The capacity of promoter DNA of two yeast genes to be unwound was studied. Both promoters, those of the CYC1 and DED1 genes, contain long oligopurine.oligopyrimidine (R.Y) tracts. The two promoters were cloned into negatively supercoiled plasmids and their sensitivity to single strand specific nuclease P1 was examined. Extensive P1 cleavage was located within the R.Y tracts and cleavage sites were mapped. The extent of cleavage was only slightly dependent on P1 concentration, indicating a slow conversion of an intermediate form of DNA into the P1 reactive state. The cleavage required negative supercoiling and was suppressed by NaCl, MgCl2 and spermine. Two-dimensional topoisomer analysis showed that six superhelical turns were opened in the plasmids examined. The results indicate that at sufficient torsional stress, the R.Y tracts can intermittently undergo a transition into an unwound, ready to separate state. The oligopurine.oligopyrimidine tracts may thus serve as DNA unwinding centers in the gene promoters where they reside.

The extent of DNA unwinding in preparation for transcription and replication has been studied for a considerable time, but many details remain unclear. A transition from a "closed" to an a largely unwound, "open"initiation complex is well established (Chamberlain, 1976; McClure, 1985; Amouyal and Buc, 1987; Borowiec and Gralla, 1987; Borowiec et al., 1990) and the size of the initiation bubble for E. coli RNA polymerase has been determined by topoisomer analysis, as 17-20 bases per polymerase molecule (Saucier and Wang, 1972; Gamper and Hearst, 1982). Evidence for the formation of unwound DNA regions ahead of the polymerase binding site was brought by Larsen and Weintraub (1982) when they found that certain regions in the promoters of chicken globin genes were susceptible to attack by single strand specific nucleases (SSN). This susceptibility to SSN was found in the intact nuclei of transcriptionally activated cells as well as in negatively supercoiled plasmids carrying the promoter region. Detailed mapping of the chicken b-globin promoter region (Nickol and Felsenfeld 1983) revealed that the main susceptible region (# -192 to -176) is a tract of 21 G on the coding strand, with one interrupting C. Many other promoter regions in higher eukaryotes have since been found to be SSN sensitive, i.e. at least intermittently unwound. A feature common to all these sensitive regions is that they are entirely or mostly homopurine on one strand and homopyrimidine on the complimentary one (Wells et al., 1988; Yagil, 1991). The more intensively investigated of these purine.pyrimidine tracts (R.Y tracts) include the promoters of the b-globin gene (Larsen and Weintraub, 1982; Nickol and Felsenfeld, 1983), the U1 gene (Htun and Dahlberg, 1989), several artificial analogs (Mirkin et al., 1987), the hsp26 (Glaser et al., 1990) and myc (Michelotti et al., 1997) genes, immunoglobulin genes (Collier et al., 1987; Kohwi-Shigematsu and Kohwi, 1990) and several other genes (Wells et al., 1988; Palecek, 1991; Yagil 1991).

In Saccharomyces cerevisiae several regions sensitive to SSN were discovered. These include the 2m plasmid, the H4 gene ARS region (Umek and Kowalski, 1987;1988), the C2G1 ARS region (Natale et al., 1992), sub-telomeric (Gilson et al., 1993) and centromeric (Tal et al., 1994) regions. Regions sensitive to KMnO4 have been identified in the promoters of the GAL1, GAL10 and HSP82 genes (Giardina and Lis 1993; 1995), which implies that these regions are either melted or highly distorted. The dominating compositional theme of most of these yeast sequences was, however, A,T richness, rather than the R.Y theme. A,T rich DNA regions constitute the classical DNA unwinding motif, which led Umek and Kowalski (1988) to propose that these SSN sensitive regions serve as DNA unwinding elements (DUE). Nevertheless, an extremely high prevalence of R.Y tracts (46 times that of random DNA!) was revealed when the 184 promoter regions of the first fully sequenced yeast chromosomes III (Yagil, 1994) and XI (Yagil, unpub.) were surveyed for these tracts. We therefore thought it of interest to determine whether R.Y rich regions in yeast promoters can also undergo a transition into an unwound state and thus serve as unwinding centers in DNA transcription or replication processes.

