Molecular Cloning and Characterization of the Promoter for the Chinese Hamster DNA Topoisomerase IIα Gene

To investigate the mechanisms governing the expression of DNA topoisomerase IIα, the Chinese hamster topoisomerase IIα gene has been cloned and the promoter region analyzed. There are several transcriptional start sites clustered in a region of 30 base pairs, with the major one being 102 nucleotides upstream from the ATG translation initiation site. Sequencing data reveal one GC box and a total of five inverted CCAAT elements (ICEs) within a region of 530 base pairs upstream from the major transcription start site. Sequence comparison between the human and Chinese hamster topoisomerase IIα gene promoter regions shows a high degree of homology centered at the ICEs and GC box. In vitro DNase I footprinting results indicate protection by binding proteins at and around each ICE on both DNA strands. However, no obvious protection was observed for the GC box. Competition gel mobility shift assays with oligonucleotides containing either the wild-type or mutated ICE sequences suggest that identical or similar proteins specifically bind at each ICE, although with different affinities for individual ICE sequences. Chloramphenicol acetyltransferase assays employing nested 5′-deletions of the 5′-flanking sequence of the gene demonstrate that the sequence between −186 and +102, which contains three proximal ICEs, is sufficient for near wild-type level of promoter activity. When these three ICEs were gradually replaced with sequences which do not interact with the binding proteins, reducing promoter activity of the resulted constructs was observed. In conjunction with results from footprinting and gel mobility shift studies, the transient gene expression finding suggests that the ICEs are functionally important for the transcriptional regulation of the topoisomerase IIα gene.

To investigate the mechanisms governing the expression of DNA topoisomerase II␣, the Chinese hamster topoisomerase II␣ gene has been cloned and the promoter region analyzed. There are several transcriptional start sites clustered in a region of 30 base pairs, with the major one being 102 nucleotides upstream from the ATG translation initiation site. Sequencing data reveal one GC box and a total of five inverted CCAAT elements (ICEs) within a region of 530 base pairs upstream from the major transcription start site. Sequence comparison between the human and Chinese hamster topoisomerase II␣ gene promoter regions shows a high degree of homology centered at the ICEs and GC box. In vitro DNase I footprinting results indicate protection by binding proteins at and around each ICE on both DNA strands. However, no obvious protection was observed for the GC box. Competition gel mobility shift assays with oligonucleotides containing either the wild-type or mutated ICE sequences suggest that identical or similar proteins specifically bind at each ICE, although with different affinities for individual ICE sequences. Chloramphenicol acetyltransferase assays employing nested 5-deletions of the 5-flanking sequence of the gene demonstrate that the sequence between ؊186 and ؉102, which contains three proximal ICEs, is sufficient for near wild-type level of promoter activity. When these three ICEs were gradually replaced with sequences which do not interact with the binding proteins, reducing promoter activity of the resulted constructs was observed. In conjunction with results from footprinting and gel mobility shift studies, the transient gene expression finding suggests that the ICEs are functionally important for the transcriptional regulation of the topoisomerase II␣ gene.
Mammalian DNA topoisomerase II (Top II) 1 is an essential nuclear enzyme which changes the topology of DNA by passing an intact helix through a transient double-stranded break made in a second helix followed by religation of the DNA break (reviewed in Refs. 1 and 2). The enzyme functions as a homodimer and in an ATP-dependent manner (3). A feature of Top II function is the covalent attachment of the enzyme to the 5Ј-termini of DNA breaks via a tyrosine-DNA phosphodiester linkage. Top II has been implicated in a number of cellular processes such as synthesis and transcription of DNA (4) and chromosomal segregation during mitosis (5). Top II enzyme also plays a structural role in organizing both mitotic chromosomes and interphase nuclei (6,7). Use of specific antibodies has demonstrated that Top II is a major component of the mitotic chromosomes and the interphase nuclear-matrix fractions (7). Moreover, specific DNA scaffold-attachment sites have been found to contain the consensus cleavage sequence for Top II (8).
Top II is also the target of several classes of anti-cancer drugs such as anthracyclines, amsacrine, and epipodophyllotoxins. These drugs stabilize the cleavable complex formed between Top II protein and DNA, resulting in increased DNA scission and concomitant inhibition of the rejoining reaction (9). The drug-induced DNA breaks are reversible after drug removal. However, most of the cells are arrested in the G 2 phase and eventually die (10).
Resistance to agents that target Top II is a major problem in cancer chemotherapy. In addition to the classical multidrug resistance, which is due to overexpression of the multidrug resistance transporter (mdr protein or P-glycoprotein) (11), atypical multidrug resistance (at-MDR) has been described and is associated with altered Top II activity that is due to either mutated enzyme or a decrease in the amount of the enzyme (11)(12)(13). It is likely that lower Top II levels result in fewer drug-induced DNA lesions and diminished cytotoxicity of Top II-targeting drugs (14,15). A correlation between cellular expression of Top II and the in vitro sensitivity to Top II active anti-tumor drugs has been found in a VM-26-resistant human cancer KB cell line (16), the 9-hydroxyellipticine-resistant Chinese hamster lung fibroblast cell line DC3F/9-OHE (10,17), and in a panel of seven human lung cancer cell lines (18).
