Electron microscopic studies of bacteriophage M13 DNA replication.

(1)

Electron Microscopic Studies of Bacteriophage M13 DNA

Replication

D. P.ALLISON,* A. T.GANESANt A. C. OLSON, C. M.SNYDER, AND S. MITRA BiologyDivision, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830

Received for publication13 June 1977

Intracellularforms of M13 phage DNA isolated after infection of Escherichia

coli with

wild-type

phage have been studied byelectron microscopy and

ultra-centrifugation. The data indicate the involvement ofrolling-circle intermediates insingle-stranded DNA synthesis. In additiontosingle-stranded,circular

DNA,

we observed covalently closed and nicked replicative-form (RF) DNAs, dimer

RF DNAs, concatenated RF DNAs, RF DNAs with single-stranded tails (a,

rolling circles), and, occasionally, RF DNAs with 0 structures. The tails in a

moleculesare

always

single stranded andare neverlongerthan the DNA from mature phage; the proportion of a to other RF molecules does not change

significantly

withtime after infection.Theoriginof single-strandedDNA

synthe-sishas beenmapped by electron microscopy at a unique location on RF DNA

byuse ofpartial denaturation mapping and restriction endonuclease digestion.

This location is between gene IV and gene II, and synthesis proceeds in a

counterclockwise directiononthe conventionalgenetic map.

M13 phage belongs to the group of

filamen-tous phage

specific

for male strains of Esche-richia coli and containscircular, single-stranded DNA (ssDNA) witha mass of2 x 106 daltons (26). Pratt and Erdahl (23) showed that there

are three basic stagesofM13 DNA

replication

after infection. These areanalogous tothose of ,X174, namely, (i) replication of

parental

ssDNAto

parental replicative-form (RF)

DNA; (ii)

synthesis

of progeny RF

DNA;

and

(iii)

synthesis

of progeny ssDNA from RF DNA.

Stageidoesnot

require

any

phage-induced

pro-teins; stage ii

requires

M13 gene 2

protein;

and

stage iii

requires

both gene 2 and 5

proteins

(23). Thegene 5

product,

asmall

ssDNA-binding

protein,

isessential forprogenyssDNA

synthesis

(19). After infection with M13 gene 5 amber

mutant(am5)

phage,

progenyssDNA

synthesis

is absent. Dressler (6) showed that

4X174

ssDNA

synthesis

occurs

by

the

rolling-circle (a)

model(10);Ray (25) showed

by

ultracentrifugal

analysis of

pulse-labeled

intracellularM13 DNA

that the virus strandis

continuously displaced

from the RF DNA and chased into progeny ssDNA. Our examination of intracellular M13 DNAbyelectronmicroscopyand ultracentrifu-gation indicated that ssDNA

synthesis

occurs

byaamechanism.This report includesa

partial

denaturation map of the a intermediates that

showsa

unique

origin

ofssDNA

synthesis.

The correlationof the

partial

denaturationmap and

t Permanent address: Department ofGenetics, Stanford University Medical School, Stanford, CA94305.

the genetic map establishes that the origin of

M13 ssDNA synthesis isin or near gene II and

that there is a counterclockwise direction of

replicationinreferencetothegenetic map.

Dur-ingthepreparation ofthis manuscript,Horiuchi

and Zinder (12) and Suggs and Ray (31)

inde-pendently showed the uniqueorigin and

direc-tion ofprogeny ssDNAsynthesis. Our dataagree

with theirs. A preliminary report of this work

has been

published

(D. P. Allison,A. T.

Gane-san,and S.Mitra, Fed. Proc.33:1491, 1974).

MATERIALS AND METHODS Bacteria andphage.E. coli K37 (fromD.Pratt)

wasusedasthewild-typehost. IsogenicE. coli K38 (su-), obtainedfrom P. Model,wasusedasthe

non-permissivehost.Wild-typeM13phageand M13am5,

obtained from D. Pratt, were further purified from

single-plaque isolates. The mutant phage stock had less than 0.1%revertants.

Growth media and buffers. Thegrowthand pu-rification ofphage and the extraction andpurification

ofphage DNA have been described(20).Thefollowing

buffers were used routinely: buffer A, 0.1 M NaCl,

0.05 M Tris-hydrochloride (pH 8.0), and 0.001 M EDTA; buffer B,1MNaCl,0.02MTris-hydrochloride

(pH 8.0), and0.002M EDTA;bufferC,0.01 M Tris-hydrochloride (pH8.0) and0.001MEDTA.

