Cyclins are regulatory molecules that undergo periodic accumulation and destruction during each cell cycle. By activating p34cdk2 and related kinase subunits they control important events required for normal cell cycle progression. Cyclin A, for example, regulates at least two distinct kinase subunits, the mitotic kinase subunit p34cdk2 and related subunit p33cdk2, and is widely believed to be necessary for progression through S phase. However, cyclin A also forms a stable complex with the cellular transcription factor DRTF1 and thus may perform other functions during S phase. DRTF1, in addition, associates with the tumour suppressor retinoblastoma (Rb) gene product and the Rb-related protein p107. We now show, using biologically active fusion proteins, that cyclin A can direct the binding of the cdc2-like kinase subunit, p33cdk2, to complexed DRTF1, containing either Rb or p107, as well as activate its histone Hl kinase activity. Cyclin A cannot, however, direct p34cdk2 to the DRTF1 complex and we present evidence suggesting that the stability of the cyclin A-p33cdk2 complex is influenced by DRTF1 or an associated protein. Cyclin A, therefore, serves as an activating and targeting subunit of p33cdk2. The ability of cyclin A to activate and recruit p33cdk2 to DRTF1 may play an important role in regulating cell cycle progression and moreover defines a mechanism for coupling cell-cycle events to transcriptional initiation.

Cyclins are an evolutionarily conserved group of regulatory molecules that undergo periodic accumulation and destruction during the cell cycle and are required for normal cell cycle progression (Hunt, 1989). They were initially identified in marine invertebrates (Evans et al., 1983), and act by binding to and regulating the activity of the mitotic kinase catalytic subunit p34cdk2 or related kinase subunits (Pines and Hunter, 1990a). Cyclins, which have been isolated and characterised from a variety of sources, are grouped into several classes. Cyclins of the B class, which function at mitosis, peak towards M phase and are responsible for activating the mitotic kinase subunit p34e*2 (Draetta et al., 1989; Minshull et al., 1990). Cyclin A, however, accumulates somewhat earlier (Giordano et al., 1989; Minshull et al., 1990; Pines and Hunter, 1990b) and in addition to p34cdk2 activates the related kinase subunit p33Cdk2 (Pines arid Hunter, 1990b; Tsai et al., 1991). Other cyclins have recently been defined which are thought to regulate progression through earlier phases of the cell cycle (Mat-sushime et al., 1991; Xiong et al., 1991; Lew et al., 1991; Koff et al., 1991).

Several reports have suggested that cyclin A regulates progression through S phase. In Xenopus egg extracts, p33cdk2 or a closely related kinase subunit is required for DNA replication (Fang and Newport, 1991) and is believed to fulfill a similar role in other types of cells (Pines and Hunter, 1991). Moreover, inactivation of cyclin A, either through antibody or antisense approaches, prevents cells from completing S phase (Girard et al., 1991). It is likely, therefore, that cyclin A regulates the kinase activity of p33cÆ2 which in turn modifies protein substrates that are necessary for cell-cycle progression.

The retinoblastoma (Rb) gene product is a negative regulator of the cell cycle that acts by controlling progression through Gi (Goodrich et al., 1991). This effect is widely believed to result from the activity of the un- or under-phosphorylated Rb protein, because this form of the protein predominates during Gi (Buchkovich et al., 1989; DeCaprio et al., 1989). Certain viral oncoproteins, such as adenovirus El a, SV40 large T antigen and HPV E7 bind the Rb protein (Whyte et al., 1988; DeCaprio et al., 1988; Dyson et al., 1989) and the Rb gene is frequently mutated in tumour cells (Hu et al., 1990; Huang et al., 1990). Both of these processes are thought to inactivate the growth suppressing properties of the Rb protein by preventing it from regulating cellular targets. Moreover, the Rb-related protein, p107, is also sequestered by adenovirus Ela and SV40 large T antigen (Whyte et al., 1988; DeCaprio et al., 1988) and, although there is no detailed information available yet on its regulation during the cell cycle, is likely to have an important role in cell-cycle control (Ewen et al., 1991).

It is thought that the Rb protein mediates its growth regulating properties by modulating the activity of molecules that regulate progression through the cell cycle, but only recently have potential candidates been identified (Bandara and La Thangue, 1991; Bagchi et al., 1991; Chellapan et al., 1991; Chittenden et al., 1991; Defeo-Jones et al., 1991). A particularly good candidate is the cellular transcription factor DRTF1 which is found in complexes that contain the Rb protein (Bandara and La Thangue, 1991). DRTF1, first defined in F9 embryonal carcinoma (EC) stem cells as a differentiation-regulated transcription factor (La Thangue and Rigby, 1987), is mostly uncomplexed in F9 EC cells, but in other cell-types forms a stable complex with the Rb protein (Bandara and La Thangue, 1991). The Rb-DRTFl complex is dissociated by Ela, large T antigen and E7 proteins (Bandara and La Thangue, 1991; Bandara and La Thangue, data not shown), which releases the transcriptionally active protein, the form that predominates in stem cells. Moreover, the protein products of all naturally occurring mutant Rb alleles so far studied fail to bind stably to DRTF1 (Bandara et al., 1991; Bandara and La Thangue, data not shown), underscoring the potential importance of DRTF1 in controlling cellular proliferation.

Recent experiments have shown that the Rb protein can specifically repress the transcription of promoters driven by DRTF1, whereas proteins encoded by some naturally-occurring mutant Rb alleles, which fail to bind DRTF1 in vitro, cannot. Furthermore, Rb-mediated transcriptional repression is relieved by the adenovirus Ela protein (Zamanian and La Thangue, 1992). The Rb protein, therefore, represses the transcriptional activity of DRTF1 which thus provides a potential mechanistic explanation for how the Rb protein exerts its growth regulating properties, given that DRTF1 binding sites occur in the promoters of genes that are necessary for cell cycle progression (Blake and Azizkhan, 1989; Pearson et al., 1991).

DRTF1 binds to a similar DNA sequence to that of the HeLa cell transcription factor E2F, and although the exact relationship between these two transcription factors has yet to be established, they are likely to be related because similar interactions have been defined for E2F (Bagchi et al., 1990; Chellappan et al., 1991).

Cyclin A also binds to DRTF1 (Bandara et al., 1991) and previously we have suggested that this allows cell cycle events to be coupled to transcription. We have developed an in vitro assay in which cyclin A efficiently binds to DRTF1 and established that cyclin A can direct the cdc2-like kinase subunit, p33cdk2, but not the mitotic kinase subunit, p34cdc2, to DRTF1. We suggest that such a kinase will play an important role in regulating the activity of other non DNA-binding proteins, such as the Rb protein and p107, in the DRTF1 transcription factor complex and hence in regulating its transcriptional activity. We also present evidence that the stability of the cyclin A-p33cdk2 complex is influenced by DRTF1 or a component within this transcription factor complex.