In human and probably in other mammals, Top II occurs in two isoforms, the 170-kDa ␣ form and the 180-kDa ␤ form, which are encoded by two discrete genes (19,20). These isoforms have different in vitro sensitivities to antineoplastic agents, different cleavage sites, thermal stability, and inhibition by AT-rich oligonucleotides (21). Recent work has demonstrated that the expression of the 170-kDa form is quantitatively cell cycle-regulated and cell proliferation-related (21,22). The level of expression peaks in the late G 2 to M phases and is greater in rapidly proliferating cells. In proliferating granulocyte precursors, the levels of 170-kDa in vivo were 2-3-fold higher than mature cells and approached the levels in neoplastic cell lines of the same lineage (22). In ras-transformed cells, the proportion of 170-kDa Top II is higher and depends less on growth state than in untransformed cells (23). The ras-transformed cells were also more sensitive to the cytotoxic effects of teniposide and merbarone, drugs which selectively inhibit the 170-kDa form of Top II, indicating a possible link between drug sensitivity and expression of the 170-kDa form (23). The changes in amounts of the mRNA coding for the 170-kDa enzyme were similar to the changes in the 170-kDa enzyme levels, suggesting that the regulation might be mainly at the transcriptional level (23).
In order to investigate the cell cycle-regulated expression of the top II␣ gene and the mechanisms of altered top II expression in drug-resistant cells, genomic clones for the top II␣ gene of Chinese hamster were isolated, and the 5Ј-flanking region of the gene was analyzed. These studies have identified and characterized a group of inverted CCAAT elements, which are present in the proximal promoters of both human and Chinese hamster top II␣ genes, and are functionally important for the transcriptional regulation of the top II␣ gene.

Cell Culture
Wild-type Chinese hamster ovary cells (CHO) were maintained in ␣-minimal essential media, supplemented with 10% fetal bovine serum at 37°C in the presence of 5% CO 2 .

Isolation of CHO Genomic Clones
The Chinese hamster ovary genomic library was purchased from Stratagene, which was prepared by cloning CHO-K1 genomic DNA in Lambda-Fix TM II vector. The phages were propagated in host bacteria P2-392. The transformation and plating procedures were according to the recommendation of the manufacturer. 10 6 plaques were screened under stringent conditions as described elsewhere (24) with the CHO topII␣ cDNA probes, pC431 (5Ј-end probe) or pC42 (middle and 3Ј-end probe) (25). The cDNA probes were multiprimed labeled with the Klenow fragment of DNA polymerase I in the presence of [␣-32 P]dCTP (24). Filter hybridization, recovery of recombinant phage DNA, restriction mapping, as well as the subcloning of genomic fragments into pBluescript and M13 vectors (Stratagene) were as previously described (24). Sequencing was performed on both strands of DNA by using a Sequenase kit (U. S. Biochemical Corp.). Sequencing data were analyzed with MacVector and GCG (Genetics Computer Group) sequence analysis programs.

Determination of Transcriptional Start Site
Primer Extension-A 21-mer oligonucleotide with sequence 5Ј-CTCGTGAGTCCCGAAAGCGAC-3Ј, which is complementary to the cDNA sequence 20 -40 base pairs upstream of the ATG codon, was labeled by T4 polynucleotide kinase and [␥-32 P]ATP. 5 ϫ 10 5 cpm of labeled primer were hybridized to 2 g of CHO poly(A) ϩ RNA or 5 g of control yeast RNA. The annealed primers were extended by Superscript reverse transcriptase (Life Technologies, Inc.), and the extended products were analyzed on a denaturing polyacrylamide gel.
RNase Protection Assay-The 4.0-kb HindIII-SalI genomic fragment (the SalI site was from the vector) was cloned into the pBluescript plasmid. This plasmid was used in the polymerase chain reaction with a pair of primers containing HindIII cloning sites, 5Ј-CGCTAAGCTT-GCTGCAGAAGGCAGGCGGA-3Ј (the forward primer), and 5Ј-CGCTA-AGCTTGGTGACGGTCCTGTAGGG-3Ј (the reverse primer), to amplify the 849-bp proximal fragment upstream from the ATG site. This fragment was cloned in antisense orientation into the HindIII cut pBluescript KS plasmid. Radiolabeled RNA was synthesized from this plasmid by T7 RNA polymerase in the presence of [ 32 P]UTP. Assays were performed with Ambion RPA II TM kit. 1.3 ϫ 10 5 cpm (about 260 pg) of labeled RNA was hybridized to 2 g of CHO poly(A) ϩ RNA or 5 g of control yeast RNA. The hybridized products were digested at 37°C for 30 min with a mixture of 20 units of RNase T1 and 1 g of RNase A. Protected fragments were electrophoresed on a denaturing polyacrylamide gel and visualized by autoradiography.

