(1999) 18, 2299 ± 2309 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc growth-regulated expression of mammalian MCM5 and MCM6 mediated by the factor E2F

Kiyoshi Ohtani1, Ritsuko Iwanaga1, Masataka Nakamura*,1, Masa-aki Ikeda2, Norikazu Yabuta3, Hiromichi Tsuruga3 and Hiroshi Nojima3

1Human Sciences Center, Tokyo Medical and Dental University, Tokyo 113-8510, Japan 2Department of Developmental , Graduate School of Dentistry, Tokyo Medical and Dental University, Tokyo 113-8549, Japan; 3Department of Molecular , Research Institute for Microbial Diseases, Osaka University, Suita 565-0871, Japan

Initiation of DNA replication requires the function of family (MCM2-7) that have been identi®ed in , MCM gene products, which participate in ensuring that Xenopus, and human. Mcm seem to regulate DNA replication occurs only once in the . the initiation at the replication origin where the loading Expression of all mammalian genes of the MCM family of the proteins onto the origin recognition complex is induced by growth stimulation, unlike yeast, and the (ORC) is regulated by and cyclin-dependent mRNA levels peak at G1/S boundary. In this study, we (Donovan et al., 1997; Tanaka et al., 1997). examined the transcriptional activities of isolated human However, the mechanism(s) by which Mcm proteins MCM gene promoters. Human MCM5 and MCM6 control the initiation of DNA replication remains promoters with in the E2F sites failed in unclear. regulation following serum stimulation and Xenopus Mcm proteins seem to be able to access exogenous E2F expression. In addition, we identi®ed a replicated DNA only after the breakdown of nuclear novel E2F-like sequence in human MCM6 promoter envelope in M phase. The Mcm proteins in S. which cooperates with the authentic E2F sites in E2F- cerevisiae are in the nucleus during G1 phase but dependent regulation. Forced expression of E2F1 could disappear from the nucleus upon the initiation of S induce expression of all members of the endogenous phase, whereas Mcm proteins in ®ssion yeast and MCM genes in rat embryonal ®broblast REF52 cells. mammalian cells are in the nucleus throughout the Our results demonstrated that the growth-regulated mitotic cell cycle. Unlike yeast, in which expression of expression of mammalian MCM5 and MCM6 genes, the Mcm proteins and messages is stable during the cell and presumably other MCM members, is primarily cycle, mammalian cells show a dramatic induction of regulated by E2F through binding to multiple E2F sites MCM mRNAs at G1/S boundary coupled with a in the promoters. signi®cant rise in levels upon growth stimula- tion such as stimulation of resting ®broblasts with Keywords: mammalian MCM; E2F; promoter; cell serum (Tsuruga et al., 1997a). Thus, in mammalian cycle; DNA replication cells, the transcriptional mechanism of expression of the MCM genes is, at least in part, central to the regulation of Mcm activities. To address this issue, we previously isolated the 5' ¯anking regions of human Introduction MCM5 (HsMCM5) and MCM6 (HsMCM6) genes (Tsuruga et al., 1997a,b). Both sequences were found In the cell cycle of eukaryotic cells, replication of DNA to have putative E2F binding consensus sites, occurs once per cell cycle to ensure the integrity of the suggesting E2F-dependent regulation of the . The mechanism that inhibits re-replication of HsMCM5 and HsMCM6 genes. the DNA that had already replicated in the current S Overexpression of E2F has been shown to induce phase seems to be conserved from yeast to human. A cellular DNA synthesis in serum starved ®broblasts model in Xenopus hypothesized a trans-acting factor (DeGregori et al., 1997; Johnson et al., 1993; Qin et al., (), which distinguishes replicated DNA 1994; Shan and Lee, 1994), suggesting that genes from unreplicated one (Blow and Laskey, 1988). On critically involved in regulation of DNA replication are the other hand, the MCM family genes have been targets of E2F. Indeed, E2F is known to regulate a identi®ed in as essential genes class of genes involved in DNA replication (La involved in the initiation of DNA replication. MCM Thangue, 1994; Nevins, 1992). These include genes gene products are reminiscent of a putative licensing for enzymatic machinery for DNA synthesis, such as factor. A series of experiments with Xenopus egg dihydrofolate reductase (DHFR) and DNA extract identi®ed Mcm molecules as components of a a, and regulatory molecules for the initiation of DNA putative licensing factor (Chong et al., 1995; Kubota et replication, such as HsOrc1 and HsCdc6 (Ohtani et al., al., 1995, 1997; Madine et al., 1995; Thommes et al., 1996, 1998; Yan et al., 1998). HsOrc1 and HsCdc6 may 1997). At present, there are six genes of the MCM be necessary for the initiation of DNA replication. However, overexpression of HsOrc1 or HsCdc6 is not sucient for the induction of DNA synthesis as shown in rat embryonal ®broblast REF52 cells (Ohtani et al., *Correspondence: M Nakamura Received 6 August 1998; revised 28 October 1998; accepted 29 1998), implying other E2F targets than Orc1 and Cdc6 October 1998 in inducing DNA synthesis in REF52 cells. E2F-mediated expression of mammalian MCM genes KOhtaniet al 2300 In the present study, we show that the cell growth- gene clustered about 15 bp upstream of the 5' end of regulated promoter activity of HsMCM5 and HsMCM5 cDNA (Tsuruga et al., 1997a). There are HsMCM6 is primarily regulated by E2F through two sets of putative E2F binding elements arranged in binding to the multiple E2F recognition sites of an overlapping fashion in the 5' ¯anking sequence of HsMCM5 and HsMCM6 promoters. Furthermore, HsMCM5 cDNA. One set (E2F sites III+IV) is expression