In this study two prominent yeast promoters are investigated, both containing long pyrimidine tracts. One promoter is of the DED1 gene, identified and sequenced by Struhl (1985a), containing 39Y/40N1 at # -149 to -110 nt from the first ATG. Deletion studies showed that this R.Y tract must be intact for effective transcription (Struhl, 1985b; Chen et al., 1987) and can stimulate both the in vivo and the in vitro transcription promoted by an adjacently fused CYC1 promoter (Lue et al., 1989). The other promoter studied is that of the iso-cytochrome c1 (CYC1) gene; the coding strand of this promoter contains a 32Y/37N R.Y tract, divided into one section of 21Y/22N and one section of 10 consecutive Y (Smith et al., 1979). Systematic deletion studies confirmed that these oligopyrimidine tracts (although not identified as such) reside in a promoter region necessary for high rates of initiation of the CYC1 gene (Faye et al., 1981; Guarente and Mason, 1983; McNeil and Smith 1986). In addition, one of the TATA boxes adjacent to the R.Y tracts of CYC1 was found to adopt a partly unwound conformation when cocrystallized with the TATA binding protein TBP (Kim, Y. et al., 1993; Kim, J.L. et al., 1993).

We report here that the R.Y stretches in the promoter regions of both CYC1 and DED1 genes are susceptible to SSN cleavage when in negatively supercoiled plasmids. This implies that DNA in the sensitive regions can assume, at least temporarily, either a completely strand separated state, or one of several alternative topologically unwound, paranemic states (Yagil, 1991)2. The results suggest that purine.pyrimidine rich tracts can serve as unwinding centers in yeast genes.

P1 digestion: Nuclease P1 (from P. centrinum ) was obtained from BRL Life Technologies Inc., USA. One enzyme unit converts one A260 unit of yeast RNA to acid soluble material in 15 min at 37o C. The standard digestion conditions were: 1 mg of supercoiled DNA was dissolved in 50 ml TE (10 mM tris, 1 mM EDTA pH 8.0), followed by nuclease P1 units as specified for each experiment. The mixture was incubated at 37o C for 30 or 60 min. The material was extracted once with phenol:chloroform (1:1), twice with chloroform: isoamyl alcohol (25:1), precipitated with two volumes of ethanol and washed with 70% ethanol. The precipitate was isolated by centrifugation, air-dried and redissolved in either water or in TE solution.

Primer extension. The following primers were used: Primers T7 and KS (Stratagene) were employed for the analysis of the upper strand of SK280 (see Fig. 1). To map the opposite, lower strand, primers SK and T3 were used. For KS200, see legend to Fig. 4. Primers were end labeled with [g-32P]ATP and T4 polynucleotide kinase. 1.8 ng labeled primer were mixed with 500 ng of P1 cleaved DNA sample, dNTP to 0.2 mM, "filling-in" Buffer (U.S.B.) and water to 50 ml. The mixture was warmed for 1 min at 95o, then lowered to 37o for 10', followed by 2.5u of Klenow fragment of DNA polymerase and incubated for 15' at 37o. Subsequently ammonium acetate was added to 2M, DNA was precipitated with ethanol, washed and dissolved in formamide loading buffer. To identify the cleavage sites, the four dideoxy sequencing reactions were run by extending with the same primer, using an U.S.B. Sequenase sequencing kit, and loaded on the gel.

Electrophoresis. Polyacrylamide-urea "buffer gradient" electrophoresis (x0.5-x2.5 TBE) was carried out as per Biggin et al. (1982). Dried samples were dissolved in formamide, heated to 75o , cooled on ice and loaded on a 0.4 mm thick 6 % acrylamide - 5.83 M urea gel. The gel was run in a 40 cm vertical apparatus at 1200 V, dried and exposed to X-ray films for the appropriate time.

Topoisomer analysis . Two-dimensional topoisomer analysis was performed essentially as described by Wang et al. (1983). A mixture of topoisomers was prepared by incubating the negatively supercoiled plasmid with topoisomerase I in 0-12 mg/ml of ethidium bromide. Following analysis on 1.4% agarose, the various fractions were combined in appropriate proportions, precipitated with alcohol and dissolved in the desired medium (generally TE). 1:6 30% glycerol was added to the sample, and loaded on one of two holes of an 20x20 cm 1.7% agarose gel cast in TBE. The gel was run for 16 hr at 40 V in one direction, in the same buffer, then shaken with 7.5 mg/ml chloroquine in 50 mM tris-phosphate, 1mM EDTA, pH 7.2 for 6 hrs, and run in perpendicular direction for another 16 hr in the same buffer. The gel was run at room temperature (22o), with recirculating buffer. Finally, the gel was stained in ethidium bromide and photographed with a CCD or Polaroid camera.

