Cell-cycle-dependent Regulation of Human aurora A Transcription Is Mediated by Periodic Repression of E4TF1*

Human aurora A is a serine-threonine kinase that controls various mitotic events. The transcription of aurora A mRNA varies throughout the cell cycle and peaks during G2/M. To clarify the transcriptional mechanism, we first cloned the 1.8-kb 5′-flanking region of aurora A including the first exon. Transient expression of aurora Apromoter-luciferase constructs containing a series of 5′-truncated sequences or site-directed mutations identified a 7-bp sequence (CTTCCGG) from −85 to −79 as a positive regulatory element. Electromobility shift assays identified the binding of positive regulatory proteins to the CTTCCGG element. Anti-E4TF1–60 antibody generated a supershifted complex. Furthermore, coexpression of E4TF1–60 and E4TF1–53 markedly increased aurora Apromoter activity. Synchronized cells transfected with the aurora A promoter-luciferase constructs revealed that the promoter activity of aurora A increased in the S phase and peaked at G2/M. In addition, we identified a tandem repressor element, CDE/CHR, just downstream of the CTTCCGG element, and mutation within this element led to a loss of cell cycle regulation. We conclude that the transcription of aurora A is positively regulated by E4TF1, a ubiquitously expressed ETS family protein, and that the CDE/CHR element was essential for the G2/M-specific transcription of aurora A.

though ipl1 mutants are defective in chromosome segregation, they can still separate into sister chromatids. In addition, kinetochores assembled in extracts from ipl1 mutants show abnormal binding to microtubules, and the kinetochore protein Ndc10p is an excellent in vitro substrate for Ipl1p kinase (19). In Drosophila, amorphic alleles of aurora A (also known as aurora) caused mitotic arrest, with the chromosomes arranged on circular monopolar spindles, resulting in pupal lethality (12). Xenopus aurora A (also known as Eg2) contributes to egg maturation by phosphorylation of cytoplasmic polyadenylation element binding factor (20). Several members of the aurora/ Ipl1p kinase family, such as Ipl1p in S. cerevisiae, aurora B (AIR2) in C. elegans, aurora B (IAL) in Drosophila, and aurora A in Xenopus, play important roles in the entry into mitosis through the phosphorylation of histone H3 (21)(22)(23)(24)(25).
Human aurora A (also known as aurora2, STK15, BTAK, and AIK) was first identified as a human homologue of the aurora/Ipl1p kinase family: it is cell-cycle regulated, mapped to chromosome 20, and highly expressed in colon and breast cancers and several other tumors (5)(6)(7)15). The full-length aurora A cDNA contains a 1209-bp open reading frame that encodes 403 amino acids with a predicted molecular mass of 45.8 kDa. Aurora A protein shares 43 and 41% amino acid identities over its entire length and 62 and 49% identities over its kinase domain with Drosophila aurora A and yeast Ipl1p, respectively (7). The 2.4-kb aurora A transcripts are abundant in mitotically active cells, embryonic cells, and meiotically active germ cells (7,15). Aurora A expression is cell-cycle regulated. Levels of mRNA, protein, and kinase activity are low in G 1 /S, accumulate during G 2 /M, and decrease rapidly after mitosis (7,15). Aurora A protein is localized in the centrosomes of interphase cells and in the spindle of mitotic cells (6,7,15). Ectopic expression of aurora A leads to an increase in centrosome number, causes catastrophic loss or gain of chromosomes, and then results in either cell death or survival through malignant transformation (6). Indeed, aurora A is expressed at high levels in human cancer cells and in most human tumor cell lines (5-7, 25, 26). In contrast, aurora A-deficient cells can form a bipolar mitotic spindle, but chromosomes fail to align on the metaphase plate (27). These data suggest that aurora A is required for segregation of chromosomes but not for separation of centrosomes. Because this protein seems to play key roles in mitotic events such as chromosome segregation and oncogenesis, the regulatory mechanism of aurora A's G 2 /M-specific expression has been a focus of interest. Honda et al. (28) recently reported that aurora A was degraded by the anaphase-promoting complex-ubiquitin-proteasome pathway. Because the level of aurora A mRNA varies during the cell cycle, another mechanism must be involved in the cell-cycle-dependent expression of aurora A. The purpose of the present study was to investi-gate the mechanisms that regulate the transcription of the human aurora A gene.

