Expression of the AML-1 Oncogene Shortens the G 1 Phase of the Cell Cycle*

The AML-1-encoded transcription factor, AML-1B, regulates numerous hematopoietic-specific genes. Inap-propriate expression of AML-1-family proteins is oncogenic in cell culture systems and in mice. To understand the oncogenic functions of AML-1, we established cell lines expressing AML-1B to examine the role of AML-1 in the cell cycle. DNA content analysis and bromodeoxyuridine pulse-chase studies indicated that entry into the S phase of the cell cycle was accelerated by up to 4 h in AML-1B-expressing 32D.3 myeloid progenitor cells as compared with control cells or cells expressing E2F-1. However, AML-1B was not able to induce continued cell cycle progression in the absence of growth factors. The DNA binding and transactivation domains of AML-1B were required for altering the cell cycle. Thus, AML-1B is the first transcription factor that affects the timing of the mammalian cell cycle. The

Although these translocations are closely associated with acute leukemia, it is the wild-type form of AML-1 that transforms cells. Wild-type AML-1 (AML-1B) is transforming when expressed in fibroblasts, and this activity requires the C-terminal transcriptional regulatory domain (25,26). Likewise, the closely related protein PEBP2␣A1 (AML-3) is up-regulated by retroviral insertions that cooperate with c-Myc to induce T-cell lymphomas (27). Therefore, in fibroblasts and in mice, AML-1family proteins are oncogenes. Thus, AML-1 has the unusual property that both the wild-type and the translocated alleles can affect cellular proliferation and differentiation pathways.
Overexpression of the fusion protein encoded by the Inv (16) inhibited cell cycle progression (28). This protein can act as an AML-1 dominant repressor, suggesting a role for AML-1 in the G 1 phase of the cell cycle. However, the Inv (16) fusion protein could also interact with and regulate other factors. In addition, many proteins are required for cell cycle progression, but few factors actually promote transit through the cell cycle. Therefore, we have explored the functions of AML-1 in regulating the cell cycle. We have enforced the expression of AML-1B and found that it accelerates transit through the G 1 phase, whereas transcriptionally inactive forms of AML-1 do not affect the cell cycle. This phenotype is similar to that observed upon overexpression of the D-and E-type cyclins (29 -31). Thus, AML-1 is one of the first transcription factors whose enforced expression shortens the G 1 phase of the cell cycle.

EXPERIMENTAL PROCEDURES
Cell Lines-Parental 32D.3 myeloid progenitor cells were maintained as suspension cultures in RPMI 1640 containing 10% fetal bovine serum (BioWhittaker), 2 mM L-glutamine (BioWhittaker), 1% penicillin G-streptomycin (Life Technologies, Inc.), and 15 units/mL IL-3. 32D.3 cells stably transfected with AML-1B or derivatives thereof were constructed as described previously (32) using the sheep metallothionine promoter plasmid pMT-CB6. COS cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The designation pMT indicates a control neomycin-resistant clonal cell line that was co-selected with the AML-1B-expressing cell lines that are designated pMT-AML-1B. The numbers after each cell line name are the clone (e.g. pMT.11, control cell line number 11).