Cyclin A targets the cdc2-related kinase subunit p33cdk2 to DRTF1 complexes that contain either the Rb protein or p107

Cyclin A is part of the DRTF1 transcription factor complex (Bandara et al., 1991). In order to investigate the significance of this interaction we developed an in vitro association assay for cyclin A and DRTF1. Initially, we performed the assay in JM cell extracts where most DRTF1 is bound to the Rb protein (complexed DRTF1 is referred to as DRTFla) and assessed whether a biologically active cyclin A fusion protein (PA-CA), containing amino acid residues 77 to 432 (Fig. 1b), would assemble with JM DRTFla. Indeed, PA-CA efficiently assembles with DRTFla, because upon addition of PA-CA to the cell extract the Rb-DRTFl complex migrates with slower mobility (Fig.la, compare tracks 2 and 3 with 4 and 5; PACA dependent complex indicated by ▪). This effect is only apparent with DRTFla since PA-CA fails to assemble with uncomplexed DRTF1, referred to as DRTFlb (see Fig.2). The binding reaction also requires the integrity of the cyclin box because some derivatives of PA-CA that were mutated in this region fail to assemble with DRTFla (discussed later). Cyclin A, therefore, contains a domain that enables it to associate with DRTFla and moreover this association occurs efficiently in vitro.

Fig. 1.

Cyclin A targets cdk2 to complexed DRTF1. (a) Gel retardation (Shivji and La Thangue, 1991) was performed in a JM whole cell extract (about 2 μg) either before (tracks 2, 3, 14, 15, 16 and 17) or after the addition of PA-CA (about 0.5 μg; tracks 4, 5, 8, 9, 10, 11, 12 and 13) or GST-cdk2 (about 0.5 μg; tracks 6, 7, 8, 9, 10, 11,12 and 13) in the presence of an anti-Rb, C36 (Whyte et al., 1988), (tracks 12, 13, 16 and 17) or control (tracks 10, 11, 16 or 17) monoclonal antibody; the probe (containing Ad5 E2A promoter sequences from −71 to −50; La Thangue et al, 1990) alone is shown in track 1. DRTFla is shifted by the anti-Rb monoclonal antibody to ○ (compare tracks 14 and 15 with 16 and 17) and ▪ indicates the PA-CA dependent complex (tracks 4 and 5). Note that the cdk-dependent complex, ▴ (tracks 8 and 9), is supershifted to • by the anti-Rb antibody (compare tracks 10 and 11 with 12 and 13). (b) Structure of fusion proteins used in this experiment. PA-CA was purified as previously described (Bandara et al., 1991). GST-cdk2 contains the entire cdk2 coding sequence (Tsai et al., 1991) fused to glutathione-S-transferase and was purified by affinity chromatography on glutathione Sepharose followed by FPLC on a MonoQ column.

Fig. 1.

Cyclin A targets cdk2 to complexed DRTF1. (a) Gel retardation (Shivji and La Thangue, 1991) was performed in a JM whole cell extract (about 2 μg) either before (tracks 2, 3, 14, 15, 16 and 17) or after the addition of PA-CA (about 0.5 μg; tracks 4, 5, 8, 9, 10, 11, 12 and 13) or GST-cdk2 (about 0.5 μg; tracks 6, 7, 8, 9, 10, 11,12 and 13) in the presence of an anti-Rb, C36 (Whyte et al., 1988), (tracks 12, 13, 16 and 17) or control (tracks 10, 11, 16 or 17) monoclonal antibody; the probe (containing Ad5 E2A promoter sequences from −71 to −50; La Thangue et al, 1990) alone is shown in track 1. DRTFla is shifted by the anti-Rb monoclonal antibody to ○ (compare tracks 14 and 15 with 16 and 17) and ▪ indicates the PA-CA dependent complex (tracks 4 and 5). Note that the cdk-dependent complex, ▴ (tracks 8 and 9), is supershifted to • by the anti-Rb antibody (compare tracks 10 and 11 with 12 and 13). (b) Structure of fusion proteins used in this experiment. PA-CA was purified as previously described (Bandara et al., 1991). GST-cdk2 contains the entire cdk2 coding sequence (Tsai et al., 1991) fused to glutathione-S-transferase and was purified by affinity chromatography on glutathione Sepharose followed by FPLC on a MonoQ column.

Fig. 2.

Cyclin A targets cdk2 to DRTF1 complexes that contain either GST-Rb or GST-p107. (a) Gel retardation was performed in an F9 EC cell extract either before (tracks 2 and 3) or after (about 0.5 μg; tracks 4, 5, 12, 13, 14,15, 16 and 17) the addition of PA-CA, GST-Rb (about 0.06 μg; tracks 6, 7, 14 and 15), GST-p107 (about 0.02 μg; tracks 8, 9, 16 and 17) or GST-cdk2 (about 0.5 μg; tracks 10, 11, 12, 13, 14, 15, 16 and 17); track 1 shows the probe alone. The cdk2-dependent complex (δ), which migrates as a doublet in the presence of GST-Rb or GST-p107, is apparent in tracks 12 through 17. (b) Structure of fusion proteins used in this experiment. GST-p107 was constructed using a sense oligonucleotide primer for amino acid residues 249-257(5′-CGGGATCCGCAGTC-ATTACTCCTGTTGCATCAGCC) and an anti-sense primer covering amino acid residues 690-732 (5′-GGACAGTGAACTA-AAGTGAATTCAAAAACACG) in a polymerase chain reaction (PCR), using a p107 cDNA (Ewen et al., 1991) as template. In a second PCR reaction, a sense primer for amino acid residues 690-732 (5′-CGTGTTT-TGAATTCACTTTAGTTCACTGTCC) and an antisense primer for amino acid residues 928-936 (5′-CGGGATCCTTAATGATTTG-CTCTTTCACTGAC) were used. Both PCR products were purified, digested with SamHI and EcoRI, ligated and cloned into pGEX-2T. GST-p107 and GST-Rb were purified as previously described (Smith and Johnson, 1988).

Fig. 2.

