mM RNA

(1)

Proc. Nat. Acad. Sci. USA

Vol. 72, No. 1, pp. 162-166,January 1975

Evidence for De Novo Production of Self-Replicating and Environmentally Adapted RNA Structures by Bacteriophage Q3 Replicase

("6S RNA"/RNA-protein interaction/selection/ethidium bromide)

MANFRED SUMPER AND

RUDIGER

LUCE

Max-Planck-Institut fur biophysikalische Chemie, 34Guttingen-Nikolausberg, WestGermany

CommunicatedbyManfred Eigen, October 11, 1974

ABSTRACT Highly purified coliphage Qf3 replicase whenincubated-withoutadded templatesynthesizes self- replicatingRNAspecies in an autocatalytic reaction.

In this paper we offer strong evidence that this RNA production is directed by templates generated de novo during the lag phase. Contamination of the enzyme by traces ofRNA templates was ruled out by the following experimental results: (1) Additional purification steps do not eliminate this RNA production. (2) The lag phase is lengthened to several hours by lowering substrate or enzyme concentration. At a nucleoside triphosphate con- centrationof 0.15 mM no RNA is producedalthough the template-directed RNA synthesis works normally. (3) Different enzyme concentrations lead to RNA species of completely different primary structure. (4) Addition of oligonucleotidesorpreincubationwithonly three nucleoside triphosphates affects the final RNA sequence. (5) Manipulation ofconditions during the lag phase results inthe production ofRNA structures that are adapted to the particular incubation conditions applied (e.g., RNA resistant to nuclease attack or resistant to inhibitors or evenRNAs "addictedtothedrug,"inthe sense thatthey only replicate in the presence of a drug like acridine -orange).

RNA species obtained in different experiments under optimal incubation conditions show very similar fingerprint patterns,suggestingtheoperationofaninstruction mechanism. Apossible mechanism isdiscussed.

The small bacteriophage of Escherichia coli,

Q0,

induces an enzyme,

Q0 replicase,

that is

responsible

forthemultiplication of the phage RNA. This

RNA-dependent

RNA polymerase consists of one

virus-specifiedl polypeptide

subunit (I) and three host

polypeptides

a, y, anti 6 (1, 2). Blumenthal et al.

(3) have foundthat

'y

and6 arethe

protein

synthesis elonga- tionfactors EF Tu andEF^-Ts,

respectively.

Subunit awas recently identified as the

protein

component S1 ofthe ribo- somal 30S subunit

(4).

Thephage replicaseshowsa very

high template specificity

forthe complementary plus andminusstrands ofthe homol- ogous viral RNA (5, 6). Unrelated viral RNAs and most otherRNAs

examineti

(lonot serve as

templates

(5).

InadditiontoreplicatingtheQu plus

anti

minusstrandsthe enzyme will alsocopy

poly(C)

(7) as wellasotherspeciesof

self-replicating

RNAs,

including

"6S RNA"isolatedfrom

QO-

infectedE.coli cells (8)

and

"variants,"of

Q,3

RNA(9).

In this paper we offer strong evidence for a new type of

temlplate-free

(de novo) RNA

synthesis, catalyzed

by

QO

replicase, in whichtruly

self-replicating

RNAstructures are

produced.

Thesesequencesare not

homopolymeric

or

strictly alternating anti they

are

adapted

totheenvironmental conditions

applied during

their

generation.

MATERIALS AND METHODS

Q

0

replicasewasassayed according to Kamen (10). One unit is defined as the amount of enzyme that catalyzes the incor- porationof 1 nmol ofGTP in 10 min at

300. Qf3 replicase

was purifiedfrom

Qf-infected

E. coli K12 Hfrcells by the method of Kamen et al. (11)uptothedensity gradient centrifugation, but omitting the chromatography on agarose.

Qo-replicase- comitaining

fractions from thedensitygradient centrifugation (stage VI) were diluted 10-fold with a

buffer

containing 50 mM Tris^-HC1 (pH 7.5), 0.1 mM dithiothreitol, and 20%

glycerol and applied to a column (1.6 X 10 cm) of QAE- SephadexA-25equilibrated with thesamebuffer. Thecolumn waseluted with a linear gradient from 0 to 0.3 M NaCl (total volume 400 ml). This gradientensures the complete separa- tionof a-lessandholoenzYme.

