Vol. 7, 1283-1290, October 1996 Cell Growth & Differentiation 1283

Cyclin EICdk2 Activity Is Controlled by Different Mechanismsin the G0 and G1 Phases of the Cell Cycle1

Wade A. Bresnahan, lstvan Boldogh, Tianlin Ma,Thomas Albrecht, and E. Aubrey Thompson2

Departments of Microbiology [W. A. B., I. B., T. A.] and HumanBiological Chemistry and Genetics [1. M., E. A. T.], University of TexasMedical Branch, Galveston, Texas 77550

AbstractThe experiments described in this report wereundertaken to define the parameters that regulate

cyclin E/cyclin-dependent kinase 2 (Cdk2) kinaseactivity in mitotically quiescent, serum-starved

fibroblastic cells and in cells that had been stimulatedto enter the cell cycle and progress through G1 into Sphase. We have analyzed the expression of cyclin Eand Cdk2, the extent to which these two proteins formcomplexes, and the enzymatic activity of cyclin E/Cdk2kinase. Particular attention was focused uponsubcellular localization and the effect ofcompartmentalization on the association betweencyclin E and Cdk2. In addition, we have examined theinteraction of cyclin E/Cdk2 complexes with two well-characterized inhibitors of Cdk2 kinase activity, CipIand Kipi . This represents the first report in which all ofthese parameters have been measured simultaneouslyin a single, normal diploid cell line. In G0 cells, there isabundant cyclin E and Cdk2, yet there is little or nodetectable Cdk2-dependent histone HI kinase activity.After serum stimulation, there is a rapid increase in theamount of cyclin E that is bound to Cdk2, althoughthere is no significant change in the abundance ofeither the cyclin or the Cdk. Immunocytochemical dataindicate that cyclin E, CipI , and KipI are located withinthe nuclei of cells in G0, but very little Cdk2 is observedwithin the nuclei of serum-starved cells. Cdk2 rapidlyenters the nucleus upon serum stimulation. Theabundance of the cyclin E/Cdk2 complex increases tothe extent that the binding capacity of CipI isexceeded about 8-12 h after serum stimulation. Theabundance of Kipi decreases at the same time thatthe Cipi threshold is exceeded, so that cyclin EIKipl -

containing complexes decrease by 90% within 8-12 h.

Received 5/13/96; revised 7/30/96; accepted 8/6/96.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1 734 solely to mdi-cate this fact.1 This work was supported in part by NIH Grants AG1 051 4 and CA24347(to E. A. T.) and by NIH Grant DEl 1389 and EPA Grant R81-9495 (toT. B. A.). W. A. B. is a recipient of a James W. McLaughlin PredoctoralFellowship.2 To whom requests for reprints should be addressed, at Department ofHuman Biological Chemistry and Genetics, University of Texas MedicalBranch, Galveston, TX 77550-0645. Phone: (409) 772-3361 ; Fax:(409) 772-5102; E-mail: athompso@mspol.med.utmb.edu.

Cyclin E/Cdk2 kinase activity begins to increase rapidlythereafter, reaching a maximum level about 16 h afterserum stimulation. We have been unable to detecthistone HI kinase activity in complexes that containcyclin E bound to KipI or CipI. We conclude thatcompartmentalization is the predominant barrier toactivation of cyclin E-dependent kinases in quiescentcells. CipI and KipI serve to prevent prematureactivation of cyclin E/Cdk2 complexes that form duringG0 or early G1.

IntroductionProgress through the eukaryotic cell division cycle is regu-

lated by serine/threonine protein kinases, known as Cdks3

(reviewed in Refs. 1-3). Cdks function as components of

multisubunit complexes, which contain both stimulatory and

inhibitory subunits. The stimulatory subunits are known as

cyclins (reviewed in Refs. 1 , 2, and 4), and the inhibitory

subunits are called CKls (reviewed in Refs. 5-7). Among

those kinases that regulate G1 progression, Cdk4 and Cdk6

are activated by association with one or another of the D-

type cyclins (4, 8, 9), whereas Cdk2 is activated primarily by

association with cyclin E or cyclin A (1 0-1 2). The biochemical

consequences of binding of cyclins to Cdks are not corn-

pletely understood, although it is known that binding of cy-

dm5 �5 a prerequisite for covalent modifications that are

essential for catalytic activity (reviewed in Refs. 1 3 and 14).

For example, cyclin binding is required for CAK to phospho-

rylate Ti 60 in Cdk2 and Ti 74 in Cdk4 (1 5, 1 6). CKI subunits

act in part to block CAK-dependent activation of Cdks (15-

1 7), suggesting that cyclins and CKIs serve antagonistic

functions with respect to CAK-dependent activation of Cdks.

Broadly speaking, there are two classes of metazoan CKIs.

One class consists of the members of the INK family, which

inactivate Cdk4 and Cdk6 but have no direct effect on Cdk2

activity (reviewed in Refs. 5-7). A second family of CKIs iscomposed of Cipi (also called WAF1 , Cap2O, and Sdii) and

Kipi (18-21), which binds to and inhibits cyclin/Cdk corn-

plexes that contain Cdk2, Cdk4, or Cdk6 (reviewed in Refs.

5-7). It is believed that Cipi and Kipi constitute an activation

threshold for progression through G1 . According to this hy-

pothesis, activation of Cdk2, Cdk4, and Cdk6 cannot occur

until the abundance of the corresponding cyclin/Cdk corn-

plexes exceeds the concentration required to saturate the

“free” pools of Cipi plus Kipi (reviewed in Refs. 5-7 and 14).

