INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERNCES

The nature of the initiation sites for DNA replication has been investigated in many different eukaryotes, and a general model, largely based on viral systems, has been proposed (DePamphilis, 1996). This model defines an origin of replication as a core element flanked by a DNA unwinding element (DUE) and nonessential auxiliary sequences (e.g., binding sites for transcription factors); however, several origins from various organisms do not conform to this general scheme. The overall sizes of replication origins are diverse: several base pairs in some viral genomes, a few hundred base pairs in unicellular organisms such as Saccharomyces cerevisiae, Schizosaccharomyces pombe, or Physarum polycephalum (Maundrell et al., 1988; Newlon and Theis, 1993; Bénard et al., 1996; Dubey et al., 1996), and several kilobase pairs in some mammalian loci, such as the dihydrofolate reductase region of hamster cells or the human ADA gene (Vita-Pearlman et al., 1993; Dijkwel et al., 1994). The nature and the relative positions of these sequences also differ significantly from one organism to another. The core element, for example, is a short consensus sequence in S. cerevisiae (the 11-bp autonomously replicating sequence [ARS] consensus sequence [ACS]; Newlon and Theis, 1993). It is composed of one or several essential stretches of 30-50 bp in Kluyveromyces lactis (Fabiani et al., 1996) or in S. pombe (Dubey et al., 1996). In other cases, for example in P. polycephalum, it is a quite large initiation zone without any obvious conserved motifs (Bénard et al., 1996). The term core element is therefore inappropriate in these cases, and perhaps also in systems where initiation itself occurs at numerous dispersed sites (Dijkwel et al., 1994; Shinomiya and Ina, 1994; Dhar et al., 1996).

Current models for initiation of DNA replication stipulate that an initiator protein binds the origin to promote initiation of DNA replication (for review, see Diffley, 1996). The origin recognition complex (ORC) in the budding yeast S. cerevisiae is a complex of six polypeptides that binds the ACS and the adjacent B1 element (Lee and Bell, 1997, and references therein). This complex is always present at the replication origins (Diffley et al., 1994), which implies that steps other than ORC binding must be cell cycle-regulated (Diffley et al., 1995; Rowley et al., 1995). In vivo foot-printing studies have shown that the chromatin structure at the ARS elements changes during the cell cycle (Diffley et al., 1994). A so-called prereplicative complex assembles in G₁, with the loading of MCMs by Cdc6p at the origins (Tanaka et al., 1997), and initiation is triggered by cyclin-dependent kinases at the G₁/S boundary (Jallepalli and Kelly, 1997). All of these initiation factors identified in yeast have highly conserved homologues in all other eukaryotes, from K. lactis and S. pombe to the multicellular organisms Arabidopsis, Drosophila, Xenopus, Caenorhabditis, and Homo sapiens (for review, see Diffley, 1996). In many organisms, these proteins have been shown to be essential for DNA synthesis, suggesting the existence of a common mechanism controlling the initiation of DNA replication (Diffley, 1996).

In Xenopus egg extracts, ORC, Cdc6p, and the MCMs regulate initiation of DNA replication at apparently random sites (Rowles and Blow, 1997). In yeast, the same factors are involved in the regulation of precisely defined origins. It is unclear how conserved trans-acting factors interact with such a variety of cis-acting sequences. The study of simple eukaryotes with replication origins structurally different from those of S. cerevisiae may help elucidate what defines a chromosomal origin of DNA replication. In S. pombe, for example, replication origins consist of 600- to 800-bp sequences containing several essential regions, which include AT-rich stretches (Zhu et al., 1994; Dubey et al., 1996). Interference has been shown between very close origins identified within the ura4 chromosomal locus (Dubey et al., 1994), whereas a plasmid harboring any of these origins replicates autonomously in that yeast. On the other hand, discrete initiation sites have also been mapped on S. pombe chromosomes, like those in S. cerevisiae (Caddle and Calos, 1994; Wohlgemuth et al., 1994). In P. polycephalum the origins identified are located within the ribosomal DNA repeated unit (Bénard et al., 1995) or at promoter regions of actively transcribed genes (Bénard et al., 1996). They are not AT rich and do not confer autonomous replication to plasmids. In the yeast Yarrowia lipolytica, ARSs are selected on plasmids constructed from a genomic library at a very low frequency, i.e.,. one every 3 Mb (Fournier et al., 1991; Matsuoka et al., 1993). These replicative plasmids are maintained at low copy number (one to three per cell) with a relatively high mitotic stability (3% loss per cell generation). Genetic studies in meiosis and analysis of integrative transformation events for two such sequences indicate that they each contained a centromere (Fournier et al., 1993; Vernis et al., 1997); however, the replication does not start at the centromere, and an adjacent site of initiation was demonstrated in the cloned ARS. The centromere and the origin of replication can be physically separated on the plasmid, and the origin is also active and essential for initiation in its chromosomal locus (Vernis et al., 1997). These observations suggested that it should be possible to clone chromosomal origins using a centromeric vector. We report here the detailed analysis of the replication origins obtained by this strategy of screening randomly generated genomic fragments. We show that initiation sites for DNA replication are present around every 20- to 50-kb in the Y. lipolytica genome. These sites are intermediate in structure between the short and simple ARS elements of S. cerevisiae and the apparently nonconserved origin regions of other eukaryotes.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERNCES

