Proc. Nat. Acad. Sci. USAVol. 69, No. 6, pp. 1560-1564, June 1972

RNA-Primed DNA Synthesis In Vitro(DNA polymerase/DNA initiation/RNA polymerase/ribonuclease H)

WALTER KELLER

Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724

Communicated by J. D. Watson, April 10, 1972

ABSTRACT In vitro DNA synthesis on single-strandedcircular DNA can be initiated by RNA primers. RNA chainsare covalently extended by DNA polymerase II from KBcells and DNA polymerase I from Micrococcus luteus, butnot by an RNA-dependent DNA polymerase from avianmyeloblastosis virus. The reaction product consists ofDNA chains with a piece ofRNA at their 5'-ends, hydrogenbonded to the template DNA. The primer RNA is linkedto the product DNA via a 3':5'-phosphodiester bond, andcan be specifically removed by ribonuclease H. The pos-sible role of ribonuclease H in RNA-primed DNA synthesisin vivo is discussed.

Genetic or biochemical evidence for direct participation ofRNA synthesis in DNA replication has recently been re-ported for Escherichia coli (1, 2), phage lambda (3), and phageM13 (4). In each instance, the basic observation was thatlack of or inhibition of transcription of certain regions of thegenome prevented DNA replication. In what way transcrip-tion is involved in DNA synthesis is not known. In the caseof phage M13 DNA synthesis, the suggestion has been madethat a small piece of RNA, synthesized by DNA-dependentRNA polymerase on the parental plus-strand DNA, promotesthe initiation of minus-strand synthesis by acting as primerforDNA polymerase (4).The occurrence of such RNA-primed DNA synthesis within

higher cells may well be the reason for the widespread ex-istence of ribonuclease H (RNase H) (5). This enzyme, nowunder active study in this laboratory, specifically degradesRNA hybridized to DNA, and thus is capable of removing5'-terminal RNA primers from DNA chains if a comple-mentary DNA strand is present. Thus, we looked at whetherthe mammalian DNA polymerase II from KB cells wouldwork better if concomitant RNA synthesis was allowed. Weobtained positive results; other workers (6, 7) independentlyreported similar observations with other systems.

In our in vitro experiments, mammalian DNA polymeraseII and Micrococcus luteus DNA polymerase synthesize DNAcomplementary to single-stranded circular OX DNA bycovalent extension, either of preformed RNA chains annealedto the template DNA (uncoupled system), or of RNA chainsgenerated by addition of RNA polymerase to the reactionmixtures (coupled system). The RNA part of the resultingRNA-DNA chains can be specifically removed by RNase H.

MATERIALS AND METHODS

Reagents. Unlabeled ribonucleoside triphosphates anddeoxyribonucleoside triphosphates came from Calbiochemand Schwarz/Mann BioResearch, respectively, [a-'2P ]CTP(6.2 Ci/mmol) and ['H]dTTP (18.1 Ci/mmol) from New

England Nuclear Corp., [Ca-32P]dATP (18 Ci/mmol), [Cia_2P]_dCTP (22 Ci/mmol), and [a-_2P]dTTP (23 Ci/mmol) fromInternational Chemical and Nuclear Corp. The latter com-pounds were free of the corresponding ribonucleoside tri-phosphates, as judged by chromatography in isopropanol-ammonium hydroxide-boric acid 6:1:3. [a-32P]dGTP wasnot used, since it contained about 6% rGTP as a contaminant.Rifampicin was a gift of Lepetit S. A. (Milano).

Templates. DNA from phage OX174 was purified as de-scribed (8). Partial DNA-RNA hybrid was prepared byincubation of 50 jug of OX DNA for 5 min at 370 with 10 ug ofE. coli RNA polymerase (see below) in a 0.5-ml reaction mix-ture, as specified by Chamberlin & Berg (9) except that allfour ribonucleoside triphosphates were present in unlabeledform. The reaction mixture was subsequently extracted withphenol (saturated with 1 M Tris. HCl, pH 7.5) at room tem-perature. The material in the water phase was concentratedby alcohol precipitation, and further purified on a 1 X 25cm column of Sephadex G-75 equilibrated with 0.1 M NaCl-0.01 M Tris-HCl (pH 7.5)-i mM Na EDTA. The buoyantdensity of an aliquot of the synthetic product in cesium sul-fate was near 1.47 g/cm3, indicating that the RNA in the hy-brid did not amount to more than a few percent of the massof the DNA (9).