High resolution mapping of the P1 cleavage sites. The detailed pattern of the cleavage sites was determined by primer extension from oligonucleotide primers situated on two opposite vector sites (Fig. 3
It should be kept in mind that primer extension maps cleavage sites on the strand opposite to the extended strand, while marker lanes indicate sites on the extended strand itself. The P1 cleavage pattern obtained upon application of a wide range of P1 concentrations is shown in Fig. 3A for the upper strand, and in Fig. 3B for the lower strand. A certain number of cleavage bands are evident, the positions of the strongest of which are indicated to the right of the figures (the numbering system can be seen in Fig. 8). Of the strong bands apparent in Fig. 3A, the two upper bands map to positions 96 and 109 and are thus located within the 39Y/40N pyrimidine tract (Y40) on the DED1 promoter. The lower cleavage sites map either within the 22 nt (#153,159) or the 10 nt (#177) pyrimidine tracts of the CYC1 promoter (Y22, Y10 on Fig. 3A). Similarly, the stronger cleavages of the lower strand fall within the R40 (#97,107) and the R22 (#151,155,156) tracts. Other cleavages (#141,139) are located in the adjacent TATA box to be discussed later. The bands observed in this and subsequent experiments are summarized in Fig. 8.
To be cleaved by a single strand sensitive nuclease, a stretch of circularly closed DNA ought to be in a ready-to-separate state. The sensitivity to SSN observed is thus in line with the working hypothesis that long homopurine.homopyrimidine (R.Y) tracts tend to assume such an unwound state. Furthermore, it is clear from Fig. 3 that not all sites within the R.Y tracts are sensitive to cleavage. The R.Y region thus cannot be completely strand separated prior to cleavage.

In order to determine the potential unwinding of the DED1 region separately from the CYC1 region, the DED1 region was independently cloned to yield plasmid KS200 (Fig. 1). The nuclease P1 cleavage pattern of this plasmid is shown in Fig. 4.
A number of sites are cleaved by the single stranded nuclease, the strongest of which are located within R40, the long purine tract on the lower strand. Weaker cleavages occur near the Xho I site adjacent to the purine tract and at the 3' end of the insert. Most of the promoter region to the left of the purine tract remains uncleaved, which again means that the promoter region cannot be entirely strand separated. P1 Cleavage of the R.Y tract in the DED1 promoter occurs thus independently of the CYC1 region.

Effect of additives. A transition from a double-stranded to a single-stranded state can be expected to be suppressed by the addition of salts. The effect of various additives to the cleavage mixture is shown in Fig 6.
The four additives examined, NaCl (lane 8,9), MgCl2 (lanes 5-7), spermine (lane 3) and spermidine (lane 4) suppressed cleavage (MgCl2 at 0.5 mM, lane 5, had no discernible effect, but this low concentration of Mg2+ is neutralized by the EDTA present). NaCl is well known to suppress strand separation and thus raising DNA melting temperature. Its effect can, therefore, be explained by suppressing transient strand separation of either B form DNA or of a paranemic DNA intermediate. MgCl2 may act similarly. Spermine and spermidine favor cruciform extrusion (Sullivan and Lilley, 1987), which will trap negative superhelicity otherwise available for unwinding. One cleavage band, at site 151, appeared only in spermidine treated samples (Fig. 6, lane 4), indicating an interaction leading to enhanced susceptibility at that site. Cleavage in the TATA box is to be noted again. The pH dependence of the cleavage was also examined (not shown); cleavage was independent of pH between pH 6 - 8. Only faint cleavage was observed at pH 9 and pH 5 (the optimal pH for nuclease S1 and for P1 with Zn2+ (Blaho et al., 1988)). Altogether, the effect of salts is consistent with the opening of an unprotonated paranemic intermediate into a strand separated form as the rate determining step of P1 cleavage.

