Eur. J. Biochem. 207, 169-176 (1992) 0FEBS 1992

Distinct forms of human CDC2 identified by novel monoclonal antibodies Jiii LUKAS ’,*,Giulio DRAETTA’ and Jii-i BARTEK ’ ’ Department of Tumor Biology, Institute of Haematology and Blood Transfusion, Prague, Czechoslovakia *

Differentiation Programme, European Molecular Biology Laboratory, Heidelberg, Federal Republic of Germany

(Received February 3/March 31, 1992) - EJB 92 0140

Studies on the functional and structural properties of the cdc2 kinase, a key cell-cycle regulator, have been possible thanks to the availability of cdc2-specific immunoreagents. In an attempt to elucidate the biochemical regulation of the cdc2 kinase in more detail, we have raised a series of novel mouse monoclonal antibodies against human recombinant cdc2 protein. The five Mab reported here can be subclassified into two groups according to their interspecies cross-reactivity and distinct immunoprecipitation patterns. Thus, the target epitopes of Mab POH-1, POH-2 and POH-7 (group 1) appear to be limited to a few mammalian species and the fraction of cdc2 immunoprecipitable by these Mab from cellular extracts is considerably enhanced by denaturation. In contrast, the POH-3 and POH-8 (group 2) Mab recognize a denaturation-sensitive epitope on cdc2 which is present in all tested mammalian species. More importantly, each of the two groups of Mab immunoprecipitate forms of cdc2 associated with a characteristic set of cellular proteins, none of which appears to be cyclin A or cyclin B. None of the antibodies precipitated a histone-HI or casein-kinase activity, although an activity which phosphorylated some of the coprecipitated proteins was coprecipitated with the group 2 Mab. These novel Mab did not interfere with the association of cdc2 with cyclin A in vitro and efficient immunoprecipitation of a panel of cdc2 mutant proteins suggests that the target epitopes may not involve amino acid residues essential for currently known cdc2 functions. The results of the present study provide evidence for the existence of additional forms of the cdc2 protein in exponentially growing human cells, distinct from both the monomeric and the cyclin-bound cdc2 identified so far.

Cdc2 is the first identified member of a novel class of protein kinases, the cyclin-dependent kinases (cdk ; for a review see [I]). Cdc2 was discovered as the product of a fission yeast gene, which is necessary for cell progression through the division cycle. Several temperature-sensitive mutant alleles of the cdc2 gene were originally found and they all arrest in either the G I or the G2 phase of the cell cycle, depending on the composition of the growth medium at the time of the temperature increase (for a review see [2]). A homologous gene of Sacharomyces cerevisiae, cdc28, plays a similar role in the cell cycle (for a review see [3]). Further to the identification of these important yeast genes, a human homolog of cdc2/28 was discovered [4, 51. Thereafter, cdc2 homologues have been found in every eukaryotic species. Recently, cDNA species encoding proteins highly related to cdc2, but acting at stages in the cell cycle other than mitosis, have been found [6 - 81. It seems now clear that, while in yeast a single cdc2 is responsible for the GI-phase to S phase transition and the G2-phase to M-phase transition, in higher eukaryotes multiple cdc2-like proteins exist (for a review see [I]). Cdc2 is a 34-kDa protein, probably one of the smallest protein kinase catalytic subunits. In fact, when compared to the CAMP-dependent protein Correspondence to J. Bartek, Department of Tumor Biology, Institute of Haematology and Blood Transfusion, Korunni 108, CS-101 03 Prague, Czechoslovakia Abbreviations. Cdk, cyclin-dependent kinase; DMEM, Dulbecco’s modified Eagle’s medium.

kinase, cdc2 is comprised within the catalytic core of cyclicAMP-dependent kinase [9]. Cdc2 is inactive as a histone-H1 kinase unless complexed with a cyclin [lo]. Cyclins were first identified in marine invertebrates as proteins which periodically accumulate in the cell cycle of early embryos and are able to induce the G2 to M transition when injected into Xenopus oocytes (see [Ill, for a review). Two structurally similar proteins, cyclins A and B, were initially identified; several new members of the cyclin family have since been discovered, all characterized by the presence of a central amino acid domain called the cyclin box (for a review, see [12]). The function of many of these cyclins is still unknown and only the B cyclins seem to have a definite role in the G2 to M transition. A complex of cyclin B and cdc2 is active as a histone-HI kinase [lo, 13, 141. Cyclin binding to cdc2 can occur in vitro, provided that an essential phosphorylation event takes place. In order for cdc2 to efficiently bind cyclin in vitro, phosphorylation of cdc2 at Thrl61 is necessary; this appears to stabilize the interaction between cdc2 and the cyclin [15]. It is not yet clear whether this phosphorylation occurs as the result of an autophosphorylation event or not, although a cellular factor, yet to be purified, needs to be present in order to obtain efficient binding between cdc2 and cyclin. Zn vivo, the cdc2-cyclin-B complex is inactivated as its formation necessitates phosphorylation of a tyrosine located within the ATP-binding site of cdc2 [16]. A tyrusine-phos-

170 phorylated inactive cdc2 -cyclin-B accumulates in the cell until DNA replication has been completed. At that time, a specific phosphatase, cdc25, catalyses the dephosphorylation of cdc2, thereby triggering the sudden activation of this kinase complex I17 -201. Cdc2 and cyclin have been studied mainly using genetic approaches and not using a classical biochemical approach, therefore relatively little is known about the regulation of this enzyme at the molecular level. We chose to develop monoclonal antibodies against the cdc2 monomer expressed in Ekchrrichia coli with the aim of using them to identify functionally important epitopes which would allow studies on the interaction of cdc2 with its regulatory subunits and substrates. In this paper, we describe the generation and characterization of anti-cdc2 Mab and their use in various biochemical assays. MATERIALS AND METHODS Preparation of monoclonal antibodies

