Physical Interaction of the Retinoblastoma Protein with Human D Cyclins

Home , Myc, MyoD, Oncogene

Cell, Vol. 73, 499-511, May 7, 1993, Copyright 0 1993 by Cell Press Physical Interaction of the Retinoblastoma Protein with Human D Cyclins

Steven F. Dowdy,* Philip W. Hinds,’ Kenway Louie,’ into Rb- tumor cells by microinjection, viral infection, or Steven I. Reed,t Andrew Arnold,* transfection can lead to the growth arrest of these recipient and Robert A. Weinberg” cells (Huang et al., 1988; Goodrich et al., 1991; Templeton *The Whitehead Institute for Biomedical Research et al., 1991; Hinds et al., 1992). and Department of Biology Oncoproteins specified by the SV40, adenovirus, and Massachusetts Institute of Technology papilloma DNA tumor viruses have been shown to associ- Cambridge, Massachusetts 02142 ate with pRb in virus-transformed cells (Whyte et al., 1988; tThe Scripps Research Institute DeCaprio et al., 1988; Dyson et al., 1989). Oncoprotein Department of Molecular Biology binding of pRb is presumed to lead to its sequestration 10666 North Torrey Pines Road and functional inactivation. Conserved region II mutants La Jolla, California 92037 of adenovirus ElA, SV40 large T antigen, human papil- *Endocrine Unit loma E7 viral oncoproteins that have lost their ability to and Massachusetts General Hospital Cancer Center bind pflb, and other pRb-related proteins exhibit signifi- Massachusetts General Hospital cantly reduced transforming potential (Moran et al., 1986; and Harvard Medical School Lillie et al., 1987; Cherington et al., 1988; DeCaprio et al., Boston, Massachusetts 02114 1988; Moran, 1988; Smith and Ziff, 1988; Whyte et al., 1989). This suggests that binding of pRb and related proteins by these oncoproteins is critical to their transforming abilities. The detailed study of pRb structure and function has The retinoblastoma protein (pRb) functions as a regu- revealed two sequence segments that together form the lator of cell proliferation and in turn is regulated by domain responsible for its ability to bind the various viral cyclin-dependent kinases. Cyclins Dl and D3 can form oncoproteins. This domain, termed the pRb “pocket,” has complexes with pRb that resemble those formed by been defined experimentally as the minimal region of pRb several viral oncoproteins and are disrupted by the required for viral oncoprotein binding (Hu et al., 1990; Hu- adenovirus El A oncoprotein and derived peptides. ang et al., 1990; Kaelin et al., 1990). pRb also uses this These cyclins contain a sequence motif similar to the pocket to bind a series of cellular proteins, such as the pRb-binding conserved region II motif of the viral on- E2F transcription factor, a group of proteins whose genes coproteins. Alteration of this motif in cyclin Dl pre- have been isolated by affinity cloning, an unidentified nu- vents formation of cyclin Dl -pRb complexes while en- clear structure, and the MyoD and myogenin myogenic hancing the biological activity of cyclin Dl assayed in factors (Chellappan et al., 1991; Bandaraand LaThangue, vivo. We conclude that cyclins Dl and D3 interact with 1991; Chittenden et al., 1991; Kaelin et al., 1991; Huang pRb in a fashion distinct from cyclins A and E, which et al., 1991; DefeoJoneset al., 1991; Mittnacht and Wein- can induce pRb hyperphosphorylation, and that cyclin berg, 1991; Gu et al., 1993). These observations suggest Dl activity may be regulated by its association with that the viral oncoproteins may be structural mimics of pRb. these cellular proteins. This mimicry may enable the El A, T antigen, and E7oncoproteins to occupy the pRb pocket, Introduction thereby preempting interaction of pRb with its normal cellular partners. The product of the retinoblastoma susceptibility gene pRb has been shown to be phosphorylated at serine (pflb) appears to be an important regulator of cell prolifera- and threonine residues present in sequence motifs remi- tion. This gene product, a 110 kd nuclear phosphoprotein, niscent of those modified by the cyclin-dependent kinases is expressed in a wide variety of cell types (Lee et al., (cdks) (Lees et al., 1991; Lin et al., 1991). These kinases, 1967; Friend et al., 1987). However, mutation of the Rb acting with associated cyclins, form the machinery regulat- gene is associated with only a narrow subset of tumors, ing cell cycle progression (see Hunt, 1989; Nurse, 1990; including retinoblastomas, osteosarcomas, and small cell Mailer, 1991; Hunter and Pines, 1991; Reed, 1991; Mur- and some nonsmall cell lung, bladder, breast, and cervical ray, 1992). The phosphorylation of pRb is modulated dur- carcinomas (for a review see Weinberg, 1992). Several ing the cell cycle, in that pRb is present in a hypophosphor- lines of evidence have converged on the model that the ylated state in the Go and early G, phases of the cell cycle Rb gene product acts in normal cells to constrain growth and becomes hyperphosphorylated in late G1. This hyper- and that its loss permits the unconstrained growth charac- phosphorylated state is maintained through S, GP, and teristic of cancer cells. Thus, a number of mutant Rb alleles most of M phase (Lee et al., 1987; Buchkovich et al., 1989; have been isolated from these tumors, and all appear to DeCaprio et al., 1989, 1992; Chen et al., 1989; Mihara et have suffered loss-of-function mutations (Friend et al., al., 1989). Together, these facts suggest that pRb is a 1987; Harbour et al., 1988; Shew et al., 1989; Varley et direct substrate of cdks. al., 1989; Horowitz et al., 1989, 1990; Furukawa et al., A body of experimental evidence suggests that the hypo- 1991). Furthermore, theintroductionofwild-typeRballeles phosphorylated form of pRb is active in growth restraint, while the hyperphosphorylated form is inactive. This belief The in vitro mixing experiments were repeated using is supported primarily by the observation that the El A, equivalent amounts of either the wild-type pRb or the mu- T antigen, and E7 oncoproteins specifically bind to the tant A22 pRb (Figure 1C). Both cyclins Dl and D3 showed hypophosphorylated form of pRb, ignoring the hyperphos- clear binding to wild-type pRb (Figure 1C, lanes 1 and 3). phorylated forms (Ludlow et al., 1989; Templeton et al., However, neither bound to the A22 mutant protein (Figure 1991; lmai et al., 1991; Mittnacht et al., 1991). By seques- lC, lanes 2 and 4). These results suggest that D cyclins tering hypophosphorylated pRb, these viral proteins are bind to pRb via its pocket domain. thought to reduce or eliminate the pool of pRb molecules that are active in growth regulation. Moreover, the ability Mechanism of pRb Binding by D-type Cyclins of pRb to arrest the growth of human osteosarcoma cells The binding of the D cyclins to the pRb pocket raised can be reversed by overexpressed cyclin A or E that the possibility that D cyclins may associate with pRb in a causes its hyperphosphorylation (Hinds et al., 1992). Fi- fashion similar to the mechanism used by the viral onco- nally, hypophosphorylated pRb can bind and apparently proteins. This possible functional analogy caused us to regulate the activity of the E2F transcription factor (Chel- examine the sequences of D cyclins for structural similari- lappan et al., 1991; Shirodkar et al., 1992; Weintraub et al., 1992). The apparent importance of pRb phosphorylation leads in turn to the notion that the cdks can regulate pRb function by promoting its phosphorylation. We report here that certain cyclins interact directly with pRb, but in a dramatically different fashion. The present results suggest that pRb may be regulated by certain cyclins that modulate its state of phosphorylation and may in turn regulate the activity of yet other cyclins through direct binding.

Results

In Vitro Association of pRb and Cyclins To uncover functional interactions between cyclins and pRb, we first determined whether cyclin proteins can form B aRB complexes with pRb. To do so, we studied the interac- r tions of these proteins in vitro, using human pRb purified from recombinant baculovirus-infected insect cells and 35S-labeled reticulocyte lysate-expressed human cyclins A, Bl, 82, C, Di , D3, and E. Complex formation between pRb and the various cyclins was assessed by measuring the ability of the cyclins to coimmunoprecipitate with pRb following addition of an anti-pRb monoclonal antibody. We 29 performed the mixing experiments with relatively small 1 2 3 4 5 6 7 R 9 amounts of the purified pRb (50 ng per 500 PI reaction) to minimize nonspecific aggregation driven by high pRb C aRB concentrations. Analysis of the immune precipitates (Figures 1A and 1 B) demonstrated that cyclin Dl specifically associates with pRb, while cyclins A, Bl, and E did not do so under these conditions. We did note a barely detectable signal with cyclins 82 and C. We have not determined whether this weak signal represents nonspecific aggregation of these proteins to pRb or a bonafide low affinity interaction. In our further work, we focused on the avidly binding D cyclins. Because a number of viral and cellular proteins associ- Figure 1. Complex Formation between Human Cyclins and pRb In ate with pRb via its pocket domain, wedetermined whether Vitro the D cyclins also exploit this pocket to bind to pRb. To (A and B) In vitro translated (IVT) human cyclins A, Bl, 82, C, Dl, and E (10 ul) were mixed with 50 ng of purified insect cell-produced address this possibility, we studied a mutant form of pRb human pRb or buffer (control) in a 1 ml reaction. The resulting com- bearing a defective pocket. This pRb variant, termed A22, plexes were immunoprecipitated with anti-pRb monoclonal antibodies is derived from the human small cell lung carcinoma 592 (a mixture of 21C9 and XZ55) and resolved by PAGE. In vitro translated cell line and is unable to bind viral oncoproteins or tether (IVT) cyclin reactions (1 ~1) were separated on the gels as controls. (C) In vitro translated (IVT) cyclins Dl and D3 (10 ul) were tested for properly to the nucleus (Horowitz et al., 1989; Mittnacht binding to 50 ng of purified insect cell-produced mutant A22pRb (A22) and Weinberg, 1991; Templeton et al., 1991). Its defect or wild-type pRb as described above. Arrowheads in (A), (B), and (C) stems from a deletion of exon 22 of the encoding Rb allele. indicate molecular size markers in kilodaltons. Retinoblastoma Protein Interacts with D Cyclins 501