Cell Cycle Analysis-Asynchronously growing cells were washed 2ϫ in PBS and resuspended in media lacking IL-3. At the indicated times after IL-3 removal, 1 ϫ 10 6 cells were centrifuged and resuspended in 1 * This work was supported by National Institutes of Health (NCI) Grants RO1 AG13726, RO1-CA64140, RO1-CA77274 (to S. W. H.), and F32 CA77167 (to J. J. W.), by NCI Center Grant CA68485, and by the Vanderbilt-Ingram Cancer Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
For bromodeoxyuridine (BrdU) analysis, asynchronously growing cells were resuspended at 0.5 ϫ 10 6 /ml in media containing IL-3. Cells were pulse-labeled with 10 M BrdU (Sigma) for 30 min, washed twice to remove BrdU, and resuspended in normal growth media containing IL-3. Aliquots of 2 ϫ 10 6 cells were obtained at the indicated time points following the removal of BrdU. BrdU detection was performed by standard procedures (33). Briefly, cells were washed in cold PBS and fixed in cold 70% ethanol for 30 min. Cell pellets were then incubated for 30 min in 4 N HCl, washed in 0.1 M borax (pH 9.1) (Sigma), followed by a PBS wash. Cells were then resuspended in 50 l of PBS-TB (PBS containing 0.5% Tween 20 (Sigma) and 0.5% bovine serum albumin Fraction V (Sigma)), and a 1:150 dilution of an anti-BrdU antibody (BU-33) (Sigma) for 30 min at room temperature. After two washes in PBS-TB, cells were incubated with a 1:150 dilution of sheep fluorescein isothiocyanate-conjugated anti-mouse F(abЈ) 2 (Sigma) in PBS-TB for 30 min at room temperature. Cells were again washed twice in PBS-TB, resuspended in PBS containing 5 g/ml propidium iodide (Sigma), 1 g/ml RNase A, and analyzed on a FACScalibur flow cytometer (Becton Dickinson).
Protein Analysis-For protein analysis, cells were lysed in microextraction buffer, sonicated, and clarified as described previously (34,35). Total protein concentration of lysates was measured using the Bio-Rad protein assay reagent. Cell lysate (100 g) was electrophoresed on a 10% polyacrylamide gel, transferred onto nitrocellulose, and probed with the indicated antibodies. Immune complexes were detected using the Super Signal detection system (Pierce) and a peroxidase-labeled goat anti-rabbit secondary antibody (Sigma). All proteins were detected using a purified rabbit polyclonal antibody specific to the runt homology domain (RHD) of AML-1. Gel mobility shift analysis and antibodies recognizing the AML family of proteins have been described (36). The antisera are available from Calbiochem. Gel mobility shift analysis for E2F was performed as described (35). Immunoblot analysis for cell cycle regulatory proteins was performed with antisera purchased from Calbiochem.
Transcription Assays-A 1.1-kilobase pair fragment of the murine cyclin D2 promoter was amplified from genomic DNA by polymerase chain reaction based on the published sequence (37). This fragment was cloned into pGL2 basic, which contains a polylinker sequence upstream of the firefly luciferase gene. The indicated promoter deletions were made using convenient endonuclease restriction sites. pGL2-D2 was co-transfected using the Superfect reagent (Qiagen) with pCMV5-AML-1B or pCMV5-AML-1B(L175D) into COS-7 cells, and luciferase activity was measured 40 h later. Secreted alkaline phosphatase activity expressed from the cytomegalovirus immediate early promoter was used as an internal control to correct for transfection efficiency. Regulation of the human thymidine kinase promoter was performed in the same manner but using the renilla luciferase as the reporter gene (Promega). The cyclin D1-luciferase promoter plasmid was a kind gift from Dr. Ronald Wisdom (Vanderbilt University School of Medicine) (38).

Enforced Expression of AML-1B Accelerates Entry into S
Phase-The observation that the Inv (16) fusion protein inhibited cell cycle progression (28) suggested that AML-1 normally functions in the G 1 phase of the cell cycle. To test this possibility, we enforced the expression of AML-1B, the largest AML-1-encoded product, to determine its effects on the cell cycle. The diploid, IL-3-dependent myeloid progenitor cell line 32D.3 was used for this cell cycle analysis because these cells express relatively low levels of AML-1B. We established 32D.3 cell lines expressing AML-1B from the zinc-inducible sheep metallothionine promoter. Single cell clones expressing AML-1B were identified from three separate parental populations by immunoblot analysis using antibodies specific for AML-1 (Fig. 1). As a control for later experiments, we also selected clonal cell lines that did not express AML-1B (labeled pMT in later figures). In addition, we isolated cell lines that expressed mutant AML-1B proteins. AML-L175D contains a single amino acid change within the DNA binding domain (also known as the runt homology domain or RHD) that disrupts DNA binding (leucine 175 changed to aspartic acid, labeled L175D, Fig. 1) (39), whereas AML-1a lacks the C-terminal transactivation domain (AML-1a.3, Fig. 1). Although the metallothionine promoter was used, high levels of protein expression were found in the absence of zinc. To avoid complications in the cell cycle experiments, metal induction was not used. Cell lines with higher basal expression were used for further analysis (e.g. lines 7 and 28).