EXPERIMENTAL PROCEDURES
Cell Cultures, Synchronization, and Cell-cycle Analysis-HeLa, NIH3T3, and human embryonic kidney 293 (HEK293) cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal calf serum (HyClone). Jurkat cells were cultured in RPMI 1640 (Invitrogen) containing 10% fetal calf serum. All cell lines were maintained at 37°C in 5% CO 2 . Untransfected and transiently transfected HeLa cells were grown in 12-well plates to 3 ϫ 10 4 cells/well and synchronized at the G 1 /S boundary by a double thymidine block protocol previously described (29). In brief, thymidine (Sigma) was added to the media to a final concentration of 2 mM as the first block. After a 16-h incubation at 37°C, the cells were washed twice with phosphate-buffered saline (Invitrogen) and incubated in complete growth media for an additional 8 h at 37°C. Thymidine then again was added to the media as the second block. After a 16-h block, cells were washed twice with phosphate-buffered saline, then complete growth media were added to release them from the block. This time point was designated time 0. Cells were harvested at various times and used for laser-scanning cytometry, luciferase assays, electromobility shift assays (EMSAs), 1 and Northern blot analysis. To synchronize them at M phase, HeLa cells were incubated with 0.1 g/ml nocodazole for 16 h.
RNA Analysis-To obtain the aurora A cDNA probe, a 344-bp aurora A cDNA fragment was amplified from the Jurkat cDNA library (CLON-TECH) by using the oligonucleotides 5Ј-ATGGACCGATCTAAA-GAAAA-3Ј and 5Ј-GCCAGTTCCTCCTCAGGATT-3Ј as primers according to a standard polymerase chain reaction (PCR) technique (30). The oligonucleotides we used were chemically synthesized and columnpurified (Amersham Biosciences, Inc.). After gel purification, the probe was labeled with [␣-32 P]dCTP by using a Random Primer Labeling Kit version 2 (Takara). Total RNA was extracted from HeLa cells by using an Isogen RNA extraction kit (Nippon Gene), then electrophoresed and blotted onto a Nytran 0.45 nylon membrane (Schleicher & Schuell). The blots were hybridized with [␣-32 P]dCTP-labeled aurora A cDNA probe. After hybridization, the blots were washed then exposed to XAR-5 x-ray film (Eastman Kodak) with an intensifying screen at Ϫ80°C (30).
Laser-scanning Cytometry-Laser-scanning cytometry was performed as previously reported (31). In brief, HeLa cells were cultured on coverslips in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and synchronized by using the described double thymidine block and release. At the time 0, 6, 8, 10, 12, and 15 h from release, the monolayer cells on coverslips were fixed with 100% ethanol. The coverslips then were dipped in a propidium iodide solution (50 g/ml in phosphate-buffered saline) (Sigma) containing 0.2% RNase (Sigma) and placed on a slide with 90% glycerol in phosphate-buffered saline. DNA content was measured by using a laser-scanning cytometer (LSC101; Olympus). Usually, more than 5000 cells were examined in each sample. The cell-cycle stages were determined according to DNA content.
Cloning of the 5Ј-Flanking Region of aurora A-The gene encoding aurora A is mapped on chromosome 20q13.2, and the sequence is available from the DDBJ/EMBL/GenBank TM data base (accession number AL121914). To obtain the gene encoding the 5Ј-flanking region of the aurora A gene, human genomic DNA was extracted from Jurkat cells by using a method previously reported (30). By using this genomic DNA as template, 1.8 kb of the 5Ј-flanking sequence of the aurora A gene (including the first exon) was amplified using a standard PCR method with the oligonucleotide primers 5Ј-GGGGGATCCTCACAT-GAGAGATTAGAGGC-3Ј (sense strand) and 5Ј-GGGGGATC-CCTCTAGCTGTAATAAGTAAC-3Ј (antisense strand). A BamHI linker was incorporated into the 5Ј end of each primer. The resulting fragment was subcloned into the BamHI site of pBluescript II SK(Ϫ) (Strategene) to generate pBlue1486, which was subjected to sequencing and further plasmid constructions.
