Identification of the E2A Gene Products as Regulatory Targets of the G1 Cyclin-dependent Kinases*

The E2A gene products, E12 and E47, are multifunctional transcription factors that as homodimers regulate B cell development, growth, and survival. In this report, the E2A gene products are shown to be targets for regulation by the G1 cyclin-dependent kinases. Two novel G1 cyclin-dependent kinase sites are identified on the N-terminal domain of E12/E47. One site displays homology to a preferential D-type cyclin-dependent kinase site (serine 780) on the retinoblastoma susceptibility gene product (pRB) and, consistent with this homology, is more efficiently phosphorylated by cyclin D1-CDK4 than by the other cyclin-dependent kinases (CDK) that were tested. The second kinase site is phosphorylated by both cyclin D1-CDK4 and cyclin A/E-CDK2 complexes. Mutation studies indicated that phosphorylation of the cyclin D1-CDK4 site, or more potently, of both the cyclin D1-CDK4 and cyclin A/E-CDK2 sites, negatively regulates the growth suppressor function associated with the N-terminal domain of E12/E47. Transient expression studies showed that ectopic expression of cyclin D1 or E negatively regulates sequence-specific activation of gene transcription by E12/E47. Analysis of site mutants, however, indicated that inhibition of E12/E47 transcriptional activity did not require the N-terminal G1cyclin-dependent kinase sites. Together, the results suggest that the growth suppressor and transcriptional activator functions of E12/E47 are targets for regulation by G1cyclin-dependent kinases but that the mechanisms of regulation for each function are distinct.

The products of the e2a gene are basic helix-loop-helix (bHLH) 1 DNA-binding proteins that control gene expression as homodimers in B cells and as heterodimers in other tissues (1,2). Expression of the e2a gene produces two splice products, E12 and E47 (3,4). One of these two proteins, E47, was originally identified by its ability to bind specific sites (E boxes) in the enhancer regions of the immunoglobulin genes (3,5,6). The E2A gene products later were found to be part of a larger, widely expressed family of bHLH regulators (the E proteins) that function primarily as heterodimers with tissue-specific bHLH proteins (1). Functional heterodimers between the E proteins and other tissue-specific bHLH regulators have been observed in differentiating neuronal, myogenic, and pancreatic cells (7)(8)(9)(10)(11). The E2A gene products function as homodimers in B cells (10,(12)(13)(14), and mature B cells are absent from e2a gene null mutant mice (15,16). On the other hand, development of other tissues, including muscle and nerves, appears normal in these mice. This result suggests that homodimers of the E2A gene products are essential for B cell development but that their role in forming active heterodimers with tissue-specific bHLH proteins may be played by other members of the E protein family (1,3,5,17).
Homodimers of E2A gene products bind and activate transcription from a minimal reporter construct containing four tandem repeats of E5, E2, and E3 sites ( [5,2,3] x4 -TATA-CAT). In the absence of ectopically expressed E2A gene products, expression of this reporter is restricted to B cells (18). The reporter is activated in NIH3T3 cells when ITF-1 (E47/E2-5) is expressed ectopically (18). The function of E2A gene products as transcriptional activators is critical for the completion of immunoglobulin gene rearrangement and normal B cell development. In e2a gene null mice, B cell development is severed between the pre-B and pro-B cell stages, and rearrangement of the immunoglobulin genes is interrupted (15,16). The importance of the E2A gene products for initiation of immunoglobulin gene rearrangement is underscored by experiments showing that ectopic expression in T cells induces immunoglobulin gene rearrangement (6). Since transcriptionally active E2A gene products are found in B cells during most stages of their development, it is likely that they also perform critical functions in cells that are more mature than pro-B cells, although the nature and importance of these functions has not been well explored.
The DNA-binding and dimer assembly domain (the bHLH) of E12/E47 maps to the extreme C-terminal region of the proteins, leaving a large N-terminal domain. The transcriptional activation domains of the E2A gene products have been mapped to this N-terminal domain and are conserved among members of the E protein family (19,20). One activation domain has been structurally characterized as a loop followed by an ␣-amphipathic helix and is referred to as a loop-helix motif (19). Within an acidic stretch of residues immediately N-terminal to the bHLH are two sites that are phosphorylated by casein kinase II and protein kinase A (21). In electrophoretic mobility shift studies, the DNA binding ability of E47 homodimers was blocked by phosphorylation of these residues, whereas the DNA binding ability of MyoD/E47 heterodimers was unaffected (21). Consistent with this, phosphorylation of E47 at these sites was observed only in non-B cells (21), suggesting a unique mecha-nism for restricting the transcriptional activity of E12/E47 homodimers to B cells.
Recent studies have indicated that the E2A gene products activate transcription of the cyclin-dependent kinase inhibitor p21 CIP/WAF/SD11 (22), suggesting a direct mechanism through which E12/E47 can regulate cell growth. In addition, other functions for E12/E47 in the regulation of cell growth that are not directly associated with the activation of gene transcription have been described. These functions have been structurally mapped to the N-terminal segment of E12/E47 and include regulation of apoptosis (23)(24)(25)(26)(27)(28) and mitogenic growth (29). Recent studies have shown that e2a gene null mice develop T cell lymphomas and that ectopic expression of E12/E47 in these lymphoma cells induces their death (27). A role for E47 in growth suppression was first observed in NIH3T3 cells (29). In colony-forming assays, deletion mutants of E47 that contained the N-terminal region suppressed growth of NIH3T3 fibroblasts as well as wild-type E47 (29). The growth suppressor activity of E47 contrasts that of MyoD (30,31) and other bHLH proteins, as E47 does not require the bHLH domain for this function. Growth suppression by E47 is induced during entry into or progression through the G 1 phase of the cell cycle (29), suggesting that the G 1 cyclin-dependent kinases (CDKs) may negatively regulate this function under appropriate growth conditions.
In this report, we identify the E2A gene products as novel targets for regulation by G 1 cyclin-dependent kinases. Two novel G 1 cyclin-dependent kinase sites are identified on the N-terminal domain of E12/E47, and evidence is presented that phosphorylation of these sites negatively regulates the growth suppressor activity of E47. One of the sites displays a strong preference for phosphorylation by cyclin D1-dependent kinase, a characteristic previously associated only with certain kinase sites of another class of growth suppressors, the retinoblastoma susceptibility gene (RB) family. In addition, ectopic expression of cyclin D1 or E is shown to regulate negatively site-specific transcriptional activation by E12/E47 but through a mechanism that does not require the N-terminal cyclin-dependent kinase sites. Together, the results suggest that the products of the e2a gene are targets through which the G 1 cyclin-dependent kinases may regulate the proliferation and gene expression of certain cell types.  (34,35), and cyclin D1 (36) were generously provided as pRc/CMV (Invitrogen) expression constructs by P. Hinds and R. Weinberg (Whitehead Institute). The vector, pECE (37), and pECE expression construct coding for the human E47⅐(ITF-1/E2-5 (5)) cDNA with a C-terminal SV40 tag were given by B. Wold (California Institute of Technology).

Baculovirus Expression Constructs
The cDNAs of cyclin A, cyclin E, and cyclin D1 were blunt-inserted into the transfer vectors, pVL1392 or pVL1393 (PharMingen). The cDNA for mouse CDK4 (38) was given by C. Sherr (St. Jude Children's Research Hospital) and subcloned into the EcoRI site of the transfer vector, pAcGHLT-A (PharMingen).