Strains

Escherichia coli strains used (DK1 and XL1-Blue-MRF') have been described previously (Vernis et al., 1997). The Y. lipolytica recipient strain used for transformation was INAG33122 (MatB, lys2-5, leu2-35, ade1, xpr2), the reference strain for chromosome separation was E150 (MatB, his-1, leu2-270, ura3-302, xpr2-322), and the wild-type strain W29 was used to prepare genomic DNA for the gene bank. All three strains are deposited at the Collection de Levures d'Intérêt Biotechnologique (Grignon) under the numbers CLIB118, CLIB122, and CLIB89, respectively. Culture media were as described by Barth and Gaillardin (1996). Mitotic stability of plasmids was measured as described by Vernis et al. (1997).

Molecular Techniques

Y. lipolytica was transformed as described in Barth and Gaillardin (1996). Chromosomes were separated by published protocols (Zimmermann and Fournier, 1996). The methods for preparation of replication intermediates and two-dimensional (2D) gel analyses were those described by Huberman (1994). Standard techniques of DNA manipulation in E. coli and DNA hybridization were used (Sambrook et al., 1989).

Mutations in ori1068 and ori3018 were introduced by PCR amplification using primers harboring the mutation (ATCGATAA instead of TACAAGTA) and ended by an SalI site. A three-way PCR was needed in the case of ori3018, using a third central primer harboring the mutation, because the consensus was located in the central part of the origin.

An artificial I-SceI was introduced at the ClaI site of pINA989 obtained from a DNA library (Xuan et al., 1988) in E. coli by colony hybridization. pINA989 contains oriX009 on a 5.5-kb genomic fragment, and the LEU2 gene. After linearization with BamHI in the oriX009 region, this construct was introduced by targeted integration into the Yarrowia genome.

DNA Library

The construction of pINA732 used for cloning origins of replication was described by Vernis et al., 1997. We used a BamHI deletion of this plasmid, called pINA732-. Sau3a-digested genomic DNA was ligated into the BamHI site of pINA732-. A set of 5700 E. coli transformants was obtained. More than 90% of the plasmids carried genomic inserts, with a mean size of 1 kb. To check the size of the cloned inserts in yeast, we used two primers: 305 is a 17-bp oligonucleotide corresponding to the sequence starting at base 305 of pBR322, and 2216 is 28 bp corresponding to the sequence starting at base 2216 in the sequence of ARS68 (GenBank M91601).

The size of the inserts was determined to plasmids in 53 yeast transformants, and 11 short sequences were selected (<600 bp) for further analysis. Only 7 of these 11 were different, because one (the genomic repeated oriX009 sequence; see below) was found five times. Total DNA of only five of these seven yeast transformants was able to transform E. coli.

Colinearity with yeast genome was checked by PCR amplification of plasmid DNA, total DNA from a yeast transformant, and wild-type genomic DNA. A band of identical size was seen in all three cases for every cloned origin, showing that no major rearrangement had occurred during the cloning experiment (our unpublished results).

Sequences of PCR amplification products were determined after purification of the DNA with a PCR Purification Kit (Qiagen, Hilden, Germany).

DNA Sequence Analysis

Sequences were compared using the GCG package (University of Wisconsin, Madison, WI) or the Macaw software (G. Schuler, National Center for Biotechnology Information, Bethesda, MD) using the Gibbs algorithm. The 3D path of DNA molecules was calculated using the algorithm of Eckdahl and Anderson (1987) and the helical parameters of Bolshoy et al. (1991), as described previously (Pasero et al., 1993). The magnitude of DNA bending on curvature maps is expressed as the ENDS ratio, which is defined as the ratio of the contour length of a segment of the helical axis to the shortest distance between its ends. ENDS ratios were computed for a window of 120 nucleotides and a step of 10 nucleotides. The thermodynamic library of Breslauer et al. (1986), characterizing the ten Watson-Crick nearest-neighbor interactions in DNA, was used to predict the stability of DNA duplexes as described previously by Kowalski and coworkers (Natale et al., 1993). The G values presented in Figure 6 were calculated for a 120-bp DNA duplex in 1 M NaCl, pH 7.0, at 25°C. Note that the parameters used in the Thermodyn program of Kowalski are slightly different and lead to values that are proportionally lower. Variations of Roll angle and AT% were calculated as described previously (Marilley and Pasero, 1996) for a 120-bp window and a 10-bp step. The distribution of AT-rich sequences within origins was analyzed using a program developed by P. Trécourt (Mathematics Department, Institut National Agronomique-Paris-Grignon, 75231 Paris Cedex 05, France). The algorithm generates random sequences of given length and base composition and can be used to score the occurrence of a given motif. The results are saved in a format that can be exploited by the Excel software. Lines of the program are as follows (separated by//): dim a%(20)//open 0; #1, dnaseq.txt//input number of tests?,nsim%//input length of the sequence?,nls//input number of A and T?,no//if no >= nls then end//freq=no/nls//temp%=0:randomize(timer)//for l = 1 to nsim%//for i = 1 to 20 :a%(i)=0:next i//for i = 1 to nls//x=rnd(1)//if x < freq then incr temp% else fin%=1//if fin%=1 then//if temp% > 0 then//if temp% > 20 then temp%=20//incr a%(temp%)//end if//temp%=0//fin%=0//end if//next i//if temp% > 0 the, incr a%(temp%)//for i = 1 to 19//print using ###;a%(i)//print#1,using ###;a%(i)//next i//print using ##;a%(20)//print#1,using ##;a%(20)//print//next l//print results are saved in the file DNASEQ.TXT//end.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERNCES

Cloning Origins of Replication with a Centromeric Plasmid

Replicative vectors in Y. lipolytica require both a centromere and a replication origin. Genomic DNA was ligated into a CEN1-based plasmid (see MATERIAL AND METHODS) (Figure 1), and the resulting library was used to transform Y. lipolytica by electroporation. More than 90% of the transformants displayed mitotic instability. Five plasmids carrying small-sized inserts were back-transformed into E. coli and further analyzed. The colinearity of the fragments with genomic DNA was checked by PCR. Purified plasmid DNA transformed Y. lipolytica at high frequency (replicative transformants), showing that the inserts confer autonomous replication to a centromeric plasmid (Figure 1). From the transformation frequency of the plasmid pool as well as of individual cloned ARSs, we estimate that there is one putative origin of replication for every 20-50 kb of the Y. lipolytica genome.

View larger version (30K): [in this window] [in a new window]   Figure 1.   Cloning of Y. lipolytica origins with a centromeric plasmid. (A) The cloning vector pINA732- contains the LEU2 gene (dotted area), the CEN1 centromere, and flanking genomic sequences. Y. lipolytica genomic DNA digested with Sau3A was ligated into the BamHI site of the vector. (B) Characteristics of the eight ORI elements analyzed in this article. Demonstration that five of them are active origins comes from Vernis et al. (1997) and from data presented below in this article. We therefore adopt the ORI designation for all of them. Two of these sequences, marked by a star (ori1068 and ori3018), were subjected to systematic size reduction to determine the minimal functional fragment (Figure 4). The chromosomal localization of the different fragments was determined by pulsed-field gel electrophoresis and Southern blotting using reference strain E150. Chromosomal bands are numbered according to Vernis et al. (1997). Chromosomal activity (active; nt, not tested) refers to the observation of replication initiation signals after 2D gel electrophoresis.

The Cloned Sequences Represent a Wide Variety of Genomic Loci

Several types of ribosomal DNA repeats have been described in Y. lipolytica that differ in the sequence of their nontranscribed spacers (NTSs). Two of them have previously been cloned and sequenced (G and P2 units) (van Heerikhuizen et al., 1985). One of the clones obtained here in our shotgun cloning of origins displays an almost perfect sequence identity with the common part of the two known NTSs. Because rDNA is a repeated sequence, it should be found several times in the gene bank, but our selection focused on smaller DNA pieces, so that only one such clone was analyzed. It was called ori-rDNA. The fragment isolated was 597 bp, and a functional 285-bp subfragment was obtained by a HindIII deletion between a site in the pBR322 vector and an internal site. A putative replication origin is thus present in the Y. lipolytica NTSs as in the yeasts S. cerevisiae and S. pombe (Linskens and Huberman, 1988; Sanchez et al., 1998a), the slime mold P. polycephalum (Bénard et al., 1995), and many other eukaryotes. Ori-rDNA hybridizes to the five chromosomal bands on an electrophoretic karyotype (Figure 1), which is consistent with the presence of several independent clusters of rDNA in the Y. lipolytica genome (Fournier et al., 1986), and with their mapping on several chromosomes in different strains (Casarégola et al., 1997).

Two other sequences (oriX009 and oriX096) hybridize to several chromosomal bands (Figure 1). Further genomic mapping by Southern hybridization suggested that each of these sequences is part of a larger element and that both elements are present in multiple copies. We compared the sequences of oriX009 and oriX096 with the Ylt1 transposon of Y. lipolytica (Schmid-Berger et al., 1994) and found no similarity. We also compared the genomic maps of oriX009 and oriX096 loci with the known maps of rDNA (van Heerikhuizen et al., 1985) and mit-DNA (Wesolowski et al., 1981); they did not appear to originate from either of these repeated DNAs. We then investigated whether oriX009 is a subtelomeric sequence. Indeed, it has been reported that an ARS, although generally inactive in the chromosome, is closely associated with the telomere in S. cerevisiae (Newlon et al., 1993). We introduced a rare cutting I-SceI site near oriX009 sequences in the chromosomes to measure the distance between oriX009 and the chromosome end after an I-SceI cut. pINA789 containing 5.5 kb flanking oriX009 was isolated from a genomic library and tagged with an I-SceI site (see MATERIALS AND METHODS). This plasmid was linearized by BamHI within the oriX009 flanking sequences and integrated into one oriX009 region by integrative transformation. The structure of the integrated DNA was checked by Southern blotting and hybridization. Chromosomal plugs of several transformants were prepared in agarose, digested with I-SceI, and separated by field inversion gel electrophoresis. The gel was blotted and hybridized with a pBR322 probe, which revealed bands ranging from 20 to 500 kb (Figure 2). This indicates that oriX009 sequences are located much further away from chromosome ends than are subtelomeric repeats in S. cerevisiae.