The role of CKI thresholds in controlling activation of cyclin

A/Cdk2 complexes during G1 progression has been ad-

dressed in considerable detail (22-25). Less is known about

3 The abbreviations used are: Cdk, cyclmn-dependent kinase; CKI, cyclinkinase inhibitor; CAK, Cdk-activating kinase; BrdUrd, bromodeoxyuridine;EMEM, Eagle’s modified essential medium; FBS, fetal bovine serum.

1284 Control of Cyclin E/Cdk2 Activity in G0 and G1

Table 1 Cell cycle analysis following serum stimulation

Subconfiuent cultures of 18LU cells were initially synchronized by so-rum deprivation, as described in “Materials and Methods,” and subse-quently stimulated with fresh medium containing 20% serum. The cellswere harvested at intervals and stained with propidium iodine, and theDNA content was determined by flow cytometry.

Hours % cells in G,/GO % cells in S % cells in G2/M

0 93.8 (1.3) 1.6(0.2) 5.1 (1.0)4 91.8 (0.2) 2.5(0.4) 5.7(0.6)

8 91.3(0.7) 1.9(0.2) 6.8(0.9)12 87.5 (1.4) 4.5(1.7) 7.9(0.4)16 89.4(0.8) 3.8(0.8) 6.0(1.2)

24 29.2 (1.4) 32.6 (0.5) 38.2 (1.8)

the role of CKls in controlling cyclin E/Cdk2 activity. Themost detailed study of the role of CKls in activation of cyclinE-dependent kinase was carried out by Ohtsubo et a!. (26),who concluded that inactivity of cyclin E/Cdk2 kinase in

quiescent cells was unlikely to be due to CKI binding. The

available data suggest that the activity of cyclin E/Cdk2 in G0

cells is not directly or in any simple way related to the

abundance of cyclin E or Cdk2 or to the binding of CKIs to

cyclin E/Cdk2 complexes. Such data suggest a level of corn-

plexity that is not presently understood or the existence ofregulatory mechanisms that have not been appreciated to

date.

We have undertaken a series of experiments to test as-

pects of the threshold hypothesis. Our studies have focused

upon induction of cyclin E and Cdk2, the rate of formation of

cyclin E/Cdk2 complexes, the abundance Kipi and Cipiproteins and the extent to which they associate with cyclin

E/Cdk2 complexes, histone Hi kinase activity of cyclin

E/Cdk2 complexes, and the subcellular localization of Cipi,Kipi , cyclin E, and Cdk2 during progression from G0 to S.

This represents the first report in which all of these param-

eters of cyclin E-dependent kinase activation have been

measured simultaneously in a single cell line undergoingprogression from G0 to S phase. The results of these studies

revealed that the activity of cyclin E/Cdk2 kinase is deter-

mined only in part by the binding capacities of Cipi and Kipi

and only in part by induction of cyclin E. Our data indicate

that subcellular localization of the components of cyclin E-

dependent kinase is likely to play a very significant role inprecluding kinase activation in mitotically arrested cells. Ingeneral, our results indicate that cornpartrnentalization of

cyclin E and Cdk2 is the major determinant of cyclin E-de-pendent kinase activity in G0, whereas CKls are the majorconstrains upon kinase activity during early G1.

ResultsFormation of Cyclin E/Cdk2 Complexes Precedes Induc-tion of Cyclin E and Activation of Cyclin E-dependentKinase. The studies described below were carried out in

serum-stimulated human diploid fibroblasts, which exhibitsynchronous progression through G1, as illustrated by the

data shown in Table 1 . The majority of these cells maintain a

G1 DNA content for 1 6 h after addition of serum. BrdUrd

labeling of such cultures indicates that <5% of the cells have

accumulated detectable amounts of BrdUrd-labeled DNAwithin 1 6 h after stimulation (data not shown). The cells

rapidly enter S phase between 16 and 24 h after addition of

serum. As shown in Table 1 , at least 70% of the cells exhibit

>2N DNA content at 24 h, and BrdUrd labeling studies

indicate that >85% of the cells in these cultures have initi-

ated DNA replication by between 1 6 and 24 h (data not

shown).

Cyclin E protein was induced after serum stimulation of

quiescent human diploid fibroblasts, as shown in Fig. 1 . Fig.

1A contains data from a representative experiment, and Fig.

lB contains quantitative data representing the average of

two such experiments. The abundance of cyclin E [Fig. 1 B,

CcnE (0)] increased slowly during the first 8 h after serum

stimulation. Thereafter, the amount of cyclin E increased

rapidly from 8-i 2 h to 1 6 h, increasing about 5-fold and

remaining relatively constant for the duration of the experi-

ment. Cdk2 expression was more or less constant, increas-

ing slightly between 1 6 and 24 h after stimulation (Fig. iB, V).Although both cyclin E and Cdk2 remained relatively con-

stant during early G1 (the first 8 h after serum stimulation), the

abundance of the cyclin E/Cdk2 complex increased signifi-

cantly within 4 h after addition of serum [Fig. 1B, CcnE/Cdk2(#{149})].Despite the rapid increase in cyclin E/Cdk2 complexes

during early G1 progression, there was little or no increase in

cyclin E-dependent histone Hi kinase activity [Fig. iB, E

Kinase (t] during the first 8 h after serum stimulation.