Poly(dA) * oligo(dT) and poly(rA) * oligo(dT) were pre-pared by incubation of poly(dA) or poly(rA) (both from MilesLaboratories, Elkhart, Ind.) at a concentration of 0.2 mg/mlwith 0.05 mg/ml of oligo(dT) (chain length: 12-18 nu-cleotides, from Collaborative Research, Inc., Waltham, Mass.)in 0.1 M NaCl-1 mM Na EDTA (pH 7.6), for 30 min at 250.

Enzymes. DNA polymerase II from KB cells, grown insuspension cultures, was isolated by a method to be publishedelsewhere. The purification procedure included chromatog-raphy of the solubilized enzyme on DEAE-cellulose, phos-phocellulose, Agarose, and DNA-cellulose. 1 mg of purifiedenzyme converted 150 nmol of dTTP into acid-insolublematerial in 30 min at 370 when assayed on poly(dA) * oligo-(dT) as template. Its properties were similar to the 6-8 SDNA polymerase from calf thymus (10) and DNA poly-merase II from HeLa cells (11).

DNA Polymerase from Micrococcus luteus (12), specificactivity 80 units/mg of protein, was purchased from Miles.

DNA Polymerase (RNA-dependent) from avian myelo-blastosis virus (kindly provided by Drs. Dorothy and J.Beard of Duke University, Durham. N.C.) was partiallypurified by chromatography on DEAE-cellulose and hada specific activity of 80 nmol of dTMP incorporation/mg ofprotein in 30 min with poly(rA) * oligo(dT) as template.

1560

Abbreviations: OX DNA,DNA of phage OX174; Me2SO, dimethylsulfoxide.

RNA-PrimedDNA Synthesis 1561

DNA-Dependent RNA Polymerase from E. coli was puri-

fied by the glycerol-gradient procedure (13), with the fol-lowing modification: the DNase treatment of the crude ex-

tract was omitted and the supernatant after high-speed cen-

trifugation was instead passed through a column of DEAE-cellulose in the presence of 0.3 M KCl to remove nucleicacids (Mangel, W. F. & Chamberlin, M. J., personal com-

munication). The specific activity was 100 units/mg withnative calf-thymus DNA (Worthington) as template.

Single-Strand Specific Nuclease from Neurospora was puri-fied as described (14), and was a gift of Dr. J. F. Sambrook. 1ug of enzyme protein hydrolyzed about 1 ug of heat-denaturedcalf-thymus DNA when assayed as described in Table 3.

RNase H from KB cells was prepared by a proceduresimilar to the one outlined above for KB-DNA polymerase(W.K., to be published). When assayed as described in Table 3with 4X DNA- [3H]RNA hybrid, 1 mg of protein converted40 nmol of RNA into acid-soluble material.

RESULTS

Stimulation of DNA Synthesis by Concomitant RNA Syn-thesis. Mammalian DNA polymerase II is inactive with single-stranded circular DNA as template. Like all known DNApolymerases, it requires a priming polynucleotide with a free3'-hydroxyl group for initiation (15). However, when DNA-dependent RNA polymerase (E. coli) and ribonucleosidetriphosphates were added to reaction mixtures containing,X DNA, KB DNA polymerase II, and deoxyribonucleosidetriphosphates, DNA synthesis occurred (Table 1). That thisDNA synthesis required concomitant RNA synthesis isshown by the results of double-label experiments (Table 1),

TABLE 1. Requirements for coupled RNA-DNA synthesis

[a-32P]CMP [3H]dTMPincorporation incorporation

(pmol) (pmol)

Complete system 41 128Minus DNA polymerase 43 4Minus RNA polymerase <0. 1 0.7Minus dATP, dCTP, dGTP 41 <0. 2Minus ATP, GTP, UTP 0.1 4.6Plus rifampicin (5 Mg) 0.2 4.2Minus DNA 0.2 <0.1Minus KB-DNA polymerase