The standard incubation mixture for the template-free RNAsynthesis containedin 200

Mul:

50

mMd

Tris HC1 (pH7.5), 10 MM

MigCl2,

0.1 mM

dithiothreitol,

10% glycerol, ATP, GTP, UTP, and CTP (one ofwhich was labeled with 14Cor

32P)

andenzyme asindicatedin thelegends.

Special precautions

were taken

throughout

toavoida con- taminationofincubationmixtureswith

self-replicating

RNAs:

(a)

tiouble-distilled

water was usedthroughout, (b) only dis-

posable

plastic tubes and

plastic pipettes

were

used,

and

(c)

mix solutions (without nucleoside triphosphates) were fil- teredover acolumnof

QAE-Sephadex.

RESULTS

QBreplicase

purified according

to the

procedure

of Kamen etal. (11) ismorethan

95%

pure

and

free of

optically

detect- abletracesofnucleicacids

(stage VI).

Atthisstageof

purifica-

tion

Q,3 replicase,

when incubated with the nucleoside triphosphates

ATP,

UTP,

CTP,

and GTP in the absence of

added

RNA

template, synthesizes self-replicating

RNA inan

autocatalytic

reaction. This synthesis becomes detectable aftera

lag phase

of 20-40 min. We willdenote thisreaction in the

following

as

"template-free

RNA

synthesis." Phosphate

(10

mMI),

a

triphosphate-regenerating

system, or

rifampicin (5,4g/ml)

doesnotinfluence this RNA

production.

Millsetal.

(13) sequenced

recently

suchanRNA

species containing

218 nucleotides.

RNA

species

isolatedfrom separate reactionmixtures run

under

identical conditions exhibit

fingerprint patterns

which are very similar toeach other*. This has been

interpreted

as

* RNAspecies growingout from template-free incubation mixtures

uinder

ourstandard conditions will be denoted in the follow- ingasstandard type RNAs (ST-RNAs).

162

(2)

being caused byacontamination of

Q0

replicasebytracesof self-replicating RNA (11). Therefore, we subjected the Q,3 replicase (stage VI) to additional purification procedures known to resolve proteins and nucleic acids with high efficiency.

Cesium Chlo'ideDensityGradient

Centrifugation.

Paceetal.

(14) have developed apycnographic purificationstep for Qf3 replicase. We, therefore, bandedourpurest

Q,3-replicase

protein (stage VI) in a CsCl density gradient (p = 1.2-1.5) and collectedthe enzymeby

piercing through

thesideofthe tube, immediately below the protein band. Any contaminating RNA should have

pelleted

onthebottomofthe

tube,

dueto itshighbuoyant density (p = 2.0). The treated

Qfl

replicase, however,retained itsabilitytogenerateST-RNA.

Anion Exchange Chromatography. The effectiveness of a purification step based on anion

exchange chromatography

was tested in the following manner.

Q#

replicase (100 ug, stage VI), deliberately contaminated with highly labeled ST[32P]RNA (5

Mg,

atotal of 107

cpm)

wasappliedtoaQAE- Sephadexcolumn (10ml) andthenthe column wasdeveloped with a linear NaCl gradient (0-0.5AM). No radioactiveRNA material eluted up to a NaClmolarity of 0.5M,whereasthe enzymatic activity eluted in two sharp and completely re- solved peaks between 0.15 and0.20AilNaCl. PeakAmaterial was found to be the recently described a-less replicase

(i1) (Q0

replicaselackingthesubunit a) and

peak

Bmaterial was identified as

Qfl-replicase

holoenzyme (containing all four subunits). Since no detectable radioactivity was found ineither enzyme fraction, less than 1 out of

106

RNA molecules re- mained associated with

Q#-replicase.

Because of its excellent separation efficiency,

QAE-Sephadex

chromatography was introduced in

addition

into the routine

preparation

of

Q0

replicase. After pooling, both the a-less and holo

replicase

fractions wereconcentrated on small

QAE-Sephadex

columns (stage VII). Even at this stage ofpurity, the ability to gen- erateST-RNAwasfullyretained.