Primer Extension-An oligonucleotide (5Ј-ACCTGCGACCCAAG-GACCCAA-3Ј) that was complementary to the region from ϩ8 to ϩ28 of the aurora A cDNA (6) was radiolabeled by T4 polynucleotide kinase (Invitrogen) with [␥-32 P]ATP (Amersham Biosciences, Inc.). Aliquots (8 g) of total RNA isolated from HeLa and HEK293 cells were denatured at 65°C and incubated in a reaction buffer (50 mM Tris⅐HCl (pH 8.3), 75 mM KCl, 3 mM MgCl 2 , and 20 mM dithiothreitol) containing 0.5 ng of the labeled oligonucleotide, Moloney murine leukemia virus reverse transcriptase (200 units) (Invitrogen), and deoxynucleotide triphosphates (0.25 mM each) at 37°C for 1 h. The synthesized cDNAs were denatured at 65°C and analyzed by electrophoresis thorough 6% acrylamide gels. The same primer was used for the DNA sequence ladder run on the same gels.
Cell Transfection and Luciferase Assay-Cells (5 ϫ 10 6 per well of a 12-well plate) were transiently transfected with 1.0 g of a luciferase construct and 0.25 g of pSV-␤ galactosidase plasmid DNA (Promega) by using the Gene PORTER Transfection Reagent (Gene Therapy Systems), and at least three different batches of plasmid were tested per construct. Where indicated, 0.25 g of pCAGGS-E4TF1-60 and/or 0.25 g of pCAGGS-E4TF1-53 were co-transfected with pGL189. After a 48-h incubation, the cells were harvested with Reporter Lysis Buffer (Promega), and luciferase activity was measured by using the luciferase assay system (Promega). The resulting luciferase activity was corrected in light of the ␤-galactosidase activity of the cell extract (34).
Preparation of Nuclear Extracts and Electromobility Shift Assays-Nuclear extracts were prepared by using a method previously reported (35), and aliquots were frozen at Ϫ80°C. EMSA was conducted according to a previously reported method (30) by using a 5% polyacrylamide gel in 0.25 ϫ TBE buffer (22.5 mM Tris borate, 0.5 mM EDTA). The probes were prepared by annealing the appropriate sense oligonucleotides with the antisense oligonucleotide. The binding reaction was performed at 20°C for 30 min in 10 l 1ϫ binding buffer (20 mM HEPES (pH 7.9), 150 mM NaCl, 10% glycerol, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) containing 25 fmol of each probe endlabeled with [␥-32 P]ATP, 2 g of herring testis DNA, and 2 g of each nuclear extract. For the competition assay, 20-or 200-fold excess amounts of the appropriate unlabeled probe were added to the binding reaction mixtures. Gel electrophoresis was carried out in a prerun at 200 V for 30 min and then at 200 V for 2 h 30 min. For the supershift assay, 1 l of antibody against E4TF1-60 (Dr. Hiroshi Handa, Faculty of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama, Japan) or Ets-1 (Transduction Laboratories) was added to the reaction mixture.

RESULTS
Induction of aurora A mRNA Transcription during G 2 / M-We confirmed the cell-cycle-regulated transcription of aurora A mRNA through Northern blot analysis by using syn-chronized cell populations. HeLa cells were synchronized at the G 1 /S boundary by double thymidine block followed by release. Laser-scanning cytometry confirmed that more than 60% of the cells were arrested in G 1 . After release from the thymidine block, about 50% of the cells immediately entered into the S phase. The cells were predominantly in the G 2 phase at 8 h and entered G 1 again at 12 h (Fig. 1B). Northern analysis of the cells (Fig. 1A) showed that aurora A mRNA accumulated in late S phase (lane 2), peaked during G 2 /M (lanes 3 and 4), and decreased as the cells entered the G 1 phase (lanes 5-7). Our results confirmed those of a previous investigation (7).