The data illustrated in Fig. iB pose two questions con-

ceming the mechanisms that control activation of cyclin

E/Cdk2 complexes in G0 and early G1 (i.e., in serum-arrested

cells and during the first 8 h after serum stimulation of suchcells). Initially, cyclin E/Cdk2 complexes increase at a time

during which neither subunit evidences any significant

change in expression, suggesting that binding of cyclin E to

Cdk2 may be precluded in G0. Secondly, the rate of activa-

tion of cyclin E-dependent kinase differs from that of forma-

tion of the cyclin E/Cdk2 complex, suggesting that inhibitors

may attenuate cyclin E-dependent kinase activity during pro-

gression through early G1.Subcellular Localization of Cdk2 Constrains Activation

of Cyclin E-dependent Kinase in G0. Association betweencyclin E and Cdk2 in quiescent cells could be inhibited by

sequestration of either protein. This inhibition could be af-footed by an inhibitor that prevents formation of the binary

complex, although no such inhibitor is known for Cdk2.Alternatively, sequestration could be affected by compart-

mentalization of cyclin E and Cdk2. lmmunocytochemical

studies were carried out to determine whether cyclin E and

Cdk2 share the same intracellular location in G0 cell.

Serum-starved cells exhibited weak, diffuse cytoplasmic

staining with Cdk2 antibodies, but very little nuclear staining

was observed (Fig. 2, 0 h). Nuclear Cdk2 staining increased

after serum stimulation. Most of the nuclei stained for Cdk2

within i2 h after addition of serum, although some nuclei

were more intensely stained than others (as shown in Fig. 2).

The intranuclear distribution of Cdk2 at i 2 h after stimulationfrequently gave rise to a punctate staining pattern, but thesignificance of this pattern is unknown. Nuclei observed 24 h

Fig. 1. Cyclin E, Cdk2, and cyclin E/Cdk2 com-plexes after serum stimulation of G0-arrested cells.A, fibroblasts at 70-80% confluence were synchro-nized in a quiescent state (G0) by serum starvationfor 48 h and then stimulated by the addition of freshEMEM with 20% FBS. Cell lysates were prepared atintervals after stimulation, and these lysates wereresolved by SDS-PAGE. The resolved proteins weretransferred to nitrocellulose and probed with eithercyclin E (CcnE) or Cdk2 antibodies. The cell lysateswere also immunoprecipitated with cyclin E anti-body. The precipitates were resolved by SDS-PAGEfollowed by immunoblotting with an antibodyagainst Cdk2 (A, CcnE/Cdk2). In addition, immuno-precipitates formed with cyclin E antibodies wereassayed for the ability to phosphorylate histone Hi,as described in “Materials and Methods.” B, quan-titative data representing the mean of two inde-pendent experiments in which the abundance of aprotein or a complex or the activity of a kinase isexpressed relative to the abundance or activity thatprevailed at the time of serum addition (0 h).

A

B

LLics,iC .�00>�

ow9- C000)00’-COCu”

VC�sJC �� V

.00

6

5�

4�

3�

2

1�

0!

16

14

12

10

8

6

4

2

0

0Cu.� Cu

.�0

a) Vc#{176}0CuC

V 0

LLO

Cell Growth & Differentiation 1285

0 4 8 12 16 24

CcnE

Cdk2 -CcnE/Cdk2

CcnE Kinase �

0 4 8 12 16 20 24

Hours After Serum

after serum stimulation stained homogeneously and in-

tensely for Cdk2.

Cyclin E was located primarily in the nuclei of quiescent

cells, as shown in Fig. 2; and the localization of cyclin E did

not change after serum stimulation. These results are con-

sistent with those reported for a transfected cell line that

overexpresses cyclin E (26). The CKls Cipi and Kipi were

also localized primarily in the nuclei of quiescent cells. The

staining intensity of Kipi decreased rapidly after serum stim-

ulation, suggesting a decrease in Kipi expression. Both the

abundance and the subcellular localization of Cipi remained

relatively constant after serum stimulation (Fig. 2). However,

within 24 h after addition of serum, many of the Cipi -stained

nuclei exhibited a pronounced punctate staining pattern,

suggesting that the intranuclear localization of this inhibitor

may change during cell cycle progression. As indicated in

Fig. 2, about three-fourths of the cells exhibited this bright,

punctate staining pattern at 24 h. One notes that about

three-fourths of the cells in this culture had initiated S phase

at this time (Table i).

The data shown in Fig. 2 suggest that Cdk2 is not abun-

dant in nuclei of serum-starved cells. Alternatively, Cdk2

could be bound to a nuclear factor that sequesters the

epitope recognized by the antibodies used in these experi-

ments. Subcellular fractionation was carried out to discrim-

mate between these alternatives. Nuclear and cytosolic frac-

tions were prepared from serum-starved cells and from cellsthat had been stimulated with serum for 24 h. The abundance

of Cdk2 in these fractions was measured by Western blot-

ting, as shown in Fig. 3. Cdk2 in the cytosolic fractions

increased by <2-fold after serum stimulation (Fig. 3A), con-

sistent with the immunocytochemical data (Fig. 2), which

indicate that cytoplasmic Cdk2 remains relatively constant

after serum stimulation. Very little Cdk2 was detected in the

nuclear fractions from serum-starved i 8LU cells, whereas

the amount of nuclear Cdk2 increased dramatically in serum-

stimulated cells. Identical results were obtained with IMR9O,

Wl38, and Balb3T3 fibroblasts (Fig. 3A).