Five Balb/c female mice were immunized by repeated subcutaneous injections of gel-purified recombinant human cdc2 protein. The two mice showing the highest titers of specific serum antibodies as tested by immunoblotting, were given additional intravenous injections on days 5, 4 and 3 before fusion. Spleen cells were fused to NS-2 mouse myeloma cells using poly(ethy1ene glycol) and the hybrids selected in azaserineihypoxanthine medium. Supernatants were tested in a nitrocellulose dot blot on purified cdc2 and the positive samples re-screened immediately using immunoblotting on SDS/PAGE-separated bacterial lysates containing human recombinant cdc2. The selected hybridomas were cloned twice by limiting dilution. Other reagents

The rabbit polyclonal antiserum G6 against a C-terminal peptide of human cdc2 has been described elsewhere [21]. Anti-(human cyclin B) serum was obtained by immunizing rabbits with purified human recombinant cyclin B (V. Baldin and G. Draetta, unpublished results). Rabbit antiserum to human cyclin A and cyclin-A - Sepharose-beads have been described in [22]. Rabbit antiserum to the ‘PSTAIR’ motif, present in cdc2, was a gift of Eric Karsenti (EMBL). Human cksl-2 Sepharose-beads were prepared as follows. Purified p9 (the protein product of CKSI-2, a human homolog of Schtoscicc,huromyces pomhe sucl gene [23]) was covalently coupled to CNBr-activated Sepharose 4B (Pharmacia). Negative-control antibodies included both rabbit and mouse preimmune sera. ~

Cell culture and metabolic labelling

Mouse myeloma and hybridoma cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 20% fetal bovine serum and antibiotics. Human HeLa (cervical carcinoma), PMC-42 (breast carcinoma), HOS (osteosarcoma); Namalwa (Burkitt’s lymphoma) and 293 (adenovirus transformed) cell lines, CV-1 (monkey kidney epithelial line), BMGE (bovine mammary gland epithelial line), CCL46 (mink lung epithelial line), clone 6 (transformed rat embryo fibroblasts) and 3T3 (mouse fibroblast) lines were grown in DMEM with 10% fetal bovine serum and antibiotics. S. pomhe, S. cerrvisiuc, and Xenapus luevis cell extracts were kindly provided by M. Pagano (EMBL, Heidelberg)

and Drosophilu melunogaster extracts by R. Paro (ZMBH, Heidelberg). For metabolic labelling, 50 - 70% confluent cell cultures or fresh pellets of cells exponentially growing in suspension were rinsed with methionine-free or phosphate-free DMEM, incubated in the same media for 30 min at 37”C, washed, then incubated for 3 h in the presence of 0.5 - 1 mCi/ml [35S]methionine (Tran3%Label, ICN) or [32P]orthophosphate(Amersham). The radiolabelled cells were washed once in 10 mM sodium phosphate, pH 7.5 and 130 mM NaCl (NaC1/Pi),pelleted and processed for immunoprecipitation. Immunochemical techniques

Exponentially growing cells were washed twice in ice-cold NaCI/Pi and harvested. Pelleted cells were washed once in cold NaCI/Pi and lysed for 30 min on ice in a lysis buffer containing 50 mM Tris/HCl, pH 7.4,0.25 M NaC1,0.1% Triton X-100, 5 mM EDTA, 50 mM NaF, 1 mM dithiothreitol and 0.1 mM Na3V04. The following protease inhibitors were added : 0.1 mM phenylmethylsulfonyl fluoride, leupeptin (1 pg/ml), aprotinin (1 pg/ml), soybean trypsin inhibitor (10 pg/ml), tosylphenylalanine chloromethane (I0 pg/ml) and tosyllysine chloromethane (10 pg/ml). Denaturing lysis was performed by adding SDS to the cell lysate (1 % final concentration) and incubating for 30 min on ice. The samples containing SDS were then diluted 10-fold with lysis buffer and Triton X-100 was added to 1% final concentration. Lysates were finally cleared by centrifugation in an Eppendorf microcentrifuge at 15000 rpm for 10 min at 4°C. The protein concentration of the cell lysates was determined with the BioRad Protein Assay kit, using bovine y globulin as standard. For immunoprecipitation of metabolically radiolabelled cells, 1 mg total proteinlreaction was used. In immunoprecipitation experiments of unlabelled cell lysates followed by immunoblotting, 5 mg total protein was used for each reaction. Cell lysates were subjected to protein-A - Sepharose preabsorption then immunoprecipitated with 2 3 p1 ascites (containing mouse Mab) or rabbit antisera for 1 h on ice. Immunoprecipitates were collected on protein-A - Sepharose [in the case of mouse IgGl antibodies, 1 p1 rabbit anti-(mouse IgG) serum diluted 1 : 1000 was added]. After several washes in lysis buffer, the packed beads were resuspended in Laemmli sample buffer [24] and loaded onto SDS/PAGE. Polyacrylamide gels were either dried and exposed for autoradiography or transferred to nitrocellulose and subjected to immunoblotting as described by [5] for 1251-labelledconjugates and by [25] for the peroxidase-detection system. ~