Since the a-lessreplicase representsthe "core enzyme" for generation and replication ofST-RNAit wasselected for fur- ther studvof thetemplate-free RNA synthesis. The contamination

hypothesis

explains the ST-RNA synthesis in any sample as being caused by the presence of at least one RNA strand. On thisbasisanestimate for the minimum number of ST-RNA strands hypothetically contaminatingourstandard

incubation

volume of 200

Al

(enzyme concentration 25-70 units/ml) wasmade by scaling

(lown

ourincubation volumes to values as small as 0.02

Ml

(without changing enzyme and substrate concentrations). After elimination of experimental difficultiessuch as surface

denaturation

of the enzymeetc., it turnedoutthat the template-freeRNAsynthesis worked even inthese small volumes. It follows that in our usualincubation volume theminimum number would be 10,000RNAstrands.

Pirst Contradiction to the Contamination Hypothesis. As shown inFig. 1A the rate of theST-RNA-directed R.NA

syn-

thesis by a-less replicase is only slightly influenced as the levels of the nucleoside

triphosl)hate

concentrations drop from 0.5m-Mto 0.15

mMX

each. The rate ofsynthesis at 0.15

mM\

is about

80%

of maximum. In sharp contrast, the lag times of the template-free RNA

synthesis

are dramatically

lengthened

by lowering the nucleoside

tril)hosphate

concen- trationsto0.15

mMAl

(Fig.

iB):

under theconditions used, the

E

C

a E

cL 00.

0 CL

a.

0 l l l

0.1 0.3 Q5 mM NTP

r"

0

E₄

La

0.

3 B

2/

0

1 2 3 4 5 6

hours

Fi. 1. Effect of substrate concentration on the rate of the template-directed RNAsynthesis (A)andonthelengthofthelag phase ofthetemplate-free RNAsynthesis (B). A: The standard incubation mixture contained 4 units of a-less replicase (stage VII), nucleotides as indicated (GTP labeled with 14C, specific activity2Ci/mol)andinaddition0.5,MgofST-RNA.Incubation was at300 foi 10min.Acid-insolubleradioactivitywasmeasured by the -Millipore filter technique. B: The standard incubation mixture contained 10 units ofa-lessreplicase (stage VII)andde- creasing concentrations of nucleoside triphosphates (GTP was labeled with 'IC, specific activity 2 Ci/mol): curve 1: 0.5

mM\

(each); curve2: 0.3

mM\

(each); curve3:0.15m.M (each). Incu- bation at300. Aliquots (20 Mul)wereremovedafterdifferent times and theincorporation ^wasmeasured.

length

ofthe

lag phase

increased from 60minat0.5mM, to 200-300minat0.3miM, and

finally

at0.15

mMI

noRNAsyn- thesis atallwasdetectable (luring^anincubationperiod of 15

hr, although

the enzyme retained

25%

of itsinitial

activity

after this

period.

Since thetemplate-(lirected RNA

synthesis

isnot

suppressed

at0.15

mMA,

thenonappearanceofST-RNA

production

atthis

low

substrate concentration hasto be at- tributed to a lack of

templates,

unless a very low level of ST-RNA (the hypothetical contamination of at least

10,000

strands, as shown above) fails to initiate synthesis for un-

known reasons. This possibility was easily ruled out by a

serialdilution experiment

(Fig. 2):

AST-RNA solution, con-

taining

1 * 1011 strands perM1 was diluted in steps of 1 :10or

1:100up toanoveralldilution of 1:1012.Then5,M1ofa

given

dilution were added to thestandard incubation mixtureat a

lowsubstrate concentration (0.15 m.M) and incubatedat

300.

AsseeninFig. 2, alltubesinoculated with RNA in dilutions up to 1011 initiated extensive RNA synthesis, whereas the 1012dilution and all controls (a total of 15) failed to

synthe-

size RNA. Therefore, as few as 5 ST-RNA strands initiate RNA synthesis at the low substrate concentration. The observation that no RNA is synthesized at 0.15 m.M NTP without the addition of atleasta fewtemplate strands con-

tradicts the contamination hypothesis. We conclude, therefore, that the RNA synthesis observed at normal substrate concentrations is directed by template synthesized de novo

duringthelagphase.

Effect

ofOligonucleotides. The addition ofanoligonucleotide (2 A260

units/ml)

such as C-C-C-C-OH or A-A-A-A-OH to thetemplate-free RNA synthesis incubationmixture restores the ability of a-less replicase to produce RNA even at our lowest substrate concentration (0.15 mMI). The important point is, however, that different oligonucleotides induce the production of different RNA species,asjudged by theirfinger-

print patterns. -Moreover,

the

fingerprints

from RNA

prod-

ucts isolated from separate but otherwise identical incuba-

(3)

164 Biochemistry: Sumperand Luce

4000 /

a3000- a

- 2000-

0.