Cyclin A targets cdk2 to DRTF1 complexes that contain either GST-Rb or GST-p107. (a) Gel retardation was performed in an F9 EC cell extract either before (tracks 2 and 3) or after (about 0.5 μg; tracks 4, 5, 12, 13, 14,15, 16 and 17) the addition of PA-CA, GST-Rb (about 0.06 μg; tracks 6, 7, 14 and 15), GST-p107 (about 0.02 μg; tracks 8, 9, 16 and 17) or GST-cdk2 (about 0.5 μg; tracks 10, 11, 12, 13, 14, 15, 16 and 17); track 1 shows the probe alone. The cdk2-dependent complex (δ), which migrates as a doublet in the presence of GST-Rb or GST-p107, is apparent in tracks 12 through 17. (b) Structure of fusion proteins used in this experiment. GST-p107 was constructed using a sense oligonucleotide primer for amino acid residues 249-257(5′-CGGGATCCGCAGTC-ATTACTCCTGTTGCATCAGCC) and an anti-sense primer covering amino acid residues 690-732 (5′-GGACAGTGAACTA-AAGTGAATTCAAAAACACG) in a polymerase chain reaction (PCR), using a p107 cDNA (Ewen et al., 1991) as template. In a second PCR reaction, a sense primer for amino acid residues 690-732 (5′-CGTGTTT-TGAATTCACTTTAGTTCACTGTCC) and an antisense primer for amino acid residues 928-936 (5′-CGGGATCCTTAATGATTTG-CTCTTTCACTGAC) were used. Both PCR products were purified, digested with SamHI and EcoRI, ligated and cloned into pGEX-2T. GST-p107 and GST-Rb were purified as previously described (Smith and Johnson, 1988).

Fig. 3.

cdk2-dependent complexes have similar stabilities. Gel retardation was performed in an F9 EC cell extract either alone (tracks 2 and 7) or after the addition of GST-Rb (about 0.06 μ g; tracks 3, 4, 8 and 9) or GST-p107 (about 0.02 μ g; tracks 5, 6, 10, and 11), together with PA-CA and GST-cdk2 (both at 0.5 μ g; tracks 3, 4, 5, 6, 8, 9, 10 and 11), and incubated for either 60 min (tracks 2 to 6) or 20 min (tracks 7 to II). The cdk2-dependent complex (δ) is indicated. Fusion proteins were as described in Fig. 2.

Fig. 3.

cdk2-dependent complexes have similar stabilities. Gel retardation was performed in an F9 EC cell extract either alone (tracks 2 and 7) or after the addition of GST-Rb (about 0.06 μ g; tracks 3, 4, 8 and 9) or GST-p107 (about 0.02 μ g; tracks 5, 6, 10, and 11), together with PA-CA and GST-cdk2 (both at 0.5 μ g; tracks 3, 4, 5, 6, 8, 9, 10 and 11), and incubated for either 60 min (tracks 2 to 6) or 20 min (tracks 7 to II). The cdk2-dependent complex (δ) is indicated. Fusion proteins were as described in Fig. 2.

Fig. 4.

cdc2 cannot engage with DRTF1. Gel retardation was performed in a JM cell extract either alone (tracks 2 and 3) or after the addition of PA-CA (about 0.5 μ g, tracks 4, 5, 10, 11, 12 and 13), GST-cdk2 (about 0.5 μ g, tracks 6, 7, 10 and 11) or GST-Xlcdc2 (about 0.2 μ g, tracks 8, 9, 12 and 13). GST-Xlcdc2 contains the entire coding sequence of a Xenopus cdc2 cDNA (Milarski et al., 1991) in pGEX-2T and was purified as described above. Note that the cdk2-dependent complex, •. is present only in tracks 10 and 11. For tracks 14 and 15, an F9 EC cell extract was incubated in the presence of either a preimmune (track 14) or anti-GST-cdk2 (track 15) rabbit antiserum. The shifted complex is indicated by as.

Fig. 4.

cdc2 cannot engage with DRTF1. Gel retardation was performed in a JM cell extract either alone (tracks 2 and 3) or after the addition of PA-CA (about 0.5 μ g, tracks 4, 5, 10, 11, 12 and 13), GST-cdk2 (about 0.5 μ g, tracks 6, 7, 10 and 11) or GST-Xlcdc2 (about 0.2 μ g, tracks 8, 9, 12 and 13). GST-Xlcdc2 contains the entire coding sequence of a Xenopus cdc2 cDNA (Milarski et al., 1991) in pGEX-2T and was purified as described above. Note that the cdk2-dependent complex, •. is present only in tracks 10 and 11. For tracks 14 and 15, an F9 EC cell extract was incubated in the presence of either a preimmune (track 14) or anti-GST-cdk2 (track 15) rabbit antiserum. The shifted complex is indicated by as.

Fig. 5.

Heterodimerisation of cyclin A and cdk2 fusion proteins produces an active Hl kinase, (a) PA-CA and GST-cdk2 bind in vitro. Either PA-CA (track 1), PA-CA mutants (Δ2, 18, 24 and C-16; tracks 2 to 5), GST-cdk2 (track 6), heat-denatured GST-cdk2 (track 7) or GST with no fusion (track 8) were assayed after western blotting and probing with an anti-GST antibody. The ability of these fusion proteins to complex in the presence (tracks 9 to 15) or absence (tracks 16 to 22) of the JM extract was assessed for the following combinations: PA-CA and GST-cdk2 (tracks 9 and 16), PA-CA and denatured GST-cdk2 (tracks 10 and 17), PA-CA and GST (tracks 11 and 18), PA-CA AAcc2 and GST-cdk2 (tracks 12 and 19), PA-CA Δ18 and GST-cdk2 (tracks 13 and 20), PA-CA Δ24 and GST-cdk2 (tracks 14 and 21) and PA-CA AC-16 PA-CA WT and GST-cdk2 (tracks 15 and 22). Complexes were affinity purified on glutathione Sepharose, denatured, western blotted and probed with an anti-GST antibody (which binds both to PA-CA and GST). Tracks 1 to 5 show a sample of PA-CA and derivatives loaded directly onto the gel and western blotted. The positions of PA-CA, GST-cdk2 and GST are indicated, (b) Binding of PA-CA A24 to GST-cdk2 produces an Hl kinase. The affinity purified material assayed in Fig.3a, tracks 9 to 22, was assayed in parallel for Hl kinase activity (tracks 1 to 14). The indicated PA-CA fusion proteins (about 300 ng) and GST-cdk2 (about 250 ng) were preincubated either alone or in a JM whole cell extract (about 100 μg) for 15 min at 20°C before the addition of 500 pl of 20 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% NP40 and 1% BSA. Complexes were harvested by incubating with glutathione Sepharose for 2 h and then either assayed for Hl kinase activity (Minshull et al., 1990) or denatured and resolved in a 15% SDS polyacrylamide gel. After transferring onto nitrocellulose, fusion proteins were detected with a polyclonal anti-GST antibody. GST-cdk2 was denatured by heating for 5 min. (c) Description of fusion proteins. PA-CA wt contains the C-terminal 355 amino acid residues of bovine cyclin A, A2 has residues 281 and 282 deleted and 283 mutated from serine to threonine, Δl8 lacks residues 273 to 290, Δ24 residues 270 to 293 and ΔC-16 the C-terminal 16 residues; shaded area indicates mutated region

Fig. 5.