CR II cyclin Dl, and 100 ng of purified ElA protein produced * * * in Escherichia coli and used here as competitor (Figure Ad5 Ela: EVIDLTCHEAGFPPSDDE sv40 T-Ag: E ENLFCSEEM PSSDDE 3A). Indeed, the ElA protein effectively competed with APV-16 El: ETTDLYCYEQLNDSSEE cyclin Dl for binding to pRb (Figure 3A, compare lanes Cyc Dl: MEHQLLCCEVETIRRAY 2 and 3) suggesting that all or part of the pocket region Cyc D2: MELLCHEVDPVRRAV of pRb may be involved in binding to cyclin Dl. ElA was Cyc D3: MELLCCEGTRHAPRA incapable of binding to cyclin Dl in control experiments rnut Dl: MEHQLLghEVETIREtAY (data not shown). We also used as competitors an ElA-derived synthetic Figure 2. Amino Acid Sequence Comparison of Viral Oncoproteins peptide containing both conserved regions I and II (see and D Cyclins Experimental Procedures) and a mutant ElA synthetic Single letter amino acid code alignment of the conserved region II peptide whose sequence mirrors that of the non-pRb- sequences of adenovirus 5 EIA. SV40 large T antigen, and human papilloma 16 E7 oncoproteins as compared with the N-terminal region binding Kl point mutant of SV40 T antigen (Figure 38; of human cyclins Di, D2, and D3 (Motokura et al., 1991; lnaba et al., DeCaprio et al., 1988). Concentrations of 50 and 150 uM 1992; Xiong et al., 1992a). The bold residues indicate conserved core wild-type ElA peptide effectively competed with cyclin Dl residues in the oncoproteins and the corresponding residues in the for binding to pRb, whereas the corresponding mutant ElA D cyclins. Mutation of the L, C, or E residues of the viral oncoproteins results in the disruption of pRb binding (see text). The substituted peptide had no effect on cyclin Dl binding to pRb (Figure residues present in the mutant cyclin Di (Rc-mutD1) expression vec- 38, lanes 3 and 4 versus lanes 5 and 6). The concentration tor are indicated. of ElApeptideusedforcompetingwithcyclin Dl iscompa- rable to those used by others to demonstrate competition with pRb-binding proteins in vitro (Helin et al., 1992). These results strongly support the notion that cyclin Dl ties with adenovirus ElA, SV40 T antigen, and human is interacting with the pocket region of pRb, binding to the papilloma E7 oncoproteins (Figure 2A). Specific regions same domain of pRb that is targeted by the viral oncopro- of the various viral proteins have been implicated in their teins. ability to form complexes with pRb. The otherwise distinc- tive oncoproteins are known to share two regions of common sequence, termed conserved regions I and II, which A aRb they use to bind to pRb and related proteins (Figge et r----g al., 1988). Conserved region II contains a core LXCXE sequence motif (single letter amino acid code) preceded by an acidic residue. Analysis of naturally occurring mutations and of site-directed mutations of the three viral onco- genes has demonstrated that alteration of any of the L, C, or E residues results in a dramatic decrease in the ability of the encoded viral oncoproteins to bind to pRb (Moran et al., 1986; Lillie et al., 1987; Cherington et al., 1988; DeCaprio et al., 1988; Moran, 1988; Smith and Ziff, 1988; Whyte et al., 1989). Our examination of the cyclin Dl, D2, and D3 amino acid sequences showed that all three contain an LXCXE sequence motif preceded by an E residue near their respective amino termini (Figure 2A; Motokura et al., 1991; lnaba et al., 1992; Xiong et al., 1992a). We speculated that these sequence motifs might may play a role in the ability of D cyclins to bind pRb. Additionally, we noted a second, closely related LXCXXE motif superimposed over the LXCXE motif in cyclin Dl (Figure 2A) that may also be involved in pRb binding. While the sequence LXCXE is conserved between three D-type cyclins, the immediate downstream sequences diverge and then converge again I at the cyclin box region of the protein (see Hunt, 1989; 1 2 3 4 5 6-l

Hunter and Pines, 1991). Figure 3. Competition between El A and Cyclin Dl for pRb Binding We reasoned that if cyclin Dl uses its version of the In Vitro LXCXE motif (its LLCCE sequence) to bind to pRb in a (A) In vitro translated cyclin Dl (10 ul) was mixed with 50 ng of pRb way analogous to that used by the viral oncoproteins, then (lane l), buffer (control) (lane 2) or pRb plus 100 ng of EIA protein the adenovirus ElA oncoprotein and derived peptides (lane 3) and tested for pRb binding as described in Figure 1. might compete with cyclin Dl for binding to pRb. To test (B) In vitro translated cyclin Di (10 ul) was mixed with 50 ng of pRb, buffer (control), pRb plus 50 or 150 uM ElA peptide (lanes 3 and 4) this possibility, we performed an in vitro mixing experiment or pRb plus 50 uM or 150 pM Kl (mutant ElA) peptide (lanes 5 and using purified pRb, %-labeled reticulocyte-expressed 6) and tested for pRb binding as described in Figure 1. Effect of pRb Hyperphosphorylation on Cyclin dent on the concomitant expression of both cyclin A and Dl Binding cdk2 (data not shown). As described above, diverse lines of evidence converge The ability to modulate pRb phosphorylation in these on the model that the hypophosphorylated form of pRb is insect cells allowed us to determine whether cyclin Dl active in constraining cell growth, while the hyperphos- associates preferentially with one form of pRb or another. phorylated form is not. Included in this evidence are experi- Cellular lysates from baculovirus-infected insect cells ments showing that hypophosphorylated pRb can bind were treated with rabbit anti-D1 antisera and collected on viral oncoproteins, the E2F transcription factor, and a nu- protein A-Sepharose. The immune complexes were re- clear structure while the hyperphosphorylated forms are solved by gel electrophoresis and transferred to filters by incapable of doing so. As one means of understanding Western blotting; the resulting blots were probed with an whether the phosphorylation state of pRb also affects its anti-pRb antibody to detect the presence of associated ability to associate with cyclin Dl, we assayed cyclin Di pRb (Figure 4). binding to hypophosphorylated and hyperphosphorylated In insect cells infected with pRb and cyclin Dl baculo- pRb prepared from baculovirus-infected High 5 (581-4) viruses, coimmunoprecipitation of hypophosphorylated cabbage looper insect cells. pRb with cyclin Dl was readily detectable (Figure 4, lane In preliminary experiments, we found that the pRb ex- 3). However, when the pRb was converted almost com- pressed in insect cells infected only with a recombinant pletely to the hyperphosphorylated state by infecting with human pRb baculovirus remained in a hypophosphory- cyclin A and cdk2 baculoviruses, no coimmunoprecipitat- lated state (Figure 4, lane 6); this was confirmed by two- ing pRb was detected in the presence of cyclin Dl (Figure dimensional tryptic phosphopeptide analysis(S. Mittnacht 4, lane 4). These results support the notion that cyclin and S. F. D., unpublished data). Coinfection of cyclin Dl- Dl binds preferentially to the hypophosphorylated form of and pRb-expressing baculoviruses in these cells did not pRb. Stated differently, we conclude that the hyper- alter this hypophosphorylated state (Figure 4, lane 7). phosphorylation of pRb, seen in the late G1 phase of the However, when we coinfected cyclin A- and cdkP cell cycle, reduces (if not eliminates) its ability to bind expressing baculoviruses (provided by D. Morgan) to- cyclin Dl. gether with the pRb baculovirus in these insect cells, the resulting pRb was found almost exclusively in a hyper- Interaction of Cyclin Dl and pRb In Vivo phosphorylated state (Figure 4, lane 8). We believe that The above results suggested that pRb binds cyclins Dl this hyperphosphorylated pRb resembles that found in and D3 via a domain that also enables its binding to the mammalian ceils, as our own work and that of others has viral oncoproteins, to the E2F transcription factor, and to indicated that cyclin A-cdk complexes can hyperphos- the protein(s) tethering it to the nucleus. We were inter- phorylate pRb in vitro and in vivo (Helin et al., 1992; Hu ested in determining whether these associations found in et al., 1992; Hinds et al., 1992; V. DuliC, S. F. D., and vitro and in insect cells mirrored interactions occurring S. I. R., unpublished data). Moreover, the hyperphos- within mammalian cells. To detect the association of cyclin phorylated state of pRb in these insect cells was depen- Dl with pRb in vivo, we immunoprecipitated lysates with anti-pRb monoclonal antibody and analyzed the precipitates by gel electrophoresis and Western blots probed with an anti-cyclin Dl antiserum. Cellular lysates were prepared from Wi38 human diploid lung fibroblasts under IP: aD1 WCE mild, nonionic detergent conditions (0.2% Nonidet P-40 r--lr----J [NP-401) in a small volume (400 ~1) and incubated with 2 PI of anti-pRb monoclonal antibody (RB-02) cross-linked to Sepharose (provided by S. Gruenwald) diluted in 30 ~1 of unreacted Sepharose. The immune complexes were resolved by gel electrophoresis and transferred to a filter by Western blotting; the resulting blots were probed with rabbit anti-cyclin Dl antiserum to detect the presence of associated cyclin Dl (Figure 5). The coimmunoprecipitation of cyclin Dl was readily detectable when the Sepha- 12 3 4 5678 rose contained the cross-linked anti-pRb antibodies (Fig- ure 5, lane 2); however, in the presence of only unreacted Figure 4. Cyclin Dl Associates with Hypophosphorylated pRb In Vivo Sepharose, the cyclin Dl was not detectable (lane 1). We Insect cells were infected with control (H5), pRb-, cyclin Dl-, cyclin A-, and/or CdkSexpressing baculoviruses as indicated. Whole cell have also observed these complexes using a second anti- extracts (WCE) from these cells were prepared and subjected to immu- pRb monoclonal antibody (21C9) and a second rabbit anti- noprecipitation (IP) with an anticyclin Dl antiserum, separated by cyclin Dl antiserum (data not shown). In addition, we have PAGE, and immunoblotted with an anti-pRb monoclonal antibody (RB- observed the cyclin Dl-pRb complex using this assay in 02) (lanes l-4). One-fifth of the whole cell extracts was separated by Sift human diploid foreskin fibroblasts (data not shown). PAGE and immunoblotted with an anti-pRb monoclonal antibody as described above (lanes 5-8). The migration of the hypophosphorylated We conclude that complexes between cyclin Dl and pRb (pRb) and hyperphosphorylated (ppRb) forms of the Rb protein are exist in human cells expressing these proteins at physio- indicated. logic concentrations. Retinoblastoma Protein Interacts with D Cyclins 503