Chloramphenicol Acetyltransferase (CAT) Transient Expression Assay
Preparation of Constructs-The positive control for the experiment is the pSV2CAT plasmid, which has a simian virus 40 early promoter fused to the CAT coding sequence (26). The parent plasmid for the syntheses of other constructs is the pCAT-(HB) which was prepared in this laboratory. The CAT coding sequence was fused into the pBluescript SK plasmid at the HindIII and BamHI sites. The pCAT-(HB) plasmid has minimal CAT activity and served as a negative control in the CAT assay. To synthesize other constructs, the 849-bp HindIII fragment described in the RNase protection assay was cloned in the sense orientation at the HindIII site upstream of the CAT coding sequence of the pCAT-(HB) plasmid. The resulting plasmid is designated as pCAT-747. The number designated to this and the following plasmids represents the length of 5Ј-flanking genomic sequence (in base pairs) upstream from the major transcription site in the constructs. Exceptions to this are the pCAT-1000 and pCAT-1700 (see below), which represent only the approximate sizes of the genomic fragments.
To prepare the constructs with more adjacent 5Ј-flanking sequence, the 4.0-kb HindIII-SalI fragment in pBluescript plasmid was cut with PstI and the 1.0-kb PstI gemomic fragment (one PstI site was from the vector) was ligated to the PstI-digested pCAT-747 DNA in the sense orientation to give the construct pCAT-1700. Nested-deletions were performed on the KpnI-XhoI-digested pCAT-1700 DNA with Exo III exonuclease (Stratagene) for different time points, mung bean nuclease (Stratagene)-treated, ligated, and transformed into Escherichia coli strain BB4. Screening of transformed cells for clones with different sized 5Ј-flanking sequence yielded the pCAT-1000, pCAT-366, pCAT-223, pCAT-49, and pCAT-0 constructs.
Construct pCAT-186 was prepared by the polymerase chain reaction of the pCAT-747 template with the primer, 5Ј-CGCTCTCGAGAA-GACTCTCCCGCCTCC-3Ј, and the above reverse primer. The amplified DNA was cloned into the pCAT-(HB) plasmid at XhoI-HindIII sites. The construct pCAT-152 was prepared by PstI-BstBI digestion of the pCAT-747 DNA, recovering the vector-containing DNA, blunting the ends with T4 DNA polymerase, and self-ligation.
Site-directed Mutagenesis of the pCAT-186 Plasmid-Transformer TM site-directed mutagenesis kit (CLONTECH) was employed to mutate the three ICEs in the pCAT-186 plasmid according to the protocol of the manufacturer. The sequence of the selection primer is 5Ј-GCCAC-CGCGGTGCATATGCAGCTTTTGTTCCC-3Ј. The underlined sequence represents the NdeI recognition sequence, which replaced the SacI sequence of the plasmid during the mutagenesis reactions. The three mutagenic primers used to replace the first, second, and the third ICE sequences were: 5Ј-CGACTCGGTGCTGGATT CCTCTGAT-3Ј, 5Ј-GAC-CGTCCACGCTGGATTACTCTAAAC-3Ј, and 5Ј-CCTCC TTTACCTA-CTGGATTCATTCGAACAG-3Ј, respectively.
DNA Transfection and CAT Assay-5 g of test constructs were cotransfected with 5 g of ␤-galactosidase expression plasmid, pCH110 (Pharmacia Biotech Inc.), into 1 ϫ 10 6 growing CHO cells by the calcium phosphate precipitation method (26). After 48 h, cells were harvested, and lysates were prepared. Amounts of lysates employed for the CAT activity assays were normalized to the ␤-galactosidase activities. CAT activity assays and ␤-galactosidase activity assays were performed as described elsewhere (26,27). Nonacetylated and acetylated chloramphenicol spots on the TLC plates were quantitated with Phos-phorImager and analyzed by softwares from Molecular Dynamics Inc.

Study of Binding Activities to the 5Ј-Flanking Sequences
Preparation of Nuclear Extracts-Nuclear extracts were prepared from growing CHO cells according to Dignam et al. (28). On average, 500 g of proteins can be obtained from 10 7 cells. Protein determination was performed using the micro BCA protein assay kit (Pierce).