In vitvo kinase assays

Immunoprecipitations from 0.5 mg of the indicated cell lysates were washed once in kinase assay buffer (solution A; 50 mM Tris/HCl, pH 7.5, 10 mM MgCl2 and 1 mM dithiothreitol). Kinase reactions were carried out in solution A supplemented with 50 mM ATP and 5 mCi [ Y - ~ ~ P ] Afor T P5 min. Calf histone HI (Boehringer) at 0.2 mg/ml or dephosphorylated casein (Sigma) at 1 mg/ml was used as substrate. In the cdc2 autophosphorylation experiments, the reactions proceeded without any exogenous substrate. For depletion of the kinase activity, 20 ml packed ProteinA-Sepharose was incubated with the POH ascites or G6 antiserum (both diluted 1 : 100 in lysis buffer) at 4°C for 1 h with constant rotation. The beads were rinsed several times with lysis buffer, resuspended in cell extracts containing 100 pg

171 total protein and incubated for 30 min at 4°C on a rotating wheel. After spinning down the beads, the supernatants were removed and processed for the second round of depletion with the same antibody-coated beads as above. The resulting supernatants were incubated for an additional 30 rnin at 4°C with 20 p1 protein-A- Sepharose-beads to collect any antibody-cdc2 complexes left in the lysates. Such extensively depleted cell extracts were then used for immunoprecipitation with G6 antiserum, followed by subjection to the kinase assay as described above. The potential inhibition of the kinase activity was assayed as follows. Cdc2 from 100 pg total HeLa protein extract was bound to 20 p1 cksl-2-Sepharose beads by incubation for 1 h at 4°C on a rotator. The beads were washed three times in lysis buffer and incubated with a POH ascites or rabbit sera (diluted 1:100 in lysis buffer). After incubation for 30 min at 4°C on a rotator, the beads were centrifuged, washed three times in lysis buffer, once in solution A and assayed for histone H I phosphorylation. Zn-vitro translation and cdc2 - cyclin-A-binding assay Wild-type and different mutant human CDC2 mRNA species, prepared according to [I 51, were kindly provided by Dr Bernard Ducommun (see also Table 2). In-vitro translations were performed using rabbit reticulocyte lysates (Promega) according to the manufacturer’s instructions. %] label was purchased from ICN. Reactions proceeded at 30°C for 90 min and the translational product was used directly for immunoprecipitation or binding assays. Aliquots of the proteins translated in vitro were also used as M , markers for SDSjPAGE and autoradiography. To examine the effect of the anti-cdc2 serum on the binding of cyclin A to cdc2 in vitro, the [35S]methionine-labelledhuman cdc2, translated in vitro, ( 5 p1 of the translational mixture) was mixed with the indicated POH Mab ascites or rabbit antiserum and incubated for 30 rnin on ice. Reactions were pcrformed in antibody excess (0.5 - 5 pl ascites/reaction, 1 10 pg Mab) to ensure that the antibody concentration was not a limiting factor for cdc2 binding. 20 p1 packed human cyclinA - Sepharose-beads [22] or bovine-serum-albumin - Sepharose, used as a control, was washed twice in solution A and added to the cdc2lantibody mixture supplemented with fresh HeLa cell extract (20 p1 volume containing 200 pg total protein). Solution A or solution A with 25 mM EDTA w d S added to a final volume of 50 p1 of the total reaction volume and the mixture was incubated at 30°C for 30 rnin with occasional gentle shaking. The binding reaction was stopped by diluting with 1 ml lysis buffer, the beads collected by centrifugation, washed three times in lysis buffer, solubilized in Laemmli sample buffer [24] and loaded on 12.5% SDS/PAGE. The gel was treated with Enlightning (NEN) prior to exposure. RESULTS

Production and characterization of monoclonal antibodies to human cdc2 Balb/c mice were immunized with bacterially produced recombinant human cdc2 and standard hybridoma technology was employed to generate monoclonal antibodies. Out of 560 growing hybrids from two fusion experiments, 11 were selected for cloning and further characterization, based upon their reactivity with purified cdc2 by nitrocellulose dot blotting, and the recognition of cdc2 by immunoblotting of bacterial lysates expressing the recombinant protein. After two

*

a

1

2

3

4

5

+ -

26 -

43

b

7

8 9 10 11 12 13

+

-

26 43

Fig. 1. lmmunoblotting of recombinant human cdc2 and various mammalian cell lysates with the POH monoclonal antibodies. (a) Immunoblots of SDS/PAGE-(10% gel)-resolved human-cdc2-expressing E. colilysates were probed with Mab supernatants from POH-1 (lane 11, POH-2 (lane 2), POH-3 (lane 3), POH-7 (lane 4) and POH-8 (lane 5 ) , G6 antiserum (lane +) and a negative-control antibody (lane -). (*) India-ink staining for total protein. (b) Immunoblotting of various mammalian cell lysates probed as follows. Lane 7, POH-I (HeLa cells); lane 8, POH-2 (PMC-42 cells); lane 9, POH-2 (HOS cells); lane 10, POH-2 (Namalwa cells); lane 11, POH-2 (293 cells); lanc 12, POH2 (C6 cells); lane 13, POH-8 (C6 cells). The positive (lanc +) and negative-control (lane -) sera are shown on PMC-42 blots. The relative molecular mass (kDa) is indicated.