IP I

wRby pRb’ Figure 5. In Vivo Association between Cyclin Dl and pRb Whole cell lysates (WCE) from Wi36 human diploid fibroblasts were subjected to immunoprecipitation (IP) with an anti-pRb monoclonal antibody cross-linked to Sepharose (lane 2) or unreacted Sepharose (lane l), separated by 10% PAGE, transferred to a nitrocellulose filter by Western blotting, and probed with rabbit anti-cyclin Dl antiserum. The arrowheads indicate the positions of cyclin Dl and a 36 kd marker. Lane 4 contains one-fifth of the whole cell lysates; lane 3 was left empty to avoid contamination of lane 2 from lane 4. 1 2 3 4 5 6

Structural Basis of Cyclin Dl Binding to pRb The success in demonstrating pRb-cyclin Dl complexes in vivo enabled us to examine the structural basis of the ability of cyclin Dl to bind pRb. In particular, we explored the proposed analogy between cyclin Dl and oncoprotein binding to pRb by altering the LXCXE sequence motif of cyclin Dl through site-directed mutagenesis of its cDNA. Thus, we replaced its LLCCE (residues 5-9) sequence with the sequence LLGHE. This mutant sequence mimics the non-pRb-binding 928 mutant of the ElA oncoprotein at the central C residue (Stein et al., 1990) and residue 6 of human cyclin D2 sequence at the second C (Inaba et al., 1992; Xiong et al., 1992a) and disrupts the superimposed misspaced LXCXXE present here (see Figure 2). We Figure 6. Structural Basis of Cyclin Dl and pRb Association placed this mutant cyclin Dl cDNA under the control of Cyclin Dl binds pRb in vivo, and mutant cyclin Dl fails to bind pRb. [%]methionine-labeled lysates from COSI cells transfected with con- the cytomegalovirus (CMV) early promoter in the Rc-CMV trol (Rc-CMV), pRb (SVRb), cyclin Dl (Rc-cycDi), and/or mutant expression vector (Rc-mutD1) as we had previously done cyclin Dl (Rc-mutD1) expression vectors were divided in half and for the wild-type cyclin Dl cDNA (Rc-cycD1; Hinds et al., immunoprecipitated with either an anti-pRb monoclonal antibody (A) 1992). To assay for complex formation between pRb and or with an anti-cyclin Dl antiserum (B). The immune complexes were mutant cyclin Dl, we cotransfected COSl green monkey resolved by PAGE and exposed for 72 hr (A) or 24 hr (B) to film. Arrowheads in (A) and (B) indicate molecular size markers in kilodal- kidney cells with an SV40-driven wild-type pRb expression tons. The migration of the cyclin Dl protein and the hypophosphory- vector (SVRb; Templeton et al., 1991) and the wild-type lated (pRb) and hyperphosphorylated (ppRb) forms of the Rb protein or mutant cyclin Dl expression vectors. are indicated. The wild-type and mutant cyclin Dl alleles were each transfected singly into COSl cells to ensure that the two alleles were expressed to the same level. 35S-labeled cell cyclin Dl, mutant cyclin Dl, and control vectors, and these lysates were prepared from the transfected cells 48 hr lysates were incubated with an anti-pRb antibody. Cells after transfection of COSl cells with wild-type or mutant transfected with the mutant cyclin Dl vector alone or in cyclin Dl and control vector, and the lysates were incu- combination with the pRb vector failed to coimmunopreci- bated with a rabbit anti-cyclin Dl antiserum. Resulting im- pitate a 36 kd protein (Figure 6A, lanes 5 and 6). The mune complexes were analyzed by fluorography for the control experiments, done in parallel with wild-type cyclin presence of the cyclin Dl protein (Figure 66). The results Dl and pRb vectors, detected a coprecipitating protein of confirmed that the mutant cyclin Dl allele was indeed ex- 36 kd with pRb (Figure 6A, lane 4). Double immunoprecipi- pressed to the same level as the wild-type cyclin Dl allele tation experiments on these lysates using an anti-pRb anti- (Figure 6B, lane 6 versus lane 4). body in the first immunoprecipitation, followed by boiling The product of the mutant allele was then examined for in 2% SDS, dilution, and subsequent immunoprecipitation its ability to form a complex with pRb and compared with with rabbit anti-cyclin Dl antiserum confirmed that this 36 wild-type cyclin Dl-pRb complexes. 35S-labeled cell ly- kd protein was indeed cyclin Dl (data not shown). These sates were prepared from the transfected cells 48 hr after results demonstrate that the formation of a cyclin Dl-pRb transfection of COSl cells with combinations of pRb, complex is dependent on the presence of an intact LXCXE Cdl 504

motif in cyclin Dl. Taken together with earlier data, we days, stained, and quantitated by counting the number of conclude that this association mimics the binding shown surviving flat cells, by viral oncoproteins, in that it requires the pRb pocket As described earlier, the pRb plus control vector- and occurs preferentially, if not exclusively, with the hypo- transfected SAOS-2 cells (positive control) in Figure 78 phosphorylated form of pRb. exhibit a characteristic growth-arrested flat cell morphology. These pRb-induced flat cells were counted and their Modulation of pRb and Cyclin II Function In Vivo number normalized to 100% (Table 1). Transfection of Like cyclins Dl and D3, the E2F transcription factor ap- SAOS-2 cells with the pRb expression vector harboring a pears to associate with the pRb pocket domain and binds termination mutation at codon 78 (SV78t) plus control vec- preferentially to the hypophosphorylated form of pRb tor (as negative control) resulted in the generation of less (Chellappan et al., 1991; Shirodkar et al., 1992; Kaelin et than 2% as many flat cells (Figure 7A; Table 1). In confir- al., 1992; Helin et al., 1992). Since E2F activity is thought mation of our published results, cotransfection of pRb plus to be regulated by pRb binding, this suggested that cyclin cyclin E (Rc-cycE) expression vectors into the SAOS-2 Dl activity may also be regulated via this association. One cells resulted in 10% of the control number of flat cells test of this relationship was afforded by a tissue culture (Figure 7F; Table 1). model that we exploited in earlier work (Huang et al., 1988; The cointroduction of pRb and wild-type cyclin Dl (Rc- Templeton et al., 1991; Mittnacht et al., 1991). SAOS-2 cycD1) vectors into SAOS-2 cells resulted in 48% of the osteosarcoma cells, which express only a mutant, defec- pRb control number of flat cells (Figure 7C; Table 1). The tive pRb (Shew et al., 1990), respond to introduced wild- cotransfection of the mutant cyclin Dl (Rc-mutD1) expres- type pRb by ceasing growth, spreading out into large flat sion vector with the pRb plasmid resulted in a further sub- cells, and showing a senescent appearance for at least 3 stantial decrease in the number of flat cells, with less than weeks after transfection. pRb expression vectors encod- 9% of the pRb control number of flat cells present (Figure ing mutant, defective forms of pRb are incapable of induc- 7D; Table l), a number comparable to that seen upon ing this phenotype (Templeton et al., 1991). This response introduction of the cyclin E vector. The cointroduction of can be reversed partially or completely by cointroducing a cyclin D3 expression vector (Rc-cycD3A; see Experi- vectors expressing cyclin A, Dl , or E into these cells (Hinds mental Procedures) with a pRb vector into SAOS-2 cells et al., 1992). The observed reversal of the pRb-induced resulted in 48% of the pRb control number of flat cells senescence by cyclins A and E could be rationalized in (Figure 7E; Table l), suggesting that cyclin D3 can also terms of the demonstrated ability of these cyclins to cause disrupt pRb function in a fashion similar to cyclin Dl. In pRb hyperphosphorylation and, we presumed, attendant summary, a mutation that reduces the affinity of cyclin Dl functional inactivation. The cyclin Dl reversal of the ef- for pRb also enhances its ability to overcome or bypass fects of pRb clearly depended upon another mechanism, a pRb-induced cell cycle block. We suggest that this de- as the pRb in these cotransfected cells remained in a hypo- creased affinity reduces the ability of pRb to sequester phosphorylated state. cyclin Dl, thereby enhancing the growth-promoting func- Two mechanisms might be invoked to explain the ability tion of this cyclin. of overexpressed cyclin Dl to reverse the growth-sup- Implicit in the above hypothesis is the notion that cyclins pressing effects of pRb. Cyclin Dl might titrate out pRb, Dl and D3 do not overcome pRb function by altering its perhaps by occupying all of the available pRb pockets. state of phosphorylation. To address this, we monitored, in This in turn would preclude interaction of pRb with other parallel, the state of pRb phosphorylation in the transfected associated cellular proteins and free them to promote cell cells described above by analyzing the electrophoretic mi- cycle progression. Alternatively, cyclin Dl itself may be a gration rate of metabolically labeled protein (Figure 8). In centrally important promoter of cell cycle progression and agreement with earlier results, cyclin E was able to force at the same time an object of sequestration by pRb. In this the hyperphosphorylation of pRb while cyclin Dl , mutant event, any of the overexpressed cyclin Dl that exceeds cyclin Dl, and cyclin D3 were unable to do so, leaving pRb the ability of pRb to bind it would remain free to promote in a hypophosphorylated, rapidly migrating form (Figure 8, downstream events favoring cell cycle progression. compare lane 8 with lanes 3, 4, and 5). We conclude that If the first hypothesis above were correct, then the rever- cyclin E reverses pRb function by causing its functional sal of the effects of pRb by cyclin Dl should depend on inactivation through hyperphosphorylation. Cyclins Dl its ability to bind pRb and thus on the intactness of its and D3, in contrast, do not affect pRb by modulating its LXCXE motif. If the second were correct, then a greatly state of phosphorylation. decreased affinity of cyclin Dl for pRb should free it from pRb control and thus potentiate its cell cycle-promoting D Cyclins Promote an Increase in Cell Cycling activity. To test which of these opposing theories was cor- A corollary of the above model is that mutant cyclin Dl rect, SAOS-2 cells were cotransfected with a pRb expres- should also be more effective than its wild-type counterpart sion vector (CMVRb), a selectable puromycin expression in releasing SAOS-2 cells from a pRb-imposed G, phase vector (pBABEpuro), and either the control vector (Rc- growth arrest. To address this, we performed flow cytome- CMV), the wild-type cyclin Dl vector (Rc-cycDl), or the try analysis on SAOS-2 cells transfected with a pRb vector mutant cyclin Dl vector (Rc-mutD1). These transfected plus control vector or a pRb vector in combination with cells were placed under puromycin selection for lo-14 either a cyclin Dl, mutant Dl, cyclin D3, or cyclin E vector Retinoblastoma Protein Interacts with D Cyclins 505