DNase I Footprinting-DNase I footprinting experiments were performed according to Goding et al. (29), except that no ammonium sulfate was added in the incubation buffer. DNA was labeled at one end by a fill-in reaction with the Klenow fragment of DNA polymerase I and the respective radioactive nucleotides. For 3Ј-coding strand labeling, both the 535-bp EcoRI-NcoI DNA and the 285-bp EcoRI-BstBI DNA were labeled with [␣-32 P]dCTP. For 3Ј-noncoding strand labeling, the 535-bp EcoRI-NcoI DNA was labeled with [␣-32 P]dATP and the 383-bp BstBI-BamHI DNA was labeled with [␣-32 P]dCTP. After the incubation with nuclear extract and the DNase I digestion steps, the DNA was phenol and chloroform-extracted, ethanol-precipitated, and analyzed on a 6% denaturing polyacrylamide gel. The (A ϩ G)-sequencing ladders were prepared with the same labeled DNA by the Maxam and Gilbert method (30).
Gel Mobility Shift Assay-Gel mobility shift assays were performed according to Goding et al. (29) with minor modifications in the incubation buffer. The concentrations of the KCl and EDTA were 50 mM and 0.1 mM, respectively. After binding, the DNA-protein complexes were resolved by electrophoresis on a native 4% polyacrylamide gel and exposed to autoradiographic film. For the competition gel mobility shift assay, the competitor oligonucleotide duplexes were as follows. The first ICE-containing oligonucleotides were 5Ј-AGCGACTCGGTGATTGGT-TCCTCTGAT-3Ј and its complementary strand. The second ICE-containing oligonucleotides were 5Ј-AGACCGTCCACGATTGGTTACTC-TAAA-3Ј and its complementary strand. The third ICE-containing oligonucleotides were 5Ј-CCTCCTTTACCTAATTGGTTCATTCGAAC-AGG-3Ј and reverse strand 5Ј-TTCGAATGAACCAATTAGGTAAAG-GAGGCGGG-3Ј. The fourth ICE-containing oligonucleotides were 5Ј-ACAGGAATAGACTATTGGTCTATCCTGAAGAC-3Ј and reverse strand 5Ј-TCAGGATAGACCAATAGTCTATTCCTGTAGCA-3Ј. The fifth ICE-containing oligonucleotides were 5Ј-TGGGCCTTTCTCATTG-GCCAGATTTCCTGTA-3Ј and reverse strand 5Ј-GGAAATCTGGC-CAATGAGAAAGGCCCATGTG-3Ј. The mutated ICE-containing oligonucleotides are the first ICE-containing oligonucleotide with the core 5Ј-ATTGG-3Ј sequence mutated to 5Ј-CTGGA-3Ј and its complementary sequence.

Cloning of the 5Ј-Flanking Region of the Chinese
Hamster topII␣ gene-The Chinese hamster genomic library was screened under stringent conditions with either the pC431 (5Ј-end of the cDNA) probe, or the pC42 (middle and 3Ј-end) probe, of the CHO top II␣ cDNA (25). A total of six overlapping genomic clones were isolated. (The structure of the top II␣ gene will be described in detail elsewhere.) The clone Top II-93 is the only clone which hybridized to the 5Ј-end cDNA probe and not to the pC42 probe. This and another overlapping clone, Top II-21, were further characterized. The 5Ј-end of the cDNA was mapped to a 4-kb HindIII-SalI fragment of Top II-93. This fragment was subsequently subcloned into pBluescript vector and analyzed. Sequencing data revealed the location of the first, second, and third exons and the 5Ј-flanking region of the top II␣ gene (Fig. 1). The ATG translation initiation codon of the cDNA is located at the NcoI site of the genomic fragment. The coding sequence ends abruptly after 21 nucleotides at the PstI site and is followed by the 1.01-kb first intron. The second and the third exons are 153 and 91 bp, respectively. Sequencing of the 0.87-kb EcoRI and the 4.5-kb EcoRI fragments from the clone Top II-21 confirmed the Top II-93 sequence.
Determination of the Transcriptional Start Site of the topII␣ Gene-To locate the transcriptional start site and confirm the translation initiation site, primer extension was performed. A 32 P-labeled 21-mer oligonucleotide complementary to the cDNA sequence 20 -40 base pairs upstream of ATG was employed in the experiment. Several extension products were observed ( Fig. 2A). The major transcript, as deduced from the predominant extension product, is being initiated from the cytosine 102 nucleotides upstream from the ATG codon. Other minor transcripts are initiated at adenine 99, cytosine 110, and thymines 119 and 129, respectively. To confirm the primer extension result, an RNase protection assay was carried out. A chimeric clone containing the genomic sequence upstream of The arrows indicate the minor extension and RNase protection products, while the arrowheads represent the major reaction products. In A, the sequence ladder next to the reaction lanes was produced by sequencing a 5Ј-genomic clone with the same 21-mer primer. There were consistent GC compressions observed in the sequence ladder at positions close to the major transcription site. Another sequence ladder produced by dITP sequencing is shown on the left to resolve the sequence around this region. In B, the sequence ladder produced by sequencing M13 mp18 DNA with universal primer is marked as size marker.