To compare the rates of cell cycle progression of these cell lines, DNA content analysis was performed. Enforced expression of AML-1B appeared to affect the cell cycle, as fewer cells were present in the G 1 phase and more cells accumulated in S phase in AML-1B-expressing cells versus a control cell line that was established at the same time (designated pMT, Fig. 2A).
The accumulation of AML-1-expressing cells in S phase could be due to a shortening of another cell cycle phase or to a lengthening of S phase. To address this issue, we determined the timing of the cell cycle phases by performing pulse-chase analysis using the nucleotide analogue BrdU. This method avoids the use of cell cycle poisons that grossly disrupt cellular morphology and physiology. A 30-min pulse of BrdU was followed over a 24-h chase period by flow cytometry using an anti-BrdU antibody to quantitate BrdU incorporation and propidium iodide staining of DNA to measure DNA content (Fig. 2, B and C). By plotting BrdU incorporation versus DNA content we were able to observe both BrdU labeled (Fig. 2B, panel pMT.3, 0 Hour area above the horizontal line, and see arrows in   Fig. 2C) traversing the cell cycle during the chase period. This method of analysis gives rise to a characteristic horseshoeshaped plot (Fig. 2B, panel pMT.3, 0 Hour). During the 30-min labeling period, all cells that were synthesizing DNA incorporated BrdU (Fig. 2B, panel pMT.3, O Hour, ϩBrdU). This includes cells at the G 1 to S phase transition that appear to have a 2 N DNA content, and cells at the S to G 2 phase transition that have a 4 N DNA content (Fig. 2B, panel pMT.3, 0 Hour). To measure the time required for cells to traverse the G 2 /M and G 1 phases, we measured cells re-entering the S phase from 8 h to 16 h after labeling. Cells that were labeled with BrdU and that were re-entering the S phase had divided and contained one-half the levels of BrdU of undivided cells. This difference ensured that we were observing cycling cells (note that the plots in Fig. 2, B and C use a logarithmic scale, making this 2-fold shift small). To measure cells that were clearly in S phase versus those cells that had just begun to synthesize DNA, the gate was set to quantitate the central portion of the S phase plot. This area is equivalent to the area between the unlabeled G 1 and G 2 /M phase cells immediately after the labeling (below the horizontal line and labeled -BrdU at 0 Hour, Fig. 2B).
At 8 to 9 h post-BrdU labeling, labeled cells had exited S phase and divided as judged by a reduction in fluoresence intensity, and the first cells were beginning to re-enter S phase (Fig. 2, B and C). By 10 h after labeling, only 17% of control 32D.3 cells containing BrdU were present in early S phase (Fig.  2B, pMT.3, 10 Hour). By 12 h after BrdU labeling, 30% of cells that were originally in S phase had traversed G 2 /M and G 1 and were re-entering S phase (Fig. 2B, pMT.3, 12 Hour). In addition, 10 h and 12 h after labeling with BrdU, most of the unlabeled cells that were initially in the G 1 and G 2 /M phases had moved into the S phase and the G 1 phase, respectively (e.g. Fig. 2B, compare pMT.3 0 Hour and 10 Hour, and note areas marked by arrowheads in Fig. 2C). Cells expressing E2F-1 and its co-factor DP-1 showed similar cell cycle profiles (38% S phase 12 h post-BrdU labeling; Fig. 2B, pMAM-Neo-E2F-1.1/ DP-1.2, and see Hiebert et al. (32)).