The principal observation reported in this study is that the homopurine.homopyrimidine (R.Y) tracts in the two promoter regions examined are susceptible to single strand specific nucleases (SSN) at a number of sites along their length, as summarized in Fig. 8.
The yeast promoters studied here behave thus similarly to those in higher Eukaryotes, where SSN cleavage of R.Y tracts is commonly observed (see Introduction). To be susceptible, the attacked regions should be in a single stranded state, or assume, at least temporarily, a topologically unwound, ready to separate state. We shall refer to such a paired yet unwound intermediate form of DNA as "the paranemic intermediate". Before discussing the evidence for such an intermediate, the behavior of the vector must be considered. The 2d analysis (Fig. 7) shows that at more than 13 negative superturns (s = ~136/3220 = - 0.042), about six superhelical turns are opened, i.e., the bulk of plasmid DNA is converted into a form in which 60-66 bases assume a formally unwound (paranemic) state. This transition occurs both in the promoter containing plasmid and in the vector. The six turns opened are, therefore, likely to reside primarily in the ori region common to all col E1 derived plasmids (Eguchi and Tomizawa, 1990), either as cruciforms (Lilley, 1980) or in the known mung-bean nuclease sensitive region (Sheflin and Kowalski, 1985). This does not exclude the intermittent migration of the opened supercoils to the CYC1 and DED1 promoter regions, enabling the cleavage within the R.Y tracts. The migration of the unwound region between vector and promoter regions can take place either by bubble migration or by an unwinding-rewinding mechanism.

The transition in the promoters into an unwound state does not necessarily imply immediate separation into single strands. Indeed, many sites within the R.Y tracts of the DED1 and CYC1 regions remain uncleaved, implying that the DNA is not all strand separated and that an alternative, formally unwound form of DNA, is intermittently formed, and is attacked by the single strand specific nuclease when occasionally opened. Other pieces of evidence favor the presence of such a paranemic intermediate in the negatively supercoiled plasmid: First, negative supercoiling greatly facilitates the cleavage of the two promoter regions (Fig. 5) as expected for an unwinding process. Second, the practical independence of P1 cleavage on nuclease concentration (Fig. 3) implies that a slow, internal transition of the target DNA precedes SSN cleavage. The rate limiting step may be either the slow opening of the paranemic form, or the slow migration of the zone unwound in the bulk to the R.Y regions of the promoters. The third piece of evidence, the suppressive effect of NaCl and MgCl2 (Fig. 6), which is known to oppose DNA melting, is in favor of strand opening as the rate limiting step. The data thus support thus the idea that paranemic intermediates, topologically but not structurally equivalent to single strands, are formed in the two homopyrimidine regions. While the existence of such intermediates will have to be verified by direct observation, consideration of its possible structure could be of help in this task.

One appealing candidate for the unwound intermediate is a paranemic duplex, i.e. a duplex in which the bases are conventionally stacked and paired, but with an average base twist angle of zero instead of ~36 degrees. A sterically acceptable structure for such a duplex was arrived at by a molecular mechanics analysis (Yagil. and Sussman, 1986). That structure was locally stable because of a sterically hindered, energy requiring rotation of the backbone around the O5'-C5' bond (anti to gauche). this structure has no special symmetry or compositional requirements at variance with the experiments reported here. More recently, the crystal structure of a DNA - TBP (TATA binding protein) complexes was determined (Kim Y. et al., 1993; Kim, J.L. et al., 1993). The TATA DNA analyzed by Sigler and coworkers (Kim Y. et al., 1993) was the same TATA box of CYC1, which is found to be P1 sensitive in the present study, located adjacent to the sensitive R.Y tract (#138 -145, boxed in Fig 8) . Ten consecutive bases of the DNA component were found to be half unwound by the X-ray analysis (~20 base pairs per double helical turn). Moreover, a detailed analysis of helical parameters of the complex (Guzikevitch and Shakked, 1996) shows that the partly unwound structure shares important features with the paranemic duplex we proposed in 1986, in particular the high incline (tilt) of the bases relative to the backbone axis (40-50o). On the basis of all these observations, the formation of a paranemic duplex as an intermediate in the P1 cleavage process remains an appealing possibility.

