Molecular cloning and sequence analysis of the promoter region of mouse cyclin D1 gene: implication in phorbol ester‐induced tumour promotion

Library screening

Approximately 4 × 10⁶ plaque‐forming units (pfu) of a mouse genomic library 129SV (containing 2.0 × 10⁶ primary plaques) in λ FIX II vector (Stratagene, La Jolla, CA, USA) were screened using a nonradioactive PCR‐based method (Amaravadi & King 1994). The primer sequences used were as follows: sense (30‐mer), 5 ′‐CCTCCAGAGGGCTGTCGGCG CAGTAGCAGA — 3′ antisense (30‐mer), 5′‐TCTTACCTCCAGCATCCAGGTGGCCACGAT — 3′. This primer set covers entire 5 untranslated region and coding sequence of mouse cyclin D1 exon 1 (Smith, Peters & Dickson, 1995). The 5′ and 3′ flanking sequences of the exon 1 were included (underlined) in the sense and antisense primers, respectively, to enhance the specificity of the PCR product. The expected size of the product was 348 bp.

In the primary screening, 20 150‐mm plates were used and approximately 200 000 plaque‐forming units (pfu) were plated on each plate using XL1‐Blue MRA (P2) as host E. coli. Plates were incubated until plaques began to contact each other (6–8 h). To soak phage from the plates, SM buffer (10 ml) was added to each plate. The plates were then placed on a slowly rotating shaker and incubated overnight at 4 °C. One milliliter of the SM buffer was collected from each plate and centrifuged at 5000 × g for 10 min to remove debris. E. coli were lysed by the addition of CHCl₃ to a final concentration of 0.3% (v/v), and the phage stocks were stored at 4 °C.

A 1‐µl aliquot from each phage stock was added to a 12.5‐µl PCR master reaction mixture and PCR amplifications were performed on a TwinBlock System (ERICOMP, San Diego, CA) using the following procedure: 35 cycles with a denaturation step at 95 °C for 30 s, annealing at 50 °C for 30 s, extension at 72 °C for 60s, followed by a final extension at 72 °C for 7min. For each amplification, Taq DNA polymerase (0.5 U; Promega Corporation, Madison, WI, USA) was used with 5 pmol of each primer, 0.2 mM of dNTPs and 1.5 mM of MgCl₂. An aliquot (8 µl) of each amplification reaction was analysed by electrophoresis in a 1.8% agarose gel stained with ethidium bromide.

The positive phage stocks from the primary plates were subjected to additional rounds of phage plating (initially on 150‐mm and 100‐mm plates and then on 60‐mm plates) and analysis by PCR. Final purification of the positive plaques was performed by three rounds of phage plating (on a 100‐mm plate), picking individual plaques, soaking each in a 1‐ml SM buffer, and analysing by PCR. The purified positive phage stocks were stored at −80 °C in SM buffer containing 7% (v/v) DMSO.

Subcloning and DNA sequencing

A positive recombinant phage recovered after multiple rounds of screening and plaque purification carried an insert (approximately 23 kb in size) containing both exon 1 and promoter region of mouse cyclin D1 gene. This insert was subcloned into pBluescript vectors (Stratagene) as outlined in Fig. 1.

The recombinant λ FIX II phage DNA (λmCD1; see Fig. 1) was prepared by plate lysate method using a combination of λ Quick Spin Kit (BIO 101, Inc., Vista, CA, USA) and QIAGEN Lambda Midi Kit (QIAGEN, Inc., Valencia, CA, USA). Digestion of the entire λmCD1 DNA by EcoRI produced three fragments (E1, E2 and E3). E1 was the largest fragment containing λ FIX II phage vector. Of the other two fragments (E2 and E3 with approximately 14—15 kb and 7—8 kb, respectively, in size), E3 carried both exon 1 and the 5′ flanking region immediately upstream of the exon 1.