II, plus viral DNA poly-merase (20ug) 40 2.4

Minus KB-DNA poly-merase II, plus M. luteus DNApolymerase (1 Mg) 32 60

The complete system contained in 0.1 ml: 0.05 M Tris HCl(pH 7.9); 0.02 M KCl, 0.01 M MgCl2, 1 mM dithiothreitol; 0.1mM (each) dATP, dCTP, and dGTP; 0.05 mM ['H]dTTP(800 cpm/pmol); 0.25 mM (each) ATP, GTP, and UTP; 0.01mM [a-32P]CTP (2000 cpm/pmol); 1 ,g of OX174 DNA; 3 Mg

of DNA-dependent RNA polymerase (E. coli); 2,g of KB DNApolymerase; 10% glycerol.

After incubation of the reaction mixtures for 30 min at 370,cold 5% Cl3CCOOH was added. Acid-precipitable material was

collected on Whatman GF/C glass-fiber filters. The filters were

washed with 2%0 ClCCOOH-0.01 M Na4pyrophosphate, rinsedwith 10 ml of 95% ethanol, air-dried, and counted in a toluene-based scintillation fluid.

TABLE 2. Response of cellular and viral DNA polymerasesto various template primers

Template

Partial XDNA*

DNA RNA poly(dA) poly(rA)polymerase XX DNA hybrid oligo(dT) oligo(dT)

KB <1 13 75 <1Viral <1 <1 22 670

The numbers represent pmol of dTTP converted into acid-precipitable material after 30 min of incubation at 37°. Reactionmixtures contained in 0.1 ml: 0.05 M Tris HCl (pH 7.9), 0.05 MKCl, 0.01 M MgCl2, 1 mM dithiothreitol, 10% glycerol, 0.1 mM(each) dATP, dCTP, and dGTP; 0.05 mM [3H]dTTP (200 cpm/pmol); 1 ug of KB DNA polymerase II or 10 jg of viral DNApolymerase, and template.Templates were prepared as described in Methods and used

at the following concentrations: OX DNA, 10 gg/ml; partialOX DNA -RNA hybrid, 15 Mg/ml; poly(dA) *oligo(dT), andpoly(rA) oligo(dT), 20 Mg/ml.

in which the inhibition of RNA synthesis by addition ofrifampicin or omission of ribonucleoside triphosphates re-duced the stimulatory effect of RNA polymerase on DNAsynthesis more than 30-fold. All four deoxyribonucleosidetriphosphates were required for DNA synthesis, and neitherRNA nor DNA was made in the absence of DNA templates.Under the conditions specified in Table 1, RNA synthesis waslinear for 30 min and then leveled off. The synthesis of DNAstarted after a lag period of 2 min, proceeded at a constantrate for 30 min, and then slowed down gradually. WithoutRNA polymerase, no RNA and very little DNA was made.

Uncoupling DNA Synthesis from RNA Synthesis. OX DNAwas incubated for 5 min with RNA polymerase and unlabeledribonucleoside triphosphates to form DNA partially coveredby complementary RNA (see Methods). This partial DNARNA hybrid was then purified by phenol extraction and gelfiltration. When the hybrid was incubated with KB DNApolymerase II and deoxyribonucleoside triphosphates, DNAsynthesis occurred (Table 2). However, the extent of thissynthesis in the 'uncoupled' system was much less than inreactions where RNA was synthesized simultaneously.DNA polymerase from Micrococcus luteus could replace

the KB DNA polymerase both in coupled (Table 1) and inuncoupled reactions. However, the viral DNA polymerasewas completely inactive (Tables 1 and 2).Further differences between cellular and viral DNA poly-

merases in the response to various templates are recorded inTable 2. When assayed with the homologous templateprimers poly(dA) oligo(dT) and poly(rA) - oligo(dT), the KBpolymerase was active only with the deoxy template, whereasthe viral enzyme displayed a very pronounced preference forthe ribo polymer as template strand.