Figure 7. Phenotypic Alteration of SAOS-2 6 Cells Transfected with the Rb cDNA in the Presence and Absence of D Cyclin Constructs SAOS-2 cells were transfected with a puromy Rb tin resistance vector plus the negative control vectors 761 and Rc-CMV (76t + CMV; [A]), the c&f pRb expression vector and Rc-CMV (Rb + CMV; [B]), the pRb vector and appropriate D cyclin vector ([Cl. [D], and [E] as indicated), or the pRb vector and cyclin E vector (Rb + E, positive control; [F]). The transfected cells were D subjected to selection in puromycin for 14 days and subsequently fixed and stained. Photomi- crographs of the stained cells were made at Rb 35.6fold magnification. Colonies expressing only the puromycin resistance marker and demonstrating the characteristic small cell n&l morphology of pRb SAOS-2 cells are repre- sented in (A), (D), and (F). F Rb

;

and scored for the cell cycle position of pRb-positive cells of a smaller number of GdG, arrested cells in cells (Table 2). We found that 44%-48% of asynchronous, un- transfected with the mutant cyclins Dl and E as compared transfected populations of SAOS-2 cells are routinely in with the cells transfected with wild-type cyclins Dl and the S, GP, and M compartments. In contrast, only 10% of D3. These results, like those reported above, indicate the the pRb-positive cells were present in the S, GP, and M enhanced ability of the mutant cyclin Dl protein to over- phases in cells transfected with pRb plus control vectors. come the growth-suppressing effects of pRb in SAOS-2 The added presence of a cyclin E-expressing vector in- cells. creased the number of S, GP, and M phase pRb-positive cells from 10% to 58%. Discussion The introduction of wild-type cyclin Dl or D3 vectors into pRb-expressing SAOS-2 cells resulted in an increase Regulation of Cell Cycle Progression in the number of pRb-positive cells in the S, GP, and M by pRb Hyperphosphorylation phases from 10% to 18% and 17%, respectively, while Cyclins and associated cdks appear to be key regulators the introduction of the mutant cyclin Dl vector resulted in of critical check points within the eukaryotic cell cycle. an increase to 26% (Table 2). We also note the presence In the cell cycle’s simplest form, developed largely from genetics and biochemical studies in both yeast and Xeno- pus laevis (see Reed, 1980, 1991; Hunt, 1989; Nurse, 1990; Maller, 1991; Murray, 1992), the cdks are present Table 1. Flat Cell Production of pRb plus Cyclin-Transfected SAOS-2 Cells in essentially constant amounts throughout the cell cycle in cycling cells; their catalytic activity and substrate speci- Number of Flat Cells ficity are determined largely by the presence of associated CMVRb + (x 10-d) Control (o/o)” cyclins that act as regulatory subunits. Cyclins are expressed in specific points in the cell cycle, and this in turn Rc-CMV 4.07 100 Rc-cycD1 2.24 46 allows phosphorylation by the cyclin-cdk complexes of Rc-mutD1 0.45 9 specific substrates that, when modified, cause progres- Rc-cycD3 2.33 48 sion through these check points. Rc-cycE 0.50 10 This model is transferable at least in part to mammalian (No Rb) 0.05 1 cells. Among the apparent key cellular substrates of cdk ’ Transfected SAOS-2 cells (6 x 10’) were plated, subjected to drug phosphorylation is pRb, which acquires a substantial num- selection for 14 days, and stained. b Percentage of flat cells relative to CMVRb plus AC-CMV. ber of phosphate groups 2-4 hr before the cell enters into S phase (Lee et al., 1987; Buchkovich et al., 1989; De- Cdl 506

aRb and S. I. R., unpublished data). However, cyclins C and Dl are capable of complementing triple CLN mutants of Saccharomyces cerevisiae, thus confirming their ability to form enzymatically active complexes with CDC28 kinase (Lew et al., 1991; Xiong et al., 1991). Although negative biochemical results deriving from in vitro incubations must be interpreted with caution, we suggest that the above results do not support the role of cyclin Dl in driving pRb pWy (116 pRb/ hyperphosphorylation in the living cell. Moreover, all three 497 D-type cyclins demonstrate minimal cell cycle-dependent modulation of messenger RNA (mRNA) expression levels I 1 1 2 3 4 5 6 in human fibroblasts and keratinocytes (Won et al., 1992; Y. Geng and R. A. W., unpublished data). In keeping with Figure 6. Phosphorylation State of pRb in Transfected SAOS-2 Cells this result are those of Hinds et al. (1992) and Ewen et [?S]methioninalabeled proteins derived from cells transfected with al. (1993 [this issue of Ce//)), which demonstrated that the the indicated expression constructs were subjected to immunoprecipitation with the anti-pRb monocional antibody 21 C9 46 hr after transfec- introduction of cyclin Dl and pRb into SAOS-2 cells leads tion. Cells used to prepare the control lysate were transfected with to minimal phosphorylation of pRb. the negative control plasmid Rc-CMV (lane I), which fails to express pRb or any cyclin; the positive control of pRb plus Rc-CMV vector A Novel Relationship between pRb and Cyclins (lane 2); the pRb vector plus the indicated D cyclin vectors (lanes 3, In the context of cell cycle progression, the G, hyperpho- 4, and 5); or the positive control for pRb hyperphosphorylation of the pRb plus cyclin E vectors (lane 6). The migration of the hypophosphory- sphorylation of pRb, possibly driven bycyclin E-cdkecom- lated (pRb) and hyperphosphorylated (ppRb) forms of the Rb protein plexes, seems to remove a pRb-imposed block, permitting in 7.5% PAGE are indicated. transit into the late G, phase of the cell cycle. In the present work, we provide evidence pointing to a quite different relationship between cyclins and pRb. The data presented Caprio et al., 1989, 1992; Chen et al., 1989; Mihara et al., here strongly suggest that cyclins Dl and D3 can be bound 1989). Several sites of pRb phosphorylation have been by pRb and that this binding, in the case of cyclin Dl, mapped biochemically and correspond to cdk consensus appears to compromise its activity. Work of Ewen et al. phosphorylation sites (Lees et al., 1991; Lin et al., 1991). (1993) points to an alternative, contrasting interaction be- Moreover, in a variety of cell types, the schedule of pRb tween cyclin D2 and pRb that results in the hyperphosphor- hyperphosphorylation is close to and is perhaps congruent ylation of pRb. with the sudden appearance of cyclin E late in the G1 phase Cyclin Dl does not appear to regulate pRb function by (Lewetal.,1991;Koffetal.,1991,1992;DuliCetal.,1992) causing its hyperphosphorylation. We propose that cyclin and the G1 restriction point as defined by Pardee (1974). Dl is an object of regulation by pRb. Implicit in this model Together, such observations provide evidence that this is the notion that the activity of cyclins can be determined late G1 pRb hyperphosphorylation is caused directly by a not only by the expression of their respective genes, but cyclin-cdk complex, perhaps by the recently described also by posttranslational regulation, including the pRb as- cyclin E-cdk2 complex (Dulic et al., 1992; Koff et al., sociation proposed here. 1992). Indeed, there is ample evidence suggesting that post- We note that in vitro kinase assays of immune com- translational processes affect cyclin activity. Cyclin Dl has plexes of cyclins A, Bl, 82, and E and associated cdks been reported to have a short half-life, suggesting an abil- from asynchronous HeLacell extracts demonstrate signifi- ity to regulate its activity by rapidly modulating its stability cant pRb kinase activity, while immune complexes of (Matsushime et al., 1992). Furthermore, cyclin Dl is cova- cyclins C and Dl and associated cdks demonstrate only lently modified by phosphorylation (Matsushime et al., background levels of pRb kinase activity (V. Dulic, S. F. D., 1991). However, data on binding of the FAR1 protein of

Table 2. Flow Cytometry Analysis of pRb plus Cyclin-Transfected SAOS-2 Cells Total Number Number in Number in Percentage in CMVRb + of Cells’ GdGP S, G2, and M S, G2# and MC Rc-CMV 1109 996 113 10 Rc-cycD1 1391 113s 253 18 Rc-mutD1 656 490 166 26 Rc-cycD3A 1456 1203 253 17 Rc-cycE 612 256 356 5s Rc-CMV (only) 12 4 S a Positive cells were those demonstrating green fluorescence (pRb expression) above background. b DNA content (2n and >2n) determined by propidium iodide fluorescence intensity. c The percentage of cells with >2n DNA content in the mass populations ranged from 44%-48% (30,000 cells counted). Retinoblastoma Protein Interacts with D Cyclins 507

