data not shown). Under identical conditions,
p18INK4c
remained stable with the half-life over 4 h
(Figure 2b, lower panel). We have reproduced these
results using other cell lines, consistently detecting rapid
degradation of p19INK4d
, and slow protein turnover of
p16INK4a
, p18INK4c
, and 15INK4b
(data not shown).
Proteolysis of the INK4 CKIs has not been system-
atically studied so far, and its potential impact on the
cell cycle control is unknown. In contrast, protein
stability of the CIP/KIP inhibitors has been investi-
gated thoroughly, giving rise to a concept of their
ubiquitin/proteasome-mediated degradation as a fun-
damental mechanism coordinating G1/S transition
(Carrano et al., 1999; Montagnolli et al., 1999;
Sutterluty et al., 1999). To see whether the rapid
turnover of p19INK4d
is also dependent on the ubiquitin/
proteasome machinery, we inhibited the proteasome
function either in exponentially growing cells or in
human diploid ®broblasts synchronized by release from
contact inhibition. Treatment of asynchronous cells
either with LLnL or Lactacystin, two drugs capable of
interfering with the proteasome function resulted in
approximately threefold accumulation of the p19INK4d
protein in several di€erent human cell lines (data not
shown). More signi®cantly, proteasome inhibition in
quiescent and to lesser degree also in G1 cells with
LLnL lead to stabilization and clear accumulation of
the p19INK4d
protein (Figure 3a), reaching nearly the
maximal levels observed in later stages of the cell cycle.
In contrast, LLnL treatment at the time-points
enriched for S and G2 cells induced only a moderate
increase of the p19INK4d
protein compared to G0 and
G1 cells. (Figure 3a), suggesting at least its partial
stabilization compared to G0 and G1 phases. The
residual response of S and G2 cells to LLnL could be
potentially explainable by the inherent problem of
incomplete synchrony in primary cells released from
quiescence. Indeed, ¯ow cytometry analyses con®rmed
that despite signi®cant enrichment of S and G2 cells in
the later time-points, we consistently detected variable
proportions of `contaminating', less eciently stimu-
lated G1 cells (see Figure 1a, and data not shown).
Alternatively, we cannot exclude that some degree of
the proteasome-dependent degradation of the p19INK4d
protein operates also in S and G2 phases, albeit with a
somewhat lower rate compared to G1 and especially
G0. In either case, proteasome-dependent degradation
could signi®cantly contribute to the ®ne adjustment of
of the p19INK4d
protein levels during the cell cycle, likely
acting in concert with other regulatory mechanisms
including periodic transcription (Hirai et al., 1995).
To corroborate the data obtained upon proteasome
inhibition, we directly tested whether p19INK4d
could be
ubiquitinated and thus marked for ecient recognition
by the proteasome. For that reason, we ®rst chose the
U-2-OS cell line because of its high transfection
eciency required for the ubiquitination assay. Indeed,
co-expression of p19INK4d
together with ubiquitin lead
to speci®c appearance of the high-molecular weight
polyubiquitinated species in an in vivo ubiquitination
assay (Figure 3b), the formation of which was
Figure 2 p19INK4d
, unlike the other INK4 inhibitors, is a highly
unstable protein. (a) Determination of p19INK4d
and p16INK4a
protein half-lives in HaCat cells. [35
S] pulse-chase assays were
performed essentially as described (Welcker et al., 1996). Labeled
p19INK4d
and p16INK4a
proteins were immunoprecipitated with
speci®c monoclonal antibodies DCS 100 (Thullberg et al., 2000)
and DCS 50 (Lukas et al., 1995c), respectively. The signal of the
autoradiogram (top) was quanti®ed by phosphorimager (bottom).
(b) U-2-OS cells were incubated in the presence of cycloheximide
(CHX; 25 mg/ml) for the indicated time and the levels of the
p19INK4d
and p18INK4c
proteins were analysed by immunoblotting
with monoclonal antibodies speci®ed in Figure 1(b)
Proteasome-dependent turnover of p19
INK4d
M Thullberg et al
2872
Oncogene

4. dramatically increased by co-expressing CDK4, the
speci®c interacting partner of p19INK4d
(Figure 3b).