Covalent Attachment of the RNA Primer to the DNA Product.The most plausible explanation for the RNA-mediated stimu-lation of DNA synthesis described above is that RNA chainscomplementary to the DNA template are serving as primersfor the covalent extension by DNA polymerase. Such a reac-tion would yield mixed RNA-DNA product chains, whichare complementary to the template DNA and have RNA

Proc. Nat. Acad. Sci. USA 69 (1972)

Proc. Nat. Acad. Sci.-USA 69 (1972)

at their 5-end and DNA at their 3'-end. Evidence for theexistence of such molecules has been obtained as follows.

Sedimentation of the contents of a reaction mixture froma coupled RNA-DNA synthesizing system in dimethylsulfoxide (Me2SO)-sucrose gradients separated the templateDNA from the newly synthesized RNA and DNA (Fig. 1).The product DNA was smaller and more heterogeneous insize than the template DNA. The RNA was found in the sameregion of the gradient as the newly synthesized DNA, witha peak positioned toward the lighter side of the DNA peak.The fractions containing the bulk of the new DNA and theRNA from the Me2SO gradient were combined and analyzedfurther by equilibrium sedimentation in cesium sulfate. Theresults are shown in Fig. 2. The new DNA banded close to adensity of 1.46 g/cm3, which is characteristic for single-stranded OX DNA (9). Of the RNA, 30% banded at the den-sity of free RNA (1.63 g/cm3); the rest' was found at lowerdensities, with a peak corresponding to the DNA peak.Similar results were obtained when the products of a coupledreaction were denatured by heating to 870 in the presenceof 0.3M formaldehyde, and subsequently analyzed by equilib-rium centrifugation in cesium sulfate, containing 0.25 Mformaldehyde (not shown). Since treatment with Me2SO orheating in formaldehyde disrupts all hydrogen bonds, the

7

i8

10

9o I-

8 .

r-

3E62

5

4L

0103 E

a

O

Fraction numter

FIG. 1. Separation of newly synthesized RNA-DNA from thetemplate DNA by sedimentation in Me2SO-sucrose. RNA-DNAwas synthesized in a 4-mi reaction mixture, as described in Table 1(complete system). The mixture was extracted with phenol, andthe material in the aqueous phase was concentrated and purifiedon Sephadex G-75, as described in Methods. After ethanolprecipitation, the sample was dissolved in 20 Ml of 0.01 M TristHOl (pH 6.8)-i mM EDTA, heated for 2 min to 800, and quicklycooled. The sample was mixed with 10 gl of NN-dimethylformamide and 180,ul of Me2SO, and layeredon two Me2SO-sucrosegradients (24). The gradients were 0-10% sucrose in 10-99%[U-2H]Me2SO (BioRad Laboratories) in Me2SO, containing 1%(v/v) 0.1 M EDTA in water (pH 7.2). Centrifugation was for 12.5hr at 45,000 rpm and 270 in a Spinco SW50.1 rotor. Fractions (0.2ml) were collected and 10-,ul aliquots were counted after acidprecipitation, as described in Table 1. RNA-DNA in peak B wasprecipitated with ethanol and passed through Sephadex G-75 asdescribed above.

banding of RNA at the density of DNA in equilibrium gra-dients must be due to a covalent linkage between RNA andDNA in the synthesized product. From the relative amountsof radioactivity incorporated into synthetic products, it wasestimated that the RNA linked to DNA amounts to 3-5%of the mass of DNA in the peak fractions from cesium sulfategradients (peak "B" in Fig. 2). If a molecular weight (ob-tained by centrifugation in Me2SO-sucrose gradients) of150,000-300,000 is assumed for the product chains, it followsthat the average DNA molecule formed in this experimentcontained 20-50 ribonucleotides.To confirm the presence of a covalent bond between the