The fragment E3 was subcloned into EcoRI site of the pBluescript KS(+) vector (pmCD1E3; see Fig. 1) using chemically competent One Shot TOP10 cells (Invitrogen, Carlsbad, CA, USA). The cells transformed with pmCD1E3 were identified by ampicillin resistance, blue‐white selection with X‐gal, direct PCR screening of colonies (Lee & Cooper 1995) and restriction analysis. Direct PCR screening of the colonies was performed using the same PCR protocol described above except the reaction mixture was first incubated at 95 °C for 10 min prior to the standard 35‐cycle PCR program.

The pmCD1E3 was then digested with either BamHI or Spe I. The resulting fragments containing both exon 1 and the promoter region (approximately 1.6 kb in size after BamHI digestion and 3—4 kb after Spe I digestion) were subcloned into pBluescript II KS(+) vector (pmCD1E3B3 and pmCD1E3S2, respectively; see Fig. 1).

The plasmid DNAs were isolated using a combination of QIAprep Spin Miniprep Kit and QIAGEN Plasmid Mini Kit (QIAGEN, Inc.) and dissolved in water. The exon 1 and the 5′ flanking region immediately upstream (approximately 1.02 kb in size) of the exon 1 was sequenced for both strands using automated ABI Prism Model 377 DNA sequencer available at the University of Alabama at Birmingham Automated DNA Sequencing Core Facility.

The sequence was submitted to the GenBank database under accession number AF212040.

Computer‐aided analysis of DNA sequence

Genetics Computer Group (GCG) Sequence Analysis Programs (Version 10) (Genetics Computer Group, Inc., Madison, WI, USA) were used to analyse the DNA sequence. GCG provides Web‐based (SeqWeb) interface for access to its program suite. Default conditions were used to run all programs.

Two programs (Compare and MFold) were used to analyse the global structure of the DNA sequence. Compare Program was used to compare mouse and human or mouse and rat cyclin D1 promoter sequences and create dot plot graphs that show where the two sequences are similar. A window of 21 bases was used with a stringency of 14. MFold Program was used to analyse the DNA secondary structure of mouse cyclin D1 promoter. Default conditions consisted of the following: folding temperature of 37 °C, colour levels of suboptimality of 1, radius of circle of 45.0, maximum number of structures to display was 25, and plotted foldings differ by a distance of at least 3. The result was displayed graphically as an energy dot plot.

In order to locate and identify various promoter motifs, MotifSearch Program was first run to search the overall consensus segments between mouse and human or rat and human cyclin D1 promoter sequences. Once identified, the overall consensus segments were analysed further by visual inspection and using BestFit Program to locate and align individual consensus motifs.

PCR‐TA cloning

The cyclin D1 promoter sequences of promotion‐sensitive (P+, clone 41.5a) and resistant (P−, clone 30.7a) JB6 mouse epidermal cells were determined by PCR‐TA cloning method.

The cells were grown as monolayer cultures (using 100‐mm dish) as described previously (Eto 1998) in Eagle's Minimum Essential Medium containing 5% heat‐inactivated fetal bovine serum, 2% l‐glutamine and antibiotics. Genomic DNAs were isolated from these cells using Puregene Genomic DNA Isolation Kit (Gentra Systems, Minneapolis, MN, USA) and digested with EcoRI.

Three sets of primers were used to amplify the cyclin D1 promoter region by PCR: (a) sense (30‐mer), 5′‐TCTGC CCGGC TTTGA TCTCT GCTTA ACAAC‐3′ and antisense (30‐mer), 5′‐TCTTA CCTCC AGCAT CCAGG TGGCC ACGAT‐3′ (b) sense (30‐mer), 5′‐AAATA ATGGC CACCA TCTTG AGCTG TTGCT‐3′ and antisense (27‐mer), 5′‐GAGTT TTGTT GAAGT TGCAA AGTCC TG‐3′ and (c) sense (20‐mer), 5′‐AATTC TAAAG GTGGG GGAAC‐3′ and antisense (27‐mer), 5‐TTACA CGTGT TGATG AAATT GAAAG AA‐3. The expected sizes of the products were 432, 727 and 525 bp, respectively. These three products overlap 81 bp (primer sets a and b) and 229 bp (primer sets b and c) in size and cover entire exon 1 and the 5′ flanking region immediately upstream (approximately 1.02 kb in size) of the exon 1. PCR amplifications were performed using the standard 35‐cycle PCR protocol described above.