When AML-1B-expressing cells were assayed, nearly twice as many BrdU-labeled cells expressing AML-1B were re-entering S phase 10 h after labeling as control cells (34% and 30% versus 17% for control cells; Fig. 2B compare panels pMT-AML-1B.7 and pMT-AML-1B.28 with pMT.3 at 10 h). At 12-h postlabeling, nearly 50% more AML-1B-expressing cells were in mid-S phase (45% and 54% S-phase, pMT-AML-1B.7 and pMT-AML-1B.28, respectively; Fig. 2B). In addition, at the 12-h time point, the unlabeled AML-1B-expressing cells showed accelerated cell cycle kinetics ( Fig. 2B; note also the unlabeled cells in the 9-and 13-h panels marked with arrowheads in Fig. 2C). 12-13 h after labeling the AML-1B-expressing cells had exited the G 1 phase and were in late S phase (Fig. 2, B and C) (note the reduction in the number of cells in early S phase of the unlabeled cell portions of the plots of cells expressing AML-1B as compared with control and E2F-1-expressing cell plots (Fig.  2B, and marked with brackets Fig. 2C)). Similar results were obtained from more than 10 independently isolated single cell clones that were selected from three different parental populations of 32D.3 cells. The percentages of cells entering S phase 10 or 11 h after BrdU labeling from three different experiments are shown in Table I. Each AML-1B-expressing cell line entered S phase faster than the control cell lines (Fig. 2B and Table I).
The results of DNA content analysis ( Fig. 2A) and BrdU analysis of S phase suggested that AML-1B caused accelerated entry into S phase. In principle, this could be due to a short-ening of the G 1 or the G 2 /M phases. However, DNA content analysis indicated that the total number of AML 1B-expressing cells in the G 2 /M phase were similar to control cells, and the number of cells in the G 1 phase was reduced, suggesting a shortening of the G 1 but not G 2 /M phases ( Fig. 2A). To extend this result and to directly determine whether the G 2 /M phase was shortened by AML-1B expression, we used BrdU pulsechase analysis to determine the time required to traverse G 2 /M ( Fig. 3 and Table II). This was accomplished by measuring the time required for the first BrdU-labeled cells (those cells in late S phase at the time of labeling) to traverse G 2 /M and enter the G 1 phase. Both control G418-resistant cells (pMT) and AML-1B-expressing cells passed through the G 2 /M phase and began entering the G 1 phase after 3-4 h ( Fig. 3 and Table II). Analysis of further time points confirmed this result. We conclude that enforced expression of AML-1B does not affect G 2 /M but shortens the G 1 phase, consistent with the ability of an AML-1 inhibitor to arrest cells in the G 1 phase (28).

AML-1B Cannot Induce G 1 to S Phase Progression in the Absence of Growth Factors-
The observation that AML-1Bexpression accelerated S phase entry in myeloid progenitor cells was unexpected. Cell cycle regulatory transcription factors and oncogenes that act in the G 1 phase of the cell cycle such as E2F-1 and c-Myc do not alter the timing of the cell cycle (32,33,40). However, both E2F-1 and c-Myc can induce S phase progression of 32D.3 cells in the absence of IL-3 (32,33,40). Therefore, we tested whether AML-1B could also induce continued cell cycle progression in the absence of cytokine. Cell lines were cultured in media lacking IL-3 for 12 h, and cell cycle progression was measured using propidium iodide staining to determine their DNA content. E2F-1-expressing cells displayed no apparent cell cycle defects in the presence of IL-3 (Fig. 4). In the absence of IL-3, E2F-1 induced continued S phase entry in the absence of the cytokine (Fig. 4). By contrast, there were fewer cells expressing AML-1B in the G 1 phase and more cells in S-phase in the presence of IL-3, but in the absence of IL-3 these cells accumulated in the G 1 phase (Fig. 4). Thus, AML-1B

AML-1 Regulates the G 1 Phase of the Cell Cycle
requires cooperating proliferative signals from IL-3 to accelerate S phase entry and affect the cell cycle.