the ATG site of the cDNA sequence was subcloned into pBluescript vector (see "Materials and Methods"), and 32 P-labeled RNA in antisense orientation was synthesized. Fig. 2B shows that several protected RNA fragments with sizes ranging from 100 to 130 bases were detected. The pattern of the RNase protected probes was similar to the primer extension pattern, and the sizes of the protected bands agree with the primer extension data, considering that there is a slight difference in the electrophoretic mobility of the RNA probes and the DNA marker. Downstream of the transcriptional start sites, the ATG initiation codon deduced from the cDNA sequence (25) is the first ATG codon in the genomic sequence, and the sequence, ACCATGG, is a perfect match to the optimal sequence for initiation by eukaryotic ribosomes as suggested by Kozak (31). The major transcriptional start site is designated hereafter as ϩ1 unless otherwise stated.
Sequence Analysis of the 5Ј-Flanking Region of the CHO top II␣ Gene-The region between the transcriptional and translation start sites and the 200-bp region immediately upstream of the transcriptional start site have a moderately high GC content of 64 and 49%, respectively. There is no canonical TATA box sequence, although an imperfect sequence of AAT-GAA was located 26 bp upstream of the predominant transcriptional start site. Further upstream at the Ϫ122 position, there is a TATA-like sequence (AATAAA). The 541-bp 5Ј-flanking sequence from the first upstream EcoRI site to the translation initiation site was searched for potential binding sites for transcription factors. The most prominent sequence motifs in this immediate upstream region are one GC box with potential for Sp1 binding on the coding strand and five CCAAT sequences on the opposite strand (refer to Fig. 1B). These five inverted CCAAT elements (ICEs) are designated one to five according to their proximities to the ATG start codon. There are also two sequences at (60 to 66) and at (Ϫ339 to Ϫ333), which are a one-nucleotide mismatch to a canonical Ap1 sequence T(T/ G)AGTCA. A pair of perfect direct repeat sequences of AGAGCTGAG are located at positions Ϫ327 to Ϫ319 and Ϫ317 to Ϫ309. Downstream from them there is a pair of single mismatched inverted repeats at positions Ϫ300 to Ϫ294 and Ϫ293 to Ϫ287. The 560-bp 5Ј-flanking and first exon sequence was submitted to a search for homologous sequences in the GenBank TM data base. The Chinese hamster sequence shares extensive homology with the human top II␣ promoter sequence (32) and the 5Ј-end of the mouse top II␣ cDNA sequence (33) (Fig. 3A). The most homologous parts between human and Chinese hamster promoters are the region around the transcriptional start site (78% identity from Ϫ18 to ϩ47 of the Chinese hamster sequence), the region immediately upstream of the ATG codon (85% from 59 to 106), and the region around the fifth ICE (80% from Ϫ260 to Ϫ215). The Chinese hamster and the human promoter sequences were aligned by the Bestfit program and are presented in Fig. 3B. The GC box, TATA-like element, and the first three ICEs of the hamster sequence can be aligned to the corresponding elements of the human sequence. However, the fourth hamster ICE element is positioned at a sequence having one mismatch with the human gene. The fifth hamster ICE is aligned with the fourth ICE of the human sequence, whereas the fifth human ICE does not have a homologous counterpart in the hamster sequence. Except for the fourth hamster ICE, regions around the aforementioned sequence elements share a relatively high homology between the hamster and human genes. The overall similarity between the human and Chinese hamster promoter sequences is 67%. The sequence similarity diminishes upstream of the Chinese hamster fifth ICE. Interestingly, the translation start site sequence including part of the first exon of the Chinese hamster top II␣ gene shows a 21/23 identity with the translation start site of the gene for human IgM heavy chain (Fig. 3A). It is not known though if the translational regulation of both genes have any similarity.

Analysis of the Proximal cis Elements in the 5Ј-Flanking Region by in Vitro DNase I Footprinting-
The 535-bp EcoRI-NcoI fragment encompassing the GC box and all of the five ICEs was labeled at either the EcoRI or NcoI ends and subjected to in vitro DNase I footprinting. To resolve the footprints away from the labeled ends, two smaller fragments, the 285-bp EcoRI-BstBI fragment and the 383-bp BstBI-BamHI fragment, were also labeled, respectively, at the BstBI site for the footprinting analysis of coding and noncoding strands. Analysis on both coding and noncoding strands of the 535-bp region is presented in Fig. 4. Footprints are observed at all the inverted CCAAT sequences and the regions juxtaposed to them on both coding and noncoding strands. The footprints of some flanking sequences, for example, around the fifth ICE on the coding strand, are more pronounced than the inverted CCAAT sequences. The fourth ICE has a lesser extent of protection than the others. Whereas there were no footprints observed at the TATA-like element and the GC box of the coding strand, analysis of the noncoding strand demonstrated a marked footprint at the TATA-like element and a small footprint at the 3Ј-end of the GC box (Fig. 4B). There are some enhanced cleavages with nuclear extracts observed at positions upstream of the fifth ICE and the TATA-like sequence, respectively, as well as downstream of the third ICE region on the noncoding strand foot-printing. This may suggest protein-induced DNA bending at these regions.