RNA primer and the product DNA, RNA-DNA chains weresynthesized as specified in the legend to Fig. 3. These condi-tions were similar to those used before (legend to Table 1),except that three of the four deoxyribonucleoside triphos-phates were labeled in the a-position with 82p, and the ribo-nucleoside triphosphates were all unlabeled and raised to aconcentration of 1 mM. The product of this reaction was ex-pected to be RNA-DNA chains, labeled only in the DNApart. If the RNA primer was covalently linked to DNA bya 3': 5'-phosphodiester bond, the 5'-phosphorus atom of thefirst deoxyribonucleotide should have been transferred tothe adjacent ribonucleotide upon alkaline hydrolysis, re-sulting in a 2'- or 3'-[82P]ribonucleotide (N-1 in Fig. 3). Thiswas exactly the result obtained. The products of alkalinehydrolysis were analyzed by two-dimensional chromatographyon polyethyleneimine thin-layer plates, and subsequentautoradiography. As illustrated in Fig. 3, the only labeledcompounds found were the four ribonucleoside monophos-phates 2'(3')-AMP, 2'(3')-CMP, 2'(3')-GMP, and 2'(3')-UMP (determined by cochromatography with unlabeled

Fraction number

FIG. 2. Equilibrium centrifugation in cesium sulfate of theproduct of coupled RNA-DNA synthesis. Aliquots of purifiedpeak-B material from Me2SO-sucrose gradients (Fig. 1) werediluted to 3 ml and adjusted to a density of 1.53 g/cm3 withcesium sulfate. After centrifugation for 70 hr at 38,000 rpm and20° in a Spinco SW50.1 rotor, fractions were collected and 10-,ulaliquots were counted. One of the gradients contained in addi-tion to the sample, [3H]RNA made in vitro with Simian virus 40(SV40) DNA (25) and [3'P]DNA from adenovirus 2 (gift of Dr. U.Pettersson). The arrows RNA and DNA indicate the position ofthese markers in the gradient.

1562 Biochemistry: KeHer

RNA-Primed DNA Synthesis 1563

N. N-2 N. N. 12 N3LOH LOH -OH IH hH LH

a a a a

C

S

FIG. 3. Radioautograph of two-dimensional fractionation ofacid-soluble material after alkaline hydrolysis of RNA- [32P] DNA.The synthesis conditions as specified in Table 1 (complete system)were modified as follows. The four ribonucleoside triphosphateswere present unlabeled at a concentration of 1 mM; [a-'2P]dATP,[a-32P]dCTP, and [Ca-32PJdTTP (4 Ci/mmol) were used at 6.25MM, unlabeled dGTP at 5 MM. Product containing 3 - 106 cpm of32p was purified as described in Fig. 1, precipitated with ethanol,and incubated for 18 hr at 370 in 0.4 ml of 0.3 N KOH. 60 Ml of14% HC104 was added at 00, and the precipitate removed by low-speed centrifugation; 50 Ag each of 2'(3')-AMP, 2'(3')-CMP,2'(3')-GMP, and 2'(3')-UMP was added to the supernatant andthe mixture was adsorbed to Norit. Material eluting from Noritwith 3 N NH40H in 50% ethanol was taken to dryness, dissolvedin 50 Al of water, and subjected to two-dimensional chromatog-raphy on a polyethyleneimine-cellulose thin layer (26). The sam-ple was applied at the starting point S, and the chromatogram wasdeveloped in the first dimension (from right to left) with 0.2 M, 1.0M, and 1.6 M LiCl, and in the second dimension (from bottom totop) with 0.5 M, 2.0 M, and 4.0 M Na formate buffer (pH 3.4).The chromatogram was exposed for two days to Kodak RPR x-rayfilm. A, C, G, and U, and the dotted lines indicate the position ofthe corresponding 2'(3')-mononucleotide markers. N-1 indicatesthe ribonucleotide linked to the first deoxyribonucleotide N1; aindicates points of cleavage by alkali; asterisks represent 32P.

markers). The identity of these products was further verifiedafter elution from the chromatography plates by electro-phoresis on cellulose sheets in pyridine-acetate buffer (re-sults not shown). In parallel experiments, the products ofuncoupled DNA synthesis with either OX DNA RNA hy-brid (as in Table 2) or activated calf-thymus DNA as tem-plate primers (in the absence of RNA polymerase and ribo-nucleoside triphosphates) were analyzed. No labeled ribonu-cleotides were detected in hydrolysates of the product ofcalf-thymus DNA-mediated reactions. However, as in thecoupled system, all four ribonucleotides became labeled inthe reactions with OX DNA RNA hybrid as template. Theamount of label transferred to ribonucleotides was 1-3% ofthe total acid-precipitable 32p in the coupled reactions, and0.01-0.05% in uncoupled reactions. The relative frequencywith which the four ribonucleotides occurred at the RNA-DNA link was determined by measurement of the radio-activity in the corresponding areas of the thin-layer sheets.For the coupled system, 28% AMP, 13% CMP, 23% GMP,and 36% UMP was observed. The product from an uncoupledreaction contained 25% AMP, 23% CMP, 32% GMP, and20% UMP at RNA-DNA links.