of E2F1 induces all genes of the endogen- located about 190 bp upstream of the 5' end of cDNA ous MCM family in REF52 cells. Our results indicate while the other (E2F sites I+II) 2 bp downstream of that the MCM family genes are critical E2F target the 5' end of cDNA (Figure 2B); i.e., each one set lies genes in the regulation of DNA replication. downstream and upstream of the transcriptional initiation sites. Similarly, RNase protection assay of the HsMCM6 5' ¯anking sequence indicated multiple transcriptional start sites in 20 to 1 bp upstream of the Results 5' end of cDNA (Tsuruga et al., 1997b). Our independent search found, in addition to two sets of -regulated expression of MCM family overlapping E2F sites (E2F sites I ± IV) recognized genes in REF52 cells previously (Tsuruga et al., 1997b), another putative We have recently reported that serum stimulation E2F site (E2F like site V), which overlapped with the induced expression of all genes of the MCM family clustered transcriptional start sites. The 5' ¯anking in human lip ®broblast KD cells and normal rat kidney sequences of the HsMCM5 and HsMCM6 genes ®broblast NRK cells (Tsuruga et al., 1997a,b). Serum- contained multiple Sp1 sites but not typical TATA dependent expression of the MCM family genes was box. These characteristics are similar to those of further examined in rat embryonal ®broblast REF52 previously reported genes, such as B-myb, E2F1 and cells, which were used for a series of transfection HsOrc1, which are primarily regulated by E2F experiments. REF52 cells were made quiescent by (Johnson et al., 1994; Lam and Watson, 1993; Ohtani culture in 0.1% fetal bovine serum (FBS) for 48 h then et al., 1996). stimulated to re-enter the cell cycle by the addition of serum. The level of mRNA expression of each MCM gene post-serum stimulation was monitored by North- ern blotting with poly(A)+ RNA. The latter was puri®ed from cells harvested at quiescence and at various time intervals after the addition of serum. Serum-starved quiescent REF52 cells produced lower levels of mRNAs for all MCM genes compared with asynchronously growing cells (Figure 1). Serum stimulation induced a transient marked inhibition of expression of MCM mRNAs at 4 h, which was followed with a profound increase for up to 12 h. The reduction at 4 h after stimulation may be due to the paucity of general transcription factors which are required for many genes activated at this stage. The levels of MCM mRNAs remained persistently high at least until 28 h post-stimulation. At 16 h post-serum stimulation REF52 cells commenced DNA synthesis which was determined by [3H]thymidine uptake (data not shown). These results indicate that MCM mRNA expression parallels proceeding to G1/S boundary of the cell cycle progression upon the addition of serum, implying the involvement of cell growth regulation in the expression of mammalian MCM genes.

HsMCM5 and HsMCM6 promoters Our previous results showed multiple putative E2F sites in 5' ¯anking sequences of the HsMCM5 and HsMCM6 genes (Tsuruga et al., 1997a,b). We wished to determine the functional signi®cance of these sequences in the regulation of HsMCM5 and HsMCM6 genes. Transcriptional start sites were mapped by RNase protection assays. By using RNA Figure 1 Growth-regulated expression of MCM family genes in probes spanning sequences upstream of HsMCM5 and REF52 cells. REF52 cells were serum-starved and re-stimulated HsMCM6 cDNAs, fragments with multiple sizes were with 20% FBS. Cells were harvested at indicated time intervals protected with mRNAs from human T cell lines MT-2 after the addition of serum. Poly(A)+ RNA prepared from and Kit 225 (Figure 2A), indicating multiple transcrip- 660 mg of total RNA was subjected to electrophoresis on a 1% agarose gel and transferred to a nylon membrane. The blot was tional start sites around which no typical splicing sequentially probed with HsMCM2 through HsMCM7 and acceptor consensus sequences were found. These GAPDH cDNAs. Lane A indicates mRNA expression in possible transcriptional start sites of the HsMCM5 asynchronously cultured REF52 cells E2F-mediated expression of mammalian MCM genes KOhtaniet al 2301 To determine whether the 5' ¯anking regions of the and HsMCM6 promoter-driven luciferase reporter HsMCM5 and HsMCM6 genes were indeed respon- plasmids were transfected into REF52 cells together sible for the growth-dependent regulation of expression with a plasmid expressing E2F1, E2F2 or E2F3. of HsMCM5 and HsMCM6 genes, we fused these Following transfection, the cells were grown in a sequences to a luciferase reporter gene and assayed the medium containing 0.1% FBS for 36 h. Cell extracts promoter activities in serum-starved and serum- were prepared and assayed for luciferase activity. The stimulated REF52 cells. REF52 cells were transfected activity of HsMCM5 promoter increased seven- and with HsMCM5- and HsMCM6-luciferase (Luc) repor- sixfolds by exogenous expression of E2F1 and E2F2, ter plasmids and a plasmid containing the b- respectively, and to a lesser extent (threefolds) by E2F3 galactosidase gene as an internal control, and then (Figure 4A). Introduction of mutation in two sets of the made quiescent by serum starvation. Serum was added overlapping E2F sites abolished E2F1-dependent to stimulate transfected REF52 cells to re-enter into activation (Figure 4C), indicating a direct involvement the cell cycle. Cell extracts were prepared at indicated of E2F in the activity of HsMCM5 promoter. The time intervals and assayed for luciferase activity, which mutant exhibited high levels of promoter activity, was corrected by b-galactosidase activity. As shown in compared with the