Expression of Cell Cycle-related Proteins Is Higher in AML-1B-expressing Cells-One of the major targets for regulation of cell cycle progression is the Rb pathway (41). To determine whether this pathway was affected either directly or indirectly by AML-1B expression, we measured the levels of cyclin-dependent kinase-2 (cdk2), cdk4, cyclin D2, and E2F. Cell lysates were prepared from control and AML-1B-expressing cells, and antibodies to each of these proteins were used to probe immunoblots to determine their levels. Each cell line expressing AML-1B contained more cdk2, slightly elevated cdk4, and substantially more cyclin D2 (Fig. 5A). Consistent with this observation, the levels of histone H1 kinase activity present in cdk2 immunoprecipitates was 2-3-fold higher in the AML-1B-expressing cells (data not shown). Electrophoretic mobility shift analysis indicated that AML-1B-expressing cells also contained more overall E2F DNA binding activity relative to control cells, with most of this increase corresponding to the "free" or active form of E2F (Fig. 5B). This was in contrast to Sp1 DNA binding activity that did not change upon AML-1B expression (Sp1, Fig.  5B).
AML-1 Activates Cell Cycle-regulated Promoters-The shortening of the G 1 phase and the higher levels of cyclin D2 suggested that AML-1 might activate key cell cycle regulatory genes that affect cell cycle progression. Therefore, we scanned the promoters for the G 1 cyclins and found several AML-1 binding sites in the cyclin D2 promoter. To determine whether AML-1 could activate the cyclin D2 promoter, we used this promoter linked to a luciferase reporter gene in transient assays. In COS-7 cells, the cyclin D2 promoter was activated by wild-type AML-1B but not by an AML-1B containing a point mutation that ablates DNA binding (L175D, changing residue 175 from leucine to aspartic acid, Fig. 6A). Expression from the internal control promoter was unaffected by AML-1B (data not shown, but used to correct for transfection efficiency). Promoter deletion analysis indicated that 5Ј promoter sequences containing several potential and known AML-1 binding sites were required for full AML-1B-dependent transactivation, whereas a perfect AML-1 binding site 3Ј to the first transcriptional start site was not sufficient for full activation (Fig. 6B). The caveat in this analysis is that cyclin D2 is regulated during the cell cycle. Thus, the activation observed with AML-1B could be indirect if AML-1B was stimulating cell cycle progression. Therefore, we compared the ability of AML-1B to activate the cyclin D2 promoter with other cell cycle regulated promoters. Although the cyclin D2 promoter was activated to a greater extent than the cyclin D1 or human thymidine kinase promoter (which are not known to contain AML-1 binding sites), these promoters were also activated by AML-1B (Fig. 6C). Once again, the internal control was unaffected by AML 1B (data not shown).
AML-1B Effects on the Cell Cycle Require DNA Binding and Transcriptional Regulatory Domains-Because we could not determine whether AML-1B directly activated cell cycle regulatory promoters, we probed the mechanism of AML-1B cell cycle control by asking whether expression of AML-1a, which lacks the C-terminal transactivation domain (3,42), or AML-1B(L175D) could affect cell cycle progression. The clonal 32D.3 cell lines expressing AML-1a or AML-1B(L175D) ( Fig. 1 and 7A) were compared with control and AML-1B-expressing cells using BrdU pulse-chase analysis (Fig. 7B). Once again, the AML-1B-expressing cells entered S phase with accelerated kinetics as compared with G418-resistant control cells. Cells expressing either AML-1a or AML-1B(L175D) did not show gross alterations in the timing of S-phase entry, indicating that both the C-terminal domain and DNA binding activity are required for AML-1B cell cycle functions. We conclude that AML-1B shortens the G 1 phase by regulating the expression of downstream target genes. DISCUSSION AML-1 is unusual in that it is an oncogene, yet it is disrupted by chromosomal translocations in acute leukemia. We observed that enforced expression of AML-1B induced accelerated S phase entry (Figs. 2 and 7 and Table I). Cultures of cells expressing AML-1B had fewer cells in the G 1 phase and more cells in S phase (Fig. 2). The time needed to traverse G 2 /M remained constant in these cells (Fig. 3, Table II), which is consistent with expression of AML-1B accelerating transit through the G 1 phase of the cell cycle. Thus, AML-1B is one of the first transcription factors described that shortens the length of the G 1 phase of the mammalian cell cycle. Furthermore, this work points toward a cell cycle mechanism for the ability of AML-1 to transform cells.