Table 1. Inter-species cross-reactivity of the POH antibodies. n. d., not determined; +, reactive; -, not reactive. ~

~~

Species

Monoclonal antibody (Ig subclass) POH-1 POH-2 POH-3 (IgG2a) (IgG2a) (IgG1)

Monkey cow Mink Rat Mouse Xenupus Drosophila S. pombe S . cerevisiae

+ + + -

POH-7 POH-8 (IgG2b) (IgG1)

+ +

n.d. -

+ + + + -

rounds of limited-dilution cloning, five independent hybridoma clones were established (designated POH-1, POH-2, POH-3, POH-7 and POH-8) and the corresponding monoclonal antibodies tested by immunoblotting on both bacterial and a variety of eukaryotic cell lysates. As can be seen from Fig. 1a, all Mab specifically recognized a M , 34000 band on immunoblots from cdc2-expressing bacteria. The same band was also positive with the control rabbit antiserum G6 (Fig. l a , lane +), while no obvious reaction was observed with either a negative-control Mab on a parallel nitrocellulose strip (Fig. l a , lane -) or any of the new Mab species on extracts from the same bacterial strain, but lacking the human cdc2-expressing plasmid (data not shown). A specific M ,

172

a

1

96

2

3

4

5

6 7 8

9 1 0 1 1 1 2 13 14 15

16

-

665645

-

364 p34

30-

2014-

b

1

2

3

4

5 6 * 7

8

9 1 0 1 1 1 2 1 3 1 4 1 5 16

-

96 6656453630-

2014-

Fig. 2. Immunoprecipitates from exponentially growing HeLa cells metabolically labelled with [35S]methionine (a) or [3zP]orthophosphate (b). Lanes 1 - 6 in both (a) and (b) represent immunoprecipitations from non-denaturing extracts, lanes 10- 16 from SDS-denatured extracts. The antibodies used were POH-1 (lanes 1 and 10); POH-2 (lanes 2 and 11); POH-3 (lanes 3 and 12); POH-7 (lanes 4 and 13); POH-8 (lanes 5 and 14); C6 (lanes 6 and 15); rabbit pre-immune serum (lanes * and 16). Lanes 7-9 show the internal standards translated in vitro; human cdc2 (lane 7), human cyclin A (lane 8) and human cyclin B (lane 9). The proteins were resolved on 12.5% SDS/PAGE; the molecular mass of the markers (kDa) run in parallel is indicated.

34000 band was recognized by the POH Mab on immunoblots of crude lysates from human cell lines derived from tumors of various histogenesis, including carcinomas, sarcomas and lymphomas (see Fig. 1b, lanes 7 - 11 for representative examples). These immunoblotting data, together with the lack of reactivity with the human cdc2-related cdk2 kinase [7, 81 (our unpublished results) suggest that cdc2 is the only protein recognized by the POH antibodies. Since cdc2 function is highly conserved during evolution, it was of interest to examine whether the new Mab species cross-reacted with cdc2 homologs from other eukaryotes. To assess the phylogenetic conservation of the target epitopes, the POH antibodies were tested on immunoblots of cultured cells from a range of organisms including rodents, Xenopus, Drosophila and yeast. An example of the immunoblotting results is shown in Fig. 1b and the data of this comparative study are summarized in Table 1. According to their interspecies cross-reactivity, the POH antibodies could be subdivided into two groups, the Mab POH1, POH-2 and POH-7 which reacted with human, monkey, bovine and mink cdc2 homologs and the antibodies POH-3 and POH-8, reactive also

with mice and rat (but not mink) cdc2. None of the new Mab showed any obvious reactivity with the Xenopus, Drosophila or yeast proteins, suggesting that the target epitopes are unlikely to be localized in the most conserved regions of the cdc2 molecule. Target forms of cdc2 recognized by the POH antibodies

Cdc2 protein forms specific complexes with cyclins A and B and consequently it was interesting to investigate whether the new Mab were able to coprecipitate cyclins through their interaction with cdc2 and whether denaturation of cellular extracts had any effect upon the recognition of cdc2 by the individual POH antibodies. To address these questions, exponentially growing HeLa cells were metabolically labelled with [35S]methionine,cell extracts prepared and immunoprecipitations performed with the POH antibodies under both native and denaturing conditions (see Materials and Methods). All POH antibodies immunoprecipitated cdc2 from the nondenaturing extracts (Fig. 2a, lanes 1- 6), although it appeared to be only a fraction of the cdc2 precipitated by the control