Analysis of the Binding Activities to the Inverted CCAAT Elements by Gel Mobility Shift Assay-The binding of protein  1-5), and with the labeled distal fragment, the 285-bp EcoRI-BstBI DNA (lanes 6 -10) are presented. The labeled DNAs were incubated without nuclear extracts (lanes 1 and 6) or with 5 g of nuclear extracts (lanes 2 -5 and 7-10). The binding of proteins to the labeled DNAs was competed with a 50-fold molar excess of proximal fragment DNA (lanes 3 and 8), distal fragment DNA (lanes 4 and 9), or the multicloning region of pBluescript DNA (lanes 5 and 10). The free labeled DNAs (F) and the nucleoprotein complexes are indicated. factors to the proximal DNA elements including the ICEs was analyzed further by gel mobility shift assay. To facilitate an initial assay of the entire 535-bp EcoRI-NcoI region, this region was split into two fragments, the 383-bp BstBI-BamHI (proximal) fragment, and the 285-bp EcoRI-BstBI (distal) fragment, and were used separately for gel mobility shift assays (Fig. 5). These two fragments bound to protein factors from the nuclear extracts and migrated as distinct complexes on the native polyacrylamide gel. There are in total three distinct DNAprotein complexes observed in the assay with the 383-bp BstBI-BamHI proximal fragment (Fig. 5, complexes A, B, and C of lane 2). There are also three bands of DNA-protein complexes observed in the assay with the 285-bp EcoRI-BstBI distal fragment (complexes AЈ, BЈ, and CЈ of lane 7), although the complexes BЈ and CЈ are not well separated. All the complexes were competed out when a 50-fold molar excess of the same unlabeled fragments were included in the binding reaction (Fig. 5,  lanes 3 and 9). The complexes formed are sequence-specific because they were not depleted by the addition of a 50-fold molar excess of nonspecific pBluescript competitor DNA (Fig. 5,  lanes 5 and 10). Interestingly, when the distal fragment was used as competitor DNA in the assay with labeled proximal fragment (Fig. 5, lane 4), complex A and most of complex B were depleted. If the proximal fragment was used as competitor DNA in the assay with labeled distal fragment, the AЈ and CЈ and some of the BЈ complexes were depleted (Fig. 5, lane 8).
These results indicate that some proteins bound to the proximally located elements may be identical or similar to those binding to the more distally located elements.
Since footprints were mostly observed at the ICEs, the ICEs are probably the binding sites for the nuclear factors in the complex formation of both fragments. To demonstrate this, we synthesized five pairs of oligonucleotide duplexes which encompass the first through five ICEs and their complementary sequences, respectively (Fig. 6B). These oligonucleotide duplexes were used in the competition gel mobility shift assays (Fig. 6A,  lanes 2-7 and lanes 11-18). In an experiment employing the proximal fragment, the first, second, and the third ICE-containing oligonucleotides were effective in competing for the formation of complexes A and B (Fig. 6A, lanes 2-7). Experiments with the fourth and the fifth ICE-containing oligonucleotides also demonstrated competition for formation of complexes A and B (data not shown). In the assay with the labeled distal fragment (Fig. 6A, lanes 10 -18), excess amounts of ICE-containing oligonucleotides could deplete the formation of com-FIG. 7. Gel mobility shift assays of the radiolabeled ICE-containing oligonucleotides. The reactions with the 32 P-labeled first (lanes 1, 2, and 13-15), second (lanes 3 and 4), third (lanes 5 and 6), fourth (lanes 7 and 8), fifth (lanes 9 and 10), and mutated (lanes 11 and 12) ICE-containing oligonucleotides are shown. The incubations were carried out without (lanes 1, 3, 5, 7, 9, and 11) or with 5 g (lanes 2, 4,  6, 8, 10, and 12-15) of nuclear extracts. The binding of nuclear proteins to the labeled first ICE-containing oligonucleotide was competed with a 50-fold molar excess of the first (lane 13) and mutated (lane 14) ICEcontaining oligonucleotides or the nonspecific multicloning region of pBluescript DNA (lane 15). The specific DNA-protein complexes are indicated by the arrow.