Bacterial Expression Constructs
The cDNA for human p21 (39 -43) was given by D. Beach (Cold Spring Harbor Laboratory) and inserted into the NcoI and EcoRI sites of pTAT-HA (44), generously provided by S. Dowdy (Washington University School of Medicine). The His 6 -tagged expression constructs were prepared by inserting PCR fragments containing coding sequence into the BamHI and HindIII sites of pQE30 (Qiagen). Wild-type and mutant E47 His 6 -tagged expression constructs were prepared by using sequences coding for residues 9 -559. Sequence coding for residues 1-912 of mouse CDK4 was used to prepare the His 6 -tagged CDK4 expression construct.

Site-directed Mutagenesis
A method was devised to generate site mutants based on two separate PCRs. The gene was divided into two fragments flanking the mutation site and extending to vector sequences at the 5Ј and 3Ј ends of the gene. To generate the 5Ј fragment, a forward (sense) primer containing a vector restriction site 5Ј to the gene (e.g. BglII) and a backward (antisense) primer starting with the mutant nucleotide and extending into the gene were used in one PCR. To generate the 3Ј fragment, a forward (sense) primer starting 1 nucleotide down from the mutant and a backward (antisense) primer starting from a vector restriction site 3Ј to gene (e.g. XbaI) were used in a second PCR. A protocol modified from that described by Liang and Pardee (45) was used to perform the PCRs. Reactions were carried out by a Techne Progene thermal cycler programmed to perform 35 cycles of a threestage process and a final elongation step. The 1st stage was set at 95°C for 40 s, 2nd stage set at 59°C for 2 min, and 3rd stage set at 72°C for 1 min. Reactions were extracted with phenol/chloroform to remove proteins and dNTPs. The 5Ј and 3Ј PCR fragments were digested by BglII and XbaI, respectively, whereas the vector was digested by both BglII and XbaI and then further digested with CIP (New England Biolabs). The two fragments and vector were combined and incubated with ligase, which resulted in a blunt-end ligation at the site of the mutation and sticky-end ligations at the BglII and XbaI sites. Vent DNA polymerase (New England Biolabs), a thermophilic DNA polymerase that generates blunt ends instead of overhangs, was used for the PCRs. Competent XL-1 blue cells were transformed with the ligation reactions as described elsewhere (46). Plasmid DNA was isolated from randomly selected clones with the Wizard SV Mini-prep DNA Isolation Kit (Promega) and digested with both BglII and XbaI to check for inserts. Positive clones were then sequenced to confirm mutations and correct junctions.

DNA Sequencing
Plasmid DNA was denatured under alkaline conditions and used as template for sequencing by the Sanger method (Sequenase version 2.0 DNA Sequencing Kit supplied by Amersham Pharmacia Biotech/U. S. Biotechnology Corp.). Sequence reactions were then run on a 6% polyacrylamide/urea gel in TBE buffer (46). The gel was fixed, dried under vacuum, and then exposed to x-ray film. Sequences were read directly from the autoradiographs.

Transient Expression Assays
Cells were transfected at 80% confluency with a standard calcium phosphate method (55) and incubated for 18 h. After transfection, cells were grown in low mitogen medium (DMEM supplemented with 3% horse serum) for another 48 h. Cells were then harvested for CAT activity. CAT assays were performed as described elsewhere (47). CAT activities were then quantified from TLC plates by PhosphorImager analysis. To control for nonspecific effects on [E5 ϩ E2 ϩ E3] x4 -TATA-CAT reporter activity, parallel experiments were performed with a constitutively active reporter, pRSV-CAT. Relative CAT activities were then determined by normalizing CAT values from [E5 ϩ E2 ϩ E3] x4 -TATA-CAT reporter with respective values obtained with the control reporter (pRSV-CAT).

Generation of Recombinant Proteins
Proteins were prepared by protocol modified from that described elsewhere (Qiagen His 6 Protein Expression Manual). TB-1 competent cells expressing pREP4 were transformed with expression constructs by a standard protocol, and small scale protein preparations were performed to isolate high protein-expressing clones. Clones were grown overnight in LB medium containing both ampicillin and kanamycin and were then used for large scale protein preparations. For large scale protein preparations, bacteria were grown for 1 h in LB medium at 37°C and then induced to express protein with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 4 -5 h. Cells were harvested, washed with ice-cold TES (10 mM Tris, pH 7.5, 1 mM EDTA, pH 8.0, 150 mM NaCl), and lysed in sonication buffer (25 mM HEPES, pH 8.0, 150 mM NaCl, and 8 M urea). Lysates were sonicated, clarified, and then loaded onto nickel-nitrilotriacetic acid-agarose (Qiagen) columns that were prewashed with wash buffer (25 mM HEPES, pH 8.0, 150 mM NaCl, 8 M urea). Columns were washed several times with wash buffer, and His 6tagged proteins were eluted with 150 mM imidazole in wash buffer. Proteins were then dialyzed twice against 10 mM HEPES, pH 8.0, 100 mM NaCl. Samples were aliquoted and flash-frozen.

Preparation of Active Cyclin-CDK Complexes
The four steps used in the preparation of active cyclin-CDK complexes included generation of baculoviruses, titration of baculoviruses, coinfection with cyclin and CDK baculoviruses, and purification of active cyclin-CDK complexes. The baculovirus expressing HA-tagged CDK2 (34, 48 -50) was generously provided by C. Prives (Columbia University). Baculoviruses expressing human MO15/CDK7 (51-54) and His 6 -tagged cyclin H (52) were gifts from D. Morgan (University of California, San Francisco).
Generation of Baculoviruses-The baculoviruses expressing cyclin A, cyclin E, cyclin D1, and CDK4 were generated with the Baculovirus Protein Expression Kit from PharMingen. Monolayers of Sf9 insect cells (2 ϫ 10 6 ) passaged into T25 flasks were cotransfected with transfer vectors (driven by the polyhedrin promoter) that contain the gene of interest and BaculoGold TM DNA (linearized viral DNA) by a calcium phosphate method for 4 h. The use of BaculoGold TM DNA results in the generation of baculoviruses that express only the protein of interest, thus eliminating the need to select positive expressing clones. After transfection, cells were grown in fresh medium and incubated for 5 days. The medium was collected, and 1 ml was used to infect 2 ϫ 10 7 Sf9 cells seeded on a 15-cm tissue culture plate. After 3 days of infection, the medium was collected, and the titer of the virus was initially measured by end point dilution assay. Multiple rounds of amplification were performed until viral stocks reached titers greater than or equal to 1 ϫ 10 8 plaque-forming units/cell.
Titration of Baculoviruses-Since cells are infected with two viruses simultaneously, infection efficiencies may differ from one virus to another. Infection efficiencies of cyclin and CDK viruses were normalized by comparing relative protein expression levels at different virus concentrations. Sf9 cells were passaged into 24-well plates containing 2 ϫ 10 5 cells/well. Amplified virus was added to the wells at dilutions values of 1:10, 1:20, 1:40, 1:100, 1:200, 1:400, and 1:1000. After 2 days of infection, cells were then harvested, and lysates were analyzed by SDS-PAGE (55). Proteins were visualized by staining gels with Coomassie Blue R-250 solution (55).
Coinfection with Cyclin and CDK Viruses-The dilution value where maximum protein expression levels were observed with each virus was used in coinfections to generate cyclin-CDK complexes. Cyclin and CDK viruses were mixed together and then added to T75 flasks containing 1 ϫ 10 7 Sf9 cells. After 48 h of infection, cells were detached from flasks by pipetting up and down and gently spun at 4 o C. Cell pellets were then washed with ice-cold wash buffer (25 mM HEPES, pH 7.5, 150 mM NaCl), spun again at 4 C, and then resuspended with 2 volumes of Sonication Buffer (25 mM HEPES, pH 7.5, 10 mM NaCl, 40 g/ml aprotinin, and 40 g/ml leupeptin). Lysates were sonicated and then flash-frozen.
Purification of Cyclin-CDK Complexes-Cyclin A/E-CDK2 complexes were affinity-purified with 12CA5-protein A-agarose beads. Briefly, 12CA5-protein A-agarose beads were generated by coupling 12CA5 monoclonal antibody (provided by T. Moran, Mt. Sinai School of Medi-cine) to protein A-agarose beads (Roche Molecular Biochemicals) with dimethyl pimelimidate (Sigma). Frozen lysates were thawed, clarified, and incubated with 12CA5-protein A-agarose beads. Purified cyclin A/E-CDK2 complexes were then eluted from the beads with HA peptide (Roche Molecular Biochemicals) containing Arg-insulin (Sigma) as carrier.