173 anti-cdc2 serum G6 [21] under identical conditions. Each of the new Mab also coprecipitated additional proteins. Interestingly, the spectrum of the coprecipitated cellular proteins was very similar for POH-1, POH-2 and POH-7, while POH-3 and POH-8 coprecipitated different proteins. Furthermore, the POH Mab did not coprecipitate cyclins A or B, as judged by the fact that none of the additional proteins present in the immunocoinplexes comigrated with cyclins translated in v i m (Fig. 2a, lanes 8 and 9), or the cyclins present in the G6 immunocomplexes (lane 6). As indicated by the relative intensity of the cdc2 bands in lanes 6 and 15 (Fig. 2a), denaturation of the cell extract had no obvious effect on the amount of the cdc2 precipitated by the G6 antiserum. In contrast, it resulted in a substantial increase of the cdc2 fraction precipitable by POH-1 (Fig. 2a, lanes 1 and lo), POH-2 (lanes 2 and 11) and POH-7 (lanes 4 and 33). The opposite effect, namely a total loss of the precipitable cdc2 fraction upon denaturation, was observed with the antibodies POH-3 (Fig. 2a, lanes 3 and 12) and POH8 (lanes 5 and 14). These results suggest that the target epitopes of the POH-1, POH-2 and POH-7 are likely to be linear, but not accessible in a large fraction of the native cdc2 protein. However, the POH-3 and POH-8 appeared to recognize conformation-dependent epitopes which were accessible in a fraction of the cdc2 molecules present in cell extracts and were destroyed by denaturation. The results of a similar immunoprecipitation experiment, performed using extracts of HeLa cells radiolabelled with [32P]orthophosphate (Fig. 2 b), confirmed the effects of denaturation upon the individual epitopes and demonstrated that at least some of the protein bands coprecipitated with cdc2 under non-denaturing conditions are phosphoproteins (lanes 1- 6). Perhaps the most surprising aspect of the immunoprecipitation experiments described above was the observation that the new anti-cdc2 Mab apparently did not coprecipitate cellular proteins corresponding to cyclin A and B (Fig. 2). To verify the lack of cyclin A and B in the protein complexes identified by the POH antibodies, the HeLa immunoprecipitates were separated by SDS/PAGE, the proteins transferred to nitrocellulose membrane and the blots probed with specific antisera to human cyclin A and cyclin B (Fig. 3). The results of this experiment confirmed that cyclins A or B were not coprecipitated by POH-2 or POH-8. The new Mab seemed to recognize distinct forms of the cdc2, complexes with some cellular proteins different from cyclin A and B. The proteins precipitated by POH-1, POH-2 and POH-7 were similar and included two prominent bands migrating at an approximate M , 95000 and 56000-57000 (between the cyclin A and B bands), while the complexes precipitated by POH-3 and POH8 contained three prominent bands of M , 80000, 65000 and 50000 (see Fig. 2a, lanes 1-6). Effect of POH antibodies on the in-vitvo association between cdc2 and cyclin A The results described above demonstrate that the POH Mab immunoprecipitated distinct forms of the cdc2 protein not associated with cyclins A or B, despite the fact that significant amounts of the cdc2-cyclin complexes were present in the extracts. One possible explanation of the data obtained is that the POH antibodies compete with cyclins for binding to cdc2 and could even dissociate (at least in part) the already existing complexes. In order to test this hypothesis, we took advantage of an efficient in-vitro binding assay, based upon the association of

1 2 3 4 5 6 7 6 9 1 0

56 66

36

-

30

-

+B +A

-

P34

Fig. 3. Immunoblotting of the POH antibody immunoprecipitates. Extracts from exponentially growing HeLa cells ( 5 mg protein/reaction) were immunoprecipitated with POH-2 (lanes 1 and 7), POH-8 (lanes 2 and 8), pre-immune mouse serum (lane 3) or G6 antiserum (lanes 4 and 9). 1 mg cell protein was immunoprecipitated with the rabbit anti(human cyclin A) serum (lane 5) and anti-(human cyclin B) serum (lane 10). Lane 6 contains 2 p1 in-vitro-translated human cyclin A. The upper part of the blot was divided and probed with either rabbit anti-(human cyclin A) serum (lanes 1 - 6) or rabbit anti-(buman cyclin B) serum (lanes 7-10). The lower part of the membrane shows cdc2 present in blots of the same immunoprecipitates but probed with the anti-(PSTAIR) rabbit antiserum. All reactions were detected with the '251-labelled conjugates and autoradiography. A, B and p34 indicate the positions of the human cyclin A, B and cdc2, respectively. The relative molecular masses (kDa) are given.

*

1

2

3

4

5

6

7

8

91011121314

Fig. 4. Binding of cdc2 to cyclin A in the presence of POH antibodies. The human cdc2, translated in vitro, was incubated with cyclin-A Sepharose (lanes 3 - 14) or bovine-serum-albumin - Sepharose (negative control, lanes 1 and 2). The antibodies preincubated with the translated cdc2 included POH-2 (lanes 13 and 14), POH-8 (lanes 11 and 12), G6 (lanes 9 and 10) and the pre-immune serum (lanes 7 and 8). The reaction buffer was supplemented with 25 mM EDTA in lanes 1,3,5,7,9,11 and 13 (see Materials and Methods for details). In some reactions, the human cyclin-A - Sepharose beads were pretreated with anti-(cyclin A) serum (diluted 1 : 100, lanes 5 and 6) or pre-immune control serum (lanes 3 and 4). The Sepharose-bound material was electrophoresed on 12.5% SDSjPAGE and the cyclin-A-bound cdc2 visualized by autoradiography (lanes 1-14); lane * shows the invitro-translated cdc2 run in parallel as an internal marker.

Sepharose-immobilized human recombinant cyclin A with invitro-translated human cdc2. Using the protocol described in Materials and Methods, we compared the effects of POH-2 and POH-8 (as representative examples of the two groups of our new Mab) with those of G6 and other control antibodies on the ability of the cdc2 protein to form complexes with human cyclin A, both in the absence and the presence of EDTA, which has been shown to inhibit binding [15]. The specificity of this assay is demonstrated by the lack of detectable cdc2 binding to bovine-serum-albumin - Sepharose, used as a control (Fig. 4, lane 2) and by the fact that addition of excess soluble cyclin A prevented the association of cdc2 with the Sepharose-bound cyclin A ([26] and our unpublished results). EDTA significantly reduces binding between cyclin A and cdc2 in virtue of its ability to prevent cdc2 phosphorylation on Thrl61, which has been shown to be necessary for binding [26] (compare lanes 3 and 4, 5 and 6, 7 and 8 in Fig. 4). Preincubation of in-vitro translated cdc2 with the POH Mab did not affect binding to cyclin A. In similar experiments