TA cloning of the PCR products was performed using TOPO TA Cloning Kit (Invitrogen) with pCR®2.1‐TOPO as the plasmid vector and chemically competent One Shot TOP10 cells as the host E. coli. The linearized TA cloning vector provides 3′ T overhangs with covalently bound topoisomerase I. Screening of the transformed cells and the DNA sequencing of the inserts were performed as described above.

Genomic library screening, subcloning and DNA sequencing

Overall structure of the mouse cyclin D1 gene is shown in Fig. 2. A mouse genomic library 129SV was screened by a nonradioactive PCR method (Amaravadi & King 1994) using a primer set covering the entire 5 untranslated region and coding sequence of exon 1 (Smith et al. 1995). Only one positive clone was identified for each 2.0 × 10⁶ primary plaques suggesting that cyclin D1 is a single copy gene. The positive phage DNA was digested with EcoRI and analysed by PCR to identify a fragment containing both exon 1 and the 5′ flanking region immediately upstream of exon 1. The positive fragment (approximately 7–8 kb in size) was subcloned into pBluescript KS(+) vector (pmCD1E3; see Fig. 1). The plasmid thus obtained was further digested with either BamHI or Spe I and the positive fragments (approximately 1.6 kb and 3–4 kb, respectively, in size) were further subcloned into pBluescript II KS(+) vector (pmCD1E3B3 and pmCD1E3S2, respectively; see Fig. 1). The nucleotide sequence of the 5 flanking region approximately 1150 bp upstream of the translation initiation codon ATG was determined for both strands (GenBank accession number AF212040; Fig. 3).

Dot matrix homology analysis of the promoter sequence

The dot plot graphs of the similarities in sequence between mouse and human (Herber et al. 1994; Muller et al. 1994; Philipp et al. 1994) and mouse and rat (Yan & Ziff 1997; Kitazawa, Kitazawa & Maeda, 1999) cyclin D1 promoters are shown in Fig. 4(a),(b), respectively. A single, but broken stretch of similarity is evident along the diagonal between mouse and human sequences (Fig. 4a) and an almost unbroken stretch is visible between mouse and rat sequences (Fig. 4b). These graphs indicate that the mouse promoter sequence is homologous to the human and more so to the rat sequences.

Overall organization of mouse cyclin D1 promoter

Since several descriptions of the organization of human cyclin D1 promoter region are available in the literature (Herber et al. 1994; Muller et al. 1994; Philipp et al. 1994; Fukami‐Kobayashi & Mitsui 1998), MotifSearch Program was first run to search the overall consensus segments between human and mouse cyclin D1 promoter sequences. These and other segments were analysed further by visual inspection and using BestFit Program to locate and align individual consensus promoter motifs. The same strategy was also used to analyse the organization of rat cyclin D1 promoter region (Yan & Ziff 1997; Kitazawa et al. 1999).

As Fig. 5 shows, the overall organizations of mouse, rat and human cyclin D1 promoters are fairly similar to each other. All three promoters lack canonical TATA‐box or TATA‐like sequence, but they contain initiator (Inr) (human) or Inr‐like element (mouse and rat). When the nucleotide position + 1 was assigned to this Inr element (Philipp et al. 1994; Yan & Ziff 1997), 5′ end of the ATG start codon in exon 1 was positioned at + 233 in mouse and rat and + 251 in human (due to the insertion of additional 18bp at approximately + 230). The 5′ end of 5′ untranslated sequence of mouse exon 1 was located at + 96 (Smith et al. 1995).