wild type with E2F1, even without Figure 2C the promoter activities of both genes were E2F1 and little, if any, activation by E2F1. These results low in quiescent and early G1 cells but markedly show that the HsMCM5 promoter is activated by E2F increased as the cell cycle progressed through G1 to S through multiple E2F sites. phase in response to serum stimulation. The increase in Similarly, exogenous expression of E2F1, E2F2 and promoter activities paralleled accumulation of MCM5 E2F3 in quiescent REF52 cells induced eight- to and MCM6 mRNAs in Northern blotting (Figures 1 threefold activation of the HsMCM6 promoter and 2C). We therefore conclude that the 5' ¯anking (Figure 4B). Mutation was introduced into two sets regions of the HsMCM5 and HsMCM6 genes contain of the putative overlapping E2F sites (Mut I ± IV). The promoter sequences responsible for the growth- basal activity of the mutant promoter without E2F regulated expression of those genes. increased threefolds relative to that of the wild type, and the activity with E2F1 was similar to that of the wild type (Figure 4D), resulting in a decrease in fold Activation of HsMCM5 and HsMCM6 promoters by activation from 8 to 2. These results clearly indicate E2F that the putative overlapping E2F sites are, at least in Given the presence of multiple putative E2F recogni- part, responsible for activation by E2F, and suggest the tion sequences within the HsMCM5 and HsMCM6 presence of another E2F responsive element(s) in the promoters, we then examined whether the putative E2F HsMCM6 promoter because the mutant still retains recognition sequences are functional elements in the responsiveness to E2F1. The newly identi®ed E2F like regulation of promoter activities. sequence (E2F like site V: 714 to 721), which is The ability of these putative E2F sites to bind E2F diverged from typical E2F site in terms of nucleotide was examined by the gel mobility shift assay. Typical sequence and non-preferential binding of E2F, was a E2F sites from the DHFR gene showed a speci®c likely candidate. To examine this notion, we mutated binding with total cell extract from F9 cells which are this site and examined the response to E2F1. shown to contain a large amount of free E2F molecules Introduction of mutation in E2F like site V in (La Thangue et al., 1990). This complex formation was addition to the two overlapping typical E2F sites completely inhibited by any wild type oligonucleotide (Mut I ± V) completely abolished the ability of the of the putative E2F sites in HsMCM5 and HsMCM6 promoter to respond to E2F1, implying involvement of promoters, with the exception of the HsMCM6 E2F E2F like site V in E2F-dependent response (Figure like site V which was less e€ective in inhibiting the 4D). E2F like site V single mutant (Mut V), however, complex formation (Figure 3A). Their mutant forms showed reduced promoter activity either with or with two-base substitution did not show appreciable without E2F1, resulting in activation by E2F1 which e€ects on complex formation. Furthermore, all of the was essentially similar to the wild type. putative E2F recognition sequences of HsMCM5 and HsMCM6, with the exception of HsMCM6 E2F like Critical involvement of E2F in cell cycle-dependent site V, were able to form complexes with the F9 cell regulation of HsMCM5 and HsMCM6 promoter extract. The complex formation was competed away activities with either a typical E2F site or own oligonucleotides, whereas mutant forms of the typical E2F site and In view of the growth-dependent control of the MCM gene E2F sites failed to inhibit complex HsMCM5 and HsMCM6 (Figure 1), formation (Figure 3B). The combination of we further investigated the role of E2F sites in growth- HsMCM6 E2F like site V and the F9 cell extract dependent regulation of HsMCM5 and HsMCM6 generated a complex whose formation was not promoters. For this purpose, we compared the kinetics interfered by typical E2F site oligonucleotides. These of the activity of E2F mutant promoter with that of results indicate that the putative E2F recognition the wild type in REF52 after inducing cell cycle sequences (sites I ± IV) in the HsMCM5 and progression by stimulation with serum (20% FBS) of HsMCM6 promoters are authentic E2F binding sites. quiescent cells cultured in 0.1% FBS for 48 h. On the other hand, E2F like site V of HsMCM6 The mutant of E2F sites (Mut I ± IV) in the preferentially binds to a factor(s) di€erent from E2F. HsMCM5 promoter failed to control expression in These ®ndings prompted us to examine whether response to stimulation of cell growth. The activity of HsMCM5 and HsMCM6 promoters could respond to the mutant promoter in quiescent cells was profoundly E2F in the transient transfection assay. The HsMCM5 higher than that of the fully stimulated wild type E2F-mediated expression of mammalian MCM genes KOhtaniet al 2302 promoter and remained persistently high after growth activity in serum-starved conditions is not known at stimulation. This result indicates that growth-regulated present. promoter activity of HsMCM5 is absolutely dependent Introduction of mutation in two sets of the on E2F. The reason of the extremely high mutant overlapping E2F sites (Mut I ± IV) in HsMCM6 E2F-mediated expression of mammalian MCM genes KOhtaniet al 2303 promoter increased the basal activity in serum-starved responsiveness to serum stimulation with the promoter conditions. Addition of serum stimulated the mutant activity as low as that of the wild type under serum promoter in a manner similar to the wild type starvation conditions, in contrast to the elevated basal promoter, although less activation was associated with activity of the E2F site I ± IV mutant. The E2F like site the mutant. However, the addition of mutation in E2F V mutation (Mut V) reduced the activity of the like site V to the I ± IV mutant completely abolished promoter in serum-starved and serum-stimulated