None of the other INK4 proteins tested in parallel
showed any signi®cant ubiquitination (negative data,
not shown). Taken together, our results revealed for
the ®rst time, that a member of the INK4 family of
CDK inhibitors could be targeted for rapid degrada-
tion by the ubiquitin/proteasome pathway. Moreover,
our results suggested that for an ecient ubiquitination
to occur, p19INK4d
must ®rst associate with CDK4.
In order to provide direct evidence for ubiquitination
of p19INK4d
, we ®rst attempted to construct a mutant of
p19INK4d
which would speci®cally uncouple the protein
from modi®cation by ubiquitin ligases. Covalent
attachment of the polyubiquitin chain to the target
protein strictly requires lysine residues (reviewed in
Ciechanover, 1998). Inspection of p19INK4d
amino acid
sequence revealed three lysines, residues 43, 62, and 91
of the human protein, respectively. Signi®cantly, none
of these residues were conserved in any of the
remaining three, more stable INK4 proteins. Con-
comitant substitution of the three lysines to arginines
resulted in a p19INK4d
protein resistant to ubiquitination
even upon coexpression of ectopic CDK4 which
otherwise greatly potentiated ubiquitination of the
wild-type protein (Figure 4a). Subsequent analysis of
single point mutations demonstrated that the lysine 62
likely represents the major ubiquitin recipient, since its
substitution to arginine was sucient to dramatically
reduce polyubiquitination of p19INK4d
(Figure 4a).
The observed phenomenon of CDK4-promoted
ubiquitination of p19INK4d
could have at least three
plausible explanations: (i) E€ects of cell cycle position;
(ii) Priming of p19INK4d
for ubiquitination by CDK-
mediated phosphorylation; and (iii) Changes in
p19INK4d
protein conformation upon attachment to
CDK4. It has been demonstrated that overproduction
of the INK4 inhibitors imposes a G1 arrest in cells
harboring wild-type pRb, such as U-2-OS cells used in
our in vivo ubiquitination experiments (Lukas et al.,
1995c). Co-expression of CDK4 can revert such arrest
via sequestration of the excessive INK4 protein(s)
(Koh et al., 1995). Alternatively, S phase entry in the
G1 arrested, INK4-overexpressing cells can be rescued
by ectopic expression of cyclin E/CDK2, which
physiologically operates downstream of pRb phos-
phorylation (Alevizopoulos et al., 1997; Lukas et al.,
1997). To test the potential e€ect of cell-cycle position
as explanation of the CDK4 impact on p19INK4d
ubiquitination, we tested the ability of p19INK4d
to
become ubiquitinated under either of the above
conditions allowing cell cycle progression in the
presence of ectopically expressed p19INK4d
. Multipara-
meter ¯ow cytometry con®rmed that co-expression of
either CDK4 or cyclin E/CDK2 restored DNA
replication in cells overexpressing p19INK4d
(data not
shown). However, only CDK4 was able to stimulate
ubiquitination of p19INK4d
(Figure 4b), ruling out the
indirect e€ect of cell cycle progression. To indepen-
dently con®rm these results, we also studied p19INK4d
ubiquitination in SAOS-2 cells, another human
osteosarcoma cell line which however lacks wild-type
pRb, and therefore cannot be arrested even by high
levels of the INK4 proteins (Lukas et al., 1995c, and
our unpublished observation). Also in this cell line,
CDK4 markedly induced p19INK4d
ubiquitination, while
CDK6 was much less ecient (Figure 4c). Parallel
experiments with U-2-OS cells con®rmed that CDK6
could not trigger massive ubiquitination of p19INK4d
despite its ability to interact with p19INK4d
to an extent
indistinguishable from CDK4 under these experimen-
tal conditions (data not shown). Experiments in both
U-2-OS and SAOS-2 cell lines further showed that co-
expression of cyclin D1, the natural activator of
CDK4, completely abolished the CDK4-mediated
stimulation of p19INK4d
ubiquitination (Figure 3b).
Together with the fact that p19INK4d
-associated CDK4
possesses no kinase activity (data not shown), and
that a catalytically inactive mutant of CDK4 (Van den
Heuvel and Harlow, 1993) stimulated p19INK4d
ubqui-
tination to the same degree as the wild-type allele
(Figure 4d), these data indicate that neither the cell
cycle position, nor phosphorylation of p19INK4d
by
active cyclin D/CDK4 complexes determine ubiquiti-
nation of p19INK4d
.
