DNA binding, condensing and unwinding properties of yeast RNase H1.

Two distinct RNases H have recently been purified from yeast cells. One of these proteins, very abundant in the cell, called RNase H,, is shown here to be an interesting nucleic acid binding protein. The affinity of RNase H, for various nucleic acids was compared by membrane filtration and sedimentation techniques. Competition experiments indicated the following order of affinity: single-stranded RNA > single-stranded DNA > double-stranded DNA > DNA. RNA hybrid. Binding of RNase H, to T, DNA or SV40 DNA resulted both in the condensation of the DNA, observed by electron microscopy, and in the destabilization of the double helix. However, there was no permanent modification of the DNA structure. Concomitant with these structural changes, a drastic stimulation of transcription of double-stranded DNA templates was obtained with yeast RNA polymerases A or B. As evidenced by electron microscopy and sedimentation studies of the transcription complex, there was a competition between the DNA template and the RNA product for binding RNase H, during the course of the transcription. These results suggest the involvement of RNase H, in chromatin structure and function.

Two distinct RNases H have recently been purified from yeast cells. One of these proteins, very abundant in the cell, called RNase H,, is shown here to be an interesting nucleic acid binding protein. and in the destabilization of the double helix. However, there was no permanent modification of the DNA structure. Concomitant with these structural changes, a drastic stimulation of transcription of double-stranded DNA templates was obtained with yeast RNA polymerases A or B. As evidenced by electron microscopy and sedimentation studies of the transcription complex, there was a competition between the DNA template and the RNA product for binding RNase H, during the course of the transcription.
These results suggest the involvement of RNase H, in chromatin structure and function.

DNA Binding
Properties of Yeast RNase HI pH 8, in the presence of Mg'+ and Na+ ions, at a protein to nucleic acid ratio around 0.5 (2). When incubated under these conditions with a labeled polynucleotide, RNase H, was able to form a complex which could be recovered on nitrocellulose membrane (Fig. 1). Binding of RNase H, to T7 [3H]DNA occurred under a wide variety of conditions between pH 6 and pH 8, with or without divalent cations or salt, but this depended on the ratio of RNase H, to DNA. At protein to DNA ratios higher than 1 (w/w) complex formation was maximal and unaffected by the addition of divalent cations (10 mM Mg"+ ions) and salt up to 0.3 M NaCl ( Figs. 1 and 2). At lower protein to DNA ratios, however, the presence of divalent cations, as well as the addition of salt, was inhibitory. Practically no retention of DNA was observed at pH 8 with 5 mM Mgz+ or 0.1 M NaCl at ratios lower than 0.5 (by weight). Retention of DNA at high salt was more efficient at pH 6 than pH 8l ( Figs. 1 and 2).
Formation of the complex does not appear to be temperature-dependent between 0 and 50" (results not shown). Assuming a molecular weight for native RNase H, of 48,000 (21, it can be calculated from the above results that 1 molecule of RNase H, can cause the retention of 250 to 300 base pairs under optimal conditions. Binding of RNase HI to Various Polynucleotides and Competition with DNA -RNase H, also formed a complex with RNA (Escherichia coli RNA) or with RNA.DNA hybrids which were retained on membrane filters. In this case, a lofold excess of protein was required when compared with T, DNA. However, these figures were difficult to compare since the size of the labeled polynucleotides was much different. Furthermore, the direct filtration technique could not be applied to single-stranded DNA. Therefore, competition experiments were performed to compare the affinity of RNase H, for various nucleic acids. RNase H, was incubated under optimal conditions with 3H-labeled T, DNA in the presence of increasing amounts of unlabeled competitor. As shown in Fig. 3, RNase H, was preferentially bound to single-stranded nucleic acids, RNA or DNA. Especially single-stranded RNA displaced very efficiently RNase H, from double-stranded DNA. Surprisingly, the RNA.DNA hybrid was the least efficient competitor (Fig. 3). The order of affinity inferred from the competition experiments was: single-stranded RNA > single-stranded DNA > DNA duplex > DNA. RNA hybrid. To explore further the complex formation, a mixture of RNase H, and T, [3H]DNA was sedimented through a glycerol gradient and the fractions were analyzed for the presence of DNA and protein. At a protein to DNA weight ratio of 2, sedimentation of T, DNA was drastically modified (Fig. 4). One-third of the DNA was found as a pellet, in spite of the presence of a 80% glycerol cushion, one-half migrated approximately twice as fast as T7 DNA alone, and only 10% migrated as free DNA. RNase H, co-sedimented with the DNA in the pellet and at the level of the rapid sedimenting fraction. Practically no protein was detected on top of the gradient, at the level where the protein alone stayed under the same conditions. At an RNase H, to DNA ratio of 0.5, sedimentation of the DNA was not significantly affected, but again the protein co-migrated with the DNA in the gradient.