The kinetics of nuclear accumulation are illustrated in Fig.

3B. There was a significant increase in nuclear Cdk2 within4 h after addition of serum to quiescent i 8LU cells. Theamount of nuclear Cdk2 increased in a more or less linear

fashion for 24 h after serum stimulation. The kinetics of

nuclear accumulation of Cdk2 paralleled those of formation

of the cyclin E/Cdk2 complex, shown in Fig. iB, at least for

the first i6 h after serum stimulation.

Activation of Cyclin E/Cdk2 Kinase during Early G1Progression Is Constrained by a CKI Threshold. The datashown in Fig. lB indicate that neither induction of cyclin E,

nuclear uptake of Cdk2, nor formation of cyclin E/Cdk2 corn-

plexes is sufficient to account for the kinetics of activation of

cyclin E/Cdk2 kinase; although all three of these parameters

are clearly important to kinase activation. For example, cyclin

E-associated histone kinase activity (Fig. i B, ) did not begin

to increase until about 8 h after serum stimulation, although

cyclin E/Cdk2 complexes (Fig. i B, #{149})had achieved nearmaximum levels within this period of time. The delay inactivation of cyclin E-dependent kinase relative to formation

Cdk2 Cipi

4�.

. .� %�

‘�

1O

Kipi

.

0

12

24

0

24

0

12

24

0

4

8

Cyclin E

p.(�

.�. :

‘.4

�

A 024024

I8LU � - ...�

IMR9O � ��-�- � . �_

W138 �W

Balb3T3

Cytosol Nuclei

B � 4 8 121624

I8LU � - � - -

Nuclear Cdk2

Fig. 3. Subcellular fractionation of Cdk2. Fibroblastic cells were arrestedand subsequently stimulated with serum as described in “Materials andMethods.” Cytosolic and nuclear fractions were prepared. Aliquots ofeach fraction were resolved by SDS-PAGE, transferred to nitrocellulosemembrane, and probed with Cdk2 antibody. A, abundance of Cdk2 inboth nuclear and ctyosolic fractions in quiescent cells (0) and 24 h afterstimulation. B, abundance of Cdk2 in the nuclear fractions derived from18LU cells 0, 4, 8, 12, 16, and 24 h after serum stimulation.

4A). The binding of Cipi to cyclin E (Fig. 4C, #{149})increasedwith kinetics that were notably different from those of Cipiinduction but similar to the kinetics of formation of cyclinE/Cdk2 complexes (Fig. i). The data are consistent with theconclusion that binding of Cipi to cyclin E increased as theabundance of cyclin E/Cdk2 complexes increased during thefirst 8-iO h after addition of serum. However, it appears thatthe binding capacity of Cipi was exceeded within about12 h, and there was little or no significant increase in theamount of Cipi -containing complexes thereafter.

The abundance of Kipi decreased rapidly after addition ofserum to quiescent 18LU cells (Fig. 4A). This observation isconsistent with the immunocytochemical data shown in Fig.2. It is significant that the amount of Kipi bound to cyclin E

increased during the first 8 h after stimulation, although the

total amount of Kipi decreased during this time. This obser-vation suggests that the binding capacity of Kipi is in excessin G0 and during the first few hours of G1 . The amount of Kipibound to cyclin E began to fall after 8 h, and there was verylittle Kipi bound to cyclin E 16 or 24 h after stimulation.

Although cyclin A/Cdk2/CKI complexes have histone ki-nase activity under certain circumstances (22, 24), we wereunable to detect any kinase activity associated with Kipi or

Cipi , as shown in Fig. 4B. In this experiment, Cipi or Kipiwas depleted from extracts prepared from quiescent (0 h) or

late G1 (i6 h) cells. The amount of Cipi and Kipi in the

Controls

� i....� IFig. 2. Subcellular localization of cyclin E, Cdk2, Cipi , and Kipi afterserum stimulation of G0 fibroblasts. Fibroblasts at 70-80% confluencewere grown on glass coverslips, synchronized in a quiescent state (G0) byserum starvation for 48 h, and stimulated by the addition of fresh EMEMwith 20% FBS. Coverslips were fixed at different times after stimulationand stained with either cyclin E, Cdk2, Cipi , or Kipi antibodies.

of cyclin E/Cdk2 complexes suggests that kinase activity inearly G1 may be constrained by a CKI threshold.

The abundance of the two principal Cdk2 inhibitors, Cipi

and Kipi, as well as the formation of Cipi/cyclin E corn-

plexes and Kipi/Cyclin E complexes, was measured in se-

rum-stimulated i8LU cells, as shown in Fig. 4. Serum stim-ulation caused a modest induction (about 2-fold) of Cipi (Fig.

1286 Control of Cyclin EJCdk2 Activity in G0 and G1

II #{149}�#{149}��

�

-�

a,� C

A

B

C

Cipi

Cipl/CcnE

Kipi

Kip 1/CcnE

Cipi

Kinase

Kipi

Kinase

6

5

4

3

2

I

0

0 4 8 12 1624� � a.

- . �m 1 �m .

0 4 8 12 1624

_.#{149}�- “

Ppt. Sup

0 16 0 16

�

2 3 4

Ppt. Sup

0 16 0 16

1 2 3 4

14

12

10

8

6

4

2

0

Cl)‘.- 0)

ox�O.