Figure 2 HsMCM5 and HsMCM6 promoter regions. (A) RNase protection analysis of HsMCM5 and HsMCM6 transcripts. HsMCM5 (positions +43 to 7481) and HsMCM6 (positions +35 to 7587) probes were hybridized with 1 mg of yeast tRNA (lanes 2 and 6), MT-2 mRNA (lanes 3 and 7) and Kit 225 mRNA (lanes 4 and 8), and digested with RNase. Protected fragments were electrophoresed in a 7% polyacrylamide sequencing gel next to products of sequencing reactions of the HsMCM5 and HsMCM6 genomic DNA as size markers. Lanes 1 and 5 contain only original probes for HsMCM5 and HsMCM6, respectively. (B) Nucleotide sequences of 5'-¯anking regions. Functional E2F sites determined by the gel shift assay and reporter assays are indicated by horizontal arrows. Transcriptional start sites determined by the RNase protection assay are indicated by vertical arrows. Sp1 sites which are often observed with E2F regulated promoters are boxed. The +1 positions are chosen arbitrarily according to the 5' ends of the HsMCM5 and HsMCM6 cDNAs (Tsuruga et al., 1997a,b). (C) Cell growth control of HsMCM5 and HsMCM6 promoter activity. REF52 cells were transfected with 5 mg of pHsMCM5-Luc(71384) or pHsMCM6-Luc(7754) together with 2 mg of pCMV-b-gal as an internal control. Transfected cells were brought to quiescence and then stimulated with serum for the indicated time intervals. At each time point, cell extracts were prepared, and luciferase and b-galactosidase activities were measured. Luciferase activity was normalized to that of b-galactosidase E2F-mediated expression of mammalian MCM genes KOhtaniet al 2304 conditions with the same fold activation as the wild Activation of expression of all endogenous MCM genes type. It is possible that the low activity of this mutant by E2F1 was due to the destruction by mutation of E2F like site V of the native transcriptional start sites while sparing In order to examine the role of E2F in the regulation cryptic transcriptional start sites, thus resulting in a of expression of MCM family genes in a more loss of regulation by E2F. In any event, E2F seems to physiological state, we examined the expression of be the main regulator of growth-dependent activation endogenous MCM genes in response to E2F1. E2F1 of the HsMCM6 promoter. recombinant adenovirus (Ad-E2F1) was used to

Figure 3 E2F binding to HsMCM5 and HsMCM6 promoters. (A) Competitive activity of HsMCM5 and HsMCM6 E2F sites in E2F binding. Gel mobility shift assays were performed with F9 cell extract and the end labeled E2F fragment from the DHFR promoter containing two overlapping E2F recognition sites as a probe. Competition experiments were performed by adding 100-fold molar excess of unlabeled oligonucleotides to the gel shift reaction mixtures. Competitor oligonucleotides used were the E2F recognition sites from the adenovirus E2 promoter (lane 3), a mutant form of this sequence (lane 4), the putative E2F recognition sites I+II from the HsMCM5 promoter (lane 5), a mutant form of this sequence (lane 6), the putative E2F recognition sites III+IV from the HsMCM5 promoter (lane 7), a mutant form of this sequence (lane 8), the putative E2F recognition sites I+II from the HsMCM6 promoter (lane 9), a mutant form of this sequence (lane 10), the putative E2F recognition sites III+IV from the HsMCM6 promoter (lane 11), a mutant form of this sequence (lane 12), E2F like site V from the HsMCM6 promoter (lane 13), and a mutant form of this sequence (lane 14). Lane 1 has the probe only, and lane 2 has no competitor. (B) Speci®city in E2F binding to the HsMCM5 and HsMCM6 promoters. Gel mobility shift assays were performed as described for (A). End-labeled double stranded oligonucleotides containing putative E2F recognition sites I+II or III+IV from the HsMCM5 promoter or putative E2F recognition sites I+II, III+IV or E2F like site V from the HsMCM6 promoter were used as probes. Competition assays were performed by adding 100-molar excess of the same unlabeled oligonucleotides to the gel shift reaction mixtures E2F-mediated expression of mammalian MCM genes KOhtaniet al 2305 express E2F1 in serum-starved quiescent REF52 cells. Moreover, the fact that the 5' ¯anking sequences of Cells were harvested for preparation of RNA, which murine MCM3 (Kimura et al., 1994) and human was subjected to Northern blot analysis to monitor MCM7 genes (Kiyono et al., 1996) contain putative levels of MCM5 and MCM6 mRNAs. The levels of E2F sites suggests that expression of MCM3 and mRNAs of both genes were markedly increased upon MCM7 is also a result of direct e€ect of E2F1. infection with Ad-E2F1 virus compared to those with control virus infection (Figure 6). In addition, exogenous expression of E2F1 signi®cantly increased Discussion the levels of mRNAs for MCM2, MCM3, MCM4 and MCM7. Although no information on rat MCM5 and The present study provides evidence that the growth- MCM6 promoters is available, considering that the regulated expression of mammalian MCM5 and MCM6 HsMCM5 and HsMCM6 promoters have E2F sites genes is mediated by E2F. Accumulating evidence that primarily regulate promoter activities in response suggests that the expression of mammalian genes to growth stimulation, it is highly likely that regulatory to DNA replication is growth-dependent, endogenous MCM5 and MCM6 genes were induced mainly through the action of the through a direct action of E2F1 in REF52 cells. E2F (Ohtani et al., 1996, 1998; Yan et al., 1998).