These experiments clearly show that stable complexes were formed with probably more or less compact structures, depending on the amount of ligand. In a parallel experiment, RNase H, was preincubated with T, [3H1DNA for 2 min at 30", then competitor RNA was added, and the mixture was centrifuged I S. Dezelee and F. Wyers, unpublished results. The mixtures were incubated for 2 min at 30" then diluted with 0.5 ml of the corresponding buffer, without divalent cations, then immediately filtered on nitrocellulose membranes. The filters were washed twice with 1 ml of warm buffer and dried and their radioactivity was determined.
Results are given as the percentage of input DNA. SV40 DNA were prepared at varying protein to DNA ratios, and at two pH values, 6 or 8. Binding of RNase H1 to DNA was clearly observed (Fig. 5); interestingly, there was nonrandom distribution of the protein along the DNA molecules. Regions of naked DNA alternated with parts where proteins accumulated and created multiloops, hairpins, clover-leaf, and more complicated structures. Bubbles were also seen in these dense entanglements, which might correspond to denatured DNA with bound RNase H,.
The apparent complexity of RNase HI-induced structures increased with the weight ratio of protein to DNA. The central core seen in some complexes probably arose with protein-protein interactions. Although this suggested some cooperative binding, all DNA molecules had some protein bound even at a low ratio of protein to DNA. Fifty per cent of the complexes appeared with a central core at a ratio of 1 and at pH 6. At higher protein to DNA ratio, every DNA molecule was highly collapsed and the regular dense nucleoprotein core was surrounded by very small DNA loops. The association of several such condensed complexes was also observed. These collapsed structures probably corresponded to the RNase H,.DNA complexes found in the pellet after centrifugation (see Fig. 4). Within these nucleoprotein cores, the length of the DNA was reduced 50-fold as compared with fully extended T7 DNA. Observations made at pH 8 were very similar except that the corresponding figures were observed at a higher protein to DNA ratio. With SV40 DNA (3.5*106 daltons), complexes had a similar shape but were considerably smaller than those formed with T, DNA, suggesting that most of the /  I  I  I  I,  I I Table 1). The mixture contained 20 rnM Tris/HCl (pH 8), 2 rnM MnC12, 25 mM ammonium sulfate, and 5 mM dithiothreitol.
After 5 min incubation at 30", the samples were diluted to a concentration of DNA around 0.5 to 0.8 pg/ml with the incubation buffer, at 30". Specimens  was hydrolyzed by the nuclease (Fig. 6). The relative amount of DNA rendered acid-soluble was roughly proportional to the amount of protein added. At a weight ratio of 2, up to 30% of the radioactive DNA was solubilized in 10 min. There was no additional degradation during a prolonged incubation of 20 min, with a 6-fold excess of RNase HI over DNA (w/w). The experiment shown in Fig. 6 was performed at neutral pH but also at pH 5 and 4.3, with essentially the same results. Such an extensive degradation of the DNA by nuclease S, was surprising, however, in view of the comparatively small amount of possibly denatured DNA bubbles seen under the electron microscope.
The effect of RNase HI, on supercoiled SV40 DNA was further investigated by the electrophoretic technique which allows the resolution of the different supercoiled states as well as the relaxed and linear forms of the DNA (Fig. 7). Using this highly sensitive technique, one can estimate the degree of contamination of RNase HI with endonucleases but also investigate whether RNase HI caused some structural constraint on the DNA molecule which might interfere with the activity of a relaxing enzyme (6) or change the final DNA products.