/1..

u 1000 I

1 1,1.78.9 L

L)o 0 0+*'--1+ "i

12 3 lo 5 6

hours

7 8

FIG. 2. Serial dilution experiment. An ST-RNA solution, containing 1 X 1011strands per1ul, wasdiluted in steps 1:10 or 1:100upto anoverall dilution of 1: 1012 [dilution buffer: 10

mMN

sodium acetate (pH 5.4), 1 mM EDTA]. Then 5

1.d

of a given dilution (curves 1-7) were added to the standardincubation mix- turecontaining 10 units of a-less replicase (stage VII) and 0.15 mM nucleoside triphosphates (each). CTP was labeled with '4C specific activity 2 Ci/mol. Curve 1: 1: 102 dilution; curve 2:

1:104 dilution; curve 3: 1:106 dilution; curve 4: 1:108

dilution;

curve5: 1:109 dilution;curve 6: 1:10"1dilution; curve 7: 1:1012 dilution; curve 8 and 9: no addition. Incubation was at 300.

Aliquots (20 Ml) were removed after different times and incor- poration was measured by theMilliporefiltertechnique.

tion mixtures containing A-A-A-A-OH differed significantly fromeach other and from ST-RNA patterns.

Effect

ofEnzyme Concentration. Thelengthof thelag phase isalso influenced by theenzymeconcentration. At an enzyme concentration of 70 units/ml, the length of the lag phase is about 50-70 min. When the enzyme concentration is lowered to20units/ml,thelagtime increases to 2-3

hr,

andfinally at enzyme concentrations of 5units/ml^orlessno RNA is synthesized for atleast 7 hr. The RNA products generated at different enzyme concentrations were compared by their fingerprintpatterns (Fig.3).Remarkably, differentsequences were produced at different enzyme concentrations, although exactlythe same absolute amount ofenzyme wasapplied in each experiment. However, repetition of template-free incubation under identical and optimal conditions (high sub- strateand enzyme concentration suchasin experimentA of Fig. 3) several times resulted in theproduction of verysimilar RNAs, asjudgedbytheir fingerprintpatterns.

Second Contradiction to the Contamination

Hypothesis.

Pre- incubation of a-less replicase (stage VII) in the presence of only three nucleoside triphosphates for 2 hr and furtherin- cubationafteradditionof the omitted nucleotide resultedin the production of RNA species completely different in

pri-

mary structure from ST-RNAs. Fig. 4 presents the

finger-

printpatterns of the obtained RNA species when GTP (A), CTP (B), orATP (C) wasomitted duringthe

preincubation period.

Thechainlengthsof these RNAswereestimatedtobe 180 for

species

A and 140 for

species

C. These RNAs were

replicated slower

(20-60%)

than ST-RNA. Toexplain these results on the basis of the contamination hypothesis, two conclusionsmustbetrueabout thesystem. First,inaddition toST-RNAs, manydifferentRNAspecies are presenteither ascontaminationsof the enzymeorasproductsderivedfrom ST-RNAs during the

preincubation

period. Second, one of these

species

is selectively favored

by

thepreincubation con-

ditions andsuppressesthe

replication

of the normallyfavored ST-RNA. This unlikely interpretation was shown to be

..

A

4

.411,

.0 jk

0 * f

a 1'

B

cellulose acetate pH 3,5

-i&

-,A

i. E C

m m a

3CT

a

CL.

C

FIG.3. Fingerprints (ribonuclease T, digests) of RNA species produced atdifferent enzymeconcentrations. Five units of a-less replicase (stage VII) were mixed withdifferent volumes of st an- dard incubation mixture containing 0.5

mM\

nucleoside triphosphates (each) andincubated at

300

untilautocatalytic growth of

BNA.

The radioactive label was [a-12P]UTP. A: incubation volume was 80,ul; B: incubation volume was 200

yl;

C: incuba- tionvolume was 400 ml. The

IINA

species were isolated by exclu- sion chromatography on Sephadex (4-50 (H20) andconcentrated by lyophilization. After heat-denaturation

(1000,

3 min) the RNAs wereprocessed according to ref. 12. The arrow designates theposition of the bluemarker.

wrongby the following control experiment: As few as five to tenST-RNA strandswere added from the verybeginning of anexperiment identicaltotheoneabove (Fig. 4). The fingerprint pattern of the RNA growing out in this

experiment

was identical with the

pattern

of theST-RNA added. Therefore,

ST-RNA

ifpresent ab initioisabletogrowoutunder theconditions used.Weagain concludethatST-RNAcannotbe pre- sent ab initio inour a-less replicase

preparation

(stage VII).