Figure 4 Activation of HsMCM5 and HsMCM6 promoters by E2F. REF52 cells were transfected with 5 mg of pHsMCM5- Luc(71384) (A) or pHsMCM6-Luc(7754) (B) and 0.2 mg of CMV promoter-driven expression vectors for either E2F1, E2F2, E2F3, or a control CMV vector, pcDNA3 along with 2 mg of pCMV-b-gal as an internal control. Transfected cells were serum- starved for 36 h and harvested for preparation of cell extracts. Luciferase and b-galactosidase activities in cell extracts were measured. Luciferase activity was normalized to that of b-galactosidase, and was plotted for each cotransfected CMV expression vector. REF52 cells were cotransfected with 5 mg of either wild type or E2F site mutant forms of HsMCM5-Luc(71384) (C)or pHsMCM6-Luc(7754) (D) with the E2F1 vector or the control CMV vector, together with 2 mg of pCMV-b-gal as the internal control. Luciferase activity was determined E2F-mediated expression of mammalian MCM genes KOhtaniet al 2306 The 1384-bp and 754-bp fragments upstream of HsMCM5 and HsMCM6 genes, respectively, mimic the kinetics of promoter activities of the rat endogenous MCM5 and MCM6 genes in response to serum stimulation (Figure 2c vs Figure 1). These results indicate not only that the 1348-bp and 745-bp sequences encompass the transcriptional regulatory region of the genes, but also that humans and rats have a similar mechanism for the regulation of expression of MCM5 and MCM6 genes. As is the case with genes whose expression is primarily regulated by E2F (Johnson et al., 1994; Lam and Watson, 1993; Ohtani et al., 1996), the E2F sites of the HsMCM5 and HsMCM6 genes are located near the RNA start sites in a multiple copy form in transcriptional regulatory regions. We previously found two sets of overlapping E2F consensus sequences in the 5' ¯anking regions of the HsMCM5 and HsMCM6 genes (Tsuruga et al., 1997a,b). One of the two sets of HsMCM5 lies

Figure 5 E2F-dependent control of HsMCM5 and HsMCM6 promoters in response to cell growth. REF52 cells were Figure 6 Induction of all members of endogenous MCM family transfected with 5 mg of either the wild type or mutant forms of genes by E2F1. REF52 cells were brought into quiescence by pHsMCM5-Luc(71384) (A) or pHsMCM6-Luc(7754) (B) along serum starvation for 48 h and infected with either Ad-E2F1 or with 2 mg of pCMV-b-gal. Cells were brought to quiescence and Ad-Con. Infected cells were harvested at 21 h post-infection, and then stimulated with serum for the indicated time intervals. poly(A)+ RNA was prepared and then separated by agarose gel Luciferase and b-galactosidase activities of cell extracts were electrophoresis. The were transferred to a nylon membrane measured at each time point. Luciferase activity was normalized and probed sequentially with either HsMCM2 through HsMCM7 to that of b-galactosidase or GAPDH cDNA E2F-mediated expression of mammalian MCM genes KOhtaniet al 2307 upstream of the clustered transcriptional start sites mutant did not show higher activity than that of the determined by RNA protection assay and the other wild type is not known at present. One possibility is downstream of those. The latter is thus in the 5' that the mutation of E2F like site V destroyed some of untranslated region of the HsMCM5 mRNA. Two sets the transcriptional start sites. We assume, however, of MCM6 are upstream of the clustered transcriptional that this is less likely because the E2F like site V single start sites. We demonstrated in this study that these mutant showed essentially the similar response to E2F sites can potentially bind to E2F and to regulate stimulations as the wild type promoter. Thus, the expression of genes. We also identi®ed an although both E2F sites I ± IV and E2F like site V additional E2F-like site in the HsMCM6 gene, which are responsible for regulation of HsMCM6 expression, overlaps the transcriptional start sites. which is activated by serum stimulation or over- Growth stimulation-dependent regulation of expres- expression of E2F1, they may mediate such regulation sion of HsMCM5 was abolished by disruption of both independently. Collectively, expression of HsMCM6 is sets of the overlapping E2F sites in the promoter, assured by combinatorial action of multiple E2F sites. clearly demonstrating the critical function of E2F in In any event, the E2F sites in HsMCM5 and HsMCM6 expression of the HsMCM5 gene. The E2F site mutant promoters are major elements regulatory to growth showed higher