Production of Purified p21 Protein
BL-21 (DE3)-competent cells were transformed with pTAT-HA-p21, and small scale protein preparations were performed to isolate high expressing clones. High expressing clones were grown overnight in LB medium with ampicillin and then used for large scale protein preparations. Bacteria were grown in Superbroth at 37 o C with shaking and induced to express protein with 0.8 mM isopropyl-1-thio-␤-D-galactopyranoside. After 4 -5 h of growth, bacteria were harvested, washed with ice-cold TES, and lysed in lysis buffer (8 M urea, 25 mM HEPES, pH 8.0, 100 mM NaCl). Lysate was then sonicated, clarified, and loaded onto a nickel-nitrilotriacetic acid-agarose (Qiagen) column pre-washed with wash buffer (8 M urea, 25 mM HEPES, pH 8.0, 50 mM NaCl). The collected flow-through was then loaded again onto the column for a second round. The column was washed with several column volumes of wash buffer and then with 1 column volume of HBS (25 mM HEPES, pH 8.0, 50 mM NaCl). The p21-TAT protein was eluted with elution buffer (25 mM HEPES, pH 8.0, 2% SDS) and dialyzed twice in 10 mM HEPES, pH 8.0, 50 mM NaCl, 0.5% Triton X-100. After dialysis, p21-TAT protein samples were aliquoted and frozen in liquid nitrogen.

Preparation of Active Cyclin D1-CDK Complexes
Sonicated cyclin D1-CDK4 baculovirus lysate (prepared as described above) was thawed slowly on ice, and purified p21-TAT protein was added (molar ratio of 1:50). The lysate was incubated at 4°C for 15 min and then clarified by spinning at 13,000 rpm for 15 min at 4°C. Glutathione-agarose beads pre-washed with wash buffer (25 mM HEPES, pH 8.0, 50 mM NaCl) were added to the collected supernatant and incubated overnight with shaking at 4°C. Beads were washed 3 times with wash buffer and then resuspended in 25 mM HEPES, pH 8.0, with 20 mM MgCl 2 , 500 M cold ATP, and clarified CAK baculovirus lysate prepared from coinfection of Sf9 cells with cyclin H and CDK7 baculoviruses. The kinase reaction was incubated at 37°C for 15 min and then incubated at ambient temperature for an additional 45 min with shaking. The beads were washed 3 times with wash buffer and then stored in wash buffer at 4°C.

In Vitro Kinase Assays
Protein substrates were phosphorylated by purified cyclin-CDK complexes in a reaction mixture containing 50 mM HEPES, pH 7.5, 20 mM MgCl 2 , 50 M cold ATP, and 10 Ci of [␥-32 P]ATP. The reactions were incubated at 30°C for 1 h and then stopped by the addition of EDTA to a final concentration of 50 mM. The samples were mixed with an equal volume of Sample buffer (100 mM Tris, pH 6.8, 2% SDS, 10% glycerol, 0.025% phenol red, 5% ␤-mercaptoethanol), boiled at 100°C for 5 min, and then analyzed by SDS-PAGE. Gels were fixed, rehydrated, dried under vacuum, and phosphorylation was detected by autoradiography.

Peptide Phosphorylation Assays
A list of the E47 peptides (generated by Research Genetics, Inc.) used in kinase assays with cyclin-CDK complexes is shown in Table I. Stock solutions were prepared by dissolving lyophilized peptides in sterile water at a final concentration of 2 mg/ml. Peptides were phosphorylated in a reaction volume of 20 l containing 4 g of peptide, 25 mM HEPES, pH 7.5, 20 mM MgCl 2 , active kinase (cyclin A/E-CDK2 complexes, or cyclin D1-CDK4 bound to glutathione-agarose beads), 50 M cold ATP, and 10 Ci of [␥-32 P]ATP. Reactions containing cyclin A/E-CDK2 were incubated at room temperature for 1 h, whereas those containing cyclin D1-CDK4 bound to beads were first incubated at 37°C for 15 min and then at room temperature for 45 min with shaking. Kinase reactions were stopped with the addition of EDTA to a final concentration of 50 mM. Kinase reactions were loaded onto cellulose/polyethyleneimine TLC flexible plates with fluorescent indicator (J. T. Baker Co.) and allowed to air-dry overnight at room temperature and kept away from light. TLC plates were placed upright in distilled water until running front reached the line of origin and dried at room temperature away from light. Plates were then placed upright in TLC glass chambers saturated with a solvent mixture containing n-butyl alcohol:ethanol: 30% ammonium hydroxide:chloroform (4:5:9:2). Plates were taken out after the running front reached 1 inch below the top of the plates and allowed to air-dry for 15 min. Plates were wrapped with plastic wrap and exposed to x-ray film. Phosphorylated peptide bands were quantified by PhosphorImager analysis. Total peptide bands were visualized by ninhydrin staining of TLC plates.