TheGeneration

of Environmentally Adapted

RNAMolecules.

Conditions caneasilybe found where

replication

ofST-RNA is completely inhibited without affecting the enzymatic activity of the

replicase.

For instance, the

replication

of ST- RNA can be entirely halted by the addition of acridine orange, ethidium bromide, .In++ions, or ribonucleases. We haveincubated

Q0

replicasetogether with the fournucleoside triphosphates under various conditions which

completely suppressed

the

replication

of ST-RNA. In all these

experi-

ments, after lag phases of2-12 hr, RNA

species

resistantto theinhibitory conditions

being applied

were

produced.

Inthe followingparagraphsafew

examples

of these

experiments

are

presentedinmore

detail.

Acridine Orange and

Ethidium

Bromide. The inhibition of

ST-RNA-directed

RNA

synthesis

by

increasing

amounts of ethidium bromide is shown in

Fig.

5. The a-less

replicase

when incubated in the standard incubation mixture in the presence of10-50

Mgo/ml

ethidium bromide

generates,

after

lag phases

of 2-6 hr, RNA species which are resistant to this drug, as shown in

Fig.

5.

Interestingly,

these RNA

prod-

uctswere

"addicted

to the

drug,"

inthe sense that

they

re-

quired

itspresence for

replication

withmaximum rate. This observation^was

particularly

trueof RNA

species

resistantto

acridine orange, which would

only reproduce

in thepresence Proc. Nat. Acad. Sci. USA 72

(1975)

(4)

a.

A (-GTP)

cellulose acetate pH3.5

.0.' :0*

B (-CTP)

T

C (-ATP)

FIG. 4. Fingerprintsof RNAspeciesproduced bya-lessreplicaseafterpreincubationinthe presence of

only

three nucleosidetriphosphates. A:GTPomitted;B: CTPomitted;C: ATP omitted. The standard incubation mixture contained 15 units of a-lessreplicase(stage VII) butonlythreenucleotides (0.5mMeach)asindicated. After incubation for 2 hrat

300,

the omittednucleotidewasadded and the incubation was continued overnight. Fingerprints (ribonuclease T1, radioactive labelwas

[a-32PJUTP)

of theoutgrowing RNAs were obtainedaccording toref. 12. Thearrowdesignates the positionofthe blue marker.

of thedrug.Again,thefingerprintpatternsoftheseresistant RNAs differ completely from that of ST-RNA. Moreover, theseresistant RNAsareverysmall

self-replicating

molecules;

weestimate chainlengthsofonly90to100nucleotides.

Ribonuclease

Ti.

The ST-RNA-directed RNA synthesis is strongly affected bythe presence ofribonuclease T1. A concentration of2

,4.g

ofnuclease per ml in thestandardmixture completely eliminates the synthesis of ST-RNA, as shown in Fig. 6.

Q0

replicase when incubated in thestandard in- cubationmixturein the presence of nucleasegenerates, after lagtimes of4-10

hr,

RNAspecies which proved to be rather resistant tonsuclease attack,asdemonstratedinFig.6.

In a series of analogous experiments, we obtained RNA species which grew in the presence of Mn++ ions or high ionic strength (0.4 M NaCI), conditions not allowing the

'0"- 1

0 20 40 60

ethidiumbromide [mg ml]

FIG.5. Effectofethidium bromide^onthe replication rateof ST-RNA (curve 1) andaresistantRNA (curve2).The standard incubation mixture contained 6 units of a-less replicase (stage VII), 0.15 mM nucleotides ([14C]GTP, specific activity 2.5 Ci/mol) and, in addition0.5,MgofST-RNAorresistant RNAand ethidium bromide as indicated. The replication rate at agiven ethidium bromide concentration was determined by measuring theGIMPincorporation after incubationperiods of 10,20,and30 min. The resistant RNA was obtained by incubating a-less replicase (stage VII) inthe standard incubation mixture(0.5mit\

nucleosidetriphosphates)in thepresenceof 50

/ug/ml

ofethidium bromide (lag time about4hr).

replication of ST-RNA. In all cases, the resistant RNAs showed new oligonucleotide fingerprint patterns

differing

completelyfrom theST-RNApatterns.