Do the unwound regions have a role in regulating the expression of the genes concerned? Earlier deletion studies do support such a role: A deletion (D99) which includes all of the Y tract has been shown to reduce transcription initiation of the CYC1 gene over 20-fold compared with a deletion containing only the first 16 bases (Faye et al., 1981). Also, a deletion extending from the third base of Y22 down to 9 bases 3' of the Y10 region, reduces transcription from 50% to only 4% of wild type (Guarente and Mason 1983). Both deletions include, however, the TAAATAT element that serves as the TATA box for initiation at sites -47 - -22 of the CYC1 gene. Still, a deletion by McNeil & Smith (1986) in which only 5 Y from the middle of Y22 were deleted and the TATA box was left intact, eliminated all transcription initiating below # -47. This implies that the Y tract is no less essential for in vivo initiation than the TAAATAT element. As for DED1, deletion of the R.Y region causes a five-fold decrease in gene expression (Struhl 1985b). Also, deleting 18 bases from the R.Y region eliminates the enhancing effect of DED1 on CYC1 promotion both in vivo and in an in vitro transcription system (Lue et al., 1989). The deletion studies thus support the notion that R.Y tracts play a role in transcription control in yeast.

For even longer transcription trains, topoisomerases may join in to permit the unwinding of additional DNA stretches; topoisomerase I has been repeatedly implicated in transcription control (Gilmour et al., 1986; Cook et al., 1992; Merino et al., 1994). in the native DED1 promoter, the Y40 tract is followed by a 36R/40N tract immediately after the Xho 1 site (Struhl, 1985a); in CYC1, the two adjacent TATA boxes may join the Y22 and Y10 tracts to permit entry of additional polymerases. The controlled transition of a DUE between the wound and unwound forms of DNA may thus serve as a switch determining whether a particular gene turned on or off, and regulate in this way the frequency of gene transcription.

Fig. 1. Plasmids SK280 and KS200. Hatched region: iso-1-cytochrome c (CYC1), bases -139 - +5. Stippled region: DED1, Bases 765-880. Dark stippled areas: homopyrimidine (Yxx) tracts in the upper strand (i.e. homopurine on the lower strand). PL: polylinker region (see Lue et al., 1989). SK, KS, T3, T7: Sites of Primers used for primer extension.

Fig. 2. Nuclease P1 cleavage of plasmids SKII+, SK1000 and SK280, electrophoretic analysis on agarose gel. Nuclease P1 was applied to the negatively supercoiled plasmids in 50 ml TE (10 mM Tris, 1 mM EDTA, pH 8.0) for 60 min at 37°C. P1 concentrations were as indicated on the plot, in units/50 ml sample. The DNA was purified, opened with BamH I and end labeled with [g-32P]ATP . The purified DNA was loaded and run on an 2% agarose gel. Lane M is an Hinf I digest of pBR322. The lengths of some of these bands are marked on the right. Lengths of observed SK280 bands are marked on the left.

Fig. 3. High resolution mapping of nuclease P1 cleavage sites of plasmid SK280. The sites of the principal cleavage sites are indicated at the right of each gel; the position of the R.Y tracts are indicated on the left. Increasing amounts of nuclease P1, as specified at the top of the gel, were applied to negatively supercoiled SK280 DNA (s = -0.052) in 50 ml TE for 60 min at 37°C. The samples were processed and analyzed on acrylamide by the primer extension procedure described in Materials and Methods. (A) Upper (coding) strand, extension of SK primer (sic, see Fig. 1). (B) Lower strand, extension of KS primer.

Fig. 4. Nuclease P1 cleavage pattern of KS200 (DED1) at different P1 concentrations. The lower strand was analyzed (SK primer). Different amounts of nuclease P1, as specified on the top of the gel, were applied to negatively supercoiled KS200 DNA in 50ml TE for 60 min at 37°C and processed as described.

Fig. 5. Nuclease P1 digestion of supercoiled vs. linearized SK280. sc - a sample of negatively supercoiled DNA (s = -0.052) incubated with the indicated amount of nuclease P1 (units/50ml) at standard conditions. lin - a sample of SK280 DNA was linearized with Kpn I prior to application of P1 and incubation as above. After cleavage by P1, DNA from both samples was purified, recleaved with BamH I, end-labeled with [g-32P]ATP and analyzed on acrylamide (lower strand). Lanes ACGT, AC: KS500 plasmids (Tal et al., 1994) were end labeled at their BamH I site and sequenced with a T7 primer, to serve as band length markers.

Fig. 6. Nuclease P1 cleavage pattern of SK280 - effects of spermine (spm, lane 3); spermidine (spd, lane 4); MgCl2 (lanes 5-7) and NaCl (Lanes 8-9). Concentrations are indicated on top. 0.5 units of nuclease P1 were applied to 1 mg supercoiled DNA in 50 ml TE, pH 8, for 60 min at 37°C. The reaction was stopped and the samples were processed as described. The upper strand was analyzed (SK primer).