Identification of the Transcriptional Initiation Site of aurora A-We then amplified the 5Ј-flanking sequence of the aurora A gene by means of a standard PCR method. Primers were designed on the basis of the sequence of chromosome 20q13.2 (DDBJ/EMBL/GenBank TM accession number AL121914), on which the human aurora A gene is located (5, 7). The resulting 1.8-kb fragment was cloned in pBluescript II SK(Ϫ) and sequenced. The sequence was identical to that of chromosome 20q13.2 except for 12 substituted bases. At least four pseudogenes were revealed by a BLAST homology search, and the aurora A gene comprised nine exons and nine introns ( Fig. 2A).
One pseudogene, on chromosome 1, has an entire open reading frame in a single exon. Two pseudogenes were mapped to chromosome 10, both consisting of exon 6 and its downstream sequence. The last pseudogene is on chromosome 20 and also consists of exon 6 and its downstream sequence. We confirmed that the 5Ј-flanking sequence of the aurora A gene was indeed different from those of the pseudogenes (Fig. 2B). These findings indicated that the cloned 1.8-kb fragment was the 5Јflanking region of the aurora A gene. Using HeLa and HEK293 cells as sources of total RNA, we next performed primer extension analysis to define the transcriptional initiation site of the aurora A gene and detected one major band (Fig. 3). We were unable to identify the putative TATA box around the transcriptional initiation region. Our results indicate that transcription of the aurora A gene is initiated from the site designated ϩ1 in Fig. 2B.
Identification of a Positive Regulatory Element in the aurora A 5Ј-Flanking Sequence-To clarify the regions responsible for transcriptional regulation of aurora A, we transiently transfected HeLa and NIH3T3 cells with several luciferase constructs containing the progressively deleted 5Ј-flanking region of the aurora A gene and then measured the luciferase activity of the resulting cell extracts (Fig. 4). Luciferase activity was normalized in light of the activity of ␤-galactosidase; this gene was co-transfected into the HeLa and NIH3T3 cells. Extracts of both cell lines transfected with pGL1486 (containing the fulllength 5Ј-flanking region) had high levels of luciferase activity. A deletion of the 5Ј-flanking sequence from Ϫ1486 to Ϫ124 did not affect the luciferase activity. An additional deletion to Ϫ90 resulted in about a 50% decrease of activity in HeLa cells, whereas it did not affect the activity in NIH3T3 cells. Further deletion to Ϫ75 resulted in about 90% decrease of activity in both HeLa and NIH3T3 cells. These results suggest that the transcription of the aurora A gene was positively regulated by a region (positive regulatory element; PRE) located between Ϫ90 and Ϫ75. A tissue-specific factor might regulate aurora A transcription through a cis-element located between Ϫ124 and Ϫ90.
Four Specific Nuclear Proteins Bound to the PRE of the aurora A Promoter Region-To characterize the positive regulatory element-binding proteins (PREBs), EMSA was performed with the nucleotide probe Ϫ90/Ϫ61, which corre-sponded to the sequence from Ϫ90 to Ϫ61 (Table I). We identified four PREBs (PREB-1, -2, -3, and -4) that bound to Ϫ90/Ϫ61 by using nuclear extracts from HeLa cells (Fig. 5, lane  2). The Ϫ90/Ϫ61-PREB1 was the major band; the other bands seemed to be minor. Formation of these complexes was inhibited by the addition of excess (20-and 200-fold) amounts of unlabeled probe (lanes 3 and 4). No specific protein binding was detected when we used a nucleotide probe from Ϫ124 to Ϫ91 (data not shown).
Effects of Mutations in the PRE-To locate the precise sequence essential for binding, we performed a competition assay using cold mutated oligonucleotides as competitors (Table I). The binding of PREBs to Ϫ90/Ϫ61 was reduced by competition with excess amounts of the probes M1, M4, M5, and M6 (Fig.  6A). In comparison, 200-fold excess amounts of M2 and M3 did not affect the binding.