DISCUSSION

Inouropinion, the experiments presented in thispaperleave nootherinterpretation butasynthesis denovoofRNA byQ83 replicase. Denovosynthesis of nucleic acids in the absence of template is a common property of RNA and DNA poly-

merases (15-19). However, all examples reported previously gaveonly highly orderedsequences.Incontrast,thetemplate- free synthesis of RNA by Qj3replicase described in thispaper leads to truly self-replicating RNA molecules with defined

10 20 30

minutes

FIG. 6. Kinetics of RNA synthesis directed by ST-RNA (curves 1 and 2) or a "T,-resistant" RNA (curves A,B, andC) inthe absence or presence of ribonucleaseTi.The standardincubation mixture (without dithiothreitol) contained 6 units of a- less replicase (stage VII), 0.15 mM\ nucleotides ([14C]GTP, specificactivity 2.5 Ci/mol) and in addition 0.5,MgofST-RNA orresistantRNA. RibonucleaseT, was added as follows:curve1 andA:noribonucleaseTi; curve2andB:

2,4g/ml

ofribonuclease TI; curveC: 4,g/mlofribonuclease T,. The resistant RNAwas obtained byincubatingQflreplicase (stageVII) in the standard incubation mixture (0.5mMl nueleosidetriphosphates, but with- outdithiothreitol) inthe presence of 2pg/ml ofribonuclease T, for 8 hr at 300.

(1975)

(5)

166 Biochemistry: Sumper and Luce

and nonrepetitive structures. A puzzling fact is the observation that under optimal conditions (high substrate and enzyme concentration)-but only at these-the products formed in separatebut otherwise identical template-free RNA synthesis experiments give verysimilar fingerprint patterns.

On the other hand, theseRNA products have variousreplica- tion kinetics, which indicates that the sequences cannot be identical. Moreover, the interpretation of the fingerprint patternsis hamperedby thefact that usually more than half of the radioactive label remains in the core material, even though heat-denatured ST-RNAs were used. Nonethe- less,thesesimilaritiesof thefingerprints demandsome sort of instruction upon the outcome of the RNA sequence. How this can be achievedin the absence of a template for

RNA

moleculesaslong as 200nucleotidesisverydifficultto imag- ine, but there areseveral observations favoring the idea that

Qo-replicase

protein itself couldexert thisinfluence.

This enzyme contains several subunits with specific RNA recognition sites. The viral subunit fi is responsible for the template specificityof the enzyme,since the analogousRNA replicases containing the same host subunits but a different viralsubunit exhibit adifferent template specificity. Another subunit ofQi3 replicase,theprotein synthesis elongation factor EF* Tu, forms specific ternary complexes with GTP and aminoacylated tRNAs (20) andensurestheproperbinding of the tRNA to the ribosome. Possibly the subunits 3 and EF^-Tufavor theoutcomeofnucleotidesequencescontaining

the

elements of recognition. Studying the sequence of one ST-RNApublished by Millsetal. (13), wewereimpressed by thehigh abundance of thenucleotide sequence UUCG and its complementCGAA. UUCG appearsasoften as seventimesin this RNA andislocatedintheunpaired regionsofthemole- cule. Remarkably, this sequence UUCG is common to most tRNAs (as T\IVCG in the "T'C

loop")

and involved in the binding process of tRNA to theribosome (21-23), which is controlled

by

EF * Tu. Assumingarandompolymerizationto

oligomers

for thefirst

phase

of the denovosynthesisaspostu- latedforotherpolymerases (24),Qt3 replicasewould

probably

make a

preferential

use of certain oligonucleotide sequences forthe final assembly of RNA molecules. Sucha discrimina- tion wouldlimit thenumberof

possible

sequences.Moreover, duringthe

replication

phasethe

produced

sequences are sub- jectedto a strongselection. First, onlyself-replicating RNAs (plus and minus strands are

recognized

by theenzyme) can multiply. The more

specific

the

recognition

mechanism the higheristhis sortofselection pressure.

Second,

from several

self-replicating

RNAs

produced

simultaneously during the lag phase, the fastest

replicating species

outgrows its com-

petitors.