Radioisotope Labeling and Immunoprecipitations
Suspension cultures of MO2058 lymphocytes grown in T75 flasks were lipofected with LipofectAMINE reagent (Life Technologies, Inc.) and expression constructs containing wild-type or mutant E47 as described by the manufacturer (Life Technologies, Inc., LipofectAMINE Lipofection Kit). Cells were incubated with lipofection mixture for 1.5 h at 37°C in Opti-MEM low serum medium (Life Technologies, Inc.), and then incubated at 37°C in supplemented DMEM with 15% FBS. After 18 -24 h, cells were washed with phosphate-free medium. Cells were labeled with 3 mCi/5 ml of medium of inorganic 32 P (PerkinElmer Life Sciences) in phosphate-free medium for 2 h. Alternatively, cells were washed with methionine/cysteine-free medium and labeled with 1 mCi/5 ml medium [ 35 S]methionine and [ 35 S]cysteine. After labeling, cells were washed two times with PBS, and lysed with IP buffer (25 mM HEPES, pH 7.5, 50 mM NaCl, 0.5% Triton X-100, 25 mM EDTA, pH 8.0, 20 g/ml aprotinin, 20 g/ml leupeptin, 10 g/ml bacitracin, 100 g/ml benzamide, 1 mM NaF, 1 mM Na 3 VO 4 , 10 mM glycerophosphate, and 500 g/ml acetylated bovine serum albumin). Lysates were clarified by centrifugation in a microcentrifuge at 13,000 rpm for 10 min (4°C). Protein G-agarose (Roche Molecular Biochemicals) beads pre-washed with wash buffer (25 mM HEPES, pH 7.5, 50 mM NaCl, 0.5% Triton X-100) were added to the collected supernatants and incubated at 4°C for 1 h with gentle shaking. Samples were centrifuged in a microcentrifuge at 13,000 rpm for 2 min, and the supernatants were collected. Antibody to the SV40 tag (KT-3, Babco) was added, and the samples were incubated overnight at 4°C. Samples were then incubated with washed protein G-agarose beads for 1 h at 4°C with gentle shaking. The beads were washed five times with wash buffer and then resuspended in sample buffer (100 mM Tris, pH 6.8, 2% SDS, 10% glycerol, 0.025% phenol red, and 5% ␤Ϫmercaptoethanol). Samples were heated in a boiling water bath for 5 min and then analyzed by SDS-PAGE and autoradiography.

Western Blot Analysis
After SDS-PAGE, proteins were electrophoretically transferred to nitrocellulose filters. Filters were fixed with 25% isopropyl alcohol, 10% acetic acid, washed with distilled water, and then washed three times with Tris-buffered saline (TBS, 25 mM Tris-Cl, pH 7.4, 100 mM NaCl). The filters were then incubated overnight (4°C) in Blocking Solution (5% nonfat dry milk in TBS). Primary antibody was added to Blotting Solution (25 mM Tris-Cl, pH 7.4, 500 mM NaCl, 0.5% Tween 20, 5% nonfat dry milk, 5% horse serum) and incubated overnight with the filters (4°C). The blots were washed three times with TBST (25 mM Tris-Cl, pH 7.4, 500 mM NaCl, 0.5% Tween 20) and then incubated with secondary antibody at 1:10,000 in TBST for 1 h at room temperature. Three washes with TBST and one wash with TBS were performed on the blots after incubation with secondary antibody. The blots were laid on plastic wrap and incubated with chemiluminescent reagent (Amersham Pharmacia Biotech ECL Western blot Protein Detection Kit). After 1 min, the blots were drained of excess reagent and wrapped in new plastic wrap and then exposed to x-ray film. Primary antibodies used for detection of E47, cyclins, and p21 proteins were obtained from Santa Cruz Biotechnology.

Colony-forming Assays
Cells passaged at 80% confluency in T25 flasks were cotransfected with a plasmid conferring G418 resistance (pRc/CMV or pSV2-Neo) and an expression construct of interest by a standard calcium phosphate method described elsewhere (55). After 24 h of transfection, cells were trypsinized and passaged into 15-cm plates. G418 (Life Technologies, Inc.) was added to the medium at an effective concentration of 600 g/ml the following day. Medium was replaced, and G418 was replenished every 3 days until macroscopic colonies were visible. Cells were washed 3 times with PBS and fixed with EFA (70% ethanol, 10% formaldehyde, 5% acetic acid) for 10 min at Ϫ20 o C. After fixation, cells were rehydrated with water and stained with 0.15% crystal violet solution for 5 min. Cells were then washed three times with water to remove excess crystal violet stain. Colonies were counted and photographed.

Preparation of Samples for Flow Cytometry
Cells were trypsinized, diluted with medium in a 50-ml Falcon conical tube, and spun at 1,000 rpm for 5 min. Cells were washed once with PBS and then resuspended with 10 ml of PBS. While vortexing gently, 30 ml of ice-cold ethanol was added dropwise to the resuspended cells. The cells were fixed at Ϫ20°C for at least 2 h. After fixation, cells were spun at 1,000 rpm for 5 min and rehydrated with PBS. Cells were then incubated with RNase A and incubated on ice for 10 min. Cells were spun down, washed once with Wash Buffer (PBS, 50 mM EDTA pH 8.0), and resuspended in PBS with propidium iodide at 25 g/ml. Cells were incubated for at least 15 min before processing for flow cytometry. FACS analysis was performed with a Becton Dickinson FACS Calibur cell sorter.