S. cerevisiae (Chang and Herskowitz, 1990) to the CLNP 1992); this is in contrast with the sudden shift in pRb modifi- cyclin and CDC28 kinase suggest that regulation of cyclin cation much later in G,. Moreover, as shown here and function by association with a second protein may be used in our earlier work, transfection of cyclin Dl- and D3- to arrest these cells in response to a negative growth factor expressing plasmids have no effect on pRb modification (Peter et al., 1993). (Hindset al., 1992). Ewen et al. (1993) reportsimilar results The model that pRb can regulate cell cycle progression with murine cyclins Dl and D3. However, these research- by binding a series of growth-promoting proteins via its ers provide evidence that coexpression of murine cyclin pocket domain is already well developed through the study D2 and pRb in SAOS-2 cells leads to hyperphosphoryla- of the E2F transcription factor, which associates with the tion of pRb, suggesting important functional distinctions hypophosphorylated but not hyperphosphorylated forms between cyclin D2 and the Dl and D3 cyclins examined of pRb (Chellappan et al., 1991; Shirodkar et al., 1992; here. Kaelin et al., 1992; Helin et al., 1992). E2F may represent None of the above observations rule out the attractive only one of a cohort of such captives; the existence of at possibility that cyclin Dl (and perhaps D3) in association least seven other pRb pocket-binding proteins is implied with cdk2, cdk4, and/or cdk5 (Matsushime et al., 1992; by the recent literature (Defeo-Jones et al., 1991; Huang Xiong et al., 1992b) are responsible for causing a low, et al., 1991; Kaelin et al., 1991; Mittnacht and Weinberg, basal level of pRb phosphorylation through the early por- 1991; Kim et al., 1992; Rayet al., 1992; Rustgi et al., 1992; tion of G, and that this basal level phosphorylation is prepa- Gu et al., 1993). The occupation of the pocket domain by ratory, indeed prerequisite, to the hyperphosphorylation viral oncoproteins may preempt association of pRb with seen later. This may be accomplished by the binding of other cellular proteins, including E2F. Hyperphosphoryla- the LXCXE motif of cyclin Dl to pocket of pRb, resulting tion of pRb is presumed to sterically alter the protein, caus- in the positioning of a cdk in close proximity to pRb. Such ing it to release bound associated proteins. a process would explain the observed dependence of pRb Cyclins Dl and D3 fit precisely into this model of pflb- hyperphosphorylation on the intactness of its pocket, as associated protein interactions. They contain a conserved mutant forms of pRb contain little, if any, phosphate resi- region II-like sequence motif that is essential for their bind- dues (Templeton et al., 1991). Indeed, mutant A22pRb, ing to pRb, the same motif that is used by the viral onco- which fails to bind cyclins Dl and 03, contains less than proteins for such binding. Cyclin Dl binding to pRb is two-thirds of the wild-type pRb tryptic phosphopeptides blocked by the viral ElA oncoprotein and a derived pep- by two-dimensional analysis when both proteins are over- tide, and the structural intactness of the pRb pocket do- expressed in insect cells; however, it retains all of the cdk main is necessary for cyclin D binding. In fact, such associ- consensus sites (S. Mittnacht and S. F. D., unpublished ation between cyclin Dl and pRb is compatible with two data). quite distinct models. On the one hand, this binding may Our own data indicate a functional parallelism between indicate a viral oncoprotein-like role for cyclins Dl and D3, E2F and cyclin Dl (and perhaps cyclin D3) and suggest in which they occupy most if not all of the pRb pockets that these cyclins, like E2F, are centrally acting agents of the cell, thereby inhibiting the ability of pRb to associate promoting cell cycle progression. The D cyclins in general, with other cellular proteins. On the other hand, this associ- and cyclin Dl in particular, may play especially critical ation may point to the control of cyclins Dl and D3 by roles in regulating cell cycle progression. This is evidenced pRb, in effect regulating cyclin D activity through binding. by the observation that the cyclin Dl (Pt?AD7/DllS287) We favor the latter model for several reasons. To begin mRNA is overexpressed in a variety of tumors, including with, pRb is an abundant nuclear protein, while the D parathyroid adenomas (Motokura et al., 1991; Rosenberg cyclins under most conditions of cell growth are present et al., 1991a), breast carcinoma (Lammie et al., 1991), in low concentration. By our imprecise estimates, pRb is head and neck squamous cell carcinomas (Lammie et al., present in amounts that are at least 5 to lo-fold higher 1991; Schuuring et al., 1992) and B cell lymphomas (Ro- than the D cyclins. Consequently, it is highly unlikely that senberg et al., 1991 b; Withers et al., 1991; Seto et al., the D cyclins can titer out all available pRb pockets during 1992), as a consequence of chromosomal translocations the course of normal cell cycle progression. In addition, or gene amplification events that directly affect the cyclin we cite evidence presented here that indicates that cyclin DlIPRAD7 gene. In the case of the B cell malignancies, Dl activity is potentiated when its ability to associate with the deregulated cyclin DlIPRAD7 expression is associ- pRb is compromised. These lines of evidence cause us ated with a t(l1;14)(q13;q32) translocation that brings the to propose that pRb and perhaps other pRb-related pocket immunoglobulin enhancer in close proximity to the cyclin proteins, such as ~107 and ~130 (Harlow et al., 1987; DlIPRAD7 gene and argues strongly that cyclin Dl is the Ewen et al., 1991), may act to regulate the activity of D long-sought bcl-7 protooncogene (Williams et al., 1993; cyclins by reversible sequestration. Rosenberg et al., 1993). We suggest that in such tumor We consider it unlikely that cyclins Dl and D3, acting cells, as in cells transfected here with cyclin Dl and D3 with associated kinases, are responsible for the massive expression vectors, the overexpressed D cyclins outpace hyperphosphorylation of pRb seen several hours before the ability of pRb to bind and regulate them. As a conse- the end of the G1 phase of the cell cycle. In most cell quence, the surfeit of cyclin D molecules escapes pRb types, cyclin Dl appears to accumulate in ever-increasing control and triggers important late GI events in a growth- amountsfromatimeshortlyaftertheonsetof G, (Motokura promoting regulatory cascade. In addition, the resultant et al., 1991, 1992; Matsushime et al., 1991; Won et al., increase in cyclin Dl expression may result in a shift in Cdl 508

the equilibrium of pRb-cyclin Dl complexes. Indeed, data 105 to 8 x lo5 cells plated. SAOS-2 cell transfections for immunopre- from Ewen et al. (19933,demonstrate a decrease in the cipitation were treated as above, except they were replated I:1 24 hr amount of E2F associated with pRb in the presence of after washing and %S labeled at 48 hr after washing the calcium- phosphate precipitation. cotransfected cycl&r Dl in SAOSQ cells, supporting the notion that cyclin Dl is occupying the pocket of pRb and Flow Cytometry Analysis preventing E2F binding to pRb. These associations, to- SAOS-2 cells (2 x 1oB) were transfected with indicated combinations gether with the demonstrated overexpression of cyclin Dl of Rc-CMV, SVRb, CMVRb, Rc-cycDl, and Rc-mutD1 expression vectors as described above. The transfected cells were trypsinized in human tumors, suggest a.central role of this cyclin in 48 hr posttransfection, fixed in -20°C methanol for 5 min and -20°C regulating cell cycle progression. acetone for 2 min in suspension, and rehydrated for IO min in PBS plus 0.1% bovine serum albumin. The rehydrated cellswere incubated Experlmrntel Procedures with an anti-pRb monoclonal antibody (RB-02; Pharmigen) at reverse transcription for 45 min, washed, incubated with a donkey anti-mouse Cell Culture immunoglobulin G fluorescein isothiocyanate-conjugated antiserum SV40-transformed African green monkey kidney COSl cells were (Jackson Immunoresearch) for 30 min, and then washed. The anti- maintained in Dulbecco’s modified Eagle’s medium (DMEM) with 10% body-incubated cells were then treated with an RNAase A and propid- calf serum plus penicillin and streptomycin in 5% CO* at37OC. SAOS-2 ium iodide solution for 20 min at reverse transcription prior to sorting osteosarcoma cells and Wi38 human diploid lung fibroblasts were on a Becton-Dickinson FACStarflow cytometer, allowing simultaneous obtained from the American Type Culture Collection and maintained detection and quantitation of cells producing pRb (green fluorescence) in DMEM with 15% heat-inactivated fetal bovine serum plus penicillin and determination of DNA content (red fluorescence intensity). and streptomycin in 3% CO,. Sift human foreskin fibroblasts were obtained from J. Rheinwold (BioSurface Technology, Cambridge, Immunopreclpltatlons Massachusetts) and were maintained as above. SAOS-2 cells consist Whole cell extracts were prepared from methionine-starved cells la- of a mixed aneuploid population and were therefore single cell sub- beled for 4 hr in DMEM-methionine minus plus 10% dialized heat- cloned by limiting dilution to isolate a subclone with a stable DNA inactivated fetal bovine serum medium plus 75 t&i/ml Tran%S-Label content for fluorescence-activated cell sorting analysis (see below). methionine (ICN Pharmaceuticals). The cells were washed in PBS and Several resultant subclones were subjected to pRb transfection and lysed In situ by the addition of 1 ml of ELB (50 mfvl HEPES [pH 7.21, were found to be identical to the parent culture in their response to 250 mM NaCI, 2 mM EDTA, 0.1% NP-40, 1 mM dithiothreitol, 1 pg/ pRb-induced growth arrest (Hinds et al., 1992). One such subclone ml aprotinin [Sigma], 1 pglml leupeptin [Sigma], and 50 pglml phenyl- (2.4) was used for all of the experiments described herein. methylsulfonyl fluoride). The lysates were precleared with 50 pl of 10% Staphylococcus aureus cells (Zymed) and clarified by centrifugation. Plasmld Constructions lmmunoprecipitations were performed by the addition of 1 ul of rabbit The phRbc-SVE (herein referred as SVRb), SV76t, Rc-cycDi, Rc- anti-D1 antiserum to cyclin Dl (Hinds et al., 1992) or by the addition cycD3, and Rc-cycE cDNA expression vectors have previously been of 200 pl of tissue culture supernates from the anti-pRb monoclonal described (Templeton et al., 1991; Hinds et al., 1992). pBABEpuro antibody 21C9, raised against a peptide representing residues 248- (Morgenstern and Land, 1990) encodes a puromycin resistance gene. 2620f pRb (Whyte et al., 1988) or XZ55(Hu et al., 1992). The immune CMVRb was generated by inserting the human Rb cDNA into the complexes were collected on protein A-Sepharose (Bio-Rad) and BamHl site of pCMV-Neo-Barn (Hinds et al., 1990). Rc-cycD3A was washed three times with extraction buffer prior to the addition of 2 x generated by digestion of Rc-cycDS with Mscl, resulting in the deletion sample buffer and boiling, followed by polyacrylamide gel electropho- of a 736 bp in the 3’ untranslated region including several AUUUA resis (PAGE) and fluorography. For coimmunoprecipitation experi- destabilizing sequences (Shaw and Kamen, 1986). The Rc-mutD1 ments from COS cell transfections, the primary antibody was allowed expression vector was generated by polymerase chain reaction (PCR)- to bind for 1 hr in the presence of protein A-Sepharose beads; small mediated mutagenesis. An 85mer oligonucleotide was synthesized numbers of samples ((6) were washed three times with ELB very that spanned the initiating methionine to residue 13, including G and rapidly. H residues at amino acid positions 6 and 7, and a negative strand For immunoprecipitations followed by Western blot analysis from 21-mer oligonucleotide that was 3’ prime of the Hindlll site in the 3’ insect cells, anti-cyclin Dl immune complexes were prepared from untranslated region, generating a 1 kb PCR DNA fragment. The 1 kb lysates as described above, separated by 7.5% PAGE, and transferred PCR DNA fragment was digested with Hindlll, inserted into the Hindlll by Western blotting to nitrocellulose filters. For immunoprecipitations site of Rc-CMV (Invitrogen), and the orientation was determined by followed by Western blot analysis from human fibroblasts, lysates were digestion with Apal (which was introduced into the 5’ untranslated prepared from two semiconfluent 10 cm dishes in 200 pl of ELB/250 region of the 85-mer PCR primers) and the insert sequenced. (250mM NaCI)andthendiluted I:1 with ELBlSO(50mM NaCI), yielding a final NaCl concentration of 150 mM. The lysates were precleared Transfectlon and Flat Cell Analysis for 30 min with 100 ul of Sepharose CMB (Pharmacia) and clarified COSl cells (5 x 103 plated on 6 cm dishes were transfected by the by centrifugation. The lysates were then incubated for 2 hr with either DEAE-dextran method (Lopata et al., 1984) with 3 pg of Rc-CMV, Rc- 30 pl of Sepharose (negative control) or 28 pl of Sepharose plus 2 pl of cycD1, Rc-mutD1, and/or SVRb vector for 4 hr, washed extensively in anti-pRb (RB-02, Pharmingen) cross-linked Sepharose. The immune phosphate-buffered saline (PBS), and fed 5 ml of DMEM plus 10% complexes were washed three times with ELBH50, separated by 10% calf serum. COS cell transfections for immunoprecipitation were PAGE, and transferred by Western blotting to nitrocellulose filters. Y&labeled at 48 hr after transfection. The filters were blocked in 5% nonfat milk, incubated with a 1:lOOO SAOS-2 cells (2 x lg) on 10 cm dishes were transfected for 16 dilution of an anti-pRb monoclonal antibody (RB-02, Pharmingen) or hr by the calcium-phosphate method (Chen and Okayama, 1987) in rabbit anti-cyclin Dl antiserum in PBS-T (PBS plus 0.05% Tween 20) 3% CO,. Transfections were performed with 15 ug of CMVRb plus 15 for >l hr, then washed three times in PBS-T, and incubated with a pg of Rc-CMV, Rc-cycDl, Rc-mutD1, Rc-cycDIA, or Rc-cycE vec- 1 :lOOO dilution donkey anti-mouse immunoglobulin G or donkey anti- tor and 1 ug of pBABEpuro, washed once in PBS, washed twice in rabbit immunoglobulin G conjugated to horseradish peroxidase (Jack- PBS plus calcium, washed once in DMEM plus serum, and fed DMEM son Laboratories) for 1 hr. The filter was washed, developed by chemi- plus serum for 24 hr prior to subsequent plating. For flat cell analysis, luminescence (Amersham), and exposed to XAR5 film (Kodak). 5 x 105 to 8 x 105 cells were plated in 10 cm dishes in duplicate, and puromycin (0.5 mg/ml) (Sigma) was added 24 hr later. Selection Recombinant Rb Beculovlrus Constructlon and was maintained for 10-14 days, and the plates were then stained with pRb Purlflcatlon crystal violet. The number of flat cells was quantitated by using a 1 The recombinant wild-type and A22 human Rb cDNAs were placed mm grid under 20 x magnification. Cells from 1.0-2.0 cm* of each under control of the polyhedrin promoter of the Autographa californica plate were counted and normalized to the number of cells per 5 x nuclear polyhedrosisvirus(AcNPV) bycotransfection of 1 ugof AcNPV Retinoblastoma Protein Interacts with D Cyclins 509