These ideas

might

explain the similarity of RNAs producedunder the

optimal

conditionsand would also

predict

the

experimental

conditions

resulting

inthe denova

synthesis

of newRNAtypes:

(1) Suppression

ofST-RNA

growth (ex-

perimentsofFigs. 5and6) or (2) a

supply

ofchanged

oligo- nucleotide

sequences, e.g., caused

by

the omission ofonenu-

cleoside triphosphate in the preincubation phase (experiments of Fig. 4) or by theexternal additionofoligonucleotides (see Results).

Asshownbyourexperiments,oligonucleotides influencethe outcomeof the final RNAspecies. Though thepresence of a contaminating templatein

our.

enzymepreparationwas ruled out, the possibilityof a contaminationby anoligonucleotide still exists. Thispossibilitymight beanother basis toexplain the similarities of RNAspecies produced under optimalcon- ditions.

We wish to thank Dr. M. Eigen for his interest and support.

We are indebted to Dr. C. Biebricher for numerous valuable sug- gestions, to B.

Kuppers

for discussions, and Dr. P. Rawlings for correcting our English.

1. Kamen, R. (1970) Nature 228,527-533.

2. Kondo, M., Gallerani, R. & Weissmann, C. (1970) Nature 228,525a-27.

3. Blumenthal, T., Landers, T. A. & Weber, K. (1972) Proc.

Nat. Acad.Sci. USA 69, 1313-1317.

4. Wahba, A. J., Miller, M. J., Niveleau, A., Landers, T.A., Carmichael,G.C.,Weber,K.,Hawley,D. A.&Slobin, L.I.

(1974)J. Biol. Chem. 249, 3314-3316.

5. Haruna, I. & Spiegelman, S. (1965) Proc. Nat. Acad. Sci.

USA 54,

579-587.

6. Feix, G., Pollet, R. & Weissmann, C. (1968) Proc. Nat.

Acad. Sci. USA 59,

145-132.

7. Hori,K., Eoyang, L., Banerjee,A.K. &August, J. T. (1967) Proc.Nat.Acad.Sci. USA 57, 1790-1797.

.8. Banerjee, A.K., Rensing,U. &August,J. T. (1969)J. Mol.

Biol. 45, 181-193.

9. Mills, D.R., Peterson, R.L.& Spiegelman,S. (1967) Proc.

Nat.Acad. Sci. USA 58, 217-224.

10. Kamen, R. (1972) Biochim. Biophys. Acta262, 88-100.

11. Kamen, R.,Kondo, M.,R6mer,W.&Weissmann,C. (1972) Eur. J. Biochem. 31,44-51.

12. Sanger, F.,Brownlee, G. G. &Barrell,B.G.

(1965)

J. Mol.

Biol.13,373-398.

13. Mills,D.R., Kramer, F.R.&Spiegelman,S. (1973)Science 180, 916-927.

14.

Pace,

N. R.,Haruna, I.&Spiegelman,S. (1968)inMethods inEnzymology,eds.Grossman,L.&Moldave,K. (Academic Press, NewYork),Vol. 12,pp. 540-555.

15. Mehrota, B.B.&Khorana, H. G.

(1965)

J.Biol. Chem. 240,

1750-1753.

16. Gomatos, P.J., Krug, R. M. & Tamm, I. (1964) J. Mol.

Biol. 9, 193-207.

17. Krakow,J. S. & Karstadt, M\. (1967) Proc. Nat. Acad. Sci.

USA 58, 2094-2101.

18. Schachman, H. K., Adler, J., Radding, C.

M., Lehman,

I.

R.&Kornberg, A. (1960)J. Biol. Chem.235,3242-3249.

19. Radding, C. M. &Kornberg,A.

(1962)

J. Biol. Chem.

237,

2877-2882.

20. Lucas-Lenard, J. & Lipmann, F.

(1971)

Annu. Rev. Bio- chem. 40,409-448.

21. Ofengand, J. & Henes, C.

(1969)

J. Biol.Chem. 244, 6241-

6253.

22.

Shimizu,

N.,

Hayashi,

H. & Miura, K.

(1970)

J. Biochem.

(Tokyo) 67, 373-387.

23. Erdmann, V. A.,

Sprinzl,

M. &Pongs, 0.

(1973) Biochern.

Biophys. Res. Commun. 54,942-948.

24. Kornberg, A., Bertsch,L.L., Jackson,J. F.&Khorana, H.

G. (1964) Proc.

Nat.

Acad. Sci. USA 51, 315-323.

Proc. Nat. Acad. Sci.

USA

72