174 Table 2. Properties of the in-vitvo-translated cdc2 mutants. n.d., not determined. Data for cdc2 mutants adapted from [15, 29, 301. ~~

~~

Binding of

Sequence

Wild type E2A, D3A, K6A ESA, K9A, E12A Y15F K20A. R20A, H23A, R24A K33A, K34A, R25A K33A, E40A, E41A, E42A RSOA, ESIA D86A, K89, D92A T161V

1

2

3

4

5

cyclinA

cyclin B

100 105 18 126 134 13 13 0 54 5

100 n. d. 23 n. d. n. d. 27 9 20 35 10

6

7

8

9

1

0

%

a

b

C

Fig. 5. Wild-type and various mutant cdc2 proteins immunoprecipitated with the POH-2 (a), POH-8 (b) and G6 (c) antibodies. 5 pl each invirvo-translated product was immunoprecipitated with 3 pl ascites (POH-2 and POH-8) or serum ((36) and the bound proteins electrophoresed and visualized by autoradiography. Wild-type (lane 1) and the following mutants of human cdc2 were used (see Table 2 for a more detailed description of the mutants): K20A, R22A, H23A, R24A (lane 2); E2A, D3A, K6A (lane 3); E8A, K9A, E12A (lane 4); Y1SF (lane 5 ) ; K31A, K34A, R36A (lane 6); E38A, E40A, E41A, E42A (lane 7); RSOA, E51A (lane 8); D86A, D88A, KWA, D92A (lane 9); T l 6 l V (lane 10. Lane *represents direct loading (2 PI) of invitro-translated wild-type cdc2 run in parallel as an internal marker.

1221, addition of anti-(cyclin A) antibodies effectively inhibited cyclin A binding to cdc2. Interestingly, the POH-8 Mab prevented the EDTA-induced inhibition of cdc2 and cyclin binding (Fig. 4, lanes 11 and 12). Preincubation of cdc2 with POH8 also enhanced complex formation with clam cyclin B (J. Lukas, unpublished results) and it is possible that binding of POH-8 to its target epitope resulted in a conformational change of the cdc2 protein which stabilizes its interaction with cyclin. In general, the results of the cdc2-cyclin-Aassociation experiment suggest that the target epitopes recognized by the POH antibodies are located on the surface of the cdc2 molecule. Since antibody binding does not prevent the association between cdc2 and cyclin, it is likely that the epitope

1

a

2

3

4

5

6

7

c

H1

casein

- 96 - 56

- 45 - 36 - 30 Fig. 6. Kinase activity of the immunoprecipitatedforms of cdc2. HeLa cell extracts were immunoprecipitated with POH-1 (lane I), POH-2 (lane 2), POH-3 (lane 3), POH-7 (lane 4), POH-8 (lane S), G 6 (lane 6 ) or rabbit pre-immune serum (lane 7) and the kinase activity associated with the immunoprecipitates assayed using either histone H I (a), casein (b) or no additional substrate (c). The reaction mixtures were resolved by SDSjPAGE and the phosphorylated proteins visualized by autoradiography. M , (kDa) is indicated.

recognized by the antibody is not directly hindered by the cyclin, but is masked due to a conformational change caused by the binding to cyclin.

The PQH antibodies immunoprecipitate various mutants and truncated forms of cdc2 Another approach to examine the potential location of target epitopes within functionally important protein sequences is to study relative binding efficiency of various defined mutants with the antibodies under study. To this end, we prepared several mutant human cdc2 proteins by in-vitro translation of mutant cdc2 mRNA [26]. We tested two representative Mab for their ability to recognize mutant cdc2, focusing in particular on cdc2 mutants which are impaired in their interaction with cyclins [26]. The list and basic description of the mutants used in the present study is given in Table 2. As it can be seen from the experiments shown in Fig. 5 , POH-2 (a) and POH-8 (b) recognized all the mutant cdc2 proteins to a similar extent as the control G6 antiserum, whose target sequence is localized at the C-terminus of the cdc2 molecule, outside any of the areas modified by the mutations (c). In addition, the POH Mab reacted not only with the full-length wild-type cdc2, but also with the N-terminal truncated forms resulting from translation at internal methionine codons (Met32, Met71, Met85 and Met100) [26]. The signal intensity of the individual forms corresponds to the relative amounts of each truncated form in the bulk translation assay (Fig. 5). Therefore, the POH target epitopes are unlikely to be localized to the N-terminal part of the cdc2 molecule and do not involve Thrl61 (Fig. 5 , lane lo), a key residue, the integrity of which is essential for binding of cdc2 to cyclins A and B.