View larger version (52K): [in this window] [in a new window]   Figure 2.   Localization of the repeated oriX009 sequence in the genome of Y. lipolytica. (A) Map of a chromosomal integration at an oriX009 locus of the pINA989 containing oriX009 within its genomic environment, the LEU2 gene, and an I-SceI site. pBR322 vector sequences are shown in light gray. (B) Chromosomes from various transformants (lanes 1-5) were embedded in agarose plugs, digested with I-SceI, and separated under the following field inversion gel electrophoresis conditions: 1% pulsed field agarose (Bio-Rad, Bio-Rad, Hercules, CA) in 0.5× TAE buffer, run at 6 volts/cm for 14 h at 14°C with an angle of 106° and a variation of pulse from 0.41 sec to 2 min 6.52 sec. The DNA was transferred to a membrane and hybridized with a pBR322 probe (black box). A nontransformed strain is shown as a control in lane 6. The Raoul marker (Appligene, Heidelberg, Germany) is in lane 7. S. cerevisiae chromosomes and  concatemers from Bio-Rad (not visible on the autoradiogram) were also used as molecular weight markers. As a comparison, the smallest chromosome among the six from Yarrowia is 2.6 Mb, and the largest is 4.9 Mb.

Ori3068 and ori4021 each hybridized to a single chromosomal band. Genomic restriction digestion and Southern hybridization experiments indicated that these were only single copies of these origins in the genome. A restriction map of each region was established, which confirmed the colinearity between the cloned fragment and the genomic locus.

Various genomic regions, either repetitive (ori-rDNA, oriX009, oriX096) or unique (ori3068, ori4021), from different chromosomes are thus able to confer extrachromosomal maintenance to a CEN1 plasmid. This confirms our previous interpretation of our initial ARS assay (Vernis et al., 1997): only centromere-proximal origins were then cloned, not because of some specific features displayed by these sequences but because of the absolute requirement for a centromere to maintain plasmids in Y. lipolytica.

Ori4021 Is Active in the Chromosome

To test the chromosomal activity of these sequences, replication intermediates were prepared and separated on a 2D gel and then hybridized with an origin probe (see MATERIALS AND METHODS). In the case of the single copy ori4021 sequence, a 5.7-kb BamHI restriction fragment centered around the putative origin was analyzed (Figure 3). A typical bubble arc was visible on the 2D gel, indicative of an initiation in the central third of the restriction fragment. A faint Y signal, which may occur if the origin fails to fire at every cell cycle, was also detected; however, this Y arc is unusual because large Y-shaped intermediates are less visible. Possibly this was due to 1) the superposition of two Y arcs resulting from partial digest with BamHI or as 2) a diffuse termination signal in this zone as shown on Figure 3A. On a BglII restriction fragment (control), only the Y arc was present, confirming that the ori4021 initiation site corresponds to a discrete locus and is not part of a large initiation zone, as is the case for some higher eukaryote loci (for review, see DePamphilis, 1996). An accumulation of molecules is also visible as a spot on the ascending part of the Y arc, which may indicate the presence of a replication pause site in that region.

View larger version (35K): [in this window] [in a new window]   Figure 3.   Ori4021 is active in the genome. (A) Sketch showing the migration patterns of various types of replication intermediates during neutral/neutral 2D gel electrophoresis. B) A restriction map of the genomic locus shows the two BamHI and BglII restriction fragments that are separated on a 2D gel and hybridized with an ori4021 probe. The pause signal observed along the Y arc on the BglII digest is not discussed here.

On the basis of existing restriction maps of ribosomal DNA units (van Heerikhuizen et al., 1985), a 2D gel analysis of EcoRI-digested genomic DNA, using ori-rDNA as a probe, was expected to reveal ori-rDNA activity. The pattern obtained was actually complex, with degradation products masking some of the replication intermediates; however, at least two bubble arcs indicative of initiation sites were detected, as well as Y arcs (our unpublished results). This complex pattern may reflect a length polymorphism of the NTS and possibly initiation at only a subset of the rDNA units, as is the case for S. cerevisiae (Linskens and Huberman, 1988).

Thus ori4021 and ori-rDNA confirm that the use of a centromeric vector to clone genomic DNA allows the isolation of active chromosomal origins of replication. The centromeric origins described previously (Vernis et al., 1997) are therefore not the only replication loci on the chromosome.