activity than the wild type in serum- stimulation. This notion is supported by the induction starved conditions and the activity remained un- of endogenous MCM5 and MCM6 genes by over- changed after serum stimulation. Many of E2F expression of E2F1 in REF52 cells. responsive genes such as B-myb, E2F1 and HsOrc1, Overexpression of E2F1 induced all endogenous are actively suppressed in serum-starved conditions. genes of the MCM family in REF52 cells. Considering Mutation of the E2F sites in these genes relieves the the direct action of E2F on HsMCM5 and HsMCM6 suppression, resulting in a higher basal activity in promoters and putative E2F binding consensus sites in serum-starved conditions (Johnson et al., 1994; Lam murine MCM3 and human MCM7 5' ¯anking and Watson, 1993; Ohtani et al., 1996). The function of sequences (Kimura et al., 1994; Kiyono et al., 1996), E2F sites in the HsMCM5 promoter seems similar to it is possible that the induction of those genes in that of the E2F responsive genes. E2F complex with a REF52 cells is directly mediated by E2F. Although we member of the Rb family on the E2F sites is thought do not have any information on 5' ¯anking sequences to be suppressive in transcription from promoters in of MCM2 and MCM4, the induction of both genes serum-starved conditions (Smith et al., 1996). Our might be a direct e€ect of E2F. knowledge at present is inadequate to account for the The Mcm proteins are rather abundant proteins even extremely high activity of the HsMCM5 promoter in in serum-starved resting ®broblasts (Tsuruga et al., serum-starved conditions relative to that of the wild 1997a). Thus, expression of the MCM genes may not be type promoter fully stimulated with serum. One rate limiting for serum-starved ®broblasts to initiate possibility is that mutation of the 3' set of the DNA replication after stimulation with serum. How- overlapping E2F sites, which are present in the 5' ever, in certain cellular circumstances such as prolonged untranslated region of HsMCM5, increases the stability serum starvation and treatment with anti-sense of the mRNA. Alternatively, the mutation in the 5' oligonucleotides speci®c to HsMCM7, the Mcm family untranslated region might facilitate the eciency of proteins seem to decrease their levels sucient enough from mRNA molecules. for rate limiting (Fujita et al., 1996). This may be similar The mutant of two sets of the overlapping E2F sites to the in vivo situation in which most cells in vivo are in a (Mut I-IV) of HsMCM6 is di€erent from that of resting state while only a very small proportion of the HsMCM5; the basal activity in serum-starved condi- cells, e.g., in cases of immune responses and injury, re- tions is higher than that of the wild type but enters the cell cycle after a prolonged period of resting signi®cantly lower than that of the wild type state. Under such conditions, transcriptional activation stimulated by growth signals. Moreover, the MCM6 of the MCM family genes is important in the regulation mutant could still be activated by E2F1 and serum of DNA replication. similar to the wild type, indicating that E2F sites I ± IV Overexpression of E2F has been shown to induce are partly responsible for the cell cycle-regulated cellular DNA synthesis in serum-starved ®broblasts expression and are presumably suppressed by binding (DeGregori et al., 1997; Johnson et al., 1993; Qin et al., of E2F complex with Rb family proteins in serum- 1994; Shan and Lee, 1994), suggesting that genes starved conditions. This conclusion implies another critically involved in the regulation of DNA replication E2F responsive element(s). In this context, it is are targets of E2F. It is possible that other E2F target intriguing to imagine that newly identi®ed E2F like genes may exist and mammalian homologs of CDC7, site V plays a pivotal role in the regulation of DBF4 and CDC45 are potential candidates (Sato et al., expression in response to E2F. Indeed, mutation of 1997). Identi®cation of other E2F target genes should E2F like site V with mutation of E2F site I ± IV provide insight into the mechanisms that regulate the completely abolished E2F-dependent regulation. Ob- initiation of DNA replication in mammalian cells. viously E2F like site V is involved in E2F-dependent regulation of HsMCM6 expression. Nevertheless this E2F like site V element seems to be di€erent from the typical E2F sites identi®ed so far in various genes in