The Action of Neurospora Nuclease and RNase H. Furtherinformation on the secondary structure of the RNA-DNA

TABLE 3. The effect of nucleases on synthetic RNA-DNAchains in the presence or absence of template DNA

Neurospora nuclease RNase H

% Resistance % Resistance

[ap]-_ [sH]- [32p] _ [3'H]-RNA DNA RNA DNA

Native product* 100 100 <2 100Isolated RNA-DNA chainst 15 20 70 100

RNA-DNA chainsrehybridized toOX DNAt 75 92 5 100

Treatment with Neurospora nuclease was performed in reac-tion mixtures containing in 0.3 ml: 0.1 M Tris -HCl (pH 7.3);0.08 M NaCl; 6 mM Mg(acetate)2; sample containing [32P]_-RNA-[3H]DNA (10,000-20,000 cpm total, prepared as describedin the legends to Figs. 1 and 2) and 20 ug of Neurospora nuclease.Duplicate assays were incubated for 30 min at 37°.Reaction mixtures for RNase H treatment contained in 0.1 ml:

0.05 M Tris-HCl (pH 7.9); 0.1 M NaCl; 10 mM MgCl2; 5-10lig of RNase H, and sample as above. After 15 min of incubation,acid-precipitable radioactivity was determined.

* Sephadex G-75 eluate, as described in the legend to Fig. 1.t Peak B from cesium sulfate gradient (Fig. 2), concentrated

by alcohol precipitation and purified on Sephadex G-75 as de-scribed in Fig. 1.

t Aliquots of peak B from a cesium sulfate gradient (Fig. 2),concentrated as described above, were incubated in reactionmixtures containing 1 M NaCl, 0.01 M Tris HCl (pH 7.5),1 mM EDTA, and 10 Mg/ml of OX DNA for 5 hr at 680, andsubsequently treated with Neurospora nuclease and RNase H.

made in vitro was obtained by comparison of the effects oftwo specific nucleases, Neurospora nuclease and RNase H.Neurospora nuclease degrades both single-stranded RNAand DNA (14, 16, 17). RNase H only degrades the RNAstrand of complementary DNA RNA hybrids (5).The product of coupled RNA-DNA synthesis reactions was

purified as described in the legend of Fig. 1, and treated withnuclease (Table 3). The synthetic RNA-DNA at this stagewas completely resistant to the Neurospora enzyme, an in-dication that both the RNA and the DNA sections were hy-drogen bonded to the complementary template DNA. RNaseH almost completely degraded the RNA part of the syntheticmixed polymers, but had no effect on the new DNA. Afterdenaturation of the native DNA RNA-DNA complex, re-moval of the template DNA by sedimentation in Me2SOgradients, and equilibrium centrifugation in cesium sulfate(see legends of Fig. 1 and 2 for details), the RNA-DNAchains became sensitive (80-85%) to Neurospora nuclease,showing that they were now in a single-stranded configuration.At the same time, the majority (70%) of the RNA had be-come resistant to RNase H, indicating the absence of com-plementary DNA. When it was incubated for 5 hr at 68°in 1 M NaCl with 10 Mg/ml of XX DNA, at least 80% of thepurified product RNA-DNA could be reannealed to the tem-plate DNA, resulting in a reversal of the nuclease sensitivity.The results of these experiments are summarized in Table 3.