Using the same nucleotide coordinate, all three promoters were found to contain NF‐κB response element, CRE (cAMP response element) and Inr or Inr‐like element at + 57 to + 65, + 38 to + 45 and + 1 to + 8, respectively (Figs 5 and 6b). In addition, all three promoters contain Oct 1, Sp1 and E2F binding sites (Figs 5 and 6c). They also contain TRE (TPA response element or AP‐1 binding site) and a tandem silencer element (Figs 5 and 6a).

On the other hand, the BRE (butyrate response element) located at − 242 to − 252 in human promoter (Lallemand et al. 1996; Tran et al. 1998) could not be identified in either mouse or rat promoter. Also, the enhancer element or SRE (serum response element) located at either − 832 to − 839 (Herber et al. 1994) or − 836 to − 843 (Fukami‐Kobayashi & Mitsui 1998) in human promoter could not be detected in either mouse or rat promoter (Fig. 5). In addition, E2F binding motif located within approximately − 43 to − 54 in human promoter (Herber et al. 1994; Phillipp et al. 1994; Fukami‐Kobayashi & Mitsui 1998) was not found in mouse or rat promoter (Fig. 5). Finally, rat (Yan & Ziff 1997; Kitazawa et al. 1999) and human (Herber et al. 1994; Albanese et al. 1995; Fukami‐Kobayashi & Mitsui 1998; Kim & Miller 1998; Watanabe et al. 1998) promoters contain two Sp1‐binding sites between − 9 and − 24, but only one Sp1‐binding site was identified within this segment in the mouse (Figs 5 and 6c)

Silencer and TRE

As shown in Fig. 6a, the silencer and TRE sequences are located within the most 5′ distal region (nucleotide positions from approximately − 789 to − 877) of all three promoters.

The tandem silencer elements (Herber et al. 1994; Fukami‐Kobayashi & Mitsui 1998) of mouse (5′‐TTTAAT‐3′ and 5′‐TTTACT‐3′), rat (5′‐CTTAAT‐3′ and 5′‐TTTACT‐3′) and human (both 5′‐TTTAAT‐3′) are embedded in a stretch of AT‐rich sequence (Fig. 6a). Herber et al. (1994) suggested that this unusual structure might produce DNA bending in this region, thereby contributing to its inhibitory effect on transcription. Fukami‐Kobayashi & Mitsui (1998) determined the consensus oligonucleotide sequence of this element and identified a 35‐kDa protein binding to this site.

In the human promoter, a perfect TRE consensus sequence (5′‐TGAGTCA‐3′) (Herber et al. 1994; Albanese et al. 1995; Fukami‐Kobayashi & Mitsui 1998; Watanabe et al. 1998; Albanese et al. 1999) is located immediately downstream (separated by 9 bp) of the tandem silencer element (Fig. 6a). In the mouse and rat promoters, TRE has the nucleotide sequence 5′‐TGTCTCA‐3′ and is located approximately 30 bp downstream of the silencer sequence. The AP‐1 (activating protein‐1) family of transcription factors consists of homodimers and heterodimers of Jun (v‐Jun, c‐Jun, JunB, JunD), Fos (v‐Fos, c‐Fos, FosB, Fra1, Fra2) or activating transcription factor (ATF2, ATF3/LRF1, B‐ATF) bZIP (basic leucine zipper) proteins (Karin, Liu & Zandi, 1997). Jun‐Jun and Jun‐Fos dimers preferentially bind to the TRE. Albanese et al. (1999) has recently reported that the coactivator protein p300 activates the human cyclin D1 promoter through this site and that it was identified within a DNA‐bound complex with c‐Jun at this site.

Inr or Inr‐like sequence, CRE and NF‐κB response element

These three sites are clustered together within the most 3′ proximal region (nucleotide positions from approximately + 1 to + 65) of all three promoters (Fig. 6b).