.�o

.o�

.2 �-Cu

-�w�l�Q

0 4 8 12 16 20 24

Hours After Serum

Cell Growth & Differentiation 1287

or Kipi antibodies during the first i6 h after serum stimula-

Fig. 4. Cipi and Kipi abundance and as-sociation with cyclin E after serum stimula-tion. A, cell lysates were prepared at inter-vals after serum stimulation. These lysateswere resolved by SDS-PAGE, transferred tonitrocellulose, and probed with either Cipior Kipi antibodies. The lysates were alsoimmunoprecipitated with Ciplor Kipi anti-body; the precipitates were resolved bySDS-PAGE followed by immunoblottingwith cyclin E antibodies to determine theextent of complex formation. B, extractswere prepared from quiescent (0) cells orfrom cells that had been stimulated with se-rum for 1 6 h. These extracts were immuno-precipitated with antibodies against Cipi orKipi, and the abundance of either CKl wasmeasured in the pellets and the supema-tants. Histone kinase activity was measuredin the Cipi or Kipi immunoprecipitates(Lanes 1 and 2). The supematant fractionsfrom these immunoprecipitates were subse-quently immunoprecipitated with cyclin Eantibodies, and histone kinase activity wasmeasured in the resulting precipitates(Lanes 3 and 4). C, quantitative abundanceof Cipi/cyclmn E and Kipi/cyclin E complexformation and cyclin E-dependent histoneHi kinase activity as a function of time afterserum stimulation.

appropriate immunoprecipitates (Lanes 1 and 2) and their

corresponding supernatant fractions (Lanes 3 and 4) mdi-

cates quantitative immunodepletion of both CKls. Little or no

histone kinase activity could be precipitated with either Cipi

tion (Lanes 1 and 2). The small amount of histone kinase

activity that is observed in Cipi immunoprecipitates at 16 h(Lane 2) is probably due to cyclmn AJCdk2/Cipi complexes,which begin to appear about i6 h after serum stimulation

(data not shown), and the kinase activity that is observed in

CcnE ->

I 2

4 1 6 residual cyclin E that was not precipitated when CKI anti-bodies were added to i 6 h extracts (Fig. 5, Lane 2). In

parallel, the supernatant fractions from CKI immunodepleted

3 4IP aCipi Ip aCdk2

& aKlpl

Fig. 5. Immunodepletion of Cipi and Kipi . Cell lysates were prepared 4and 1 6 h after serum addition. The lysates were immunoprecipitated twicewith a mixture of Cipi and Kipi antibodies under the conditions used inthe experiment shown in Fig. 4B. Aliquots of the immunodepleted super-natant fractions were resolved by SDS-PAGE, transferred to nitrocellulosemembranes, and probed with antibodies against cyclin E or Cdk2 (Lanes1 and 2). In the experiment shown in Lanes 3 and 4, aliquots of theimmunodepleted supematant fractions were immunoprecipitated withCdk2 antibodies. The resulting immunoprecipitate was resolved by SDS-PAGE, transferred to nitrocellulose membranes, and probed with antibod-ies against cyclin E of Cdk2 (Lanes 3 and 4).

1288 Control of Cyclin E/Cdk2 Activity in G0 and G1

4 16

Cdk2 ->

Kipi immunoprecipitates (Lanes 1 and 2) is not significantlyhigher than that obtained when preimmune serum is addedto extracts. The immunodepleted supernatant fractions were

subsequently immunoprecipitated with cyclin E antibodies,and cyclin E-dependent kinase activity was measured in theprecipitates (Lanes 3 and 4). A significant amount of cyclin

E-dependent kinase activity was detected in the CKI immu-nodepleted supematant fractions from late G1 cells (Lane 4).

These data indicate that complexes that contain cyclin E,Cdk2, and Cipi or Kipi are inactive.

The quantitative data shown in Fig. 4C emphasize thetemporal relationship between association of cyclin E withthe two CKls (Cipl/CcnE and Kipl/CcnE) and activation ofcyclin E-dependent kinase (CcnE Kinase). The binding ca-

pacity of Cipi was saturated 8-i2 h after addition of serum

(Fig. 4C, #{149}).The binding capacity of Kipi decreased rapidly

from 8-i2 h after stimulation, as the abundance of Kipi fell

(Fig. 4C, #{149}).As a result, the threshold of Cipi was exceeded,

whereas the Kipi threshold decreased. Cyclin E-associated

kinase activity accumulated rapidly at this time (Fig. 4C, A).

The data shown in Fig. 4 are consistent with the hypothesis

that activation of cyclin E-dependent kinase during early G1

progression is attenuated by the binding of CKls to cyclin

E/Cdk2 complexes. This hypothesis predicts that all or mostof the cyclin E/Cdk2 should be bound to Cipi or Kipi at early

time points (e.g. , at 4 h after stimulation). On the other hand,

there should be a significant amount of cyclin E/Cdk2 that isfree of inhibitors in late G1 (e.g., 16 h after stimulation). An

immunodepletion experiment was carried out to test these

predictions. Extracts were prepared 4 and 16 h after serumstimulation. The extracts were immunoprecipitated twice

with a mixture of Cipi and Kipi antibodies under conditions

in which either antibody depletes >90% of its antigen in a

single immunoprecipitation (as shown in Fig. 4B). The super-

natant fractions were resolved by electrophoresis and as-

sayed for cyclin E, as shown in Lanes 1 and 2 of Fig. 5. There

was no detectable cyclin E in the CKI-depleted supematantfraction from 4 h extracts (Fig. 5, Lane 1), although suchsupematant factions contained Cdk2. However, there was

extracts (as shown in Fig. 5, Lanes 1 and 2) were subse-quently precipitated with antibodies against Cdk2, and theimmunoprecipitates were assayed for cyclin E and Cdk2