Heterodimerisation of cyclin A and cdk2 fusion proteins produces an active Hl kinase, (a) PA-CA and GST-cdk2 bind in vitro. Either PA-CA (track 1), PA-CA mutants (Δ2, 18, 24 and C-16; tracks 2 to 5), GST-cdk2 (track 6), heat-denatured GST-cdk2 (track 7) or GST with no fusion (track 8) were assayed after western blotting and probing with an anti-GST antibody. The ability of these fusion proteins to complex in the presence (tracks 9 to 15) or absence (tracks 16 to 22) of the JM extract was assessed for the following combinations: PA-CA and GST-cdk2 (tracks 9 and 16), PA-CA and denatured GST-cdk2 (tracks 10 and 17), PA-CA and GST (tracks 11 and 18), PA-CA AAcc2 and GST-cdk2 (tracks 12 and 19), PA-CA Δ18 and GST-cdk2 (tracks 13 and 20), PA-CA Δ24 and GST-cdk2 (tracks 14 and 21) and PA-CA AC-16 PA-CA WT and GST-cdk2 (tracks 15 and 22). Complexes were affinity purified on glutathione Sepharose, denatured, western blotted and probed with an anti-GST antibody (which binds both to PA-CA and GST). Tracks 1 to 5 show a sample of PA-CA and derivatives loaded directly onto the gel and western blotted. The positions of PA-CA, GST-cdk2 and GST are indicated, (b) Binding of PA-CA A24 to GST-cdk2 produces an Hl kinase. The affinity purified material assayed in Fig.3a, tracks 9 to 22, was assayed in parallel for Hl kinase activity (tracks 1 to 14). The indicated PA-CA fusion proteins (about 300 ng) and GST-cdk2 (about 250 ng) were preincubated either alone or in a JM whole cell extract (about 100 μg) for 15 min at 20°C before the addition of 500 pl of 20 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% NP40 and 1% BSA. Complexes were harvested by incubating with glutathione Sepharose for 2 h and then either assayed for Hl kinase activity (Minshull et al., 1990) or denatured and resolved in a 15% SDS polyacrylamide gel. After transferring onto nitrocellulose, fusion proteins were detected with a polyclonal anti-GST antibody. GST-cdk2 was denatured by heating for 5 min. (c) Description of fusion proteins. PA-CA wt contains the C-terminal 355 amino acid residues of bovine cyclin A, A2 has residues 281 and 282 deleted and 283 mutated from serine to threonine, Δl8 lacks residues 273 to 290, Δ24 residues 270 to 293 and ΔC-16 the C-terminal 16 residues; shaded area indicates mutated region

Fig. 6.

Evidence that PA-CA directed binding of p33cdk2 is influenced by DRTF1. Gel retardation was performed in a JM cell extract either alone (tracks 2 and 3) or after the addition of either wild-type PA-CA (tracks 4, 5, 12 and 13), A2 (tracks 6, 7, 14 and 15), Al 8 (tracks 8,9, 16 and 17) or ΔC-16 (tracks 10, 11, 18 and 19) in the presence (tracks 12 to 19) or absence (tracks 2 to 11) of GST-cdk2. The PA-CA dependent complex is indicated by ▪ and note that A2 causes the appearence of this complex. Both PA-CA and A2 are supershifted upon the addition of GST-cdk2 (indicated by •). * indicates a non-specific complex.

Fig. 6.

Evidence that PA-CA directed binding of p33cdk2 is influenced by DRTF1. Gel retardation was performed in a JM cell extract either alone (tracks 2 and 3) or after the addition of either wild-type PA-CA (tracks 4, 5, 12 and 13), A2 (tracks 6, 7, 14 and 15), Al 8 (tracks 8,9, 16 and 17) or ΔC-16 (tracks 10, 11, 18 and 19) in the presence (tracks 12 to 19) or absence (tracks 2 to 11) of GST-cdk2. The PA-CA dependent complex is indicated by ▪ and note that A2 causes the appearence of this complex. Both PA-CA and A2 are supershifted upon the addition of GST-cdk2 (indicated by •). * indicates a non-specific complex.

Since cyclin A regulates the activity of cell cycle kinase subunits, such as p34cdc2 (Draetta et al., 1989; Minshull et al., 1990) and p33cdk2 (Pines and Hunter, 1990b; Tsai et al., 1991), we reasoned that the cyclin A associated with DRTF1 may provide a targeting function and thus recruit such kinase subunits to a potentially important substrate. We tested this idea directly by determining if an affinity purified fusion protein that contains the entire coding sequence of p33cdk2 (amino acid residue 1 to 298; GST-cdk2, Fig.lb) would bind to DRTF1 in JM cell extracts. By itself, GST-cdk2 inefficiently binds to the endogenous DRTFl-Rb complex (Fig.la, tracks 6 and 7) although in some experiments a weak supershifted cdk2-dependent complex is apparent, presumably caused by a binding reaction with the endogenous cyclin A-DRTF1 complex. Binding is enhanced, however, when additional cyclin A is provided by adding PA-CA at the same time as GST-cdk2, when a novel, slower migrating complex is apparent (Fig.la, compare tracks 4 and 5 with 8 and 9; indicated by A). Thus, GST-cdk2 can assemble into the DRTF1 complex, but does so in a PA-CA-dependent fashion. The association of cdk2 with DRTF1 is thus dependent on cyclin A and therefore cyclin A contains domains that enable it to bind simultaneously to two distinct types of molecule, DRTF1 and cdk2.

The JM DRTFla complex contains the Rb protein (Fig.la, compare tracks 14 and 15 with 16 and 17; Rb-supershift indicated by ○). To determine if the cdk2-induced complex also contained the Rb gene product, we tested if the anti-Rb antibody caused a further supershift. Indeed, the cdk2-dependent shift can be further shifted by the Rb monoclonal antibody (Fig. la, compare tracks 10 and 11 with 12 and 13; cdk2-dependent Rb-supershift indicated by •). Thus, DRTF1, cyclin A, cdk2 and the Rb gene product can exist in the same molecular complex.

To confirm and extend this result, we tested if cyclin A and cdk2 behaved in a similar fashion in an F9 EC cell extract where most DRTF1 is uncomplexed (Fig.2a, tracks 2 and 3; indicated as DRTFlb). In F9 EC cells, the level of complexed DRTF 1 can be increased by adding an Rb fusion protein (containing Rb coding sequence from 379 to 928, GST-Rb) to the cell extract, which efficiently assembles with F9 EC DRTFlb (Bandara et al., 1991 and Fig.2a, compare tracks 2 and 3 with 6 and 7; Rb complex indicated by DRTFla). When added alone, GST-cdk2 has little effect (Fig.2a, compare tracks 2 and 3 with 10 and 11) and PACA a marginal effect (Fig.2a, compare tracks 2 and 3 with 4 and 5). When GST-cdk2 and PA-CA are added together, however, a cdk2-dependent complex with similar mobility to that produced in JM cell extracts is apparent (Fig.2a, compare tracks 2 and 3 with 12 and 13, cdk2-dependent complex indicated by A). High levels of uncomplexed DRTFlb still remain, suggesting that GST-cdk2 and PA-CA can only associate with endogenous complexed DRTFla in F9 EC cell extracts and therefore the amount of cdk2-induced complex reflects the abundance of DRTFla. This was confirmed by increasing the level of DRTFla, by adding GST-Rb at the same time as GST-cdk2 and PA-CA when the abundance of the cdk2-dependent complex is enhanced (Fig.2a, compare tracks 12 and 13 with 14 and 15, indicated by δ little uncomplexed DRTFlb remains in these conditions. These data indicate that, as in JM extracts, the assembly of p33 with F9 EC DRTF1 is cyclin A-dependent. The combined conclusion of these in vitro assembly assays is thus that the association of GST-cdk2 with DRTFla is dependent on PA-CA, and that this binding reaction only occurs efficiently with complexed DRTFla.