Materials and methods that the E2F like site V mutant did not show elevation of the basal promoter activity in serum-starved Cell culture conditions and the E2F site oligonucleotide was an Human T-cell lines MT-2 and Kit 225 were maintained in ine€ective competitor for the complex formation of a RPMI 1640 containing 10% FBS supplemented with typical E2F site. The reason that the E2F site I ± V 0.5 nM IL-2 (kindly provided by Ajinomoto, Yokohama, E2F-mediated expression of mammalian MCM genes KOhtaniet al 2308 Japan) for Kit 225. A rat embryonic ®broblast cell line REF52 cells were transfected with expression and reporter REF52 and murine teratocarcinoma cell line F9 were plasmids by the calcium-phosphate method together with maintained in Dulbecco modi®ed Eagle medium (DMEM) b-galactosidase expression vector, pCMV-b-gal, which was containing 10% FBS. For experiments of promoter activity used as an internal control to monitor the transfection assay and expression of endogenous MCM mRNAs, eciency. Luciferase activities were assayed using Lucifer- REF52 cells were serum-starved in DMEM containing ase Assay System (Promega) and normalized to b- 0.1% FBS for 48 h and stimulated by addition of 20% FBS galactosidase activities, which were determined by the for indicated time intervals. method of Rose and Botstein (1983). All assays were performed at least three times in duplicates and represen- tative data are presented. Northern (RNA) blot assay Total RNA extraction was carried out using Isogen Gel mobility shift assay (Nippon Gene) and poly(A)+ RNA was puri®ed from total RNA by PolyA Tract (Promega) according to the protocol Double-stranded oligonucleotides containing each of the recommended by the manufacturer. Gel electrophoresis, putative E2F recognition sites in the 5' ¯anking regions of transfer to nylon membrane, and hybridization were HsMCM5 (nt 7194 to 7183 and +2 to +13) and performed as described previously (Johnson et al., 1994). HsMCM6 (7169 to 7158, 7107 to 796 and 721 to The probes for MCM2 through MCM7 were the AscI/NotI 714) and mutants of these sequences were used for the fragments cut out from pGEX8T-HsMCM2, 3, 5 and 7, and assay. Sequences were: HsMCM5 wild type-I+II: 5'- pBAC3-HsMCM4 and 6. The GAPDH cDNA was used as a TTGTTTTTCCCGCGAAACTCGG-3';HsMCM5mu- control probe. Radioactivity of the signals was measured tant-I+II: 5'-TTGTTTTTCCATCGAAACTCGG-3';Hs- with an image analyzer BAS 1500 (Fuji Film). MCM5 wild type-III+IV: 5'-AGAAGTTTCGCGC- CAAATTGTT-3'; HsMCM5 mutant-III+IV: 5'-AGAAG- TTTCGATCCAAATTGTT-3'; HsMCM6 wild type-I+II: RNase protection assay 5'-TCAGGTTTGGCGCGAAATCTCC-3';HsMCM6mu- RNase protection assay was performed as described tant-I+II: 5'-TCAGGTTTGGATCGAAATCTCC-3';Hs- previously (Johnson et al., 1994). Brie¯y, 524-base and MCM6 wild type-III+IV: 5'-TTGCATTTGGCGC- 622-base anti-sense RNA probes beginning at +43 and GAAATCCCT-3'; HsMCM6 mutant-III+IV: 5'-TTGCA- +35 (relative to 5' ends of the cDNA as +1) in the TTTGGATCGAAATCCCT-3'; HsMCM6 wild type-V: 5'- HsMCM5 and HsMCM6 5' untranslated regions, respec- CGGCGGCGCGCGCAAAGCTGCA-3';HsMCM6mu- tively, were hybridized to 1 mgofyeasttRNA,MT-2orKit tant-V: 5'-CGGCGGCGCGATCAAAGCTGCA-3'. 225 poly(A)+ RNA for 16 h at 658C. RNA hybrids were These double-stranded oligonucleotides were also used as digested by RNase A (1 unit/ml) and RNase T1 (50 units/ competitors at 100-fold molar excess. A typical E2F site from ml) at 378C for 30 min, and protected fragments were the DHFR promoter (Ikeda et al., 1996) was also used as a analysed on a 7% polyacrylamide sequencing gel. Sequen- probe. Gel mobility shift assay was performed as described cing reactions using primers corresponding to the end of previously (Yee et al., 1989). In brief, radiolabeled probes the RNase protection probes were used as markers. were incubated with F9 cell extract in the presence or absence of unlabeled competitor oligonucleotides for 20 min at room temperature, followed by electrophoresis in a 5% polyacry- Construction of plasmids lamide gel in TBE (50 mM Tris-borate/1 mM EGTA) HsMCM5 (71384 to +43) and HsMCM6 (7754 to +35) containing 5% glycerol. 