Chemicals andenzymes.Pronase(B grade), pan-creaticRNase, cytochrome c,and ethidiumbromide werepurchased fromCalbiochem,LaJolla,Calif. Be-fore use, Pronasesolution(10 mg/ml) inbufferAwas

self-digested at 37°C for 30 min, and RNase was heatedat100°Cfor10min. The[methyl-3H]thymidine

(6 to 20 Ci/mmol) was purchased from Amer-673

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

sham/Searle, Arlington Heights, Ill.,orNewEngland

Nuclear Corp., Boston, Mass., and carrier-free H3

31pO4wasobtained fromAmersham/Searle. Sarkosyl

NL97was a gift of Ciba-Geigy, and sodiumdodecyl

sulfatewas aBDH Chemicalsproduct. All other chem-icalswereof reagentgrade.

The purification anddigestion withE. coli exonu-cleaseI andHaemophilus influenzaerestriction en-donuclease (HindII) have been reported earlier (4). H.aegypticusrestriction endonuclease(HaeII)was a giftof R. diLauro of the Public HealthService, Na-tional Cancer Institute, Bethesda,Md. This enzyme wasused accordingtothemethod ofvanden Hondel and Schoenmakers (35).

Isolation of intracellular phage DNA. E. coli cultures growntolog phase (2X108 to 4 x108cells/per ml) insupplemented M9medium (23) at 370C were infected (multiplicity ofinfection, 100 to 200) with wild-type M13 and maintained at 370C or infected withM13 am5and maintainedat34WC, unless other-wise stated. The cultureswerelabeled with20

jLCi

of [3H]dThd per ml 8or 45minpostinfection and, after specified times, weredumped into equal volumes of chilledpoisontostopbacterialsynthesis. Thepoison was either buffer A containing KCN (10-2 M) and pyridine (5%) (15) or 75% ethanol, 21 mM sodium acetate (pH 5.3), 2 mMEDTA, and 2%phenol (22). The cellswerethen harvestedbylow-speed centrifu-gation and washed twice with bufferAcontaining10-2

MKCN. Sometimes the cellsuspensions were shaken in aVirTis homogenizer for 2min at low setting to removeunadsorbedphage particles.

Next, the cells were suspended in 1/10 to 1/20 volumeof bufferAandincubated in the presence of lysozyme (200

tg/ml)

at 0°C for 45 min (25). The cellswerethenlysed -with Sarkosyl NL97 (0.5%) for 15minat37°CanddigestedwithPronase (200 ,ug/ml) at 37°Cfor30to 60 min. In someexperiments lysis was accomplished by adding sodium dodecyl sulfate

(0.8%)tocellssuspendedin50mMTris-hydrochloride (pH 8.0)containing5mM EDTA in 10%sucrose(11). Finally, the lysatesweregently layered on top of20%

sucrosein buffer Bin anSW27 swinging-bucket rotor tube and centrifuged at 25,000 rpm for 3 to 4 h at 15°C inaBeckman-Spinco ultracentrifuge. The liquid wastakenfrom thetopwithout disturbing the bacte-rialDNA,whichsedimnented at or near the bottom of the tube. Afterphenol extraction in the presence of 1% sodium dodecyl sulfate and subsequent alcohol

precipitation,thephage DNAwasdissolved in buffer C.Occasionally,the DNAsamplesweretreated with RNase (50,tg/ml, 37°C, 1 h) before electron micro-scopy.

Ultracentrifugalanalysis. Equilibrium ultracen-trifugation in CsCl containing ethidium bromide (1) (4.2 g of CsCl and0.4mlof 5-mg/ml ethidiumbromide per4.0ml ofDNA) was carriedout in polyallomer

tubes in a no. 40 rotor of aBeckman-Spinco ultracen-trifuge. After 40 to 50 h ofcentrifugation at 35,000 rpm and23°C, 30 to 40fractions were collected from thebottom, and portionswereassayed for radioactiv-ity. Ethidiumbromide fromthepooled fractions was removedby repeated extraction with isopropanol (sat-uratedwith 30% [wt/wt] CsClin 0.2 M Tris-hydro-chloride[pH 8.0])followedby dialysis against buffer C. Forequilibrium ultracentrifugation in alkaline CsCl

(8),7.65gof CsClwasaddedto5.5mlofDNAsolution in40mMK2PO4(pH12.1) andcentrifugedina Beck-manno.40 rotor at35,000 rpm for42hat23°C.

Bandsedimentation in neutral sucrose (5 to 20% sucrose in buffer B) and alkaline sucrose (5 to 20% sucrose in 0.8 M NaCl, 0.2 M NaOH, and 2 mM EDTA) gradients was accomplished in a Beckman SW41swinging-bucketrotor.Totrapthefast-moving, covalentlyclosed, supercoiledRFtype I (RFI) DNA,

wesometimes useda0.5-ml cushion of 60% CsCl and 60%sucrose(wt/wt)atthebottom.