The Rb-related protein, p107, which has a degree of similarity with the Rb protein (Ewen et al., 1991), also binds to DRTF1 because a fusion protein containing the p!07 coding sequence from amino acid residue 249 to 936 (GST-pl()7; Fig.2b) efficiently binds to F9 EC DRTFlb (Fig.2a, compare tracks 2 and 3 with 8 and 9). In fact, the characteristics of the p107 and Rb protein binding reactions are very similar indeed both in quality, efficiency of the complex formation and half-life (data not shown).

Since p107 also binds to DRTF1, we assessed whether cyclin A would also bind to the DRTFl-plO7 complex and, further, if it is also able to direct p33cdk2 to the p107 complex. The p107-induced F9 EC DRTFla complex is further shifted upon the addition of PA-CA (data not shown), in a similar fashion to the effect that PA-CA has on JM DRTFla and, once assembled, PA-CA is likewise able to bind GST-cdk2 because the DRTF 1-p107 complex is super-shifted upon the addition of GST-cdk2 (Fig.2a, compare 8 and 9 with 16 and 17). Cyclin A can thus bind to and recruit p33cdk2 to DRTF1 complexes containing either the Rb gene product or p107.

There are no obvious features of the cdk2-dependent complexes formed with either DRTF 1/GST-Rb or GST-p107 that distinguish between them: the mobilities of the complexes are similar (both GST-Rb and GST-p107 cdk2-dependent complexes migrate as a doublet when assayed in F9 EC cell extracts; Fig.3, compare tracks 3 and 4 with 5 and 6) and the stability of these complexes are similar, at least up to 60 min (Fig.3, compare tracks 2 to 6 with 7 to 11, cdk2-dependent complexes indicated by δ. p33edk2 can therefore, bind via cyclin A to DRTF1 complexes that contain either the Rb protein or p107 and, furthermore, the properties of these binding reactions cannot readily be distinguished in gel retardation assays.

We confirmed that p33cdk2 or a closely related subunit is present in the endogenous DRTFla complexes in F9 EC cell extracts using an antibody raised against the GST-cdk2 fusion protein. The immune but not the preimmune produces a shifted-shift in F9 EC cell extracts (Fig.4, compare tracks 15 and 14); a similar effect is also observed in JM cell extracts (data not shown). We conclude, therefore, that p33cdk2 is present in the DRTF1 complex. In this respect, DRTF1 is similar to E2F (Cao et al., 1992; Devoto et al., 1992).

We addressed the specificity of the p33cdk2/DRTFl inter-action by asking if the mitotic kinase subunit, p34cdk2, could also assemble with DRTF1. To test this, the entire coding sequence of a Xenopus p34cdc2 (Milarski et al., 1991) was expressed as a fusion protein, GST-Xlcdc2 (Fig.4b), affinity purified and then assayed for any binding activity with DRTF1. It was necessary to use Xenopus p34cdc2 because repeated attempts to express human p34cdk2 as a fusion protein have failed. Although GST-cdk2 is able to assemble efficiently with DRTFla in a cyclin A-dependent manner (Fig.4a, tracks 4 to II), GST-Xlcdc2 cannot (Fig.4, compare tracks 10 and 11 with tracks 12 and 13). This lack of effect is not simply caused by a fusion protein that lacked any biological activity because GST-Xlcdc2 is able to bind efficiently to PA-CA in the cell extract to produce Hl kinase activity (R. Y. C. Poon and J. P. A., unpublished data). We therefore conclude that p33cdk2 is selectively recruited to DRTF1 by cyclin A.

Complex formation between cyclin A and cdk2 fusion proteins produces a biologically active Hl kinase

Since heterodimerisation of an appropriate cyclin molecule with a compatible catalytic subunit is necessary to produce a biologically active protein kinase, we tested if PA-CA would form a heterodimer with GST-cdk2, and furthermore determined if the heterodimer possessed H1 kinase activity. The ability of PA-CA to bind GST-cdk2 was assessed after incubating both fusion proteins together, either in the presence or absence of JM cell extract, and then harvesting GST-cdk2 (and thus any bound PA-CA) on glutathione beads. These complexes were then denatured and assayed by immunoblotting with an anti-GST antibody which would bind to both GST-cdk2 and PA-CA (through the protein A moiety) and therefore indicate whether GST-cdk2 had formed a complex with PA-CA.

In isolation, the ability of GST-cdk2 to bind PA-CA is slight (Fig.5a, track 16, notice weak reactivity with PA-CA) although, we believe, significant, because when several derivatives of PA-CA in which the cyclin box or the C terminus were mutated (Fig.5c, A2, A18, A24 and AC-16) are tested, no detectable binding with PA-CA is apparent (Fig.5a. compare track 16 with 19. 20, 21 and 22). The binding efficiency of PA-CA with GST-cdk2 is, however, enhanced when they are incubated together in the JM cell extract and then assayed (Fig.5a, compare track 9 with 16 and notice the increased amount of PA-CA compared with track 9). This binding requires the integrity of the cyclin box and the C terminus in PA-CA because there is negligible binding when either of these regions are mutated (Fig.5a, compare track 9 through 12 with 15); it is also absolutely dependent on the presence of cdk2, since GST alone or denatured GST-cdk2 do not bind to PA-CA (Fig.5a, compare track 9 with 10 and 11, respectively).

We next determined if heterodimerisation of PA-CA and GST-cdk2 results in Hl kinase activity. Indeed, the affinity purified heterodimer formed in the presence of the JM extract possesses Hl kinase activity (Fig.5b, track 1) which is, as expected, severely compromised when the PA-CA mutants are similarly assayed (Fig.5b, compare track 1 through 4 with 7), and again is dependent on the presence of cdk2 coding sequences (Fig.5b, compare track 1 with tracks 2 and 3). These data show that the PA-CA/GST-cdk2 heterodimer forms a biologically active protein kinase and that an activity in JM cell extracts is necessary for this to occur. Since PA-CA can direct GST-cdk2 to DRTF1 in JM cell extracts, we believe that this would enable heterodimerisation of PA-CA with GST-cdk2 and thus target an active kinase to DRTF1.