RESULTS
Identification of G 1 Cyclin-dependent Kinase Sites on E47-By using the S/TPX consensus site, predicted phosphorylation sites for cyclin A-CDK2, cyclin E-CDK2, or cyclin D1-CDK4 were identified in the primary sequence of E47 (Fig. 1A). Peptides were synthesized containing these predicted proline-directed kinase sites (Table I) and assayed for phosphorylation with purified cyclin A-CDK2, cyclin E-CDK2, or cyclin D1-CDK4. The cyclin-dependent kinase preparations used for these experiments initially were characterized using recombinant RB protein and histone H1 as a substrates. Preparations of cyclin A-CDK2 and cyclin E-CDK2 phosphorylated both histone H1 and RB protein, whereas preparations of cyclin D1-CDK4 phosphorylated RB protein equivalently but lacked histone H1 kinase activity (data not shown). Peptides were incubated with kinases in the presence of [␥-32 P]ATP, and phosphorylation was detected by thin layer chromatography (TLC) and autoradiography (Fig. 1B). The relative incorporation of 32 P into phosphorylated peptides was quantified by PhosphorImager analysis (Table II). Phosphorylation by cyclin A-CDK2 and cyclin E-CDK2 was observed on peptides containing serine 48 (Ser-48) and serine 154 (Ser-154), with a strong preference for phosphorylation of Ser-48 ( Fig. 1B and Table II). Cyclin D1-CDK4 also phosphorylated Ser-48 and Ser-154; however, preferential phosphorylation by cyclin D1-CDK4 was observed at Ser-154 ( Fig. 1B and Table II).
We next confirmed that residues Ser-48 and Ser-154 are phosphorylated in vitro on full-length recombinant E47 by purified cyclin A-CDK2, cyclin E-CDK2, and/or cyclin D1-CDK4. Serine to alanine site mutants of E47 were generated at Ser-48 (S48A), Ser-154 (S154A), and both Ser-48 and Ser-154 (S48A/ S154A). Recombinant wild-type E47, E47-S48A, E47-S154A, and E47-S48A/S154A were used as substrates. An unrelated E47 site mutant, E47-S250A was used as an additional control. The results of phosphorylation experiments with recombinant E47 and E47 site mutants was consistent with the results obtained with the synthetic peptides. Recombinant wild-type E47 was phosphorylated by purified cyclin A-CDK2, cyclin E-CDK2, or cyclin D1-CDK4 ( Fig. 2A). The E47-S154A mutant was phosphorylated nearly as efficiently as wild-type E47 by cyclin E-CDK2. In contrast, cyclin E-CDK2 was strongly inhibited from phosphorylation of the E47-S48A mutant (Fig. 2A). The data suggest that cyclin E-CDK2 phosphorylates Ser-48 on E47 almost exclusively. Phosphorylation of the E47-S48A mutant by cyclin A-CDK2 also was significantly reduced from the level of phosphorylation observed with wild-type E47, the E47-S250A mutant, or the E47-S154A mutant ( Fig. 2A). A low level of phosphorylation by cyclin A-CDK2 was observed with the E47-S48A/S154A double mutant, suggesting that an additional weak phosphorylation site may be present for this kinase complex.
Cyclin D1-CDK4 phosphorylated wild-type E47, the E47-S250A mutant, and the E47-S48A mutant ( Fig. 2A). In contrast to the other cyclin-dependent kinases tested, cyclin D1-CDK4 phosphorylated the E47-S154A mutant much less efficiently than it did wild-type E47. Virtually no phosphorylation of the E47-S48A/S154A double mutant was observed using cyclin D1-CDK4 (Fig. 2A). These data are consistent with the peptide phosphorylation studies, which showed that cyclin D1-CDK4 phosphorylates Ser-154 more efficiently than Ser-48. In summary, the data identifies two residues on E47 that are phosphorylated by G 1 cyclin-dependent kinases in a type-specific manner as follows: Ser-48, which is preferentially phosphoryl-ated by cyclin A/CDK2 and cyclin E/CDK2, and Ser-154, which is preferentially phosphorylated by cyclin D1/CDK4. We next asked whether Ser-48 and Ser-154 are phosphorylated in vivo. A permanent cell line (MO2058; Refs. 56 and 57) derived from a mantle B cell lymphoma was used for these experiments. This cell line was chosen for two reasons as follows. 1) Because of a chromosomal translocation characteristic of mantle cell lymphomas (t(11;14)(q13;q32)), these cells express high levels of cyclin D1-dependent kinase activity. 2) B cells have been shown to hypophosphorylate other kinase sites on E12/E47 (21). As indicated by Western blot, mutations of Ser-48 or Ser-154 do not adversely affect accumulation of E47 protein in transiently transfected cells (Fig. 2B). This result also indicates that the viability of cells is not altered by transient transfection of the E47 and E47 site mutant expression plasmids, a conclusion that is supported by their equivalent transcriptional activity in transient expression assays (described below). MO2058 cells were transfected with expression constructs for SV40-tagged wild-type E47 and E47 site mutants S48A, S154A, and S48A/S154A. After 18 h, the transfected cultures were metabolically labeled with inorganic 32 P. Ectopically expressed wild-type E47 and mutants were immunoprecipitated with an SV40 tag monoclonal antibody (KT-3; Babco), and analyzed for phosphorylation by SDS-PAGE and autoradiography. Incorporation of 32 P was significantly reduced in the S154A and S48A/S154A site mutants (Fig. 2C). The lowest level of incorporation was observed in the S48A/ S154A double mutant, and the S154A mutant displayed a lower level of incorporation than the S48A mutant (Fig. 2C). These results indicate that both Ser-48 and Ser-154 may be phosphorylated in proliferating mantle lymphoma cells.
Ectopic Expression of G 1 Cyclins Inhibits Activation of Transcription by E47, but Inhibition Is Not Mediated by Phosphorylation of the G 1 Cyclin-dependent Kinase Sites-Regulation of FIG. 1. Identification of G 1 cyclin-dependent kinase sites on E47. A, functional domains and predicted proline-directed kinase sites of E47. The predicted primary sequence maps of three reported E47 cDNA clones is shown. Residue numbers for the cyclin D1-CDK4 and cyclin A/E-CDK2 sites for each of the three clones are shown. The cDNA clone designated E47 (61) was used for the studies presented in this report, and the relative positions of all potential proline-directed kinase sites are marked (small bars). The cDNA clone designated E2-5 was numbered from the first predicted amino acid (5). The third cDNA clone (E2A; see Ref. 27) is thought to contain an additional transcriptional activation domain on the N-terminal region (TAD1). All three cDNA clones contain a central transcriptional activation domain (TAD2) with a loop-helix motif (LH) and a C-terminal bHLH. B, phosphorylation of E47 peptides by cyclin A-CDK2, cyclin E-CDK2, and cyclin D1-CDK4. Synthetic peptides representing the 11-residue neighborhoods surrounding the potential proline-directed kinase sites in E47 (Table I) were incubated with purified cyclin A-CDK2, cyclin E-CDK2, and cyclin D1-CDK4 in the presence of [␥-32 P]ATP and Mg 2ϩ . Phosphorylated peptides were then resolved by thin layer chromatography (TLC) and detected by autoradiography. Phosphorylation at serines 48 and 154 was detected (arrows). transcription by the E2A gene products has been shown to be mediated by E box sequences in the promoter/enhancer region of the immunoglobulin heavy chain gene (3,5,10). Expression of a synthetic reporter construct containing four tandem repeats of three different E box sequences linked to the thymidine kinase TATA box region [p4R(E5 ϩ E2 ϩ E3)TATA-CAT] has been shown to be activated by E47 in fibroblasts (18). We used this reporter construct to study the effects of ectopic cyclin expression on transcriptional activation by E47. NIH3T3 cells were transfected with an E47 expression construct under control of the SV40 early region promoter (pECE-E47) and a series of cyclin expression constructs under control of the CMV promoter. These expression constructs have been shown in previous studies to be expressed at high levels in fibroblasts (58). Ectopic expression of cyclin D1 and cyclin E strongly inhibited activation of transcription by E47 in NIH3T3 cells (Fig. 3A). Moderate inhibition and/or stimulation of E47 transcriptional activity was associated with ectopic expression of other cyclins, but the significance of those results will not be explored in this report. We next determined whether phosphorylation of serine 48 and/or serine 154 affects activation of transcription by E47. Activation of p4R(E5 ϩ E2 ϩ E3)TATA-CAT transcription by E47, E47-S48A, E47-S154A, and E47-S48A/S154A was compared in NIH3T3 cells. Base-line transcriptional activity of the three mutants also did not vary significantly from that of wild-type E47 in transient expression assays (Fig. 3B). The effect of ectopic expression of cyclins on activation of p4R(E5 ϩ E2 ϩ E3)TATA-CAT transcription by the E47 mutants was also determined. In NIH3T3 cells, transcriptional activation by E47-S48A, E47-S154A, and E47-S48A/S154A was inhibited by ectopic expression of cyclin D1 and cyclin E equivalently to wild-type E47 (Fig. 3C). Together, these results indicate that phosphorylation of serine 48 and/or serine 154 is not involved in the negative regulation of E47 transcriptional activation by cyclins D1 and E. G 1 Cyclin-dependent Kinase Sites Negatively Regulate Growth Suppression by E47-Expression of the E2A gene products has been shown to suppress the growth of fibroblasts in culture (29). Consistent with previous observations, we have found that ectopic expression of E47 in NIH3T3 cells reduces colony-forming efficiency. Deletion mutants of E47 were generated that contained either the N-terminal segment of the protein but lacked the bHLH motif (residues 1-402; E47Not) or contained the bHLH motif but lacked most of the N-terminal domain (residues 1-45, 415-559; E47bHLH). Growth suppressor activity persisted, albeit at lower levels, in the mutant containing the N-terminal segment but was completely absent from the mutant lacking the N-terminal region (Table III). The data are consistent with previous results (29) showing that