(Lanes 3 and 4). That fraction of cyclin E in i 6 h extracts that

could not be precipitated with Cipi and Kipi antibodies (Fig.5, Lane 2) could be precipitated with Cdk2 antibodies, asshown in Lane 4. The data in Fig. 5 indicate that within the

limits of detection, all of the cyclmn E/Cdk2 complexes in early

G1 (4 h) are saturated with either Cipi or Kipi . However, the

CKI binding capacity is not sufficient to saturate those cyclinE/Cdk2 complexes that form in late G1 (i 6 h).

DiscussionWe have undertaken to study the mechanisms that govern

activation of cyclin E kinase during the transition from G0 into

S phase, and to this end we have focused our attention uponthose principals that we felt were most likely to play a directrole in determining the activity of the cyclin E/Cdk2 complex.These principals include Cdk2, cyclin E, and the Cipi familyof CKls. Our working hypothesis holds that the activity ofcyclin E/Cdk2 kinase should reflect the relative abundance of

these entities, their subcellular localization, and the abun-dance and/or activity of other factors that might influence the

ability of cyclin E to associate with Cdk2 and form active

complexes.

Although there is abundant cyclmn E and Cdk2 in G0 cells,

these two proteins do not reside in the same subcellularcompartment. Consequently, quiescent cells contain verylow concentrations of cyclin E/Cdk2 complexes. We proposethat sequestration of the catalytic and regulatory subunits is

the primary mechanism whereby activation of cyclin E-de-pendent kinase is precluded in G0. In addition, an excess ofCipi and Kipi binding capacity is maintained in G0 and

during early G1 . Based upon the results of immunodepletionexperiments, we estimate that about one-third of the cyclin Ein G0 cells is bound to CKIs, of which the majority is bound

to Kipi (data not shown), and we have shown that virtually all

of the cyclin E in early G1 (4 h after serum stimulation) is

bound to Cipi plus Kipi . This observation indicates that the

binding capacity of Cipi plus Kipi is sufficient to saturate not

only the low levels of cyclin E/Cdk2 that form in G0 cells, but

also all of the newly formed cyclin E/Cdk2 complexes thataccumulate during early G1 . The observation that the amountof Kipi bound to cyclin E increases in early G1 , although thetotal amount of Kipi is decreasing rapidly, is likewise con-

sistent with the conclusion that there is an excess of CKIbinding capacity that persists for about 8 h after serum

stimulation. Consequently, the small amount of cyclin

E/Cdk2 complex that forms during G0, as well as the moresubstantial amount of cyclin E/Cdk2 that forms in early G, , is

saturated with Kipi (and to a lesser extent Cipi) and is

probably unavailable for activation by CAK.Three things happen during the first few hours of G1 pro-

gression: the abundance of cyclin E increases, Cdk2 entersthe nucleus, and the abundance of Kipi decreases. Our dataindicate that nuclear recruitment of Cdk2 precedes induction

Cell Growth & Differentiation 1289

of cyclmn E by several hours, such that accumulation of cyclmnE/Cdk2 complexes is observed prior to any significant in-duction of cyclin E. Cyclin E/Cdk2 complexes accumulate forabout 8 h before the CKI threshold is exceeded. This intervalappears to be determined by the length of time required to

degrade Kipi , as well as the length of time required toaccumulate enough cyclin E/Cdk2 to saturate the Cipi bind-

ing capacity. The abundance of cyclin E/Cdk2/CKI corn-plexes reaches a maximum about 8 h after addition of serum.The abundance of cyclin E/Cdk2 complexes continues to

increase for about i 6 h, so that the most rapid accumulationof active cyclmn E/Cdk2 kinase occurs 8-i2 h after serumstimulation. These data are consistent with the hypothesis

that Cipi and Kipi define a threshold that precludes activa-

tion of cyclin E/Cdk2 kinases during early to mid G1. The

observation that most of the cyclmn E/Cdk2 complexes in G0

cells are bound to Kipi suggest that this CKI plays a moresignificant role in G0 and very early G1 , whereas Cipi may be

more important in mid G1, when Kipi levels are declining.

We conclude that the major determinant of Cdk2 kinase

activity in quiescent cells is subcellular compartmentalizationof the regulatory and catalytic subunits. That is not to say

that the CKIs are unimportant in G0. Cipi and Kipi areabundant in G0 cells and undoubtedly contribute to inhibition

of cyclmn E/Cdk2 complexes by preventing activation of any

complex that may form as a result of leakage of Cdk2 into the

nucleus. Upon mitotic stimulation, the abundance of cyclin

E/Cdk2 complexes increases, in part as a result of induction

of cyclmn E and in part due to entry of Cdk2 into the nucleus.

Both Cipi and Kipi are potent inhibitors of cyclin E-associ-ated kinases, and the inhibitory threshold set by these fac-

tors is probably the major determinant of Cdk2 activity inearly to mid G1.