We next constructed luciferase plasmids in which the PRE sequence of pGL189 carried mutated bases homologous to those in the M-series of oligonucleotides (resulting plasmids denoted pGL189M1, pGL189M2, pGL189M3, pGL189M4, pGL189M5, and pGL189M6) and transfected these constructs into HeLa cells. The mutated bases in pGL189M1, pGL189M4, pGL189M5, and pGL189M6 did not affect the luciferase activity compared with that of the wild-type plasmid (Fig. 6B). However, as we expected, the mutated bases in pGL189M2 and pGL189M3 resulted in 90 and 85% losses of activity, respectively, compared with that in cells transfected with pGL189. Taken together, these results clearly indicate that the 7-bp sequence CTTCCGG (Ϫ85 to Ϫ79) was necessary for PREB binding and transcriptional activity.
Effects of Overexpression of E4TF1-60 and E4TF1-53-We next investigated the effects of overexpression of E4TF1-60   AGGCGTCACTTCCGGGGGCCTTCACCAGTGT 33 and E4TF1-53 on aurora A promoter activity. Co-transfection of pCAGGS-E4TF1-60 and pGL189 increased promoter activity by 2.7-fold at most, and co-transfection of pCAGGS-E4TF1-53 and pGL189 led to only a Յ1.8-fold increase (Fig. 8). However, co-transfection of both E4TF1 constructs with pGL189 increased aurora A promoter activity by as much as 9-fold. These findings indicated that E4TF1-60 and E4TF1-53 together activate the aurora A promoter and that neither E4TF1-60 nor E4TF1-53 alone does. A similar mechanism has been reported for the promoters of CD18, retinoblastoma (RB), and BRCA1 (40 -44).
Transcription of aurora A Is Regulated during the Cell Cycle-We next analyzed the cell-cycle-dependent transcription of aurora A by assessing the luciferase activities of pGL189-transfected HeLa cells that had been synchronized in G 1 /S by double thymidine block and release. After the cells were released from the block, the luciferase activities at the indicated time points were measured and normalized for ␤-galactosidase activity (Fig. 9). Most cells were in G 1 /S at the start of the experiment. As the cells entered the S phase, the pGL189-associated luciferase activity rose, peaking ϳ12 h after release, when most cells were in G 2 /M. For a negative control, we transfected HeLa cells with pGL75, and the luciferase activity of these cells remained low throughout the cell cycle. These results show that the cell-cycle-dependent transcription of aurora A is regulated, at least in part, by a transcriptional mechanism.
To test whether the binding of E4TF1 to the PRE of aurora A is cell-cycle-dependent, we performed EMSAs using the Ϫ90/ Ϫ61 oligonucleotide and nuclear extracts from synchronized HeLa cells. PREBs were detected throughout the cell cycle (data not shown). In light of this finding, we speculated that E4TF1 did not play a central role in the G 2 /M-specific transcription of aurora A and that another mechanism must be involved in the G 2 /M specificity. Indeed, the transcription of thrombopoietin (TPO), B19 parvovirus, RB, CD18, BRCA1, and prolactin is controlled by E4TF1 but, at least in most of these examples, is not cell-cycle-dependent (the cell-cycle-dependent transcription of RB is controlled by E2F) (33, 40, 43-48).

Identification of a Cell-cycle-regulated Repressor Element of the aurora A Promoter-
The G 2 /M-specific transcription of many genes (e.g. cyclin A, cdc25C, cdc2, polo-like kinase, p130, cyclin B2, and rabkinesin 6) is regulated by tandem repression elements, a cell-cycle-dependent element (CDE), and a cellcycle gene homology region (CHR) (49 -54). The sequences from Ϫ44 to Ϫ40 and from Ϫ39 to Ϫ35 of the aurora A promoter, respectively, resemble the CDE and CHR consensus sequences (Fig. 10A). These results suggest that these two putative ciselements of aurora A promoter also functioned as a G 1 /S-specific repressor as previously reported for G 2 /M-specific genes. To confirm this hypothesis, we induced various mutations in the putative CDE and CHR of pGL189 (Fig. 10B). The wildtype and two mutated constructs were transiently transfected into HeLa cells, and luciferase activities were measured in cells arrested with thymidine (G 1 /S) or nocodazole (M phase) for 16 h. The luciferase activity in extract from cells transfected with pGL189 and arrested with thymidine was low (Fig. 10C), about 80% lower than that from cells arrested with nocodazole (Fig. 10D). In contrast, compared with that from cells transfected with pGL189, the luciferase activity in extract from thymidine-arrested cells transfected with pGL189-mCDE and pGL189-mCHR demonstrated increases of as much as 3.5-and 3.7-fold, respectively. However, the activity from nocodazolearrested cells transfected with pGL189-mCDE and pGL189-mCHR was only 1.4-and 1.2-fold higher than that in cells arrested with thymidine. These results suggest that the CDE and CHR in aurora A act as a G 1 /S-specific repressor and are essential for the cell-cycle-specific transcription of this gene. DISCUSSION In the present study, we investigated the transcriptional regulation mechanism of the human aurora A gene. We identified two distinct cis-regulatory elements in the 5Ј-flanking region. One positively regulated transcription, and the other was a cell-cycle-dependent transcriptional repressor.