DISCUSSION

Single-stranded circular DNA, like that of phage qX174, is apoor template for in vitro DNA synthesis, since it lacks the

Proc. Nat. Acad. SC?". USA 69 (1972)

Proc. Nat. Acad. Sci. USA 69 (1972)

complementary primer strands with free 3'-hydroxyl groupsthat are essential for initiation of the polymerizing action ofDNA polymerases. It is known that initiation can be ac-complished by the addition of oligodeoxyribonucleotides (15).However, the experiments reported here demonstrate thatshort RNA chains, synthesized either before the action ofDNA polymerase (uncoupled system) or simultaneously withDNA synthesis (coupled system) can also serve as primers forthe production of DNA complementary to the circular tem-plate. Proof has been obtained for covalent linkage of thepriming RNA to the newly made DNA via a 3': 5'-phospho-diester bond between the last ribonucleotide and the firstdeoxyribonucleotide, resulting in DNA chains with a smallpiece of RNA at their 5-ends.

All four ribonucleotides were found in RNA-DNA linkages,implying a lack of specificity for the in vitro system: RNAchains ending with any of the four ribonucleosides could becovalently extended byDNA polymerases.A profound difference was noted in the response of various

DNA polymerases: whereas KB DNA polymerase II andDNA polymerase from M. tuteus were about equally efficient,viral DNA polymerase could not operate at all in eithercoupled or uncoupled reactions. As observed by others (17),the viral enzyme, although it generally can copy either RNAor DNA templates, is unable to act on DNA with extensivesingle-stranded regions.Based on experiments with phage M13-infected E. coli

cells, Brutlag et al. (4) have proposed RNA-primed DNAsynthesis as a mechanism of DNA initiation in vivo. Rif-ampicin, a specific inhibitor of DNA-dependent RNA poly-merase, prevented the conversion of parental phage DNA intothe replicative form, as well as the subsequent synthesis ofprogeny DNA. Evidence for the direct participation of atranscriptional event in DNA replication in vivo had alreadybeen reported for phage lambda (3), and has also been ob-tained for E. coli episomes (2) and the chromosome of E. coli(1). My results obtained in vitro lend additional support tothe notion that RNA-primed initiation of DNA synthesismay be a common mechanism.

Finally, I would like to call attention to the role RNase Hmay play in RNA-primed DNA synthesis. As reported above(Table 3), RNase H is able to remove the priming RNAcovalently linked to product DNA. Preliminary experiments(R. Crouch and W.K.) with this enzyme revealed an endonu-cleolytic mode of action, resulting in mono- and oligoribonu-cleotides that carry 5'-phosphate and 3'-hydroxyl groups.The 5'-termini of the DNA, remaining after removal of thepriming RNA, also carry a 5'-phosphate. Thus, RNase Hdisplays precisely the characteristics expected of a nucleasewhose in vivo function is to remove priming RNA pieces fromnascent DNA chains. Although RNase H has been isolatedonly from eukaryotic sources, it can be expected that similarenzymes will also be found in prokaryotic organisms. [RNaseIII from E. coli (18), although it has a preference for double-stranded RNA, also does degrade RNA in DNA-RNA hy-brids, and could be a candidate for a prokaryotic RNase HI.Furthermore, the occurrence of RNase H activity in avianmyeloblastosis virus (19) suggests that this nuclease mayparticipate in the production of double-stranded viral DNAby RNA-dependent DNA polymerase. In this system, viralRNA is first copied into a RNA DNA hybrid, followed bythe appearance of double-stranded DNA (20, 21). The mecha-nism of the latter reaction is not understood, but could readily

be conceived to require RNase H action. Limited degradationof viral RNA at the RNA-DNA hybrid stage would lead tooligoribonucleotides with 3'-hydroxyl groups complementaryto the newly synthesized single-stranded DNA, which inturn could serve as primers for the viral DNA polymerase(16, 22), resulting in synthesis of double-stranded DNA.Understanding of the detailed mechanism of such concertedreactions requires further experiments. Further work willalso be necessary to elucidate the regulation of coupledRNA-DNA synthesis in vitro. It is not clear whether DNApolymerase initiates exclusively on RNA chains whose syn-thesis has first been terminated, or if it can compete withRNA polymerase. It is also of interest that under the condi-tions of coupled RNA-DNA synthesis, ribonucleoside tri-phosphates are not incorporated into DNA. Such 'ribosub-stitution', as first shown by Berg et al. (23), probably occursonly in the presence of Mn++ and in the absence of the cor-responding deoxyribonucleoside triphosphates.