The Origin Size Can Be Reduced to 125 bp on Plasmid

To define more precisely what is the minimal size required for initiating replication, we focused on the two previously characterized centromeric origins ori3018 and ori1068, which are active both on plasmids and in the chromosome location (Vernis et al., 1997). Fragments of different lengths were synthesized by PCR amplification (Figure 4) and inserted into the SalI site of the LEU2-CEN1 vector pINA732 (Fournier et al., 1993). The resulting plasmids were checked by digestion and by sequencing the origin and were subsequently used to transform Y. lipolytica. Ori1068 reduced to a 125-bp fragment and ori3018 reduced to a 144-bp fragment are each sufficient to transform yeast at high frequency (Figure 4). It is possible that ori3018 could be reduced further, but smaller amplification fragments were structurally unstable in E. coli. No significant changes in transformation efficiency were observed between the various deletions, so no functionally essential sequences appeared to map outside the minimal origins. We also checked that the yeast transformants obtained with the smallest origin sequences were still mitotically unstable like the initial ORI plasmids, and that the plasmids were present as monomeric CCC molecules. The small length of these deleted centromeric origins is comparable to the size of other genomic origins cloned above, like ori3068 (115 bp) or oriX009 (159 bp), and we turned to an extensive sequence analysis of these regions.

View larger version (14K): [in this window] [in a new window]   Figure 4.   Mapping the minimal functional fragment of ori1068 (A) and ori3018 (B). Position and orientation of the synthetic primers are shown on the top line. Nucleotides positions are according to GenBank M91601 and M91600. The strain INAG3122 was transformed by electroporation with centromeric plasmids containing the reduced origins, because only replicative transformation is obtained at high frequency with this method. Results of transformation are shown by + and . NR, Not recovered in E. coli.

Six Yarrowia Origins Share an 8-bp Motif That Is Dispensable for ARS Function

We looked for the presence of the S. cerevisiae 11-bp ARS consensus sequence (WTTTAYRTTTW) in the eight Yarrowia origins and found no perfect match. Several but not all origins displayed 9/11 and 10/11 matches. Neither the 40-bp core sequence of K. lactis ARS (Fabiani et al., 1996) nor motifs frequently found in S. pombe origins (Maundrell et al., 1988, Dubey et al., 1996) were present. We therefore looked for a sequence common to the Y. lipolytica origins. Several sequence alignment programs were used (see MATERIALS AND METHODS), and we repeatedly found copies of a short 8-bp motif TDCAAGTH (D = A or G or T; H = A or C or T), which is present one to five times in all the origins, except in oriX009 and in oriX096 (Figure 5). To assess the role of this sequence in origin function, we mutated the TACAAGTA sequence in ori1068 and ori3018 to redistribute the bases within this 8-bp stretch without affecting the overall base composition. In ori3018, the two partially overlapping motifs were both destroyed. The resulting sequences ligated into the centromeric vector still transformed Y. lipolytica with high frequency. The consensus is therefore not essential for origin function on a plasmid.

View larger version (20K): [in this window] [in a new window]   Figure 5.   Localization of the consensus sequence TDCAAGTH in six of the eight Y. lipolytica origins. Nucleotide sequences are deposited under the following GenBank accession numbers: AF038941 (oriX009), AF038942 (oriX096), AF038943 (ori4021), AF039581 (ori-rDNA), and AF038940 (ori3068).

Because Yarrowia origins apparently do not share any essential sequence consensus, we then looked for the overrepresentation of short motifs. Indeed, Zhu et al. (1994) have observed that some AT-rich hexanucleotides are abundant in S. pombe ARS, suggesting that they could be important for initiation. To perform a statistical analysis of the distribution of nucleotides in Yarrowia origins, we used a program developed in our lab (see MATERIALS AND METHODS) that generates random sequences of a given length and a given composition. Five thousand random DNA sequences of the same length and base composition as each Yarrowia origin were generated, and the occurrence of AT-rich oligonucleotides from SW₁S to SW₂₀S (where S stands for G or C) was scored. The frequency of occurrence of these motifs in a Yarrowia origin was then compared with the frequency of occurrence of the same motifs in the generated sequences using a ² test. We failed to detect any significantly unusual distribution of oligonucleotides in the Yarrowia origins. As a control, the same program was applied to scaffold-attached regions (SARs), which have been reported to be frequently associated with origins of replication (DePamphilis, 1996) and are characterized by a biased distribution of AT repeats (Roberge and Gasser, 1992). With the Drosophila fushi tarazu (ftz) SAR or with the SAR associated with the histone gene, a significant divergence from a random distribution was detected by the program (² 20.3 vs. 15.51 in a standard ² table for ftz, and 42.77 vs. 18.31 for HIST). We conclude that the eight Yarrowia origins identified so far do not contain any significant reiteration of primary sequence motifs.

We also used the same program to analyze two origins from higher eukaryotes: the 55% AT-rich 500 bp of the ori locus in the hamster DHFR origin region (Burhans et al., 1990), and the 51% AT-rich 1994 bp of the origin associated with the mouse ADA gene (Vita-Pearlman et al., 1993). Both gave a significant ² value (17.87 vs. 11.07 in the first case, and 47.07 vs. 15.51 in the second), the bias being due to an excess of the classes SWS and SWWS and to a lower representation of classes with >4 W. This result is in accordance with the presence of alternating purines, which has been described in these origins as well as in other eukaryotic origins (Bergemann and Johnson, 1992). A protein factor displays affinity for the purine motif GGNNGAGGGAGARRRR (R = purine; N = any base), but no Pur sequence was found in the Y. lipolytica origin sequences.