In order to test directly the third possibility, we
employed a melanoma-associated, point missense
mutant of CDK4, CDK4-R24C, which is impaired in
its ability to interact with the INK4 proteins (WoÈ lfel et
al., 1995; Bartkova et al., 1996). In an in vivo
ubiquitination assay, expression of the CDK4-R24C
mutant was reproducibly less ecient in promoting
formation of the polyubiquitin chains when coex-
pressed with p19INK4d
, relative to the wild-type allele
Figure 3 p19INK4d
is ubiquitinated and degraded in a protea-
some-dependent manner. (a) Skin diploid ®broblasts (strain
AG01518B) were synchronized by contact inhibition essentially
as described in Figure 1(a). After replating to lower density, the
cell cycle re-entry was monitored by ¯ow cytometry, and at the
indicated cell cycle phases, the abundance of p19INK4d
was
analysed by immunoblotting, and compared with a stable and
non-oscillating protein CDK7 (Tassan et al., 1994), used as a
loading control. Where indicated, the cells were treated with
LLnL (25 mM) for 4 h before lysis. (b) U-2-OS cells were
transfected with the indicated CMV-driven expression plasmids
(5 mg DNA of each), and subjected to the in vivo ubiquitination
assay (Treier et al., 1994). Polyubiquitinylated p19INK4d
species
were detected with the DCS 100 monoclonal antibody
Oncogene
Proteasome-dependent turnover of p19
INK4d
M Thullberg et al
2873

5. or to another CDK4 mutant, de®cient in its catalytic
activity yet capable of interacting with the INK4
proteins (Van den Heuvel and Harlow, 1993, and our
unpublished observation) (Figure 4c). Western blotting
analysis performed in parallel con®rmed that all CDK4
versions were expressed to similar levels (Figure 4d).
Collectively, our results are consistent with the
interpretation that the direct interaction with free,
presumably monomeric CDK4 determines productive
ubiquitination and subsequent rapid degradation of
p19INK4d
.
Contrary to the tremendous progress in our knowl-
edge about the molecular network controlling protein
turnover of the p27Kip1
CDK inhibitor (reviewed in
Amati and Vlach, 1999), regulation of protein stability
of the INK4 CKIs has remained unexplored despite the
evidence implicating accumulation of these inhibitors
in fundamental biological processes such as cellular
∆∆
Figure 4 Association with CDK4 promotes p19INK4d
ubiquitination. (a) Wild-type p19INK4d
or the indicated lysine-to-arginine (R)
mutants were transiently expressed together with CDK4 and ubiquitin in U-2-OS cells (trR, triple substitution of lysines 43, 62, and
91 to arginine). The in vivo ubiquitination assay was performed essentially as described in Figure 3(b) with the exception that
ubiquitin was expressed from the pCW7 plasmid (gift from C Ward) which gives lower amount of ubiquitin compared to pCMV-
neo/bam expression pasmid used in Figure 3(b). (b) In vivo ubiquitination assay of the wild-type p19INK4d
in U-2-OS (left panel) and
SAOS-2 cells (right panel) following transient transfection of the expression plasmids as indicated (note that only CDK4 alone can
eciently promote robust polyubiquitination of p19INK4d
). (c) Comparison of the eciency of the wild-type allele (wt), INK4-
binding-de®cient R24C mutant (DINK), and catalytically inactive mutant (dn) of CDK4 to induce ubiquitination of p19INK4d
. In
vivo ubiquitination assay was performed in SAOS-2 cells as described in (a) and (b). (d) Equal expression levels of the HA-tagged
CDK4 wild-type and mutants used in (c) were veri®ed by imunoblotting with the monoclonal antibody DCS 156 (Lukas et al., 1999)
Proteasome-dependent turnover of p19
INK4d
M Thullberg et al
2874
Oncogene

6. senescence or cell-cycle arrest in response to negative
cytokines (reviewed in Roussel, 1999). Our present
study brings the ®rst evidence that also a member of
the INK4 family of CKIs is targeted by the ubiquitin/
proteasome pathway, and could be regulated at the
level of protein stability. Physiologically low levels of
the p19INK4d
protein seen in rodent (Hirai et al., 1995)
and human cells (this study) could participate in
controlling timely assembly to critical threshold levels
of the cyclin D/CDK4(6) complexes during G1
progression. Conversely, accumulation of p19INK4d
in
early S phase co-incides with the disappearance of the
CIP/KIP assembly factors of cyclin D/CDK4(6)
complexes, and may help prevent de-novo formation
of these complexes before they are needed in the
subsequent cell cycle. Together with the previously
described evidence for its periodic mRNA expression
(Hirai et al., 1995), the identi®cation of the ubiquitin/
proteasome-dependent step in the control of the
p19INK4d
protein abundance helps to explain the cellular
capacity to timely and accurately adjust the actual
steady-state levels of this potentially important CDK
inhibitor. Our study already revealed one step critical
for p19INK4d
ubiquitination, namely physical association
with CDK4. Search for the speci®c ubiquitin ligase
targeting p19INK4d
, and better understanding of how
interference with the cell cycle-dependent oscillation of
p19INK4d
may a€ect cell cycle progression, are among
the intriguing issues raised by the unexpected ®ndings
we report.