The stability of the cyclin A-p33cdk2 complex is influenced by DRTFI

The data in the previous section show that PA-CA can bind to GST-cdk2 and that this results in an active Hl kinase. Regions either within the cyclin box or at the C terminus of PA-CA are necessary for this interaction because derivatives in which either region is mutated cannot bind to GST-cdk2 (Fig.5). We wished to determine if these regions were also required for the interaction of PA-CA with DRTF1 and therefore assessed if these mutants assembled with DRTF I a in JM cell extracts. As shown earlier, PACA efficiently assembles with DRTFla, producing a characteristic slower migrating complex (Fig.l, compare tracks 2 and 3 with 4 and 5; PA-CA-dependent complex indicated by ▪). Mutant A2 is also able to bind to DRTF1 (Fig.6, compare tracks 4 and 5 with 6 and 7; complex indicated by ▪) whereas Al8 and AC-16 cannot because there is no effect on the migration of DRTFla (Fig.6, compare tracks 2 and 3 with 8, 9, 10 and 11); the efficiency of A2 binding is, however, lower than the activity of the wild-type sequence. We conclude from these data that the alteration in A2 affects the interaction of PA-CA with GST-cdk2 more severely than its ability to interact with DRTFla because the A2 mutant cannot bind to GST-cdk2 (Fig.5) but can bind to DRTFla.

We next assessed if Δ2 can target GST-cdk2 to DRTF1. When both Δ2 and GST-cdk2 are added to the JM extract, the expected cdk2-dependent shift is produced (Fig.6, compare tracks 12 and 13 with 14 and 15, cdk2-dependent shift indicated by •). Thus, Δ2 can still direct GST-cdk2 to DRTF1. In contrast, but as expected, both Al8 or AC-16 fail to direct GST-cdk2 to DRTF1 (Fig.6, tracks 16 to 19). We conclude from this experiment that mutant Δ2 can still target GST-cdk2 to DRTF1 even though the ability of these two proteins to interact outside the DRTF1 complex is severely compromised (Fig.5). We suggest, therefore, that the stability of the interaction between GST-cdk2 and mutant Δ2 is influenced by DRTF1 or a component within the transcription factor complex.

It is known that cyclins regulate the activity of cell cycle kinases and thus participate in controlling the cell cycle (Hunt, 1989). In this study we have defined a new but widely predicted role for cyclin A by demonstrating that it recruits and thus dictates substrate specificity to the catalytic subunit p33cdk2. We believe that such a targeting function is likely to be a general principle that enables cyclin-dependent kinase subunits to locate protein substrates during the cell cycle.

Cyclin A directs p33cdk2-to DRTF!

The data obtained from the in vitro association assay in which cyclin A targets p33cdk2 to DRTF 1 clearly lead us to conclude that cyclin A has at least two binding interfaces: one for DRTF1 and the other for p33cdk2. The fact that mutant Δ2, which has amino acid residues 281 and 282 deleted and 283 altered from a serine to a threonine, cannot bind directly to GST-cdk2 (Fig.5a) but still engages with DRTFla (Fig.6) supports this idea and potentially defines a sequence that distinguishes between these two interaction domains. However, since an increase in the size of the deletion (for example Δ18 which lacks residues 273 to 290) prevents cyclin A from binding to DRTF1, these two binding interfaces may overlap.

Although GST-cdk2 cannot bind directly to A2 when it is studied outside the DRTF1 complex it can, nevertheless, be directed and stably bind to the DRTF1 transcription factor complex in an A2-dependent fashion (Fig.6). There are at least two possible explanations for this. Perhaps the most likely one is that, in addition to its well characterised binding with cyclin A, p33cdk2 interacts with one or more of the components that make up DRTF1 and this serves to stabilise its binding. Alternatively, cyclin A may undergo some modification upon binding to DRTF1 that then enables it to bind more efficiently to p33cdk2. Either of these models would also go someway towards explaining the selective targeting function of cyclin A, namely that cyclin A targets p33cdk2 but not p34cdc2 to DRTF1. We suggest that this selectivity is also established by an additional inter-action that occurs between p33cdk2 (but not p34cdc2) and a domain in DRTF1 which may, of course, be the same domain as that involved in enabling the association of p33cdk2 with the Δ2-DRTF1 complex.

If indeed such selective targeting and stable association also occurs for p34cdc2 substrates (because of analogous interactions between p34cdc2 and appropriate substrates), then it would lead us to make the clear prediction that, although both p33cdk2 and p34cdc2 can bind to cyclin A, they will, nevertheless, stably associate with different, perhaps overlapping, sets of substrates. This is not an altogether surprising suggestion given that p33cdk2 and p34cdc2 are thought to regulate progression through two distinct parts of the cell cycle, namely S phase and mitosis respectively (Pines and Hunter, 1991b), roles presumably mediated by modulating different substrates. In conclusion, we believe that these data are consistent with a model in which cyclin A provides a general targeting function for catalytic sub-units, such as p33cdk2 and p34cdc2, that recruits them to a variety of substrates, but that additional interactions between the substrate and catalytic subunit influence whether the catalytic subunit remains stably associated with the substrate.

Cyclin A recruits cdk2 to Rb and p107 complexes

The in vitro assay clearly demonstrates that cyclin A can bind, and recruit p33cdk2, to DRTF1 complexes that contain either the Rb protein or p107. This activity is dependent on the presence of Rb or p107, since cyclin A cannot efficiently bind to free DRTF1. However, in cell extracts, cyclin A is found exclusively in E2F complexes that contain p107 (Cao et al., 1992; Shirodkar et al., 1992). It is clear, however, from other studies that a cdc2-like kinase (containing the conserved PSTAIRE amino acid motif) coprecipitates with the Rb protein in anti-Rb immunoprecipitates (Hu et al., 1992). Furthermore, this Rb-associated kinase contains cyclin A (Hu et al., 1992). We believe, therefore, that the Rb-cyclin A-DRTF1 complex generated in our assays is likely to reflect physiological interactions. The difficulty encountered in detecting this complex in cell extracts by gel retardation may reflect, for example, its short half life in vivo. This seems likely because the active kinase (generated by cyclin A plus p33cdk2) may phosphorylate the Rb protein, and others have reported that the Rb-E2F complex contains exclusively underphosphorylated Rb protein (Chellappan et al., 1991). Our ability to recreate stable complexes in vitro may thus result from physiological interactions that have been ‘frozen’ in vitro. This in vitro assay should prove useful in elucidating the role of this kinase in regulating DRTF1.