We identified that the 7-bp sequence CTTCCGG (Ϫ85 to Ϫ79) was essential for the transcriptional activity of aurora A. The sequence contains the ets motif, (C/A)GGA(A/T). Ets proteins recognize a purine-rich (C/A)GGA(A/T) motif in the middle of 10-bp nucleotides, and the sequences flanking the motif determine the specificity of each Ets protein (55-57). We concluded that E4TF1, a member of the Ets family, was the specific nuclear protein that bound to the sequence in aurora A. This conclusion was supported by the following findings. (i) The CTTCCGG sequence contained a core motif ((C/A)GGA(A/T)) for Ets family protein binding, and an E4TF1 competitor com-pletely inhibited the binding of PREBs to the Ϫ90/Ϫ61 oligonucleotide in EMSAs. (ii) The 10-bp sequence CACTTCCGGG (Ϫ87 to Ϫ78) completely matched E4TF1 binding sequences on TPO (33) and mouse cytochrome c oxidase subunit Vb (58).  9. The luciferase activity of aurora A-containing constructs is regulated during the cell cycle. HeLa cells were transiently transfected with pGL189 (squares) or pGL75 (circles). Cells were synchronized with double thymidine block and release (time 0), and harvested at the indicated times for luciferase assays. Luciferase activities were normalized relative to the activity of the co-transfected ␤-galactosidase. Results are expressed as the ratio of luciferase activity in each extract relative to that in the extract from HeLa cells at time 0. The data are presented as the mean Ϯ 1 S.D. of four independent experiments. plex comprising two E4TF1-60-E4TF1-53 heterodimers. By itself, the 60-kDa E4TF1-60 protein binds the specific DNA sequence through the Ets domain in its C terminus but does not stimulate transcription (59,60), and E4TF1-60 associates with E4TF1-53 or E4TF1-47 through the regions flanking the C-terminal Ets domain (61,62). The 53-kDa E4TF1-53 alone neither binds DNA nor stimulates transcription (59,60). The N-terminal 332 amino acids of E4TF1-53 and E4TF1-47 are homologous, but the proteins differ at the C termini. E4TF1-53 homodimerizes through its C terminus, whereas E4TF1-47 does not (61)(62)(63). E4TF1-53 and -47 interact with E4TF1-60 through the notch-ankyrin motifs in their N termini to form heterodimers that activate transcription (61,62). Heterodimers composed of E4TF1-60 and E4TF1-53 are dimerized further through the C terminus of E4TF1-53, resulting in the formation of a tetrameric complex, which stimulates transcription (62,63). Using supershift assays, we determined that PREB-1 was E4TF1 but were unable to identify PREB-2, -3, or -4. Because the formation of Ϫ90/Ϫ61-PREB-2, -3, and -4 (like that of Ϫ90/Ϫ61-PREB-1) was abrogated by the addition of E4TF1 as a competitor, these other PREBs also might belong to the Ets family of transcription factors, especially in light of the fact that multiple Ets family transcriptional factors can stimulate the same gene (43,48,64,65). However, PREB-2, -3, and -4 alternatively may be other forms of E4TF1, such as E4TF1-60 monomer, the E4TF1-60⅐E4TF1-53 heterodimer, or the E4TF1-60⅐E4TF1-47 complex.