plexes AЈ and CЈ and partially the complex BЈ. The first ICEcontaining oligonucleotide functioned as a competitor at least as well as the ICE-containing oligonucleotides derived from the distal fragment. The addition of 50-fold molar excess of ICEcontaining oligonucleotide competitors resulted in low level complex formation of BЈ (Fig. 6A, lanes 11, 13, 15, and 17). However, the bands were not completely depleted even with a 500-fold excess of competitors (Fig. 6A, lanes 12, 14, 16, and  18), suggesting that a portion of BЈ was derived from complexes formed by non-CCAAT binding activity. These findings suggest that the complexes A, B, AЈ, CЈ, and partially BЈ were composed of proteins which recognized the common sequences of all the five ICE-containing oligonucleotide competitors. Since the sequence 5Ј-ATTGG-3Ј (its complementary sequence is 5Ј-CCAAT-3Ј) is the common sequence represented by all five ICEs, the binding proteins are likely to be CCAAT-binding proteins. The second (lanes 4 and 5) and fourth (lanes 13 and 14) ICE-containing oligonucleotide competitors were not as effective as other ICE-containing competitors in the competition. To confirm that the complexes are formed by CCAATbinding proteins, an oligonucleotide duplex containing the same flanking sequences as the first ICE but with the core ATTGG sequence mutated to CTGGA was employed in the competition gel mobility shift assay (Fig. 6A, lanes 8, 9, 19, and  20). A 50-fold molar excess of the mutated ICE-containing competitor did not compete for complex formation, while large excess amounts (500-fold) of the mutated competitor could compete for the formation of complexes A and AЈ.
To analyze the CCAAT-binding activities to the ICEs, the six pairs of oligonucleotide duplexes were radiolabeled and employed in the gel mobility shift assay (Fig. 7). A DNA-protein complex was formed with each ICE-containing oligonucleotide duplex but not with the mutated ICE-containing oligonucleotide duplex (lane 12). The mutated ICE-containing oligonucleotide duplex was also not an effective competitor for the complex formation (lane 14). These are consistent with the previous results (Fig. 6) that the bindings to the ICE-containing oligonucleotides were by CCAAT-binding proteins. All of the ICEspecific complexes comigrated in the gel, suggesting that the same CCAAT-binding factors or proteins of similar electrophoretic properties were bound to the oligonucleotides. There was less complex formation with the fourth ICE-containing oligonucleotide duplex (lane 8), despite the fact that equal amounts of radiolabeled oligonucleotides were used. There were also some faster migrating complexes observed in the gel mobility shift assays, which might be due to protein degradation or some unknown protein bindings not specific to the ICEs.
Delineation of the 5Ј-End of the CHO top II␣ Promoter and the Functional Analysis of the ICEs-To search for a DNA sequence important for in vivo promoter activity, constructs with nested 5Ј-deletions of the 5Ј-flanking sequence through the residue immediately upstream of the ATG codon were fused to the CAT coding sequence and transfected into CHO cells. CAT activities were assayed from the cell lysates (Fig. 8). pCAT-0 (construct with deletion of sequence upstream of the transcriptional start site) and pCAT-49 (construct with an in- tact GC box) did not have any measurable promoter activity (compare Fig. 8, lane 1, the negative control, with lanes 2 and 3), suggesting that the GC box alone is not sufficient for promoter activity and additional upstream elements are required. pCAT-152 (Fig. 8, lane 4) had promoter activity of about 25% that of the other constructs with more upstream sequence. With only 34 bp of additional upstream sequence, the construct pCAT-186 elicited near maximum promoter activity which is similar to constructs with more upstream sequence (compare Fig. 8, lane 5 with lanes 6, 7, 8, and 10). All constructs yielded CAT activity of about one-third that directed by the simian virus 40 early promoter (Fig. 8A, lane 11). Hence, the 5Ј-limit of the top II␣ gene promoter can be localized between the Ϫ186 and Ϫ152 region. Interestingly, the pCAT-1000 plasmid (lane 9) consistently resulted in about 50% decrease in promoter activity.
Since the pCAT-186 construct encompasses three ICEs and CCAAT-binding proteins were observed to interact with the ICEs in the previous experiments, the functional role of ICEs in promoter activation was analyzed by site-directed mutagenesis and transient gene expression (Fig. 9). When the first ICE was mutated, about 40% of pCAT-186 promoter activity was lost (Fig. 9, lane 4). Mutation of the third ICE alone also gave similar results (data not shown). When both the first and third ICEs were mutated (lane 5), 72% of the promoter activity was lost. With further mutation of the second ICE (lane 6), the promoter activity was diminished to about 23%. DISCUSSION We have isolated genomic clones that contain the promoter elements of the Chinese hamster top II␣ gene. Comparison of the available genomic sequence of the human top II␣ gene (32) with the Chinese hamster gene suggests that the genomic structures of these two genes may be similar. In both genes, the first exon is comprised of about 90 -102 nucleotides as an untranslated region, and 21 nucleotides as the coding sequence. The first and second introns are of similar sizes and are of the same class. 2 The promoters of these two genes share a high degree of homology (67% sequence identity). The highly homologous regions are centered around and between the transcription and translation initiation sites, and the ICE areas in the 5Ј-flanking region (Fig. 3). The genomic sequence for the mouse top II␣ gene is not available, but comparison of the Chinese hamster sequence with the 5Ј-end of the mouse cDNA sequence demonstrates 86% sequence identity (Fig. 3). The 5Ј-flanking sequence, however, does not share any homology to the Drosophila and yeast sequences (data not shown). This suggests that mammalian top II␣ genes may share the same transcriptional regulation machinery.