I thank Dr. J. D. Watson for his support and encouragement,and John Cairns, Bob Crouch, Joe Sambrook, Phil Sharp, andBill Sugden for help and discussions. This work was supportedby Grant 1 P01 CA13106-01 from the National Cancer Institute.

1. Lark, K. G. (1972) J. Mol. Biol. 64,47-60.2. Bazzicalupo, P. & Tocchini-Valentini, G. P. (1972) Proc.

Nat. Acad. Sci. USA 69, 298-300.3. Dove, W. F., Inokuchi, H. & Stevens, W. F. (1971) in

The Bacteriophage Lambda, ed. Hershey, A. D. (Cold SpringHarbor Laboratory, New York), pp. 747-771.

4. Brutlag, D., Schekman, R. & Kornberg, A. (1971) Proc.Nat. Acad. Sci. USA 68, 2826-2829.

5. Hausen, P. & Stein, H. (1970) Eur. J. Biochem. 14, 278-283.

6. Stavrianopoulos, J. G., Karkas, J. D. & Chargaff, E. (1971)Proc. Nat. Acad. Sci. USA 68, 2207-2211.

7. Karkas, J. D., Stavrianopoulos, J. G. & Chargaff, E. (1972)Proc. Nat. Acad. Sci. USA 69, 398-402.

8. Sinsheimer, R. L. (1966) in Procedures in Nucleic Acid Re-search, eds. Cantoni, G. L. & Davies, D. R. (Harper & Row,New York), pp. 569-576.

9. Chamberlin, M. & Berg, P. (1964) J. Mol. Biol. 8j 297-313.10. Yoneda, M. & Bollum, F. J. (1965) J. Biol. Chem. 240,

3385-3391.11. Weissbach, A., Schlabach, A., Fridlender, B. & Bolden, A.

(1971) Nature New Biol. 231, 167-170.12. Zimmerman, B. K. (1966) J. Biol. Chem. 241, 2035-2041.13. Burgess, R. R. (1969) J. Biol. Chem. 244, 6160-6167.14. Rabin, E. Z. & Fraser, M. J. (1970) Can. J. Biochem. 48,

389-392.15. Goulian, M. (1971) Annu. Rev. Biochem. 40, 885-898.16. Verma, I. M., Meuth, N., Bromfeld, E., Manly, K. F. & Bal-

timore, D. (1971) Nature New Biol. 233, 131-134.17. Hurwitz, J. & Leis, J. P. (1972) J. Virol. 9, 116-129.18. Robertson, H. D., Webster, R. E. & Zinder, N. D. (1968)

J. Biol. Chem. 243, 82-91.19. Molling, K., Bolognesi, D. P., Bauer, H., Busen, W., Plass-

mann, H. W. & Hausen, P. (1971) Nature New Biol. 234,240-243.

20. Fanshier, L., Garapin, A. C., McDonnell, J., Faras, A.,Levinson, W. & Bishop, J. M. (1971) J. Virol. 7, 77-86.-

21. Manly, K. F., Smoler, D. F., Bromfeld, E. & Baltimore, D.(1971) J. Virol. 7, 106-111.

22. Leis, J. P. & Hurwitz, J. (1972) J. Virol. 9, 130-142.23. Berg, P., Fancher, H. & Chamberlin, M. (1963) in Informa-

tional Macrooilecules. eds. Vogel, H. J., Bryson, V. &Lampen, J. 0. (Academic Press, New York), pp. 467-483.

24. Sedat, J., Lyon, A. & Sinsheimer, R. (1969) J. Mol. Biol.44,415-434.

25. Keller, W. & Goor, R. (1970) Cold Spring Harbor Symp.Quant. Biol. 35, 671-680.

26. Randerath, K. & Randerarth, E. (1967) in Methods inEnzymology, eds. Grossmann, L. & Moldave, K. (AcademicPress, New York), Vol. XII A, pp. 323-347.

1564 Biochemistry: KeHer