DNA and 2 ug of pVL3HNwtRb or pVL3HND22Rb into Spodoptera Chen, C., and Okayama, H. (1987). High-efficiency transformation of frugiperda (8%) insect cells. pVL3HNwtRb and pVL3HND22Rb were mammalian cells by plasmid DNA. Mol. Cell. Biol. 7, 2745-2752. generated by inserting a double-stranded 85-mer oligonuceotide into Chen, P.-L., Scully, P., Shew, J.-Y., Wang, J. Y. J., and Lee, W.-H. the 5’ end of the Rb cDNA and at the unique Eagl site in the open (1989). Phosphorylation of the retinoblastoma gene product is modu- reading frame (residue 11) of Rb, resulting in a deletion of the 5’ un- lated during the cell cycle and cellular differentiation. Cell 58, 1193- translated region followed by MGHHHG immediately in front of the 1198. endogenous initiating Met and inserted into the BamHl site of the Cherington, V., Brown, M., Paucha, E., St Johnston, L., Spiegelman, pVL941 baculovirus shuttle vector. The region was sequenced to as- B. M., and Roberts, T. M. (1988). Separation of simian virus 40 Iarge-T- sure that a full-length oligonucleotide had been inserted. antigen-transforming and origin-binding functions from the ability to Recombinant pRb was isolated by passing a nuclear extract from block differentiation. Mol. Cell. Biol. 8, 1380-1384. High 5 Trichoplusia nickel (581-4) cabbage looper cells (Invitrogen) infected with the Rb baculoviruses over a single nickel-charged nitrilo- Chittenden, T., Livingston, D. M., and Kaelin, W. G., Jr. (1991). The triacetic acid column (Qiagen) in nondenaturing buffer containing 1 T/ElA-binding domain of the retinoblastoma product can interact se- mM imidazole (20 mM HEPES [pH 7.2],250 mM NaCI, 0.1% NP40,l lectively with a sequence-specific DNA-binding protein, Cell 65, 1073- tug/ml aprotinin, 1 pglml leupeptin, and 50 ug/ml phenylmethylsulfonyl 1082. fluoride). The column was extensively washed in 1 mM imidazole fol- DeCaprio, J. A., Ludlow, J. W., Figge, J., Shew, J.-Y., Huang, C.-M., lowed by a step-wise gradient of increasing concentrations of imidaz- Lee, W.-H., Marsilio, E., Paucha, E., and Livingston, D. M. (1988). ole (5. 10, 20, 30, 50, and 100 mM). The recombinant Rb protein SV40 large tumor antigen forms a specific complex with the product releases from the nickel-nitrilotriacetic acid resin at 20 mM imidazole. of the retinoblastoma susceptibility gene. Cell 54, 275-283. The purified pRb was extensively dialyzed against PBS, 0.1% NP-40, DeCaprio, J. A., Ludlow, J. W., Lynch, D., Furukawa, Y., Griffin, J., 1 mM dithiothreitol and frozen in 10% glycerol. Piwnica-Worms, H., Huang, C-M., and Livingston, D. M. (1989). The product of the retinoblastoma susceptibility gene has properties of a In Vitro Protein Binding cell cycle regulatory element. Cell 58, 1085-1095. In vitro translated “S-labeled cyclins were produced by programming DeCaprio, J. A., Furukawa, Y., Ajchenbaum, F., Griffin, J. D., and reticulocyte lysates (Promega) with in vitro transcribed RNA from hu- Livingston, D. (1992). The retinoblastoma-susceptibility gene product man cyclins A, El, 82, C, Dl, D3, and E. The programmed reticulocyte becomes phosphorylated in multiple stages during cell cycle entry and lysates (10 al) were mixed with 50 ng of purified unlabeled pRb or progression. Proc. Natl. Acad. Sci. USA 89, 1795-1798. pA22Rb in 500 nl of ELB on ice for30 min. The mixwas then precleared DefeoJones, D., Huang, P. S., Jones, R. E., Haskell, K. M., Vuocolo, and immunoprecipitated as described above with anti-pRb antibody G. A., Hanobik, M. G., Huber, H. E., and Oliff, A. (1991). Cloning of 21C9, X255, or both (Hu et al., 1991). Coimmunoprecipitating cDNAs for cellular proteins that bind to the retinoblastoma gene prod- “S-labeled cyclinswere assayed by PAGE and fluorography. Competi- uct. Nature 35.2, 251-254. tion experiments with EIA conserved region I and II-derived peptides Dulic, V.. Lees, E., and Reed, S. I. (1992). Association of human cyclin werecarriedout bytheadditionof 50and 150 aMwild-typeElApeptide (residues [37-49]0[117-125]E]127-1321; Dyson et al., 1992) or mu- E with a periodic GI-S phase protein kinase. Science 257, 1958- tant Kl-EIA peptide (residues [37-49]Q]117-125]K[127-132)) to the 1961. in vitro mix just prior to the addition of the programmed %-labeled Dyson, N., Howley, P. M., Miinger, K., and Harlow, E. (1989). The reticulocyte lysates. respectively. human papilloma virus-18 E7 oncoprotein is able to bind to the retinoblastoma gene product. Science 243, 934-937. Acknowledgments Dyson, N., Guida, P., McCall, C., and Harlow, E. (1992). Adenovirus EIA makes two distinct contacts with the retinoblastoma protein. J. We are indebted to Drs. S. Gruenwald for the gift of cross-linked anti- Virol. 66, 4808-4611. pRb Sepharose; D. Morgan for the cyclin A and cdk2 baculoviruses Ewen, M. E., Xing, Y., Lawerence, J. B., and Livingston, D. M. (1991). gifts; T. Motokura for the cyclin D3 cDNA gift; V. DuliC for the anti-cyclin Molecular cloning, chromosomal mapping, and expression of the Dl antisera gift; and E. Harlow for the XZ55 monoclonal antibody cDNA for ~107, a retinoblastoma gene product-related protein. Cell gift. We thank Drs. T. Meeker, N. Dyson, M. Ewen, C. Sherr, and D. 66,1155-l 164. Livingston for communicating results prior to publication. We also Ewen, M. E., Sluss, H. K., Sherr, C. J., Matsushime, H., Kato, J.-y., and thank S. Egan and D. Cobrinik for critical comments on the work. This Livingston, D. M. (1993). Functional interactions of the retinoblastoma work was supported by a grant from the American Cancer Society protein with mammalian D-type cyclins. Cell 73, this issue. (CD-355C) (R. A. W.), by grant CA55909 from the National Cancer Figge, J., Webster, T., Smith, T. F., and Paucha, E. (1988). Prediction Institute (A. A.), by an American Cancer Society Faculty Research Award (A. A.), by Public Health ServicegrantsGM38328 andGM48008 of similar transforming regions in simian virus 40 large T, adenovirus (S. I. R), and by grant 3185 from the Council for Tobacco Research ElA, and Myc oncoproteins. J. Viral. 62, 1814-1818. (S. I. R.). S. F. D. was supported by a postdoctoral fellowship from Friend, S. H., Horowitz, J. M., Gerber, M. R., Wang, X. F., Bogenmann, the Damon Runyon-Walter Winchell Cancer Research Fund (DRG E., Li, F. P., and Weinberg, R. A. (1987). Deletions of a DNA sequence 1072). R. A. W. is an American Cancer Society Research Professor. in retinoblastomas and mesenchymal tumors: organization of the sequence and its encoded protein. Proc. Natl. Acad. Sci. USA 84,9059- Received December 11, 1992; revised March 5, 1993. 9063. Erratum: Proc. Natl. Acad. Sci. USA 85(7), 2234. Furukawa, Y., DeCaprio, J. A., Belvin, M., and Griffin, J. D. (1991). References Heterogenous expression of the product of the retinoblastoma susceptibility gene in primary human leukemia cells. Oncogene 6, 1343- Bandara, L. R., and LaThangue, N. B. (1991). Adenovirus ElA pre- 1348. vents the retinoblastoma gene product from complexing with a cellular Goodrich, D. W., Ping Wang, N., Qian, Y.-W., Lee, E. Y.-H. P., and Lee, transcription factor. Nature 357, 494497. W.-H. (1991). The retinoblastoma gene product regulates progression Buchkovich, K., Duffy, L. A., and Harlow, E. (1989). The retinoblastoma through the Gl phase of the cell cycle. Cell 67, 293-302. protein is phosphorylated during specific phases of the cell cycle. Cell Gu, W., Schneider, J. W., Condorelli, G., Kaushal, S., Mahdavi, V., 58, 1097-l 105. and Nadal-Ginard, B. (1993). Interaction of myogenic factors and the Chang, F., and Herskowitz, I. (1990). Identification of agene necessary retinoblastoma protein mediates muscle cell commitment and differen- for cell cycle arrest by a negative growth factor of yeast: FAR1 is an tiation. Cell 72, 309-324. inhibitor of a Gl cyclin, CLNP. Cell 63, 999-1011. Harbour, J. W., Lai, S. L., Whang, P. J., Gazdar, A. F., Minna, J. D., Chellappan, S. P., Hiebert, S.. Mudryj, M., Horowitz, J. M., and Nevins, and Kaye, F. J. (1988). Abnormalities in structure and expression of J. R. (1991). The E2F transcription factor is a cellular target for the the human retinoblastoma gene in SCLC. Science 247, 353-357. RB protein. Cell 65, 1053-1081. Harlow, E., Whyte, P., Franza, B. R., Jr., and Schley, C. (1987). Associ- Cell 510