A mechanism for coupling cell-cycle events to transcription

The ability of cyclin A to direct p33cdk2 to DRTF1 is likely to be an important event in controlling cell-cycle progression because some of the genes that are potentially regulated by DRTF1 are necessary for DNA synthesis (Blake and Azizkhan, 1989; Pearson et al., 1991). Moreover, the Rb protein, which acts in Gi to prevent cells from entering S phase, represses the transcriptional activity of DRTF1 (Zamanian and La Thangue, 1992) and thus could prevent cells from progressing into S phase by limiting transcription of these gene. Since this is likely to be mediated by un- or under-phosphorylated Rb protein (Buchkovich et al., 1989; DeCaprio et al., 1989) we suggest that the ability of a cyclin to direct a kinase to this complex may regulate the activity of the Rb protein within this transcription factor complex. It is possible, therefore, that the transcriptional activation of DRTF1 is necessary for DNA synthesis to occur. Importantly, we do not wish to imply that p33cdk2 is the kinase responsible for regulating the transcriptional activity of DRTF1 but merely, based on the data discussed here, that it is a strong possibility.

Moreover, we cannot rule out that the kinase has a different role from what we have suggested, for example, in modulating the activity of other transcription-related molecules. Whatever the targets for this kinase actually are, it is likely that the ability of cyclin A to target and activate such a kinase is an important event in regulating cell-cycle progression.

We thank Li-Huei Tsai, Matthew Meyerson and Ed Harlow for pGST-cdk2, Bill Kaelin for pGST-Rb, Mark Ewen and David Livingston for a p107 cDNA, Ariel Blocker for construction of cyclin A mutant AC-16 and Peter Rigby for many helpful suggestions. J. P. A. acknowledges receipt of a graduate scholarship from the Stiftung Stipendienfonds des VCIeV and L. R. B. is funded by an MRC Collaborative Studentship with Roche Products. N. B. L. T. is a Jenner Fellow of the Lister Institute of Preventive Medicine.