FIG. 2. Phosphorylation of E47 by G 1 cyclin-dependent kinases.
A, phosphorylation of recombinant histidine-tagged E47 and E47 site mutants by cyclin A-CDK2, cyclin E-CDK2, and cyclin D1-CDK4 in vitro. Purified cyclin A-CDK2, cyclin E-CDK2, and cyclin D1-CDK4 were incubated with recombinant wild-type E47 and E47 site mutants in the presence of [␥-32 P]ATP and Mg 2ϩ . Phosphorylated proteins were detected by SDS-PAGE and autoradiography. Relative phosphorylation, as quantified by PhosphorImager analysis, is indicated under the autoradiography. ND indicates not determined. Cyclin A-CDK2 and cyclin E-CDK2 were inhibited from phosphorylating E47 S48A site mutants, indicating preferential phosphorylation of serine 48. In contrast, cyclin D1-CDK4 was inhibited from phosphorylating E47 S154A site mutants, indicating preferential phosphorylation of serine 154. Lower phosphorylation of the E47 S48A/S154A double site mutant than the S154A site mutant suggests that cyclin D1-CDK4 phosphorylates serine 48, albeit more weakly than serine 154. The phosphorylated recombinant E47 substrate migrated as a single band near 54 kDa. B, site mutation of serine 48 or serine 154 does not affect accumulation of E47 protein in transiently transfected cells. CV-1 fibroblasts were transfected with expression constructs for wild-type E47 (pECE-E47), E47-S48A site mutant (pECE-E47-S48A), E47-S154A site mutant (pECE-E47-S154A), E47-S48A/S170A double site mutant (pECE-E47-S48A/S170A), and empty vector (pECE). After 36 h, a whole cell lysate was generated by dissolving the cells in SDS-PAGE sample buffer. The proteins in the lysates were resolved by SDS-PAGE and analyzed by Western blot for expression of E47 and E47 site mutants. Ectopically expressed E47 was detected as a doublet migrating at 52-54 kDa; this doublet was not detected in CV-1 fibroblasts transfected with empty vector. C, phosphorylation of E47 in a lymphoblastoid cell line derived from a mantle cell lymphoma (MO2058). MO2058 cells were transfected by lipofection with expression constructs for SV40-tagged wild-type E47, E47-S48A site mutant, E47-S154A site mutant, and E47-S48A/S154A double site mutant. After 18 h, the cells were metabolically labeled with inorganic 32 P for 2 h. Detergent lysates were then generated, and ectopically expressed E47 and E47 site mutants were immunoprecipitated with an antibody to the SV40 tag (KT-3). Immunoprecipitated proteins were resolved by SDS-PAGE, and phosphorylated proteins were detected by autoradiography. Phosphorylation of E47 was inhibited in the S154A site mutant, indicating that serine 154 on E47 is phosphorylated in MO2058 cells. Lower phosphorylation of the E47 S48A/S154A double site mutant than the S154A site mutant suggested that serine 48 also is phosphorylated in MO2058 cells, albeit more weakly than serine 154. High molecular mass phosphoprotein(s) consistently coimmunoprecipitated with wild-type E47 and the S48A site mutant, suggesting an affinity for the serine 154-phosphorylated form of E47. The identification of these bands is in progress. D, cultures were transfected in parallel as described in C and after 18 h metabolically labeled with [ 35 S]methionine and [ 35 S]cysteine for 4 h. Detergent lysates were then generated, and ectopically expressed E47 and E47 site mutants were immunoprecipitated with an antibody to the SV40 tag (KT-3). Immunoprecipitated proteins were resolved by SDS-PAGE and detected by autofluorography.
inhibition of colony formation is conferred by the N-terminal domain of E47.
We next asked whether phosphorylation of Ser-48 and/or Ser-154 regulates growth suppression by E47. Colony-forming assays were performed with C3H10T1/2 cells. These cells express endogenous E proteins and are more resistant to growth suppression by E47 than are NIH3T3 cells. Expression plasmids for wild-type E47, E47-S48A, E47-S154A, and E47-S48A/ S154A were cotransfected with a plasmid conferring G418 resistance into C3H10T1/2 cells. Colonies were selected for G418 resistance, and the colony-forming efficiency for the transfected cultures was compared. Consistent with the notion that C3H10T1/2 cells are resistant to growth suppression by E47, ectopic expression of wild-type E47 or E47-S48A in C3H10T1/2 cells did not significantly reduce the efficiency of colony formation (Table IV). In contrast, ectopic expression of E47-S154A or E47-S48A/S154A severely reduced colony-forming efficiency (Table IV). The double site mutant, E47-S48A/S154A, inhibited colony formation more strongly than E47-S154A, even though the E47-S48A single site mutant did not significantly inhibit colony formation. These observations suggest that phosphorylation of Ser-154 blocks growth inhibition by E47 and that phosphorylation of Ser-48 in conjunction with phosphorylation of Ser-154 more potently blocks growth inhibition.
Ectopic Expression of E47 Kinase Site Mutants Enhances Sensitivity of C3H10T1/2 Cells to Growth Factor Depletion and Contact Inhibition of Growth-Cultures were cotransfected with expression plasmids for either wild-type E47, E47-S48A, E47-S154A, or E47-S48A/S154A and an expression plasmid conferring G418 resistance into C3H10T1/2 cells. Colonies were selected for G418 resistance and then dissociated with trypsin and grown as polyclonal cultures. Although colonyforming efficiency was reduced in cultures transfected with E47-S154A or E47-S48A/S154A (see above), large scale experiments produced sufficient clones from each to generate polyclonal cultures. Time course experiments showed that C3H10T1/2 cells transfected with E47-S154A or E47-S48A/