Preincubation of RNase HI with supercoiled SV40 DNA at a weight ratio of 1 did not change the migration pattern of the supercoiled bands. With increasing amounts of RNase HI, the band of relaxed form seemed to increase a little, suggesting the presence of trace amounts of nuclease. However, another possibility was that the RNase HI preparation con-  8. Analysis by zone sedimentation of the transcription complexes. Assay mixtures (0.1 ml) contained T, 13HlDNA as template (10 pg; 7,000 cpm/pg), RNase H, (25 pg), yeast RNA polymerase A (6 pg) or RNA polymerase B (4.5 pg), and buffer and salts as described for A or B enzymes in Table I. RNA polymerase was added after a 2-min preincubation at 30" of RNase H, with DNA, and RNA synthesis was started by the addition of the four nucleoside triphosphates with [(u-~~PIGTP as marker (80,000 cpm/nmol). The reaction was carried out in separate tubes for 2, 30, or 90 min, then the reaction mixtures were diluted a-fold with transcription buffer and layered on 4.2 ml of a 10 to 30% glycerol gradient containing 0.025 M Tris/HCl, pH 8, with a cushion of 80% glycerol (0.75 ml) at the bottom of the centrifuge tube. Centrifugation was carried out for 75 min at 64,000 rpm in the SW 65 rotor of a Beckman ultracentrifuge. After the run, 250-~1 fractions were collected from the bottom of the tube and pellet (P) was recovered. A sample of 60 ~1 of each fraction was used for radioactivity measurements of DNA and RNA after trichloroacetic acid precipitation. 13HlDNA (cpm, O-O); [3ZP]RNA, (cpm, l . . . . *O). The arrow indicates the sedimentation of control T, DNA. Radioactivity in the pellet fraction (P) is indicated. Left panels correspond to RNA polymerase A; right panels correspond to RNA polymerase B. tained a trace of relaxing activity. This was more likely because such a relaxing activity co-purified with RNase H1 up to the last phosphocellulose chromatography step.2 The relaxing enzyme from Krebs ascites cells, on the other hand, shifted all the supercoiled DNA molecules into the fully relaxed form. The addition of RNase H, to the reaction mixture caused no important change in the final reaction product. Apparently, within the limit of the technique, enzymatic relaxation of SV40 DNA* RNase H1 complexes occurred normally. Therefore, the formation upon deprotemization of fully relaxed DNA, or even DNA with a limited number of super-helical turns, would indicate that RNase Hr did not superimpose an overall and important torsional deformation of the DNA molecule. This was in contrast with what could be expected from the nuclease S, experiment.
Effect DNA duplexes. Unpaired DNA regions are required in vitro by the yeast enzymes to perform the initiation step (14). Therefore, we investigated whether a structural alteration of the DNA brought about by RNase H, could stimulate transcription by yeast RNA polymerases.
The effect of RNase H1 on transcription of a variety of natural and synthetic templates was investigated using a weight ratio of 2 RNase H,/DNA, since it was under these conditions that DNA was most susceptible to nuclease S, (see Fig. 6). Table I shows that transcription of all double-stranded natural DNA, either phage DNA, calf thymus, or yeast DNA, was indeed markedly stimulated.
The most active template, in the presence of RNase H,, was T, DNA. The template activity was increased to the extent that yeast RNA polymerases reached a specific activity approaching that of the bacterial enzyme with the corresponding templates. On the other hand, no effect of RNase H1 was observed on transcription of the double-stranded homopolymer (dG); (dC), and (dT), . (dA), or with the alternated copolymer d(A-T), or d(G-C),. The template activity of denatured T, DNA and (rC), was also practically unaffected. It should be noted here that no change in the sedimentation of T, [3H]DNA in alkaline sucrose gradient was observed after a prolonged incubation of native DNA with RNase H1 in transcription buffer, and no acidsoluble radioactivity could be detected (less than 0.01%). These controls were performed to exclude the artifact of template activation by nucleases.
Fate We have also accumulated evidence3 in favor of the identity between RNase H, and one of the polypeptides present in RNA polymerase A (M, = 48,000) (4). These observations and further work characterizing the stimulatory activity will be presented elsewhere.
The question which remains, at this point of the in vitro studies, concerns the in. vivo significance of all these observations. It should be noted that all the properties of RNase H,, stimulation of transcription, DNA binding, and hybridase activity, can occur under similar ionic environment. Furthermore, from the high amount of RNase H, which can be extracted from yeast cells (l), a 2:l ratio, by weight, of RNase H, to DNA does not seem unreasonable.
Accurate determination of the number of RNase H, molecules per yeast genome are required to clearly establish this point. At any rate, it is clear that the biological role of the RNases H found in yeast (1, 2), and specially of RNase H,, will remain uncertain and speculative until defective proteins are isolated and the phenotype of the mutants are investigated in detail.