Thus, no identifiable short sequence motifs or consensus sequences in Yarrowia origins could be detected. This situation can be compared with domains B within S. cerevisiae ARSs, which play a crucial role in initiation and can be exchanged between different origins, although they do not display any obvious conserved sequence (Marahrens and Stillman 1992).

Yarrowia Origins Do Not Display Any Characteristic Structural Features

It has been proposed that eukaryotic replication origins are associated with clusters of structural motifs such as SARs, DUEs (Umek and Kowalski, 1988), or bent DNA (Trifonov, 1991), which can be mapped by computer modeling (Eckdahl and Anderson, 1990, Dobbs et al., 1994). As illustrated in Figure 6, A and E, bent DNA is frequently associated with S. cerevisiae origins (Snyder et al., 1986; Williams et al., 1988), although it does not seem to be a general feature of the origins in yeast, or may even be dispensable (Marahrens and Stillman, 1992). We display here ARS1 and HMRE as controls. The curvature of yeast origins was calculated with the wedge model of Trifonov (see MATERIALS AND METHODS) and is displayed as a projection of the spatial path of the molecule for small fragments (Figure 6, A and B) or as a curvature map for longer molecules (ENDS ratio; Figure 6E). When applied to the Yarrowia origins (Figure 6, B, D, and E; and our unpublished results), this analysis did not suggest significant DNA bending in the minimal origin fragment.

View larger version (50K): [in this window] [in a new window]   Figure 6.   Y. lipolytica origins do not contain structural motifs that are frequent in S. cerevisiae ARS elements. (A) S. cerevisiae ARS1 contains a strong, bent DNA region. The 2D projection of the 3D path of the molecule was calculated as described in MATERIALS AND METHODS. The position of the three elements A, B1, B2, and B3 (Marahrens and Stillman, 1992), the binding sites of ORC and ABF1 (Lee and Bell, 1997), and the site of initiation of DNA synthesis (Bielinsky and Gerbi, 1998) are indicated. (B) Y. lipolytica origins are not intrinsically curved. The 3D path of the minimal origin region of ori3018 (144 bp), ori1068 (125 bp), and oriX009 (159 bp) was calculated as described for ARS1. (C)Variation of helical stability and minor groove width along the S. cerevisiae ARS1 DNA fragment (350 bp). The free energy of the duplex or G (207-227 kcal/mol; circles of increasing size) and the mean Roll angle (1 to 0.1°; gray levels of increasing intensity) were calculated as described in MATERIALS AND METHODS for a 120-bp window sliding along the molecule. The position of the SAR mapped by Amati et al., 1988 (low Roll angle) and of the DUE (low G) are indicated. (D) Spatial path and helical parameters of Y. lipolytica minimal ori-rDNA fragment (284 bp). The mean Roll angle varies from 0.3 to 0.5°, and the G ranges from 215 to 226 kcal/mol. (E) Structural maps of large fragments containing S. cerevisiae HMRE ARS, Y. lipolytica ARS2, and Y. lipolytica ori4021. Like ARS1, the ACS of the HMRE ARS is flanked by a bent motif bearing binding sites for the auxiliary factors RAP1 (dark box) and ABF1 (open box) and by a DUE. The SAR fragment (Amati et al., 1988) maps perfectly to the region of low Roll angle. In all the graphs, the dotted lines correspond to the average value. Ori4021 is shown in its pBR322 environment, and the other two origins are at their chromosomal locus.

Regions of low helical stability that facilitate the initial unwinding of the DNA molecule are often found in the vicinity of replication origins (Umek and Kowalski, 1988). We used the thermodynamic library of Breslauer et al. (1986) to map the variation of the energy (G) required to unwind the DNA at different origins (Figure 6, C and E). A low G DUE is located close to the ACS of ARS1 and of the HMRE ARS. No such feature is detectable in any of the cloned Yarrowia origins (Figure 6, D and E; and our unpublished results), which therefore lack a strong DUE; however, it is noteworthy that easily unwound sequences have been found in the neighboring plasmid sequence that could potentially substitute for DUEs.

Finally, we looked for the presence of putative SARs in Yarrowia ARS elements, because they often colocalize with eukaryotic origins (Amati and Gasser, 1988; Brun et al., 1990; Razin et al., 1991; Du et al., 1995). These sequences are generally AT rich and present a narrow minor groove (Roberge and Gasser, 1992) that can be mapped using the Roll angle, one of the three helical parameters of the double helix (see MATERIALS AND METHODS). As shown in Figure 6, B and E, the region of low Roll angle fits perfectly with the position of the SAR mapped experimentally for S. cerevisiae ARS (Amati et al., 1990). When applied to Y. lipolytica origins, this analysis did not reveal any significant pattern (Figure 6, D and E). A putative SAR between ori4002 and CEN2 (Figure 6E), as well as a strong bending element (ENDS ratio >1.2) previously detected by gel retardation assay (Matsuoka et al., 1993), were found. Similar "low Roll" regions are predicted between the ORI and the CEN from the two other chromosomes, suggesting that it may play a role in chromosome maintenance; however this region can be deleted from the plasmids without affecting the transformation frequency (Vernis et al., 1997).