5' ¯anking sequences obtained by PCR using Promoter Finder DNA walking kit (Clontech Laboratories) were Infection with recombinant adenoviruses subcloned into the NcoI±SacI sites of pPBV2 (Wako) to construct pHsMCM5-Luc(71384) and pHsMCM6- The recombinant adenovirus for expression of E2F1, Ad- Luc(7754), respectively. The mutant of two sets of the E2F1 and control virus, Ad-Con (previously Ad-CMV), overlapping E2F sites in the pHsMCM5-Luc(71384) have been described previously (Schwarz et al., 1995). construct was made by site-directed mutagenesis which Quiescent REF52 cells cultured in DMEM containing changed TTTCCCGCGAAA (nt +2 to +13) to 0.1% FBS for 48 h were infected with Ad-E2F1 or Ad- TTTCCATCGAAA and TTTCGCGCCAAA (nt 7194 to Con at multiples of 300 plaque forming units per cell in 7183) to TTTCGATCCAAA, generating pHsMCM5- 2mlper150mmplatefor1h at378C (Schwarz et al., Luc(E2F7). The mutants of putative E2F sites in the 1995). Cells were further cultured in DMEM containing pHsMCM6-Luc(7754) construct were similarly prepared 0.1% FBS for 21 h and harvested for RNA isolation. by changing TTTGGCGCGAAA (nt 7107 to 796) to TTTGGATCGAAA and TTTGGCGCGAAA (nt 7169 to 7158) to TTTGGATCGAAA, and GCGCGCAA (nt 721 to 714) to GCGATCAA, generating pHsMCM6- Luc(E2FI-IV7) and pHsMCM6-Luc(E2FV7), respec- tively. pHsMCM6-Luc(E2FI-V7) has both . Acknowledgements The E2F1, E2F2, and E2F3 expression plasmids, as well This work was supported in part by a Grant-in-Aid for as the pCMV-b-gal, have been described previously General Scienti®c Research and Research from the (Johnson et al., 1994). Ministry of Education, Science, Sports and Culture of Japan, in part by a grant for CREST (Core Research for Evolutional Science and Technology) of Japan Science and Transfection assay TechnologyCorporation(JST)andinpartbyagrantfrom Transfection of REF52 cells and luciferase assay were the Kanehara Foundation. We thank Y Matsumura and M performed as described previously (Johnson et al., 1993). Horimoto for their excellent technical assistance. E2F-mediated expression of mammalian MCM genes KOhtaniet al 2309 References

Blow JJ and Laskey RA. (1988). , 332, 546 ± 548. Ohtani K, DeGregori J, Leone G, Herendeen DR, Kelly TJ Chong JP, Mahbubani HM, Khoo CY and Blow JJ. (1995). and Nevins JR. (1996). Mol. Cell. Biol., 16, 6977 ± 6984. Nature, 375, 418 ± 421. Ohtani K, Tsujimoto A, Ikeda MA and Nakamura M. DeGregori J, Leone G, Miron A, Jakoi L and Nevins JR. (1998). Oncogene, 17, 1777 ± 1785. (1997). Proc. Natl. Acad. Sci. USA, 94, 7245 ± 7250. Qin XQ, Livingston DM, Kaelin Jr WG and Adams PD. Donovan S, Harwood J, Drury LS and Di‚ey JF. (1997). (1994). Proc. Natl. Acad. Sci. USA, 91, 10918 ± 10922. Proc. Natl. Acad. Sci. USA, 94, 5611 ± 5616. Rose M and Botstein D. (1983). Methods Enzymol., 101, Fujita M, Kiyono T, Hayashi Y and Ishibashi M. (1996). 167 ± 180. Biochem. Biophys. Res. Commun., 219, 604 ± 607. Sato N, Arai K and Masai H. (1997). EMBO J., 16, 4340 ± Ikeda MA, Jakoi L and Nevins JR. (1996). Proc. Natl. Acad. 4351. Sci. USA, 93, 3215 ± 3220. Schwarz JK, Bassing CH, Kovesdi I, Datto MB, Blazing M, Johnson DG, Ohtani K and Nevins JR. (1994). Genes Dev., George S, Wang XF and Nevins JR. (1995). Proc. Natl. 8, 1514 ± 1525. Acad.Sci.USA,92, 483 ± 487. Johnson DG, Schwarz JK, Cress WD and Nevins JR. (1993). Shan B and Lee WH. (1994). Mol. Cell. Biol., 14, 8166 ± 8173. Nature, 365, 349 ± 352. Smith EJ, Leone G, DeGregori J, Jakoi L and Nevins JR. Kimura H, Nozaki N and Sugimoto K. (1994). EMBO J., 13, (1996). Mol. Cell. Biol., 16, 6965 ± 6976. 4311 ± 4320. Tanaka T, Knapp D and Nasmyth K. (1997). Cell, 90, 649 ± Kiyono T, Fujita M, Hayashi Y and Ishibashi M. (1996). 660. Biochim. Biophys. Acta, 1307, 31 ± 34. Thommes P, Kubota Y, Takisawa H and Blow JJ. (1997). Kubota Y, Mimura S, Nishimoto S, Masuda T, Nojima H EMBO J., 16, 3312 ± 3319. and Takisawa H. (1997). EMBO J., 16, 3320 ± 3331. TsurugaH,YabutaN,HashizumeK,IkedaM,EndoYand Kubota Y, Mimura S, Nishimoto S, Takisawa H and Nojima Nojima H. (1997a). Biochem. Biophys. Res. Commun., 236, H. (1995). Cell, 81, 601 ± 609. 118 ± 125. La Thangue NB. (1994). Trends Biochem. Sci., 19, 108 ± 114. TsurugaH,YabutaN,HosoyaS,TamuraK,EndoYand La Thangue NB, Thimmappaya B and Rigby PW. (1990). Nojima H. (1997b). Genes Cells, 2, 381 ± 399. Nucleic Acids Res., 18, 2929 ± 2938. Yan Z, DeGregori J, Shohet R, Leone G, Stillman B, Nevins Lam EW and Watson RJ. (1993). EMBO J., 12, 2705 ± 2713. JR and Williams RS. (1998). Proc. Natl. Acad. Sci. USA, Madine MA, Khoo CY, Mills AD and Laskey RA. (1995). 95, 3603 ± 3608. Nature, 375, 421 ± 424. Yee AS, Raychaudhuri P, Jakoi L and Nevins JR. (1989). Nevins JR. (1992). Science, 258, 424 ± 429. Mol. Cell. Biol., 9, 578 ± 585.