The Inr sequence of human promoter has the sequence 5′‐CATTCTCT‐3′ (Phillipp et al. 1994). As shown in Fig. 6(b), this sequence is very similar to those from several other genes. The corresponding Inr sequence of the mouse and rat promoters (5′‐TTTTCTCT‐3′) does not contain CA at the 5′ end. Rather it contains TT in place of CA. This makes the mouse and rat Inr sequences somewhat different from that of the human promoter.

The CRE is located at nucleotide positions from + 38 to + 45 in all three promoters and has the same sequence 5′‐TAACGTCA‐3′ (Fig. 6b). The CRE‐binding protein complexes include not only CREB (CRE‐binding protein) but also AP‐1 complex proteins such as Jun‐ATF dimers, ATF homodimers or CREB‐ATF dimers (Karin et al. 1997). cAMP and AP‐1 appear to exert opposing effects on the expression of cyclin D1. Elevated levels of cAMP tend to downregulate cyclin D1 expression (L'Allemain et al. 1997; Cospedal, Lobo & Zachary, 1999; Fukumoto et al. 1999; Gagelin et al. 1999; Kronemann et al. 1999; Musa et al. 1999), whereas AP‐1 complexes tend to upregulate it (Herber et al. 1994; Beier et al. 1999; Sabbah et al. 1999).

NF‐κB response element is one of the latest motifs to be identified in the human cyclin D1 promoter (Guttridge et al. 1999; Hinz et al. 1999; Joyce et al. 1999). As Fig. 6b shows, in all three promoters, the NF‐κB response element is located at nucleotide positions from + 57 to + 65 and has the same sequence 5′‐GGGGAGTTT‐3′.

Energy dot plot analysis of mouse cyclin D1 promoter

MFold Program (Zuker 1989; Jacobson & Zuker 1993; Zuker & Jacobson 1995) was used to predict secondary structure of mouse cyclin D1 promoter. This program produces a two‐dimensional triangular energy dot plot in which predicted base‐pairs are plotted as points. The energy dot plot is a useful representation of the folding or global domain structure of large DNAs or RNAs.

The plot (Fig. 7) shows three triangular regions above the diagonal, corresponding to base pairs within each of the following three domains: (a) nucleotide positions downstream of − 280 (b) from − 280 to − 700, and (c) upstream of − 700. The rectangles above and to the right of the middle triangle (the second domain) contain fewer dots, which means that base pairing between the three domains is less likely. Alternative structures are more likely to occur from alternative foldings within these domains.

Promotion‐sensitive and resistant JB6 mouse epidermal cells

The cyclin D1 promoter regions of promotion‐sensitive (P+ , clone 41.5a) and resistant (P−, clone 30.7a) JB6 cells were cloned using PCR‐TA method. The three sets of PCR primers were designed to amplify the nucleotide positions from + 431 to − 937 with overlaps of 81 bp and 229 bp in size.

DNA sequence analysis of the clones indicated that, except for a few minor differences in the nucleotide sequence, mostly within the nucleotide positions from − 56 to − 85, the cyclin D1 promoters of both promotion‐sensitive and resistant JB6 mouse epidermal cells (data not shown) were identical to the sequence determined for a clone isolated from 129SV mouse genomic library (Fig. 3).

Comparative organizations of mouse, rat and human cyclin D1 promoters

The results show that all three promoters have similar organizations — at least within 1.15 kb immediately upstream of the ATG start codon in exon 1. In this region, the same promoter motifs are located at more or less comparable nucleotide positions in all three promoters. For example, using the three domain structure predicted by the energy dot plot analysis and assigning the nucleotide position + 1 to the Inr or Inr‐like element, the most 3′ domain (nucleotide positions downstream of − 280) contains NF‐κB response element, CRE, Inr or Inr‐like elements, Sp1 binding site or sites and Oct 1. The middle domain (nucleotide positions between − 280 and − 700) contains an additional Sp1 binding site, E‐box and E2F binding sequence. The most 5′ domain (nucleotide positions upstream of − 700) contains TRE and a tandem silencer element.