175 Kinase activity of different antibody-defined forms of cdc2 The association between cyclins A or B and cdc2 has been found to be essential for activating the cdc2 histone H1 kinase [lo, 14, 271 and these complexes are able to phosphorylate a range of other substrates as well [28]. In view of the unexpected coprecipitation patterns obtained with our new Mab species, it was of particular interest to investigate whether distinct protein complexes Precipitated by the POH antibodies had an associated histone H1 kinase. Therefore, we immunoprecipitated cdc2 with the individual POH antibodies from nondenaturing extracts of exponentially growing HeLa cells and the immunoprecipitated complexes were tested for associated kinase activity on histone H1 and casein, as well as for their autophosphorylation activity (Fig. 6). While the control G6 antiserum readily precipitated a strong kinase activity (Fig. 6, lane 6a, b), the POH Mab precipitated only negligible histone H1 or casein kinase activity (Fig. 6a and b, lanes 1-5). In

a

1

2

4

3

5

6

7

8

9

H1 b

b

the absence of any exogenously added substrate, the G6associated cdc2 complexes phosphorylated cyclin B (Fig. 6c, lane 6). Close examination of Fig. 6c revealed the presence of two phosphorylated proteins in the POH-8 immunoprecipitations (Fig. 6c, lane 5), suggesting that either this form of cdc2 or a coprecipitated protein is active as a kinase, with a substrate specificity distinct from that of the classical cdc2cyclin form. Two phosphorylated bands of the same mobility (although of lower intensity) as the ones detected in the POH8 track were also seen in the POH-3 precipitate, while there was no autophosphorylation activity associated with the complexesprecipitated by POH-1, POH-2 or POH-7 (Fig. 6c). To complement the kinase experiments described above, we examined whether the POH antibodies either depleted or inhibited the histone H1 kinase activity associated with cdc2 in immunoprecipitations of HeLa cell extracts. Cell extracts were treated with POH Mab and immunocomplexes were retrieved by incubation with protein-A - Sepharose. No depletion of kinase activity was achieved even after repeated cycles of treatment with the POH-2 Mab, while the G6 antiserum depleted the histone H1 kinase activity very effectively (Fig. 7a). This confirms that a distinct sub-population of cdc2 molecules is recognized by these Mab. Furthermore, POH2, POH-3 and G6 did not inhibit histone H1 kinase when immobilized cdc2-cyclin complexes were preincubated with each antibody (Fig. 7 b). This suggests that the antibodies do not hinder substrate accession to the catalytic site.

10 11 12 13 14 15

Fig. 7. Assessment of the cdc2 kinase depletion or inhibition by anticdc2 antibodies. (a) HeLa-cell extracts were treated over two rounds of immunoprecipitation with POH-2 Mab (lane I), G6 (lane 2) or preimmune serum (lane 3). Depleted extracts were then immunoprecipitated by G6 antiserum which is known to recognize an active form of cdc2 and the immunoprecipitates assayed for histone H1 kinase activity (lanes 1 - 3). The kinase activities associated with the beads from the first and second depletion respectively are shown in lanes 4 and 7 for POH-2, lanes 5 and 8 for G 6 and lanes 6 and 9 for the negative control serum. (b) HeLa cdc2 was bound to cksl-2Sepharose beads, the bound complexes incubated with POH-2 (lane lo), POH-8 (lane 1I), G6 antiserum (lane 12), a control antibody (lane 13) or lysis buffcr (lane 14). All antibodies were diluted 1 : 100; the negative control reaction contained control Sepharose beads (lane 15). The immunocomplexes were then processed for kinase assay using calf histone H I as a substrate, followed by SDSjPAGE and autoradiography.

DISCUSSION In this paper we show the following. Five new monoclonal antibodies to the human cdc2 protein were generated. These antibodies reacted with the cdc2 protein in extracts from mammalian cells, but not from frog, insect or yeast cells, suggesting that the recognized epitopes lay outside the most conserved regions of the cdc2 molecule. The antibodies recognize a phosphorylated, non-cyclin-A-bound or cyclin-B-bound form of cdc2. They coprecipitated a subset of yet-unidentified cellular proteins, which might be novel cdc2-associated proteins. Table 3 summarizes the properties of these antibodies. Two distinct subsets of antibodies were found, one recognized a denaturation-sensitive epitope, the other did not. It is interesting to notice that, although the antibodies did not coprecipitate cyclins A or B, they did not prevent cdc2 binding to cyclin. This suggests that a conformational change occurs on the cdc2 protein upon binding to the cyclins and that this

Table 3. Distinct properties of the cdc2 forms recognized by different antibodies.