In contrast to the accepted tumor-suppressor role of
other INK4 CKIs, genetic alterations of p19INK4d
are
yet to be ®rmly implicated in human oncogenesis.
Given our present results, it is tempting to compare
p19INK4d
with p27KIP1
whose protein abundance appears
aberrantly downmodulated in common human tumor
types, through mechanisms targeting p27 turnover
(Esposito et al., 1997; Loda et al., 1997). By analogy,
it might be rewarding to examine p19INK4d
protein levels
in normal versus tumor tissues, to search for any
potential abnormalities possibly re¯ecting some de-
fect(s) in the ubiquitin/proteasome-dependent turnover
of this CDK inhibitor and candidate tumor suppressor.
Acknowledgments
We wish to thank G Evan, NE Fusenig, K Helin, M
Pagano, S Reed, M Seto, J Shay, C Ward and A Winoto
for generous gift of some reagents, and M Welcker for his
advice with the ubiquitination assays. This work was
supported by grants from the Danish Cancer Society,
The Danish Medical Research Council, and the Human
Frontier Science Program.
References
Alevizopoulos K, Vlach J, Hennecke S and Amati B. (1997).
EMBO J., 16, 5322 ± 5333.
Amati B and Vlach J. (1999). Nat. Cell Biol., 1, E91 ± E93.
Bartek J, Bartkova J and Lukas J. (1996). Curr. Opin. Cell
Biol., 8, 805 ± 814.
Bartkova J, Lukas J, Guldberg P, Alsner J, Kirkin AF,
Zeuthen J and Bartek J. (1996). Cancer Res., 56, 5475 ±
5483.
Bartkova J, Lukas J, Strauss M and Bartek J. (1998).
Oncogene, 17, 1027 ± 1037.
Carrano AC, Eytan E, Hershko A and Pagano M. (1999).
Nat. Cell Biol., 1, 193 ± 199.
Chan FKM, Zhang J, Cheng L, Shapiro DN and Winoto A.
(1995). Mol. Cell. Biol., 15, 2682 ± 2688.
Ciechanover A. (1998). EMBO J., 17, 7151 ± 7160.
Esposito V, Baldi A, De Luca A, Groger AM, Loda M,
Giordano GG, Caputi M, Baldi F, Pagano M and
Giordano A. (1997). Cancer Res., 57, 3381 ± 3385.
Guan KL, Jenkins CW, Li Y, Nichols MA, Wu X, O'Keefe
CL, Matera AG and Xiong Y. (1994). Genes Dev., 8,
2939 ± 2952.
Haas K, Staller P, Geisen C, Bartek J, Eilers M and MoÈ roÈ y
T. (1997). Oncogene, 15, 179 ± 192.
Hannon G and Beach D. (1994). Nature, 371, 257 ± 261.
Hirai H, Roussel MF, Kato J, Ashmun RA and Sherr CJ.
(1995). Mol. Cell. Biol., 15, 2672 ± 2681.
Jiang H, Chou HS and Zhu L. (1998). Mol. Cell Biol., 18,
5284 ± 5290.
Koh J, Enders GH, Dynlacht B and Harlow E. (1995).
Nature, 375, 506 ± 510.
Lees E. (1995). Curr. Opin. Cell Biol., 7, 773 ± 780.
Loda M, Cukor B, Tam SW, Lavin P, Fiorentino M, Draetta
GF, Jessup JM and Pagano M. (1997). Nat. Med., 3, 231 ±
234.
Lukas C, Jensen SS, Bartkova J, Lukas J and Bartek J.