Materials and MethodsTissue Culture and Cell Lines. Human diploid embryonic lung fibro-

blasts 18LU in passage 12-20, W138 in passage 25-30, IMR9O in passage14-i8, and murine Balb3T3 were grown in EMEM (Eagle’s medium withEarle’s salts) with iO% FBS, penicillin (100 units/mI), and streptomycin(iOO �.tg/ml) in a 37”C incubator with a 5% CO2 atmosphere until theyreached 70-80% confluence. The medium was then removed and the

cells were washed with warm, serum-free EMEM. After washing, freshserum-free EMEM was placed on the cells, and the incubation was con-tinued for another 48 h. The medium was removed after serum depnva-

tion, and the cells were stimulated by addition of 20% FBS in EMEM. Thecells were harvested at intervals and processed for immunoprecipitation,

Western blotting, immunofluorescence, or histone Hi kinase assays, asdescribed below.

Western Blotting and Immunoprecipltation. Cells were harvestedby mild trypsin digestion, collected by centrifugal sedimentation, andlysed in NP4O lysis buffer (50 mM Tris, pH 7.4, i50 m� NaCI, 0.5% NP4O,

with i m� NaVO3, 50 mM NaF, 1 mM phenylmethylsulfonyl fluoride, i m�DTT, 25 j�g/ml leupeptin, 25 �g/ml trypsin inhibitor, 25 �g/ml aprotinin, imM benzamide, and 25 j.�g/ml pepstatin A added just before use). Cellular

debris was removed by sedimentation, and the supematant fluids werereserved. Nuclear extracts were prepared as described previously (27).The protein concentration was determined by the method of Bradford (28).Equal aliquots of protein (40 �sg) were dissolved in sample buffer and

resolved by electrophoresis on polyacrylamide gels containing SDS. Theproteins were transferred to nitrocellulose membranes (Bio-Rad) andprobed with the specified antibodies (Santa Cruz Biotechnology). Proteins

were detected by the enhanced chemiluminescence system (Amersham).Exposure was calibrated to a range that ensured linearity of responsewithin a 5-10-fold variation in concentration, and specific bands were

quantified by densitometry (Applied Imaging Lynx 5000 software, Version5.5).

Immunoprecipitation was carried out using the same method as de-

scribed for Western blotting. Equal amounts of protein (1 50 �g) wereincubated with the indicated antibody for 2 h at 4”C. Antibody complexes

were recovered on Protein A-Sepharose beads (Sigma), washed 4 timeswith NP4O lysis buffer, and dissolved in 2x sample buffer followed by

separation and detection as described above.Indirect lmmunofiuorescence. Cells were grown on glass coverslips

as described above and subsequently fixed in acetone and methanol (1 :i)

at -20”C for 10 mm. Cells were then incubated with primary antibodydiluted in PBS for i h at 37#{176}Cin a humidified chamber. After two washesin PBS for is mm each, the cells were incubated with a secondary

antibody (affinity-purified goat antirabbit FITC-conjugated lgG) for 30 mm.Coverslips were washed in PBS for 30 mm, mounted in PBS and glycerol(1 :1), and examined with the aid of a Zeiss Photomicroscope using a 40xNeofluar lens. Images were recorded on Kodak Ektachrome Elite 400

color slide film.Histone HI Kinase Activity. Kinase activity was measured as de-

scribed previously (i i). Briefly, protein complexes were immunoprecipi-tated as described above. After washing the Protein A-Sepharose com-

plex with NP4O Iysis buffer, the complexes were washed three times with

2x kinase buffer (40 mM Tns-HCI, pH 7.5, 8 m�i MgCl�). Kinase reactionswere carried out in a total volume of 5 �l, which included 3 �I of 2x kinase

buffer containing 2.5 �g histone Hi (Life Technologies, Inc.) and 2 �l[�?2PJATP (iOCi/mmol; Dupont NEN) at 37”C for 30 mm. The reaction was

stopped by adding S �I of 2x sample buffer and boiling for 5 mm. Each

sample was then separated by SDS-PAGE; the gels were dried andexposed to film (Kodak XAR-5), and specific bands were quantified by

densitometry (Applied Imaging).Antibodies. All antibodies were obtained from Santa Cruz Biotech-

nology. Antibodies used for Western blotting include cyclin E (HE12),

Cdk2 (M2), Cipl (C-i9), and Kipi (C-19). Immunoprecipitations were doneusing antibodies HEi i i for cyclin E, M2 for Cdk2, C-i9for Cipi , and C-19for Kipl . Antibodies used for immumofluorescence staining include C-19for cyclin E, M2 for Cdk2, C-i9 for Cipi , and N-20 for Kipi.

References1 . Draetta, G. Mammalian G1 cyclins. Cuff. Opin. Cell Biol., 6: 842-846,i994.

2. Sherr, C. Mammalian G1 cyclins. Cell, 73: i059-i065, 1993.

3. Sherr, C. G1 Phase Progression: Cycling on Cue. Cell, 79: 551-555,i994.

4. Meyerson, M., and Harlow, E. Identification of G1 kinase activity for

Cdk6, a novel cyclin D partner. Mol. Cell. Biol., 14: 2077-2086, 1994.

5. Elledge, S., and Harper, J. W. Cdk inhibitors: on the threshold ofcheckpoints and development. Curr. Opin. Cell Biol., 6: 847-852, 1994.