Antibody

Property inhibition of cyclin-cdc2 interaction

kinase activity on

kinase activity

histone H I

coprecipitated proteins

depletion

inhibition

95 kDa 56 - 51 kDa

no

no

no

no

no

Yes

80 kDa 65 kDa 50 kDa

no

no

Yes

no

no

no

cyclins A and B

no

yes

cyclin B

yes

no

Reactivity with rodent cdc2

epitope denaturation sensitivity

coprecipitated proteins

Group I Mab

no

no

Group 11 Mab

yes

G6

yes

176 change is inhibitory for antibody binding at site(s) distinct from the cyclin-binding site itself. The Mab tested recognized monomeric cdc2, translated in vitro, to the same extent as a control antibody. Since in-vitro-translated cdc2 is fully competent for cyclin binding and kinase activation [15], it is evident that the POH Mab species did not simply recognize a denatured form of cdc2. The POH Mab coprecipitate a subset of proteins, in virtue of their association with cdc2 and not by direct interaction (since they are not seen in immunoblots and are not seen in immunoprecipitations from denatured extracts). Furthermore, the presence of these novel cdc2-associated proteins suggests that a sub-population of cdc2 molecules exists in the cell, although its function has not yet been identified. Further work on the epitope mapping for these antibodies might identify these regions in cdc2. In regard to these associated proteins, it might be possible, as it has been shown in other cases [29], to raise monoclonal antibodies to the POH immunocomplexes, trying to generate antibodies which specifically recognize any of these proteins, in order to identify them. Useful information on the nature of these novel cdc2 forms might also come from immunofluorescence microscopy studies and subcellular fractionation. The proteins associated with cdc2 might be part of cellular complexes, which target cdc2 to a specific cellular site. It has recently been demonstrated that stable complex of cdk2, a cdc2-like protein, cyclin A and the transcription factor E2F exists. The specific structures in the cell which might be highlighted by the POH Mab would give us specific indications on the role of these novel cdc2associated species. Furthermore, it might be interesting to investigate the association of these proteins with cdc2 during the cell cycle. The use of these monoclonals in immunohistochemistry (J. Lukas and J. Bartek, unpublished results) is presently being evaluated and it might be interesting to study whether the immunostaining will correlate with the proliferative status of a certain tissue. We are especially grateful to Michele Pagano for suggestions and encouragement throughout the course of these experiments. We thank Bernard Ducommun for providing us with wild-type and mutant cdc2 mRNA species, Eric Karsenti for anti-PSTAIRE serum, Zdenka Staskova and Radka Pavlovska for their help with hybridoma preparation and all members of G. Draetta’s group for their interest and helpful discussions. This research was supported by a grant No. 00953 from the Czechoslovakian Ministry of Health. J. L. was supported by a Boehringer Ingelheim Fonds short-term fellowship. G. D. was supported in part by a grant from NATO (No. CGR900652).

REFERENCES 1. Pines, J. &Hunter, T. (1991) Trends Cell. Biol. I , 117-121.

2. 3. 4. 5.

Nurse, P. (1990) Nature 344, 503-508. Reed, S. (1991) Trends Genet. 7, 95-99. Lee, M. G. & Nurse, P. (1987) Nature 327, 31 -35. Draetta, G., Brizuela, L., Potashkin, J. & Beach, D. (1987) Cell 50,319-325. 6. Paris, J., LeGuellic, R., Couturier, A,, LeGuellic, K., Omilli, F., Camonis, J., MacNeill, S. & Philippe, M. (1991) Proc. Nut1 Acad. Sci. U S A 88, 1039-1043. 7. Tsai, L.-H., Harlow, E. & Meyerson, M. (1991) Nature353,174177. 8. Elledge, S. J. & Spottswood, M. R. (1991) EMBO J . 10, 26532659. 9. Knighton, D., Zheng, J., Ten Eyck, L. F., Ashford, V., Xuong, N., Taylor, S. S. & Sowadski, J. (1991) Science 253,407-414. 10. Brizuela, L., Draetta, G. & Beach, D. (1989) Proc. Nut1 Acad. Sci. USA 86,4362-4366. 11. Murray, A. W. & Kirschner, M. W. (1989) Science 246, 614621. 12. Hunt, T. (1991) in Seminars in Cell Biology (Draetta, G., ed.) vol. 2, pp. 213-222, Saunders Scientific, London. 13. Labbe, J. C., Picard, A,, Peaucellier, G., Cavadore, J . C . , Nurse, P. & Doree, M. (1989) Cell 57, 253 -263. 14. Pondaven, P., Meijer, L. & Beach, D. (1990) Gen. Dev. 4 , 9- 17. 15. Ducommun, B., Brambilla, P., Felix, M.-A., Franza, B. R. J., Karsenti, E. & Draetta, G. (1991) EMBO J . 11, 331 1-3319. 16. Gould, K. L. & Nurse, P. (1989) Nature 342, 39-45. 17. Dunphy, W. G. & Kumagai, A. (1991) CeN67,189-196. 18. Kumagai, A. & Dunphy, W. (1991) Cell64,903-914. 19. Gautier, J., Solomon, M. J., Booher, R. N., Bazan, J. F. & Kirschner, M. W. (1991) Cell67, 197-211. 20. Strausfeld, U., LabbC, J . C., Fesquet, D., Cavadore, J . C., Picard, A., Sadhu, K., Russell, P. & Doree, M. (1991) Nature 351, 242 - 244. 21. Draetta, G. &Beach, D. (1988) CeN54, 17-26. 22. Pagano, M., Pepperkok, R., Verde, F., Ansorge, W. & Draetta, G. (1992) EMBO J. 11,761 -771. 23. Richardson, H. E., Stueland, C. S., Thomas, J., Russell, P. & Reed, S. I. (1990) Gen. Dev. 4 , 1332-1334. 24. Laemmli, U. K. (1970) Nature 227,680-685. 25. Bartek, J., Bartkova, J., Vojtesek, B., Staskova, Z., Lukas, J., Rejthar, A,, Kovarik, J., Midgley, C. A., Gannon, J. V. & Lane, D. P. (1991) Oncogene 6, 1699-1703. 26. Ducommun, B., Brambilla, P., Felix, M. A,, Franza. B. J., Karsenti, E. & Draetta, G. (1991) EMBO J . 10, 3311 -3319. 27. Giordano, A., Whyte, P., Harlow, E., Franza, B. R., Beach, D. & Draetta, G. (1989) CeN58, 981 -990. 28. Nigg, E. (1991) in Cell cycle control in eukuryotes (Draetta, G., ed.) vol. 2, pp. 261 - 270, Saunders Scientific, London. 29. Whyte, P., Buchkovich, K., Horowitz, J., Friend, S., Raybuck, M., Weinberg, R. & Harlow, E. (1988) Nature 334, 124129. 30. Ducommun, B., Brambilla, P. & Draetta, G. (1991) Mol. Cell. Biol. 11, 6178-6184.