The results also show that the cyclin D1 promoter sequences of promotion‐sensitive and resistant JB6 mouse epidermal cells are essentially identical to the sequence determined for a mouse genomic clone.

Inr or Inr‐like sequences

The core promoter region of typical eukaryotic messenger RNA coding genes contains the minimal promoter elements required for initiation by RNA polymerase II. It consists predominantly of two elements: the TATA box and/or the Inr (initiator) element, which can be present either alternately (TATA⁺Inr or TATA⁻Inr⁺) or in limited cases simultaneously (TATA⁺Inr⁺) ((1993b), (1993a)). The cyclin D1 promoters of mouse, rat and human are TATA⁻Inr⁺. They contain Inr or Inr‐like sequences, but, like those of many of the genes encoding the so‐called ‘housekeeping’ enzymes, oncogenes, growth factors and their receptors, and transcription factors (Azizkhan et al. 1993), they do not contain a canonical TATA box.

The Inr of human cyclin D1 promoter has the sequence 5′‐GCTCC C⁺¹ATTC TCTGC‐3′ (nucleotide positions from − 5 to + 10). By primer extension analysis and alignment of Inr sequences from several other genes, Philipp et al. (1994) placed the nucleotide position + 1 at C⁺¹. The consensus, however, has been to place the nucleotide position + 1 at A⁺². For example, the mutant and computer analyses have defined preferred Inr sequence as PyPyA⁺¹N(T/A)PyPy (where Py is T or C) with an A at + 1, a T or A at + 3 and a pyrimidine (Py) at − 1 as most critical for determining the strength of the Inr (Javahery et al. 1994; Kaufman & Smale 1994). Many Inr‐containing genes appear to conform to this consensus motif with the nucleotide A assigned as the nucleotide position + 1, but there are exceptions. For example, in mouse fibulin‐2 gene, the nucleotide position + 1 was assigned to two bases upstream of the nucleotide A (Grassel et al. 1999).

The Inr elements of mouse and rat promoters have the sequences 5′‐CCCCC T⁺¹TTTC TCTGC‐3′ and 5′‐CTCCT T⁺¹TTTC TCTGC‐3′, respectively. These sequences are almost identical to that of human Inr, but they contain T⁺¹ rather than C⁺¹ and they lack the critical nucleotide A⁺². It should be noted, however, that the Inr of human immunodeficiency virus‐1 (Roy et al. 1991; Weis & Reinberg 1992; Du, Roy & Roeder, 1993), mouse (but not human) KDR/flk‐1 (Wu & Patterson 1999) and mouse c‐mos (Pal et al. 1991) also do not contain the critical nucleotide A.

An alternative or possibly an additional Inr site was identified at nucleotide position + 90 in human cyclin D1 promoter by Herber et al. (1994) using RNase protection assay. This Inr element has the sequence 5′‐AGC⁺⁹⁰CTCCAGAGGG‐3′ (nucleotide positions from + 88 to + 100). The four nucleotides immediately 3′ to the nucleotide position + 90 (5′‐CTCC‐3′) conform to the consensus Inr sequence 5′‐N(T/A)PyPy‐3′. The mouse and rat promoters also have sequences identical to that of the human promoter. The nucleotide position + 90 is very close to the nucleotide position + 96, which was assigned as the 5′ end of the 5′ untranslated region (5′‐UTR) of mouse cyclin D1 exon 1 (Smith et al. 1995).