By sequencing, we identified another potential cis-element, an Sp1-binding sequence (GGGGCTGGG), from Ϫ129 to Ϫ121 in the aurora A promoter. Sp1 and E4TF1 synergically activate human B19 parvovirus promoter (45), and Sp1 activates G 2 /Mspecific genes such as cyclin A, cdc2, and cdc25C (49,50). However, the 5Ј deletion from Ϫ189 to Ϫ125 had only a small effect on the activity of the aurora A promoter, and co-transfection of an Sp1 expression plasmid with pGL189 did not affect the promoter activity (data not shown). Therefore, we concluded that Sp1 did not contribute to the activity of the aurora A promoter.
In addition, we observed that a 5Ј deletion from Ϫ124 to Ϫ91 resulted in a 50% decrease of aurora A promoter activity in HeLa cells. However, in NIH3T3 cells, pGL90 showed almost the same activity as did pGL124, and we were unable to identify a specific cis-element or transcription factor-binding site in this region by using luciferase reporter assays or EMSAs. Because deletion of Ϫ124 to Ϫ91 affected aurora A promoter activity only in HeLa cells, this region may contribute to highlevel expression in malignant cells or during malignant transformation.
Although we clarified that E4TF1 positively regulated aurora A transcription, we could not explain its cell-cycle-specific expression. In fact, EMSAs showed that E4TF1 bound to the PRE throughout the cell cycle. We therefore thought that the cell-cycle-dependent transcription of aurora A may be negatively regulated by another mechanism.
Recently, it has been reported that G 2 /M-specific genes such as cyclin A, cdc25C, cdc2, Polo-like kinase, p130, and cyclin B2 are regulated by a tandem repressor element that consists of a CDE (consensus sequence: (G/C)GCGG) and a cell cycle gene homology region (CHR; consensus sequence, TTGAA) (49 -53). Just downstream of the aurora A PRE are located two potential CDE sequences; the CGCGG sequence from Ϫ53 to Ϫ49 (CDE-(Ϫ53/Ϫ49)) and the CGCCC sequence from Ϫ44 to Ϫ40 (CDE-Cells were harvested and assessed for luciferase activity. D, relative luciferase activity of cells in G 2 /M versus to that of those in G 1 /S is given for the wild-type and mutant constructs. The data are presented as the mean Ϯ 1 S.D. of four independent experiments. FIG. 10. Effects of mutations of the CDE and CHR repressor elements on aurora A promoter activity. A, alignment of cdc25C, cyclin A, cdc2, cyclin B1, PLK, p130, rabkinesin 6, and aurora A promoter sequences in the region of the CDE and CHR elements (49,(51)(52)(53)67). Box, core sequence. B, a schematic representation of the aurora A-luciferase constructs. Box, CDE/CHR repressor element; X, a mutation site. C, transient transfection experiments with the aurora A-luciferase (LUC) constructs. Wild-type (pGL189) and mutant constructs (pGL189-mCDE and pGL189-mCHR) were transfected into HeLa cells. Cells were treated for 16 h with thymidine to block exit from G 1 /S (open boxes) or with nocodazol to block exit from G 2 /M (solid boxes).
Aurora A kinase activity was tightly regulated during the cell cycle. In parallel with increased mRNA and protein levels, kinase activity rises gradually from the S phase and peaks at the M phase. At the end of the M phase, the mRNA and protein levels decrease markedly, and aurora A protein is degraded through the anaphase-promoting complex-ubiquitin-proteasome pathway (28). In the present study, we demonstrated that cell-cycle-specific transcriptional regulation was at least in part responsible for the tight regulation of the aurora A kinase activity, but we could not clarify the mechanism by which aurora A was expressed at high levels in tumor cells and cell lines. Because several genes that correlate with malignant transformation, such as GnT-V and BRCA1, are regulated by Ets family transcriptional proteins (40,66), E4TF1 may correlate with the high-level expression of aurora A in tumor cells and cell lines. However, because GnT-V, CD18, rat prolactin, and other genes are regulated by E4TF1 and other Ets family proteins (43,48,64,65), an Ets protein other than E4TF1 may contribute to malignant transformation. More research into this question is necessary and is in progress.