TABLE III E47 growth suppression is conferred by the N-terminal domain
The growth suppressor function of E47 is conferred by the N-terminal domain. Expression constructs for E47 and deletion mutants of E47 consisting of either the N-terminal domain minus the bHLH motif (E47-Not, residues 1-402) or the bHLH motif minus the N-terminal domain (E47-bHLH, residues 1-45, 415-559) were cotransfected with pSV2Neo into NIH3T3 cells. Transfection of pECE vector with pSV2Neo was used as control (vector). After 2 weeks selection in G418, stable transfected colonies were stained with crystal violet and counted. Mean colony counts and variance are shown from duplicate transfection experiments. ND, not determined. A, ectopic expression of cyclin D1 and cyclin E inhibits activation of transcription by E47. NIH3T3 cells were transfected with p4R(E5 ϩ E2 ϩ E3)TATA-CAT reporter, and expression plasmids for E47 (pECE-E47) and cyclins A, B1, B2, C, D1, D3, and E (pRcCMV-cyclin). After 24 h in high mitogen medium, the cells were switched to low mitogen for 48 h and then assayed for CAT activity. Duplicate experiments were performed, and mean values of quantification of CAT activity by PhosphorImager analysis are shown beside the autoradiographs. B, mutation of G 1 cyclin-dependent kinase sites does not affect transcriptional activation by E47. NIH3T3 cells were transfected with p4R(E5 ϩ E2 ϩ E3)TATA-CAT reporter, and expression plasmids for wild-type E47 (pECE-E47), E47-S48A site mutant (pECE-E47-S48A), E47-S154A site mutant (pECE-E47-S154A), E47-S48A/S170A double site mutant (pECE-E47-S48A/S170A), and empty vector (pECE). After 24 h in growth medium, the cells were switched to low mitogen medium for 48 h and then assayed for CAT activity. Duplicate experiments were performed, and quantification of CAT activity was performed by PhosphorImager analysis of the TLC plates. The background value for p4R(E5 ϩ E2 ϩ E3)TATA-CAT expression in the presence of empty vector was subtracted from all points. Mean values and variance are shown. C, mutation of G 1 cyclin-dependent kinase sites does not block inhibition of E47 by ectopic expression of G 1 cyclins. NIH3T3 cells were transfected with p4R(E5 ϩ E2 ϩ E3)TATA-CAT reporter, and expression plasmids for wild-type E47 (pECE-E47), E47-S48A site mutant (pECE-E47-S48A), E47-S154A site mutant (pECE-E47-S154A), E47-S48A/S170A double site mutant (pECE-E47-S48A/S170A), and empty vector (pECE). Cultures were also cotransfected with expression plasmids for cyclin D1 (pRcCMV-cyclin D1) and cyclin E (pRcCMV-cyclin E). After 24 h in growth medium, the cells were switched to low mitogen medium for 48 h and then assayed for CAT activity. Quantification of CAT activity was performed by PhosphorImager analysis of the TLC plates. The background value for p4R(E5 ϩ E2 ϩ E3)TATA-CAT expression in the presence of empty pECE expression vector was subtracted from all points. Mean values and variance are shown. S154A expression plasmids proliferated more slowly than cells transfected with wild-type E47, E47-S48A, or vector control (Figs. 4A). In addition, the density to which cultures transfected with E47-S154A and E47-S48A/S154A grew at confluence was significantly reduced (Fig. 4, A and B). The cultures were harvested during the logarithmic phase of their growth curves, stained with propidium iodide, and processed for fluorescence-activated cell sorter (FACS) analysis. The FACS analysis revealed that during the logarithmic phase of their growth (Fig. 4A, arrow), the distribution of cells between G 1 , S, and G 2 phases of the cell cycle varied only slightly between the cultures (Table V). Growth suppression of cells expressing ectopic E47-S154A and E47-S48A/S154A does not appear to result from an increase in the time required for completion of a particular phase of the cell cycle. Instead, the protracted growth curves and reduced density at confluence suggest a greater sensitivity of the cells to contact inhibition of growth and/or to depletion of growth factors from the medium. Changes in these parameters may manifest as delayed entry into or more frequent exit from the cell cycle.

DISCUSSION
By using kinase assays with purified cyclin A-CDK2, cyclin E-CDK2, cyclin D1-CDK4, and a series of synthetic peptides as substrates, we identified two G 1 cyclin-dependent kinase phosphorylation sites on E47 as follows: one that is phosphorylated more efficiently by cyclin A and cyclin E-dependent kinase (Ser-48) and another that is phosphorylated more efficiently by cyclin D1-dependent kinase (Ser-154). Further studies using as substrates recombinant wild-type E47 or site mutants substituting alanine for serine 48 (S48A), serine 154 (S154A), or both serine 48 and 154 (S48A/S154A) confirmed the results obtained with the synthetic peptides. Finally, metabolic labeling studies in transfected MO2058 cells provided evidence that Ser-48 and Ser-154 are phosphorylated in vivo. Transient expression assays in NIH3T3 cells using the p4R(E5 ϩ E2 ϩ E3)TATA-CAT reporter indicated that wild-type E47 and the S48A, S154A, and S48A/S154A mutants sustained equivalent transcriptional activity. In addition, none of the mutants were resistant to inhibition by cyclin D1 or cyclin E, indicating that phosphorylation of these sites was not required for inhibition of E47 transcriptional activity.
Ectopic expression of cyclin D1 has been shown in many studies to inhibit activation of muscle gene transcription by the myogenic bHLH factors (59,60). We show here that ectopic expression of two G 1 cyclins, cyclin D1 and cyclin E, inhibits activation of transcription by another member of the bHLH FIG. 4. Analysis of growth suppression by E47 site mutants. A, stable expression of S154A or S48A/S154A site mutants of E47 in C3H10T1/2 cells enhances sensitivity to depletion of growth factors and/or contact inhibition of growth. Expression constructs for E47 (pECE-E47) and S48A (pECE-E47-S48A), S154A (pECE-E47-S154A), and S48A/S154A (pECE-E47-S48A/S154A) site mutants of E47 were cotransfected with pRc/CMV (confers G418 resistance) into C3H10T1/2 cells. Transfection of vector (pECE) with pRc/CMV was used a control. After 2 weeks selection in G418, colonies were removed by trypsin and grown as polyclonal cultures. Growth curves were generated after seeding T25 flasks with 3 ϫ 10 4 cells. Cells were counted at the indicated intervals. An arrow indicates when the cells were harvested for FACS analysis (Table V); another arrow indicates when the culture medium was replaced with new medium (Refresh medium). At the end point (14 days), cells had reached confluence, and no mitotic figures were observed in the cultures. B, phase contrast microscopy of C3H10T1/2 cells permanently transfected with E47 and E47 site mutants. Permanently transfected polyclonal cultures described in Fig. 5A were grown to confluence (14 days) and photographed by phase contrast microscopy. Magnification is the same in all fields (ϫ 400); approximate nuclear counts for each field are indicated below the photographs. C3H10T1/2 cells transfected with either the S154A or S48A/S154A site-mutants of E47 displayed significantly greater sensitivity to contact inhibition of growth, as indicated by the lower density of these cultures at confluence.