ation of adenovirus early-region IA proteins with cellular polypeptides. Green, M. R. (1992). Retinoblastoma gene product activates expres- Mol. Cell. Biol. 6, 1579-1589. sion of the human 7GF#2 gene through transcription factor ATF-2. Helin, K., Lees, J. A., Vidal, M., Dyson, N., Harlow, E., and Fattaey, Nature 358, 331-334. A. (1992). A cDNA encoding a pRB-binding protein with properties of Koff, A., Cross, F., Fisher, A., Schumacher, J., Leguellec, K., Philippe, the transcription factor E2F. Cell 70, 337350. M., and Roberts, J. M. (1991). Human cyclin E, a new cyclin that Hinds, P. W., Finlay, C. A., Quartin, R. S., Baker, S. J., Fearon, interacts with two members of the CDC2 gene family. Cell 66, 1217- E. R., Vogelstein. B., and Levine, A. J. (1990). Mutant ~53 DNA clones 1228. from human colon carcinomas cooperate with ras in transforming pri- Koff, A., Giordano, A., Desai, D., Yamashita, K., Harper, J. W., Elledge, mary rat cells: a comparison of the “hot spot” mutant phenotypes. Cell S., Nishimoto, T., Morgan, D. O., Franza, B. R., and Roberts, J. M. Growth Differ. 7, 571-580. (1992). Formation and activation of a cyclin E-cdk2 complex during Hinds, P. W., Mittnacht, S., Dulic, V., Arnold, A., Reed, S. I., and the Gl phase of the human cell cycle. Science 257, 1889-1894. Weinberg, R. A. (1992). Regulation of retinoblastoma protein functions Lammie, G. A., Fantl, V., Smith, R.,Schuuring, E., Brookes,S., Michal- by ectopic expression of human cyclins. Cell 70, 993-1008. ides, R., Dickson, C., Arnold, A., and Peters, G. (1991). DllS287, a Horowitz, J. M., Yandell, D. W., Park, S. H., Canning, S., Whyte, P., putative oncogene on chromosome 1 iql3, is amplified and expressed Buchkovich, K., Harlow, E., Weinberg, R. A., and Dryja, T. P. (1989). in squamous cell and mammary carcinomas and linked to BCL-1. On- Point mutational inactivation of the retinoblastoma anti-oncogene. Sci- cogene 6,439~444. ence 243,937~940. Lee, W.-H., Shew, J.-Y., Hong, F. D., Sery, T. W., Donoso, L. A., Horowitz, J. M., Park, S. H., Bogenmann, E., Cheng, J. C., Yandell, Young, L.J., Bookstein, R., and Lee, E. Y.-H. P. (1987). The retinoblas- D. W., Kaye, F. J., Minna, J. D., Dryja, T. P., and Weinberg, R. A. toma susceptibility gene encodes a nuclear phosphoprotein associ- (1990). Frequent inactivation of the retinoblastoma anti-oncogene is ated with DNA binding activity. Nature 329, 842-845. restricted to a subset of human tumor cells. Proc. Natl. Acad. Sci. Lees, J. A., Buchkovich, K. J., Marshak, D. R., Anderson, C. W., and USA 87, 2775-2779. Harlow, E. (1991). The retinoblastoma protein is phosphorylated on Hu, Q., Dyson, N., and Harlow, E. (1990). Regions of the retinoblas- multiple sites by human cdc2. EMBO J. 70, 4279-4290. toma protein needed for binding to adenovirus EIA or SV40 Large T Lew, D. J., Dulic, V., and Reed, S. I. (1991). Isolation of three novel antigen are common sites for mutations. EMBO J. 9, 1147-1155. human cyclins by rescue of Gl cyclin (Cln) function in yeast. Cell 66, Hu, Ct., Bautista, C., Edwards, G. M., DefeoJones, D., Jones, R. E., 1197-1208. and Harlow, E. (1991). Antibodies specific for the human retinoblas- Lillie, J. W., Loewenstein, P. M., Green, M. R., and Green, M. (1987). toma protein identify a family of related polypeptides. Mol. Cell. Biol. Functional domains of adenovirus type 5 El a proteins. Cell 50,1091- 11, 5792-5799. 1100. Hu, Ct., Lees, J. A., Buchkovich, K. J., and Harlow, E. (1992). The Lin, B. T.-Y., Gruenwald, S., Morla, A. O., Lee, W.-H., and Wang, retinoblastoma protein physically associates with the human cdc2 ki- J. Y. J. (1991). Retinoblastoma cancer suppressor gene product is a nase. Mol. Cell. Biol. 12, 971-980. substrate of the cell cycle regulator cdc2 kinase. EMBO J. 10, 857- Huang, H.J. S., Yee, J.-K., Shew, J.-Y., Chen, P.-L., Bookstein, R., 884. Friedmann, T., Lee, E. Y.-H. P., and Lee, W.-H. (1988). Suppression Lopata, M. A., Cleveland, D. W., and Sollner-Webb, 8. (1984). High of the neoplastic phenotype by replacement of the Rb gene in human level transient expression of a chloramphenicol acetyl transferase cancer cells. Science 242, 1583-l 588. gene by DEAE-dextran mediated DNA transfection coupled with a Huang, S., Wang, N.-P., Tseng, B. Y., Lee, W.-H., and Lee, E. Y.-H. P. dimethyl sulfoxide or glycerol shock treatment. Nucl. Acids Res. 72, (1990). Two distinct and frequently mutated regions of retinoblastoma 5707-5717. protein are required for binding to SV40 T antigen. EMBO J. 9, 1815 Ludlow, J. W., DeCaprio, J. A., Huang, C.-M., Lee, W.-H., Paucha, 1822. E., and Livingston, D. M. (1989). SV40 large T antigen binds preferen- Huang, S., Lee, W.-H., and Lee, E. Y.-H. (1991). A cellular protein tially to an underphosphorylated member of the retinoblastoma sus- that competes with SV40 T antigen for binding to the retinoblastoma ceptibility gene product family. Cell 56, 57-85. gene product. Nature 350, 180-182. Mailer, J. L. (1991). Mitotic control. Curr. Opin. Cell Biol. 3, 289-275. Hunt, T. (1989). Maturation promoting factor, cyclin and the control Matsushime, H., Roussel, M. F., Ashmun, R. A., and Sherr, C. J. of M-phase. Curr. Opin. Cell Biol. 7, 288-274. (1991). Colony-stimulating factor 1 regulates novel cyclins during the Hunter, T., and Pines, J. (1991). Cyclins and cancer. Cell 66, 1071- Gl phase of the cell cycle. Cell 65, 701-713. 1074. Matsushime, H., Ewen, M. E., Strom, D. K., Kato, J.-Y., Hanks, S. K., Imai, Y., Matsushima, Y., Sugimura, T., and Terada, M. (1991). Purifi- Roussel, M. F., and Sherr, C. J. (1992). Identification and properties cation and characterization of human papillomavirus type 18 E7 pro- of an atypical catalytic subunit (~34 PSKJ3/cdk4)for mammalian D type teinwithpreferential bindingcapacitytotheunderphosphorylatedform Gl cyclins. Cell 71, 323-334. of the retinoblastoma gene product. J. Virol. 65, 4988-4972. Mihara, K., Cao, X. R., Yen, A., Chandler, S., Driscoll, B., Murphree, Inaba, T., Matasushime, Y., Valentine, M., Roussel, M. F., Sherr, A. L., Tang, A., and Fung, Y. K. (1989). Cell cycle-dependent regula- C. J., and Look, A. T. (1992). Genomic organization, chromosomal tion of phosphorylation of the human retinoblastoma gene product. localization, and independent expression of human cyclin D genes. Science 246, 1300-I 303. Genomics 13, 585-574. Mittnacht, S., and Weinberg, R. A. (1991). GllS phosphorylation of Kaelin, W. G., Jr., Ewen, M. E., and Livingston, D. M. (1990). Definition the retinoblastoma protein is associated with an altered affinity for the of the minimal simian virus 40 large T antigen- and adenovirus ElA- nuclear compartment. Cell 65, 381-393. binding domain in the retinoblastoma gene product. Mol. Cell. Biol. Mittnacht, S., Hinds, P. W., Dowdy, S. F., and Weinberg, R. A. (1991). 70, 3781-3789. Modulation of retinoblastomaprotein activity during the cell cycle. Cold Kaelin, W. G., Jr., Pallas, D. C., DeCaprio, J. A., Kaye, F. J., and Spring Harbor Symp. &ant. Biol. 56, 197-209. Livingston, D. M. (1991). Identification of cellular proteins that can Moran, E. (1988). A region of SV40 large T antigen can substitute for interact specifically with the T/ElA-binding region of the retinoblas- a transforming domain of the adenovirus EIA products. Nature 334, toma gene product. Cell 64, 521-532. 168-170. Kaelin, W. G., Jr., Krek, W., Sellers, W. R., DeCaprio, J. A., Ajchen- Moran, E., Zerler, B., Harrison, T. M., and Mathews, M. B. (1986). baum, F., Fuchs, C. S., Chittenden, T., Li, Y., Farnham, P. J., Blanar, Identification of separate domains in the adenovirus ElA gene for M. A., Livingston, D. M., and Flemington, E. K. (1992). Expression immortalization activity and the activation of virus early genes. Mol. cloning of a cDNA encoding a retinoblastoma-binding protein with Cell. Biol. 6, 3470-3480. EPF-like properties. Cell 70, 351384. Morgenstern, J. P., and Land, H. (1990). Advanced mammalian gene Kim, S.J., Wagner, S., Liu, F., D’Reily, M. A., Robbins, P. D., and transfer: high titre retroviral vectors with multiple drug selection mark- Retinoblastoma Protein Interacts with D Cyclins 511