6. Hunter, T., and Pines, J. Cyclins and cancer II: cyclin D and Cdkinhibitors come of age. Cell, 79: 573-582, 1994.

7. Sherr, C. J., and Roberts, J. M. Inhibitors of mammalian G1 cyclmn-

dependent kinases. Genes & Dev., 9: 1149-i 163, 1995.

8. Bates, S., Bonetta, L, Macallan, D., Parry, D., Holder, A., Dickson, C.,

and Peters, G. Cdk6 (PLST1RE) and Cdk4 (PSK-J3) are distinct subset ofthe cyclin-dependent kinases that associate with cyclin Di . Oncogene, 9:7i-79, 1994.

9. Matsushime, H., Ewen, M. E., Strom, D. K., Kato, J., Hanks, S. K.,Roussel, M. P., and Sherr, C. I. Identification and properties of an atypicalcatalytic subunit (p34 PSK-J3/Cdk4) for mammalian D-type G1 cyclins.

Cell, 71: 323-334, i992.

1 0. Kidd, V. J. Cell division control-related protein kinases: putative on-gins and functions. Mol. Carcinog., 5: 95-101 , i992.

i 1 . Dulic, V., Lees, E., and Reed, S. I. Association of human cyclin E with

a periodic G1-S phase protein kinase. Science (Washington DC), 257:

i958-i96i, i992.

12. Rosenblatt, J., Gu, T., and Morgan, D. Human cyclmn-dependent

kinase 2 is activated during the S and G2 phases of the cell cycle andassociates with cyclmn A. Proc. NatI. Acad. Sci. USA, 89: 2824-2828,i992.

1290 Control of Cyclmn E/Cdk2 Activity in G0 and G1

13. Clarke, P. H. CAK-handed kinase activation. Curr. Biol., 5: 40-42,

1995.

14. Morgan, D. 0. Principles of Cdk regulation. Nature (Lond.), 374:

131-134, 1995.

15. Gu, V., Turck, C., and Morgan, D. Inhibition of Cdk2 activity in vivo by

an associated 20K regulatory subunit. Nature(Lond.), 366: 707-710, 1993.

16. Kato, J-Y., Matsuoka, M., Strom, D. K., and Sherr, C. J. Regulation of

cyclin D-dependent kinase 4 (odk4) by cdk activating kinase. Mol. Cell.Biol., 14: 2713-2721, 1994.

17. Slingerland, J. M., Hengst, L, Pan, C-H., Alexander, D., Stampfer, M.

R., and Reed, S. I. A novel inhibitor of cyclin-Cdk activity detected intransforming growth factor �3-arrested epithelial cells. Mol. Cell. Biol., 14:3683-3694, 1994.

18. Harper, J. W., Adami, G. A., Wei, N., Keyomarsi, K., and Elledge, S. J.

The p21 Cdk-Interacting protein Cipl is a potent inhibitor of G1 cyclin-dependent kinases. Cell, 75: 805-816, 1993.

19. El-Deity, W. S., Tokino, T., Velculescu, V. E., Levy, D. B., Parsons, A.,

Trent, J. M., Un, D., Mercer, W. E., Kinzler, K. W., and Vogelstein, B.WAF1, a potential mediator of p53 tumor suppression. Cell, 75: 81 7-825,1993.

20. Polyak, K., Lee, M., Erdjument-Bromage, H., Koff, A., Roberts, J. M.,

Tempst, P., and Massague, J. Cloning of p27KlPl, cyclin-dependentkinase inhibitor and a potential mediator of extracellular antimitogenicsignals. Cell, 78: 59-66, 1994.

21 . Toyoshima, H., and Hunter, T. p27, a novel inhibitor of G1-cyclmn-Cdkprotein kinase activity, is related to p21 . Cell, 78: 67-74, 1994.

22. Harper, J. W., Elledge, S., Keyomarsi, K., Dynlacht, B., Tsai, L,Zhang, P., Dobrowolski, S., Bai, C., Connell-Crowely, L, Swindell, E., Fox,M. P., and Wei, N. Inhibition of cyclin-dependent kinases by p21 . Mol.Biol. Cell, 6: 387-400, 1995.

23. U, Y., Jenkins, C. W., Nichols, M. A., and Xiong, Y. Cell cycle expres-sion and p53 regulation of the cyclin-dependent kinase inhibitor p21.

Oncogene, 9: 2261-2264, 1995.

24. Zhang, H., Hannon, G. J., and Beach, D. p21-containing cyclin ki-nases exist in both active and inactive states. Genes & Dev., 8: 17�0-1758, 1994.

25. Resnitzky, D., Hengst, L, and Reed, S. I. Cyclin A-associated kinaseactivity is rate limiting for entrance into S phase and is negatively regulatedin G1 by p27Kipl . Mol. Cell. Biol., 15: 4347-4352, 1995.

26. Ohtsubo, M., Theodoras, A. M., Schumacher, J., Roberts, J. M., andPagano, M. Human cyclin E, a nuclear protein essential for the G1 to Sphase transition. Mol. Cell. Biol., 15: 2612-2624, 1995.

27. Mahajan, P. B., and Thompson, E. A. Glucocorticoid regulation oftranscription of the c-myc cellular protooncogene in P1 798 cells. Mol.

Endocnnol., 1: 899-907, 1987.

28. Bradford, M. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye bind-ing. Anal. Biochem., 72: 248-254, 1976.