TABLE IV
Mutation of G 1 cyclin-dependent kinase sites enhances growth suppression by E47 The growth suppressor function of E47 is enhanced by conversion of serine 154 or both serines 48 and 154 to alanine. Expression constructs for E47 (pECE-E47) and S48A (pECE-E47-S48A), S154A (pECE-E47-S154A), and S48A/S154A (pECE-E47-S48A/S154A) site mutants of E47 were cotransfected with pRc/CMV (confers G-418 resistance) into C3H10T1/2 cells. Transfection of empty vector (pECE) with pRc/CMV was used as control. After 2 weeks selection in G418, stable transfected colonies were stained with crystal violet and counted. Relative growth is presented as a percentage of the colonies obtained by cotransfecting selection and empty expression vector (vector).  (63) require somewhat greater doses of cyclin D1 for complete inhibition, 2 but it is not clear that these phosphorylation sites are directly involved in the mechanism of inhibition. Some studies have suggested that phosphorylation of the retinoblastoma susceptibility protein (pRB) by D-type cyclindependent kinases may block its binding to the myogenic bHLH proteins and that the interaction between these two proteins is essential to activate transcription of muscle-specific genes (64 -67). A problem posed by this hypothesis is that pRB is efficiently phosphorylated by cyclin D3-dependent kinases (68,69), even though cyclin D3 does not inhibit transcriptional activation by the myogenic bHLH factors (59). A recent report has suggested that CDK4 directly binds and inhibits the activity of MyoD (70) and that cyclin D1 is required for translocation of CDK4 from the cytoplasm to the nucleus (71). The role for cyclin D1 in this mechanism is to mediate nuclear translocation of CDK4, thereby allowing it to bind and inhibit MyoD (70). This mechanism has been shown so far to apply only to MyoD, although future studies may reveal its role in the inhibition of other bHLH proteins by D-type cyclins.
Growth suppression of NIH3T3 cells by products of the e2a gene was first observed by Philipson and coworkers (29). Cotransfection of E47 expression constructs with a plasmid construct conferring G418 resistance into NIH3T3 cells significantly reduced the efficiency of colony formation under selective conditions. Consistent with these results, we have observed that ectopic expression of E47 in NIH3T3 cells reduces colony-forming efficiency and that the bHLH domain of E47 is not required for growth suppressor activity. Since the bHLH domain of E47 is required for dimer assembly, sitespecific DNA binding activity, and site-specific activation of gene transcription, these functions apparently are not directly associated with growth suppression by E47. Previous studies have shown that binding to Id proteins blocks growth suppression by E47, but when Id binding was abrogated by removing the HLH domain, growth suppression by E47 remained intact (29). We show here that two G 1 cyclin-dependent kinase sites (Ser-48 and Ser-154) in the N-terminal domain of E47 negatively regulate growth suppression by E47. Site-directed mutagenesis of Ser-154 strongly enhances growth suppression of C3H10T1/2 cells. Site-directed mutagenesis of Ser-48 alone did not enhance growth suppression by E47 but instead as a double mutant augmented the increase in growth suppression associated with mutation of Ser-154. As indicated above, Ser-154 is preferentially phosphorylated by cyclin D1-dependent kinases, whereas Ser-48 is preferentially phosphorylated by cyclin A-or cyclin E-dependent kinases. Together, these observations suggest that phosphorylation of E47 first by cyclin D1-dependent kinases at Ser-154 and then by cyclin A-or cyclin E-dependent kinases at Ser-48 may negatively regulate growth suppression as cells enter and/or progress through the cell cycle.
Enhanced cyclin D1 expression has been shown to be involved in the oncogenesis of several malignancies, and the targets of cyclin D1 may differ with each type. Expression of cyclin D1 is enhanced in a significant percentage of human breast cancers (72)(73)(74). It has been proposed that in breast cancer cyclin D1 does not function as a regulatory subunit of a cyclin-dependent kinase complex but rather as a transcriptional coactivator for the estrogen receptor (75,76). Recent studies have not supported this hypothesis. Although cyclin D1 null mutant mice display significant defects in mammary development (77,78), replacement of the coding region of the cyclin D1 gene with the coding region of the cyclin E gene does not appear to affect murine development (79). Since cyclin E is not a transcriptional cofactor for the estrogen receptor, this result suggests that the proposed function of cyclin D1 as a transcriptional cofactor is not essential. It is important to note that the mice with cyclin E substituted for the cyclin D1 coding region still contained intact cyclin D2 and D3 genes, and therefore their apparent normal phenotype does not prove that cyclin E expression can substitute for or bypass all of the functions of the D-type cyclin family. In addition, the results do not explain the oncogenic potential of cyclin D1, which appears to be more prominent than that of cyclin E.
Phosphorylation of RB family members by D-type cyclin-dependent kinases regulates progression during the G 1 phase of the cell cycle (80). Perturbation of this regulatory step through enhanced expression of cyclin D1, loss of p16 or other cyclin-2 C. Chu and D. Stave Kohtz, unpublished results.
FIG. 5. Homology between sites displaying preferential phosphorylation by cyclin D1-CDK4 in E47 and pRb. Phosphorylation of serine 154 on E47 displays a marked preference for cyclin D1-CDK4, similar to serine 780 on pRb. Homology between these two sites is shown. Most striking is the presence of prolines and hydrophobic residues C-terminal to the directing proline in both the E47 serine 154 site and pRb serine 780 site. In contrast, the pRb serine 795 site, which displays roughly equivalent phosphorylation by both cyclin E-CDK2 and cyclin D1-CDK4, has a C-terminal basic residue characteristic of generic cyclin-CDK phosphorylation sites. Also shown is the consensus sequence for generic cyclin-CDK phosphorylation sites, which in contrast to sites preferentially phosphorylated by cyclin D1-CDK4 contains basic rather than hydrophobic residues C-terminal to the directing proline. E47 site mutant expression constructs Polyclonal cultures derived from the stable cotransfection of pRc/ CMV (confers G418 resistance) and expression constructs for E47 (pECE-E47), E47-S48A (pECE-E47-S48A), E47-S154A (pECE-E47-S154A), and E47-S48A/S154A (pECE-E47-S48A/S154A) were harvested during the logarithmic phase of their growth, stained with propidium iodide, and analyzed for cell cycle distribution by FACS. dependent kinase inhibitors, or loss of RB has been proposed as a common event leading to malignancy in many tissue types (81,82). Studies of neuronal precursor cell cycle regulation have indicated that growth control by D-type cyclin-CDK4/6 complexes is targeted toward regulation of RB, whereas growth control by CDK2 in these cells is regulated by additional targets (83). Enhanced expression of cyclin D1 is achieved in mantle cell lymphoma (a B cell malignancy) through rearrangement of the cyclin D1 gene with the immunoglobulin heavy chain locus (84,85). The role of enhanced cyclin D1 expression in this malignancy has not been determined, although the critical importance of the E2A gene products in B cell growth and development has been well established (15,16). Some investigations have provided strong evidence that pRB is not the oncogenic target of cyclin D1-CDK4 in mantle lymphoma cells (86,87). It is possible that enhanced expression of cyclin D1 may block growth suppression of mantle B cells by products of the e2a gene, resulting in a malignant phenotype.
The similarity between regulation of growth suppression by the retinoblastoma susceptibility gene product (pRB) and E47 is intriguing. Both proteins contain G 1 cyclin-dependent kinase sites that negatively regulate their functions as growth suppressors, including sites that are preferentially phosphorylated by cyclin D1-dependent kinase (88). Although some gene products have been identified that are phosphorylated by cyclin D1-dependent as well as other cyclin-dependent kinases, to our knowledge only RB family members and now E47 have been shown to contain sites that are preferentially phosphorylated by D-type cyclin-dependent kinases. The sequence surrounding Ser-154 on E47 displays significant homology to the major preferred cyclin D1-dependent kinase site found on pRB (serine 780 (88) ; Fig. 5). Some studies have suggested that the optimal consensus for preferential phosphorylation by D-type cyclin-dependent kinases differs significantly from the more generic consensus for all cyclin-dependent kinases, and our results support this conclusion. Virtually all of the cyclin-dependent kinases identified so far are proline-directed, phosphorylating a serine or threonine immediately N-terminal to this residue. In general, inappropriate residues N-terminal to the serine/ threonine residue can abrogate phosphorylation, whereas variations in residues C-terminal to the directing proline can confer preferential phosphorylation by certain cyclin-dependent kinase types (88). Many cyclin-dependent kinase sites are phosphorylated efficiently by both D-type cyclin-dependent kinases and cyclin A/E-dependent kinases. As shown here, sites preferentially phosphorylated by D-type cyclins may be distinguished by alternating hydrophobic and proline residues Cterminal to the determining proline.