ers and a complementary helper-free packaging cell line. Nucl. Acids Stein, R. W.. Corrigan, M., Yaciuk, P., Whelan, J., and Moran, E. Res. 18, 3567-3596. (1990). Analysis of El A-mediated growth regulation functions: binding Motokura, T., Bloom, T., Kim, H. G., Juppner, H., Ruderman, J. V., of the 300 kilodalton cellular product correlates with ElA enhancer Kronenberg, H. M., and Arnold, A. (1991). A BCL-1 linked candidate repression function and DNA synthesis-inducing activity. J. Virol. 64, oncogene which is rearranged in parathyroid tumors encodes a novel 4421-4427. cyclin. Nature 350, 512-516. Templeton, D. J., Park, S.-H., Lanier, L., and Weinberg, R. A. (1991). Motokura,T., Keyomarsi, K., Kronenberg, H. M., andArnold,A.(1992). Nonfunctional mutants of the retinoblastoma protein are characterized Cloning and characterization of human cyclin D3, a cDNA closely re- by defects in phosphorylation, viral oncoprotein association, and nu- lated in sequence to the PRADllcyclin D1 proto-oncogene. J. Biol. clear tethering. Proc. Natl. Acad. Sci. USA 88, 3033-3037. Chem. 267, 20412-20415. Varley, J. M., Armour, J., Swallow, J. E., Jeffrey% A. J., Ponder, Murray, A. W. (1992). Creative blocks: cell-cycle checkpoints and feed- B. A., T’ang, A., Fung, Y. K., Brammar, W. J., and Walker, R. A. back controls. Nature 359, 599-604. (1989). The retinoblastoma gene is frequently altered leading to loss of expression in primary breast tumours. Oncogene 4, 725-729. Nurse, P. (1990). Universal control mechanism regulating onset of M-phase. Nature 344, 503-506. Weinberg, R. A. (1992). Tumor suppressor genes. Science 254,1138- 1146. Pardee, A. B. (1974). A restriction point for control of normal animal cell proliferation. Proc. Natl. Acad. Sci. USA 77, 1288-1290. Weintraub, S. J., Prater, C. A., and Dean, D. C. (1992). Retinoblastoma protein switches the E2F site from positive to negative element. Nature Peter, M., Gartner, A., Horecka, J., Ammerer, G., and Herskowitz. I. 356, 259-261. (1993). FAR1 links the signal transduction pathway to the cell cycle machinery in yeast. Cell 73, in press. Whyte, P., Buchkovich, K. J., Horowitz, J. M., Friend, S. H., Raybuck, M., Weinberg, R. A., and Harlow, E. (1988). Association between an Ray, S. K.,Arroyo, M., Bagchi, S., and Raychaudhuri, P. (1992). Identi- oncogene and an anti-oncogene: the adenovirus EIA proteins bind fication of a 60-kilodalton Rb-binding protein, RBPGO, that allows the to the retinoblastoma gene product. Nature 334, 124-129. Rb-EPF complex to bind DNA. Mol. Cell. Biol. 72, 4327-4333. Whyte, P., Williamson, N. M., and Harlow, E. (1989). Cellular targets Reed, S. I. (1980). The selection of S. cefevisiae mutants defective in for transformation by the adenovirus ElA proteins. Cell 58, 67-75. the START event of cell division. Genetics 95, 561-587. Williams, M. E., Swerdlow, S. H., Rosenberg, C. L., and Arnold, A. Reed, S. I. (1991). Gl specific cyclins: in search of an S-phase promot- (1993). Chromosome 11 translocation breakpoints at the PRADl/ ing factor. Trends Genet. 7, 95-99. cyclin Dl gene locus in centrocytic lymphoma. Leukemia, in press. Rosenberg, C. L., Kim, H. G., Shows, T. B., Kronenberg, H. M., and Withers, D. A., Harvey, R. C., Faust, J. B., Melnyk, O., Carey, K., and Arnold, A. (1991a). Rearrangement and overexpression of Dl lS287E, Meeker, T. C. (1991). Characterization of a candidate bcl-7 gene. Mol. a candidate oncogene on chromosome 1 lq13 in benign parathyroid tumors. Oncogene 6, 449-453. Cell. Biol. 77, 4846-4853. Won, K.-A., Xiong, Y., Beach, D., and Gilman, M. 2. (1992). Growth- Rosenberg, C. L., Wong, E., Petty, E. M., Bale, A. E., Tsujimoto, Y., regulated expression of D-type cyclin genes in human diploid fibro- Harris, N. L., and Arnold, A. (1991 b). PRADl, acandidate BCLl onco- blasts. Proc. Natl. Acad. Sci. USA 89, 9910-9914. gene: mapping and expression in centrocytic lymphoma. Proc. Natl. Acad. Sci. USA 88, 9638-9642. Xiong, Y., Connolly, T., Futcher, B., and Beach, D. (1991). Human D-type cyclin. Cell 65, 691-899. Rosenberg, C. L., Motokura, T., Kronenberg, H. M., and Arnold, A. (1993). Coding sequence of the overexpressed transcript of the puta- Xiong, Y., Menninger, J., Beach, D., and Ward, D. C. (1992a). Molecu- tive oncogene PRADllcyclin Dl in two primary human tumors. Onco- lar cloning and chromosomal mapping of CCND genes encoding hu- gene, in press. man D-type cyclins. Genomics 73, 575-584. Rustgi, A. K., Dyson, N. J., and Bernards, R. (1992). Amino-terminal Xiong, Y., Zhang, H., and Beach, D. (1992b). D type cyclins associate domains of C-myc and N-myc proteins mediate binding to the retino- with multiple protein kinases and the DNA replication and repair factor blastoma gene product. Nature 352, 541-544. PCNA. Cell 71, 505-514. Schuuring, E., Verhoeven, E., Mooi, W. J., and Michalides, J. A. M. (1992). Identification and cloning of two overexpressed genes, U21 BIllPRADl and EMSl, within the amplified chromosome 1 lq13 region in human carcinomas. Oncogene 7, 355-361. Seto, M., Yamamoto, K., lida, S., Akoa, Y., Utsumi, K., Kubonishi, I., Miyoshi, I., Ohtsuki, T., Yawata, Y., Namba, M., Motokura, T., Arnold, A., Takahashi, T., and Ueda, R. (1992). Gene rearrangement and overexpression of PFfADl in lymphoid malignancy with 1(11;14)(q13;q32) translocation. Oncogene 7, 1401-1406. Shaw, G., and Kamen, R. (1986). A conserved AU sequence from the 3’ untranslated region of GM-CSF mRNA mediates selective mRNA degradation. Cell 46, 659-867. Shew, J.-Y., Ling, N., Yang, X. M., Fodstad, O., and Lee, W.-H. (1989). Antibodies detecting abnormalities of the retinoblastoma susceptibility gene product (ppl10RB) in osteosarcomas and synovial sarcomas. Oncogene Res. 4, 205-214. Shew. J.-Y., Lin, B. T., Chen, P. L.. Tseng, B. Y., Yang, F. T., and Lee, W.-H. (1990). C-terminal truncation of the retinoblastoma gene product leads to functional inactivation. Proc. Natl. Acad. Sci. USA 87, 6-10. Shirodkar, S., Ewen, M., DeCaprio, J. A., Morgan, J., Livingston, D. M., and Chittenden, T. (1992). The transcription factor E2F interacts with the retinoblastoma product and a pl07-cyclin A complex in a cell cycle-regulated manner. Cell 68, 157-166. Smith, D. H., and Ziff, E. B. (1988). The amino-terminal region of the adenovirus serotype 5 ElA protein performs two separate functions when expressed in primary baby rat kidney cells. Mol. Cell. Biol. 8, 3882-3890.