(1999). Hybridoma, 18, 225 ± 234.
Lukas J, Bartkova J, Rohde M, Strauss M and Bartek J.
(1995a). Mol. Cell Biol., 15, 2600 ± 2611.
Lukas J, Bartkova J, Welcker M, Petersen OV, Peters G,
Strauss M and Bartek J. (1995b). Oncogene, 10, 2125 ±
2134.
Lukas J, Herzinger T, Hansen K, Moroni MC, Resnitzky D,
Helin K, Reed SI and Bartek J. (1997). Genes Dev., 11,
1479 ± 1492.
Lukas J, Pagano M, Staskova Z, Draetta G and Bartek J.
(1994). Oncogene, 9, 707 ± 718.
Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss
M, Peters G and Bartek J. (1995c). Nature, 375, 503 ± 506.
Lukas J, Sorensen CS, Lukas C, Santoni-Rugiu E and Bartek
J. (1999). Oncogene, 18, 3930 ± 3935.
McConnell BB, Gregory FJ, Stott FJ, Hara E and Peters G.
(1999). Mol. Cell Biol., 19, 1981 ± 1989.
Medema RH, Herrera RE, Lam F and Weinberg RA. (1995).
Proc. Natl. Acad. Sci. USA, 92, 6289 ± 6293.
Montagnoli A, Fiore F, Eytan E, Carrano AC, Draetta GF,
Hershko A and Pagano M. (1999). Genes Dev., 13, 1181 ±
1189.
Parry D, Bates S, Mann DJ and Peters G. (1995). EMBO J.,
14, 503 ± 511.
Pavey S, Conroy S, Russell T and Gabrielli B. (1999). Cancer
Res., 59, 4185 ± 4189.
Reynisdottir I, Polyak K, Iavarone A and Massague J.
(1995). Genes Dev., 9, 1831 ± 1845.
Roussel MF. (1999). Oncogene, 18, 5311 ± 5317.
Ruas M and Peters G. (1998). Biochim. Biophys. Acta.,
F115 ± F177.
Sandhu C, Garbe J, Bhattacharya N, Daksis J, Pan CH,
Yaswen P, Koh J, Slingerland JM and Stampfer MR.
(1997). Mol. Cell Biol., 17, 2458 ± 2467.
Serrano M. (1997). Exp. Cell Res., 237, 7 ± 13.
Serrano M, Lin AW, McCurrach ME, Beach D and Lowe
SW. (1997). Cell, 88, 593 ± 602.
Shapiro GI, Park JE, Edwards CD, Mao L, Merlo A,
Sidransky D, Ewen EM and Rollins BJ. (1995). Cancer
Res., 55, 6200 ± 6209.
Sherr CJ. (1996). Science, 274, 1672 ± 1677.
Oncogene
Proteasome-dependent turnover of p19
INK4d
M Thullberg et al
2875

7. Sherr CJ and Roberts JM. (1999). Genes Dev., 13, 1501 ±
1512.
Sutterluty H, Chatelain E, Marti A, Wirbelauer C, Senften
M, Muller U and Krek W. (1999). Nat. Cell Biol., 1, 207 ±
214.
Tassan JP, Schultz SJ, Bartek J and Nigg EA. (1994). J. Cell
Biol., 127, 467 ± 478.
Thullberg M, Welcker M, Bartkova J, Kjerul€ A A, Lukas J,
HoÈ gberg J and Bartek J. (2000). Hybridoma, 19, 63 ± 72.
Treier M, Staszewski LM and Bohmann D. (1994). Cell, 78,
787 ± 798.
Van den Heuvel S and Harlow E. (1993). Science, 262, 2050 ±
2054.
Welcker M, Lukas J, Strauss M and Bartek J. (1996).
Oncogene, 13, 419 ± 425.
WoÈ lfel T, Hauer M, Schneider J, Serrano M, WoÈ lfel C,
Klehmann-Hieb E, De Plaen E, Hankeln T, Meyer zum
BuÈ schenfelde K-G and Beach DA. (1995). Science, 269,
1281 ± 1284.
WoÈ l€ B and Naumann M. (1999). Oncogene, 18, 2663 ± 2666.
Zindy F, Quelle DE, Roussel MF and Sherr CJ. (1997).
Oncogene, 15, 203 ± 211.
Proteasome-dependent turnover of p19
INK4d
M Thullberg et al
2876
Oncogene