Bagchi
,
S.
,
Raychaudhuri
,
P.
and
Nevins
,
J.R.
(
1990
).
Adenovirus E1A proteins can dissociate heteromeric complexes involving the E2F transcription factor: A novel mechanism for El A trans-activation
.
Cell
62
,
659
669
.
Bagchi
,
S.
,
Weinmann
,
R.
and
Raychaudhuri
,
P.
(
1991
).
The retinoblastoma protein copurifies with E2F-1, an ElA-regulated inhibitor of the transcription factor E2F
.
Cell
65
,
1063
1072
.
Bandara
,
L.R.
and
La Thangue
,
N.B
. (
1991
).
Adenovirus Ela prevents the retinoblastoma gene product from complexing with a cellular transcription factor
.
Nature
351
,
494
497
.
Bandara
,
L.R.
,
Adamczewski
,
J.P.
,
Hunt
,
T.
and
La Thangue
,
N.B
. (
1991
).
Cyclin A and the retinoblastoma gene product complex with a common transcription factor
.
Nature
352
,
249
251
.
Blake
,
M.C.
and
Azizkhan
,
J.
(
1989
).
Transcription factor E2F is required for efficient expression of hamster dihydrofolate reductase gene in vitro and in vivo
.
Mol.Cell.Biol
.
9
,
4994
5002
.
Buchkovich
,
K.
,
Duffy
,
L.A.
and
Harlow
,
E.
(
1989
).
The retinoblastoma protein is phosphorylated during specific phases of the cell cycle
.
Cell
58
,
1097
1105
.
Cao
,
L.
,
Faha
,
B.
,
Dembski
,
M.
,
Tsai
,
L.-H.
,
Harlow
,
E.
and
Dyson
,
N.
(
1992
).
Independent binding of the retinoblastoma protein and p 107 to the transcription factor E2F
.
Nature
355
,
176
179
.
Chellappan
,
S.P.
,
Hiebert
,
S.
,
Mudryj
,
M.
,
Horowitz
,
J.M.
and
Nevins
,
J.R.
(
1991
).
The E2F transcription factor is a cellular target for the RB protein
.
Cell
65
,
1053
1061
.
Chittenden
,
T.
,
Livingston
,
D.M.
and
Kaelin
, Jr.,
W.G.
(
1991
).
The T/ElA-binding domain of the retinoblastoma product can interact selectively with a sequence-specific DNA-binding protein
.
Cell
65
,
10731082
.
DeCaprio
,
J.A.
,
Ludlow
,
J.W.
,
Figge
,
J.
,
Shew
,
J.-Y.
,
Huang
,
C.-M.
,
Lee
,
W.-H.
,
Marsillo
,
E.
,
Paucha
,
E.
and
Livingston
,
D.M.
(
1988
).
SV40 large tumor antigen forms a specific complex with the product of the retinoblastoma susceptibility gene
.
Cell
54
,
275
283
.
DeCaprio
,
J.A.
,
Ludlow
,
J.W.
,
Lynch
,
D.
,
Furukawa
,
Y.
,
Griffin
,
J.
,
Piwnica-Worms
,
H.
,
Huang
,
C-M.
and
Livingston
,
D.
(
1989
).
The product of the retinoblastoma susceptibility gene has properties of a cell cycle regulatory element
.
Cell
58
,
1085
1095
.
Defeo-Jones
,
D.
,
Huang
,
P.S.
,
Jones
,
R.E.
,
Haskell
,
K.M.
,
Vuocolo
,
G.A.
,
Hanobik
,
M.J.
,
Huber
,
H.E.
and
Oliff
,
A.
(
1991
).
Cloning of cDNAs for cellular proteins that bind to the retinoblastoma gene product
.
Nature
352
,
152
254
.
Devoto
,
S.H.
,
Mudryj
,
M.
,
Pines
,
J.
,
Hunter
,
T.
and
Nevins
,
J.R.
(
1992
).
A cyclin A-protein kinase complex possesses sequence-specific DNA binding activity: p33cdk2 is a component of the E2F-cyclin A complex
.
Cell
68
,
167
176
.
Draetta
,
G.
,
Luca
,
F.
,
Westendorf
,
L.
,
Brizuela
,
J.
,
Ruderman
,
J.
and
Beach
,
D.
(
1989
).
cdc2 protein kinase is complexed with both cyclin A and B: Evidence for proteolytic inactivation of MPF
.
Cell
56
,
829
838
.
Dyson
,
N.
,
Howley
,
P.M.
,
Munger
,
K.
and
Harlow
,
E.
(
1989
).
The human papilloma virus-16 E7 oncoprotein is able to bind to the retinoblastoma gene product
.
Science
243
,
934
936
.
Evans
,
T.
,
Rosenthal
,
E.T.
,
Youngblom
,
J.
,
Distel
,
D.
and
Hunt
,
T.
(
1983
).
Cyclin: a protein specified by maternal mRNA in sea urchin eggs that is destroyed at each cleavage division
.
Cell
33
,
389
396
.
Ewen
,
M.E.
,
Xing
,
Y.
,
Lawrence
,
J.B.
and
Livingston
,
D.M.
(
1991
).
Molecular cloning, chromosomal mapping and expression of the cDNA for p107, a retinoblastoma gene product-related protein
.
Cell
66
,
11551164
.
Fang
,
F.
and
Newport
,
J.W.
(
1991
).
Evidence that the Gi-S and G2-M transitions are controlled by different cdc2 proteins in higher eukaryotes
.
Cell
66
,
731
742
.
Giordano
,
A.
,
Whyte
,
P.
,
Harlow
,
E.
,
Franza
,
B.R.
,
Beach
,
D.
and
Draetta
,
G.
(
1989
).
A 60Kd cdc2-associated polypeptide complexes with E1A proteins in adenovirus-infected cells
.
Cell
58
,
981
990
.
Girard
,
F.
,
Strausfeld
,
U.
,
Fernandez
,
A.
and
Lamb
,
N.J.C.
(
1991
).
Cyclin A is required for the onset of DNA replication in mammalian fibroblasts
.
Cell
67
,
1169
1179
.
Goodrich
,
D.W.
,
Wang
,
N.P.
,
Qian
,
Y-W.
,
Lee
,
E.Y.-H.
and
Lee
,
W-H.
(
1991
).
The retinoblastoma gene product regulates progression through the Gi phase of the cell cycle
.
Cell
67
,
293
302
.
Hu
,
Q.
,
Dyson
,
N.
and
Harlow
,
E.
(
1990
).
The regions of the retinoblastoma protein needed for binding to adenovirus El A or SV40 large T antigen are common sites for mutations
.
EMBO J
.
9
,
1147
1155
.
Hu
,
Q.
,
Lees
,
J.A.
,
Buchkovich
,
K.J.
and
Harlow
,
E.
(
1992
).
The retinoblastoma protein physically associates with the human cdc2 kinase
.
Mol.Cell.Biol
.
12
,
971
980
.
Huang
,
H.-J.S.
,
Wang
,
N.-P.
,
Tseng
,
B.Y.
,
Lee
,
W.-H.
and
Lee
,
E.H.-H.P.
(
1990
).
Two distinct and frequently mutated regions of retinoblastoma protein are required for binding to SV40 T antigen
.
EMBO J
.
9
,
1815
1822
.
Hunt
,
T.
(
1989
).
Maturation promoting factor, cyclin and the control of M-phase
.
Curr.Opin.Cell.Biol
.
1
,
268
274
.
Koff
,
A.
,
Cross
,
F.
,
Fisher
,
A.
,
Schumacher
,
J.
,
Leguellec
,
K.
,
Philippe
,
M.
and
Roberts
,
J.M.
(
1991
).
Human cyclin E, a new cyclin that interacts with two members of the cdc2 gene family
.
Cell
66
,
12171228
.
La Thangue
,
N.B.
and
Rigby
,
P.W.J.
(
1987
).
An adenovirus ElA-like transcription factor is regulated during the differentiation of murine embryonal carcinoma stem cells
.
Cell
49
,
507
513
.
La Thangue
,
N.B.
,
Thimmappaya
,
B.
and
Rigby
,
P.W.J.
(
1990
).
The embryonal carcinoma stem cell E la-like activity involves a differentiation-regulated transcription factor
.
Nucl. Acids. Res
.
18
,
29202938
.
Lew
,
D.J.
,
Dulic
,
V.
and
Reed
,
S.
(
1991
).
Isolation of three novel human cyclins by rescue of Gi cyclin (Cln) function in yeast
.
Cell
66
,
1197
1206
.
Matsushime
,
H.
,
Roussel
,
M.F.
,
Ashmun
,
R.A.
and
Sherr
,
C.J.
(
1991
).
Colony-stimulating factor 1 regulates novel cyclins during the Gi phase of the cell cycle
.
Cell
65
,
701
713
.
Milarski
,
K.L.
,
Dunphy
,
W.G.
,
Russell
,
P.
,
Gould
,
S.J.
and
Newport
,
J.W.
(
1991
).
Cloning and expression of Xenopus cdc2, a component of MPF
.
Cold Spring Harb.Symp. Quant.Biol
.
56
,
377
384
.
Minshull
,
J.
,
Golsteyn
,
R.
,
Hill
,
C.S.
and
Hunt
,
T.
(
1990
).
The A- and B-type cyclin associated cdc2 kinases in Xenopus turn on and off at different times in the cell cycle
.
EMBO J
.
9
,
2865
2875
.
Pearson
,
B.E.
,
Nasheuer
,
H.-P.
and
Wang
,
T.S.-F.
(
1991
).
Human DNA polymerase agene: sequences controlling expression in cycling and serum-stimulated cells
.
Mol.Cell.Biol
.
11
,
2081
2095
.
Pines
,
J.
and
Hunter
,
T.
(
1990a
).
p34c*2. The S and M kinase?
The New Biologist
2
,
389
401
.
Pines
,
J.
and
Hunter
,
T.
(
1990b
).
Human cyclin A is adenovirus E1A-associated protein p50 and behaves differently from cyclin B
.
Nature
346
,
760
763
.
Pines
,
J.
and
Hunter
,
T.
(
1991
).
Cyclin-dependent kinase. A new cell cycle motif?
Trends Cell.Biol
.
1
,
117
121
.
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 p107-cyclin A complex in a cell cycle-regulated manner
.
Cell
68
,
157
166
.
Shivji
,
M.K.
and
La Thangue
,
N.B
. (
1991
).
Multicomponent differentiation-regulated transcription factors in F9 embryonal carcinoma stem cells
.
Mol.Cell.Biol
.
11
,
1686
1695
.
Smith
,
D.B.
and
Johnson
,
K.S.
(
1988
).
Single-step purification of polypeptides expressed in Escherichia coli as fusion with glutathione-S-transferase
.
Gene
67
,
31
40
.
Tsai
,
L-H.
,
Harlow
,
E.
and
Meyerson
,
M.
(
1991
).
Isolation of the human cdk2 gene that encodes the cyclin A and adenovirus El A-associated p33 kinase
.
Nature
353
,
174
177
.
Whyte
,
P.
,
Buchovich
,
K.H.
,
Horowitz
,
J.M.
,
Friend
,
S.H.
,
Raybuck
,
M.
,
Weinberg
,
R.A.
and
Harlow
,
E.
(
1988
).
Association between an oncogene and an anti-oncogene: the adenovirus E1A proteins bind to the retinoblastoma gene product
.
Nature
334
,
124
129
.
Xiong
,
Y.
,
Connolly
,
T.
,
Futcher
,
B.
and
Beach
,
D.
(
1991
).
Human D-type cyclin
.
Cell
65
,
691
699
.
Zamanian
,
M.
and
La Thangue
,
N.B
. (
1992
).
Adenovirus E1A prevents the retinoblastoma gene product from repressing the activity of a cellular transcription factor
.
EM BO J
.
11
,
2603
2610
.