The Chinese hamster top II␣ gene promoter has a moderately high GC content, no canonical TATA box sequence, and the transcriptional start sites are scattered in several discrete positions. These are the characteristics of promoters of genes that have housekeeping and growth-related functions (34). Like many housekeeping genes, the promoter of Chinese hamster top II␣ gene contains a GC box with the potential for binding of the transcription factor Sp1. However, footprinting analysis of the top II␣ promoter did not reveal any bona fide protection of the GC box. GC box elements that do not bind Sp1 are also observed in other promoters such as the herpesvirus immediate-early 3 (ICP-4) promoter (35).
Five ICEs were found scattered within the 400-bp proximal promoter region in the Chinese hamster top II␣ gene. DNase I footprinting and gel mobility shift assays demonstrated the binding of sequence-specific proteins at and around the ICEs. The CCAAT sequence is a moderately conserved transcriptional regulatory element in many eukaryotic promoters, such as histone (36), albumin (37), globin (38), major histocompatibility complex class II (39), and viral gene promoters (40,41), and has been shown to function in either orientation (40). Inverted CCAAT elements are important for the cell-cycle regulation of transcription in the human thymidine kinase gene (42,43) and the serum induction of transcription from the human DNA polymerase ␣ gene (44) and transcription from the long terminal repeat of Rous sarcoma virus (41). Several proteins that specifically recognize CCAAT elements have also been characterized (37,39,45). In the analysis of the ICEs in the top II␣ promoter, competition with ICE-containing oligonucleotides in the gel mobility shift assays employing labeled proximal and distal fragments suggests that complexes A, B, AЈ, CЈ, and part of BЈ are ICE-binding complexes (Fig. 6A). Although they may be formed by different proteins, the combined results of Figs. 6 and 7 suggest that the depletable complexes are more likely formed by the same or very similar ICE-binding proteins to the ICEs with different affinities. For example, in the gel mobility shift assay with the proximal fragment, the ICE-binding protein would bind to the first ICE with greater affinity to form complex B. Additional binding of the ICE-binding protein to the second ICE of the proximal fragment with lower affinity produced the less intense complex A. In the competition assays, addition of the ICE-containing oligonucleotides would easily compete out the binding to sites with lower affinity and disrupt the formation of higher ordered complexes. The fourth and second ICEs have less affinity to the ICE-binding proteins since the fourth ICE-containing oligonu-2 S.-W. Ng, J. P. Eder, and L. E. Schnipper, unpublished data. cleotide formed less complex (Fig. 7), and the fourth and second ICE-containing oligonucleotide competitors were less effective in the competition gel mobility shift assays (Fig. 6A). This is consistent with the result of DNase I footprinting, in which the fourth ICE exhibited lesser protection from DNase I digestion (Fig. 4). The different affinities of the ICE-binding protein to the ICEs can be a function of the interaction with the flanking sequences around the core ATTGG sequence. This may also account for the partial competition of 500-fold molar excess of mutated ICE-containing oligonucleotide for the complex formation (Fig. 6A). All five ICEs are similar in that they have a pyrimidine-rich 3Ј-flanking sequence (Fig. 6B), and like many CCAAT elements, they are asymmetrical. However, alignments of the ICE sequences did not show any obvious flanking sequence residues which suggest distinction between the high affinity ICEs and the low affinity ICEs.
Transient gene expression assays have delineated the 5Јlimit of the functional promoter to the region between 186 and 152 bp upstream of the major transcription site. 5Ј-Deletion beyond this limit significantly reduces the promoter activity. The three ICEs in this core promoter region were analyzed by site-directed mutagenesis and transient gene expression (Fig.  9). Mutations of the ICEs elicited reduction in basal promoter activity, although a residual promoter activity remained when all three ICEs were mutated. This suggests the activation role of ICEs in the transcriptional regulation of the top II␣ gene and some other elements may be present in the core promoter for the residual activity. The reduction in basal promoter activity was additive for the first and third ICE mutations, whereas the mutation of the second ICE had a minimal effect on the decrease of promoter activity. Thus, the in vitro binding activity of the ICEs is likely consistent with their in vivo activation activity.
In summary, these studies have characterized the promoter region of the Chinese hamster topoisomerase II␣ gene and the five protein-binding ICEs which have promoter activation function. Further study is required to characterize CHO topII␣ ICE-binding protein(s) and compare them to other CCAATbinding proteins, as well as other embedded elements in the promoter and the upstream regions which regulate both the basal and cell cycle-regulated expression of the top II␣ gene.