TRE, CRE, NF‐κB response element: implication in phorbol ester‐induced tumour promotion

The mouse, rat and human cyclin D1 promoters contain three elements (TRE, CRE and NF‐κB response element) that are known to respond positively to the tumour promoter TPA. TPA may use all three elements to stimulate the expression of cyclin D1. To inhibit the expression of cyclin D completely, all three elements may need to be blocked because, otherwise, some elements may still be available for stimulation. One example bears noting in this respect. The promotion‐resistant JB6 mouse epidermal cells express a low level of mitogen‐activated protein kinase (MAPK) (extracellular signal‐regulated protein kinase, ERK) which contributes to the lack of AP‐1 activation at TRE and cell transformation responses to TPA in these cells (Huang et al. 1998). The results of our previous study (Eto 1998) indicated that, in these cells maintained under anchorage‐independent culture conditions, TPA‐induced expression of cyclin A was almost completely inhibited and the TPA‐induced synthesis of DNA and cell proliferation was completely blocked. However, TPA‐induced expression of cyclin D1 was not completely inhibited. TPA was still able to induce significant — albeit attenuated — expression of cyclin D1 in these cells.

Incorporating not only TRE but also NF‐κB response element and CRE, we tentatively propose a possible sequence of events that leads to TPA‐induced, anchorage‐independent synthesis of cyclins D1 and A in the promotion‐sensitive JB6 mouse epidermal cells (fig. 8).

NF‐κB response element is the latest element to be identified in the cyclin D1 promoter (Guttridge et al. 1999; Hinz et al. 1999; Joyce et al. 1999). NF‐κB has been implicated in the regulation of cell proliferation, transformation, and tumour development. TPA has long been known to stimulate NF‐κB transactivation in a variety of cells including JB6 cells (Li et al. 1997). In JB6 cells, NF‐κB inhibitor efficiently inhibits not only TPA‐induced NF‐κB transactivation but also cell transformation. TPA appears to stimulate NF‐κB transactivation in two ways: (a) TPA participates in the upstream NF‐κB pathway and/or (b) TPA produces AP‐1 proteins which, in turn, interact with NF‐κB factors and bind to the NF‐κB response element (Ahmad, Theofanidis & Medford, 1998). Recent discovery of the importance of NF‐κB response element in the cyclin D1 promoter activity may provide evidence for a direct link between TPA‐induced NF‐κB transactivation, G1‐to‐S‐phase transition and anchorage‐independent growth of JB6 cells.

CRE appears to upregulate the expression of cyclin D1 when it interacts with AP‐1 proteins. For example, Herber et al. (1994) reported that the cyclin D1 promoter is inducible by c‐Jun and that this induction is mediated predominantly through the cyclin D1 CRE. Also, Beier et al. (1999) demonstrated that ATF‐2, as a complex with a CRE‐binding protein (CREB)/CRE modulator protein, binds to the cyclin D1 CRE and induces cyclin D1 expression. and Sabbah et al. (1999) provided evidence that ATF‐2/c‐Jun heterodimers bind to cyclin D1 CRE and mediate the activation of cyclin D1 promoter by the liganded oestrogen receptor.

Cyclins D1 and A: reversibility and threshold effects of TPA

The two most distinguishing characteristics of the promotion stage of multistage carcinogenesis are its reversibility and the threshold effects of tumour promoters. If we assume that these two characteristics are attributable to the TPA‐induced, anchorage‐independent expression of cyclins D1 and A, this appears to provide an explanation for some of the observations reported previously (Eto 1998). During the TPA‐induced proliferation of promotion‐sensitive JB6 cells maintained under anchorage‐independent culture conditions, expression of cyclin D1 begins within 12 h after the addition of TPA and the level of cyclin D1 protein increases continuously afterward until it reaches maximum at about 48 h. In contrast, expression of cyclin A is not yet detectable even at 24 h after the addition of TPA. It begins sometime between 24 and 48 h and the level of cyclin A protein reaches maximum at about 48 h. This sequence of events suggests that, under anchorage‐independent culture conditions, cyclin A protein cannot be synthesized unless the amount of cyclin D1 protein either reaches or exceeds a certain threshold level.

A corollary of this hypothesis is that, as in the case of the promotion‐resistant JB6 cells, even a partial reduction in the expression of cyclin D1 such as inhibiting only one of the TRE, CRE and NF‐κB response element may produce a nearly total attenuation of the anchorage‐independent expression of cyclin A.