Ectopic Initiation Occurs within Short DNA Fragments

We have shown that short DNA fragments presenting no obvious primary or secondary sequence similarity are able to confer extrachromosomal maintenance to a centromeric plasmid. We examined the question of whether these minimal sequences contain all the information that is necessary and sufficient to function as an origin of replication on the chromosome. We already knew that ori1068 is still active when its 3'-flanking region is modified (Vernis et al., 1997). To test whether Y. lipolytica ORIs require a particular chromosomal context or additional information to be active in the genome, several origins were moved to another locus and analyzed by 2D gel. Two noncentromeric origins (the 285 bp of ori-rDNA and the 158 bp of oriX009) and one centromeric origin (the 125-bp minimal ori1068) were inserted into the SalI site in pINA214 (pBR322 + LEU2) and integrated into the LEU2 locus in the chromosome after BstXI digestion of the plasmid. A chromosomal XhoI-SacII restriction fragment containing the origins was analyzed on 2D gel (Figure 7). A control experiment, performed with the vector alone integrated at the same locus, showed the absence of initiation in this region (Figure 7D), whereas each of the three origins tested generated an initiation signal within this restriction fragment (Figure 7, A-C). Therefore, we conclude that the short Yarrowia ORI elements contain the information sufficient to promote initiation at ectopic loci, even when embedded within heterologous (pBR322) sequences.

View larger version (41K): [in this window] [in a new window]   Figure 7.   Ectopic initiation occurs within short DNA fragments. (A-C) Three different origins (A, ori-rDNA; B, oriX009; C, ori1068) were ligated into pINA214; the resulting plasmids were digested with BstXI and integrated into the LEU2 locus. The resulting genomic restriction map is shown, with the position of the sites used to digest the DNA for 2D gel analysis. The gels were blotted and hybridized with a pBR322 probe, indicated as black boxes. (D) The same experiment was performed with pINA214 as a control: hybridization of the 2D gel showed the absence of initiation at the LEU2 locus. Notice that the diagonal signal visible on these images is present on most chromosomal 2D gels performed in Yarrowia and may correspond to degradation products.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERNCES

The Y. lipolytica Replicon

It had been shown previously that ARS sequences are very rarely obtained from this yeast (Fournier et al., 1991; Matsuoka et al., 1993). We later demonstrated that this low cloning frequency was due to both a centromere and an origin of replication being necessary to establish an ARS vector (Fournier et al., 1993; Vernis et al., 1997). Because an origin can be associated with different centromeres to confer extrachromosomal plasmid maintenance, we postulated that any chromosomal origin could therefore be cloned using a centromeric vector (Vernis et al., 1997). The results presented here confirm this hypothesis: several new ORI elements from various chromosomal loci were cloned using this strategy. Those tested were shown to be chromosomal initiation sites for DNA replication. Our data suggest that origins of replication are present at every 20-50 kb in the Y. lipolytica genome, consistent with the average replicon size of other yeasts (Maundrell et al.,1985; Newlon and Theis, 1993).

These origins of replication are well defined genetic elements: the deletion of 240 bp from the ori1068 chromosomal locus abolishes initiation, whereas the introduction of bacterial sequences nearby does not (Vernis et al., 1997). We demonstrate here that an element as short as 125-150 bp comprises all the necessary information to initiate DNA replication outside of its natural chromosomal environment. Linker scanning or internal deletion analysis would be required to establish whether all of this sequence is necessary or relevant for ORI function

The data presented here indicate that centromeric sequences are not required to initiate DNA replication on the chromosome but are absolutely essential for plasmid maintenance. They may be necessary for any distribution of plasmids at mitosis, which is not the case in S. cerevisiae. Alternatively, they may function as helper sequences for the replication origins, perhaps through a modification of their chromatin structure or of their subnuclear localization. These two possibilities are not mutually exclusive, but several lines of evidence are in favor of the second interpretation. Indeed, it has been reported that S. cerevisiae ARS elements bind the nuclear scaffold (Amati et al., 1990). These origins contain AT-rich stretches with a narrow minor grove (Figure 6) that could target the plasmids to the replication foci within the nucleus (Cook, 1991; Pasero et al., 1997). As shown by computer modeling (Figure 6) or by the ² test (our unpublished results), Y. lipolytica origins do not contain any similar structures, but the centromeric regions do. Therefore, the putative role of centromeres on a plasmid may be to direct plasmids to the replication centers. The SAR activity of Y. lipolytica ORI and CEN fragments still has to be tested to confirm this hypothesis, as well as the replacement of the CEN on the plasmid with an exogenous SAR fragment (Vernis, unpublished observations), but this major difference between Y. lipolytica and S. cerevisiae ARS will certainly prove very useful for investigating the putative role of SARs in the definition of eukaryotic origins of replication (Hyrien et al., 1997).