The ability of AML-1B to alter the cell cycle led us to examine the promoters of known components of the G 1 phase cell cycle control machinery to determine whether any of these genes could be direct targets of AML-1B. We found perfect AML-1B binding sites in the p21 Waf1/Cip1 cyclin-dependent kinase inhibitor promoter (16,43) and the cyclin D2 promoter but not in the E2F1, cdk4, or cdk2 promoters. In AML-1B-expressing cells, endogenous p21 Waf1/Cip1 protein levels were slightly elevated (approximately 2-3-fold, data not shown). Recent work indicates that p21 Waf1/Cip1 can both inhibit cdk activity and promote the assembly of cyclin:cyclin-dependent kinase complexes (44). Both the p21 Waf1/Cip1 and cyclin D2 promoters can be regulated by AML-1B (Fig. 6, data not shown, and Refs. 16 and 43), consistent with the requirement of AML-1B transcriptional activation for the cell cycle phenotype (Fig. 7). However, because AML-1B accelerates the cell cycle, we could not definitively determine whether up-regulation of these genes is directly responsible for the shortening of the G 1 phase or whether the levels of these proteins (or the activation of their promoters) are elevated because the cells are cycling faster. Increasing amounts of AML-1B or the AML-1B L175D mutant plasmid were co-transfected with 700 ng of pGL2-cyclin D2 plasmid. Fold activation was calculated after correcting for transfection efficiency using a plasmid expressing a secreted form of alkaline phosphatase (SEAP) from the cytomegalovirus IE promoter. The numbers below the bars indicate the amount of the plasmid in nanograms. For the AML-1B 800-ng sample, the internal control was not used, as this level of input plasmid suppressed the expression of SEAP, probably due to promoter competition. The levels shown are the average of duplicate experiments. In those samples lacking error bars, the error was too small to graph. For subsequent experiments, 200 -300 ng of AML-1B plasmid was used. B, schematic representation of the murine cyclin D2 promoter and fold activation by AML-1B. A perfect AML-1 binding site is shown as a hatched box, and 5 of 6 base pair matches of the consensus binding site are shown as dark boxes. Numbers above or below each box indicate the number of nucleotides from the first major transcriptional start site. Fold activation is shown on the right hand side. C, AML-1 activates the cyclin D1 and thymidine kinase (TK) promoters. The ability of AML-1B to activate the cyclin D2 promoter was compared with the TK and D1 promoters that lack perfect AML-1 binding sites. Note that both TK and D1 were activated but to a lesser degree than the D2 promoter.
Initially, we observed faster growth of the AML-1B-expressing cells, suggesting that these cells had not compensated for the acceleration of the G 1 phase and were cycling faster. However, as these lines were maintained in culture, this phenotype was lost. It is likely that these cells eventually compensated for the shortening of G 1 by lengthening the S phase, as has been observed for cells overexpressing the G 1 cyclins (29,30). We also found that the cell culture conditions were critical for revealing the AML-1B-cell cycle phenotype. For instance, at high cell densities, the shortening of G 1 was lost (data not shown), suggesting density-dependent effects. The mechanisms behind these phenotypes will require further investigation.
Transcription factors such as E2F-1 and c-Myc are capable of affecting cellular proliferation, but enforced expression of these factors did not affect the timing of the 32D.3 cell cycle (32,40). Therefore, the phenotype of enforced AML-1B expression is more similar to that of overexpression of the G 1 cyclins, accelerating the G 1 phase without affecting the entry to or exit from quiescence (29 -31). Although stimulation of cyclin D2 or p21 Waf1/Cip1 expression may contribute to AML-1B-mediated G 1 phase control, these genes are unlikely to be the only mediators of this phenotype. We are actively searching for other cell cycle control genes that may contribute to AML-1-cell cycle control.