Y. lipolytica Origins Do Not Share Any Essential Consensus Motif

In the initially cloned 1.3-kb ARS18 and 2.3-kb ARS68 fragments, we observed several 9/11 matches to the S. cerevisiae ACS, which may be why they were able to replicate in the budding yeast (Fournier et al., 1993); however, these ACS are not present within the minimal ORI sequences. We then identified a short 8-bp stretch present in all but two sequenced origins; however, three observations suggest that this sequence is not essential for origin function. 1) Mutation of this sequence does not affect transformation frequency; 2) two origins (oriX009 and oriX096), which do not harbor a perfect match to this consensus, allow extrachromosomal replication on a centromeric plasmid; and 3) one of these two origins (oriX009) is still able to initiate DNA replication after being moved to an ectopic chromosomal location, where it is flanked by bacterial plasmid sequences (Figure 7B). Most of the origins analyzed harbor several degenerated copies of this consensus, however, and in S. cerevisiae the replicative property is conserved when several partial ACS are present within the ARS (Zweifel and Fangman, 1990). We examined all exact and degenerated copies of the consensus in the Yarrowia origins (32 occurrences) and found a more degenerate motif: WDMRWNYH (R = purine; Y = pyrimidine; M = A or C; W = A or T; D = A or G or T). Because some of the analyzed sequences share only five bases, it seems very unlikely that this motif represents a biologically significant consensus. Theis and Newlon (1997) found that ORC can bind a 9/11 match to the S. cerevisiae ACS, and this lead them to redefine the consensus on a broader length (17 bp instead of 11 bp). We therefore searched the flanking sequences of all exact matches to the Y. lipolytica putative consensus and did not find any other conserved bases that could allow a broader definition of this element.

We devised a program to look at the distribution of bases within the sequences of the origins, and no bias was found. We also analyzed the 2675-bp ura4 locus of S. pombe (67.6% AT), which harbors the two initiation sites ars3002 and ars3003 (Dubey et al., 1994). The result is significant (² of 23.76 vs. 21.03 in the table). Surprisingly, when this analysis is performed on each ARS separately (821 bp of ars3002 and 543 bp of ars3003), no bias is observed. A similar observation was made for the Y. lipolytica ori3018 environment (see above). In this case, the origin efficiency should be very dependent on the nature of the flanking sequences, as postulated for some metazoan origins (Larner et al., 1997) and should be very sensitive to chromosomal position effects. In S. pombe the origin efficiency is indeed affected by the presence of exogenous sequences, and plasmid maintenance is sometimes associated with a multimerization of the origin in vivo (Zhu et al., 1994). In Y. lipolytica, we never observed such an alteration of plasmid structure, which suggests that all the information necessary and sufficient for initiation is found within the origin. This is consistent with the observation that such a sequence can be moved to the LEU2 locus and retain origin activity; however, we did not compare the initiation efficiencies at the natural and at this ectopic locus and therefore cannot completely rule out a possible effect of the genomic environment on origin activity. It seems at least likely that a DUE, if absent from the cloned origins, should be present in the vicinity of the origins on the chromosome to facilitate the entry of the replication machinery (Lin and Kowalski, 1997).

A situation similar to that of Y. lipolytica origins has been found in Physarum, where two origins have been described in the actin genes promoters (Bénard et al., 1996). In both loci962-bp actB with 53.3% AT, and 1049-bp actC with 56.8% ATno bias was observed in the distribution of AT-rich sequences. In both organisms the AT content of origins is quite low, the bases seem to be randomly distributed, and no canonical origin consensus nor sequences similarity between the origins are found. It could be interesting to check whether Y. lipolytica origins are within or near promoter regions.

Yarrowia As a Model to Analyze ORC-Origin Interaction

As mentioned above, the ORC is highly conserved among eukaryotes. It is required for the initiation of DNA replication in S. cerevisiae, S. pombe, Xenopus, or Drosophila (for review, see Diffley, 1996), although the replication origins are structurally very different in these organisms. Several explanations of this apparent paradox have been proposed (Burhans and Huberman, 1994; Diffley, 1996; Larner et al., 1997). One tempting idea is that any DNA sequence that is able to bind ORC and that is located in a chromatin environment permissive for initiation becomes an origin. In this case, the higher-order organization of chromatin would be the main determinant of origin function. Chromosome architecture was indeed shown to dictate site-specific initiation in Xenopus egg extracts (Lawlis et al., 1996; for review, see Hyrien et al., 1997), and the random initiation observed in the rDNA of early Xenopus embryos is subsequently restricted to nontranscribed regions after activation of transcription (Hyrien et al., 1995). Alternatively, the binding of ORC could be strictly sequence-specific in somatic cells but could also occur at degenerated sites when the ratio of ORC to DNA is very high, as in early embryos. It is therefore essential to characterize the sequences that are recognized by this complex in systems other than S. cerevisiae. This is technically very difficult in higher eukaryotes, where the initiation zones are large regions. Interestingly, Sanchez et al. (1998b) showed that several proteins bind to ARS elements in S. pombe; however, the two proteins identified so far do not correspond to the initiator protein. Similar work in Y. lipolytica would be valuable, because origins correspond to much shorter and well defined genetic elements. The identification at Y. lipolytica origins of a footprint that changes during the cell cycle would certainly be an important step toward the characterization of the sequences recognized by ORC on eukaryotic chromosomes. This is a further illustration that unicellular eukaryotic organisms different from S. cerevisiae can provide useful models for exploring mechanisms at replication origins.