Radioactivity measurement.Theportionswere eithercounteddirectlyin aqueous mediuminaTriton X-100-toluene scintillation solvent systemorspotted

onWhatman3MMdisks and washed with 5% trichlo-roacetic acid, alcohol, and ether, in succession, to

remove acid-soluble radioactivity. The dried disk-were counted in toluene containing

2,5-bis[2-(5-ter-butylbenzoxazolyl)]thiophene (BBOT) (20).

Electronmicroscopy. Samplesof DNAwere pre-pared for electron microscopy by the Inman and

Schnos(13) modification of theprotein filmtechnique (16). Fractions fromsucrose or CsCl gradients were

dialyzedagainstbufferC for3h atroomtemperature. A0.07-mlsamplewasmixed with0.03ml of buffer(4 ml of 37%formaldehyde,0.32ml of1MNa2CO3,and 0.4 mlof0.126 MEDTA) adjustedbeforemixing to eitherpH8.7 (neutral) or11.1 (denatured) with 1 N HClor 5NNaOH,respectively. Afterincubatingfor 10minat25°C, the DNA-buffer solutionswere trans-ferred to an ice bath, cooled for 5 min, and mixed with an equal volume of formamide (Mallinckrodt,

99%).Cytochromec wasthen addedto0.01%, and the solutionswereincubated for10min at room temper-aturebeforespreading0.005mlontothe surface ofa waterdroplet. The DNA waspickedup on nitrocel-lulose films supported on 400-mesh grids, rinsed in 95% ethanol, stained withuranyl acetate (5), rinsed inisopentane, and rotary shadowed withplatinumat a 60 angle. Electron micrographs were taken on a SiemensElmiskop 1A, with magnifications calibrated byusingadiffraction gratingreplica (Fullam,54,864 lines perinch). Themicrographswereprojectedonto paper and traced, and the tracings were measured withacurvimetertothenearest 0.05,um.Underour

conditions,therewas nostatisticallysignificant differ-ence inthe contourlengths ofM13 circularssDNA

andRFIDNA;therefore, itwasnotnecessaryto use a correction factor to correlate measurements of ssDNA and double-stranded DNA.

Alignment of partially denatured rolling cir-cles.Thepartiallydenatured circles werealigned for maximumoverlapofdenaturedregions bya correla-tionmethod similar to, but computationally simpler than, that ofYounget al. (36). Computations were carriedouton aPDP11/40 computer. Each molecule was divided into 100 segments; each segment was assignedavalue ofzeroifthesegment was more than half native,orofone ifitwas more than half dena-tured.A moleculewasalignedtoanother by shifting one relative to the other a segment at a time, flipping it, and again shifting. The measure of overlap for eachrelativealignmentwas

F

(X,+

y,)2

i=1

on November 10, 2019 by guest

http://jvi.asm.org/

(3)

where X,and Yiare the values assigned to the ith segment of each of the two molecules. Before the alignment, themolecules were ordered by their extent ofdenaturation, from most to least. The alignment procedure considered the molecules in that order, aligning the second to overlap maximally with the first. The third was then aligned to the sum of the first two,the fourth to the sum of the first three, etc. When all of the molecules had been aligned in this way, asecond cycle, for refinement, was performed. In this cycle, each molecule was considered in the same order asbefore. Its contribution was subtracted fromthe sum; the moleculewasaligned with thesum of the remaining molecules and added back to the sumin thisnewposition. The differences between the results of the first and second cycleswerequite minor. Theresultingalignmentwasthen translated to bring it into correspondence with the picture of partially denaturedHind-cut, linear M13 RF DNA molecules.

RESULTS

Intracellular forms of M13DNA. The M13

DNA isolated from cells 8 min after infection with

wild-type

phage was spread directly for electronmicroscopyorsubjectedtoequilibrium centrifugation in CsCl containing ethidium bro-midetoseparatecovalently closedDNA(heavy band) from the rest (light band) (Fig. 1). Por-tionsfrom theheavy and

light

bandswerefreed ofdye,

dialyzed,

andthensubjectedtoelectron microscopy. The different intracellular forms of

M13 DNA found in the

heavy

and

light

bands

are shown in Fig. 2 and 3, respectively. The

structuresin theheavy bandfrom DNAisolated

8 min postinfection include unit-length (2

,Am)

supercoiled RFI (Fig. 2A), dimer-length

super-coiled RF

(<1%) (Fig. 2B),

and RFconcatenates

(ca. 5%) (Fig. 2C). We have

previously

shown thatwhenRFIDNA is

spread

with

formamide-formaldehyde,

it contains a

small,

denatured region as

depicted

in

Fig.

2A

(S.

Dasgupta et

al., J. Biol.

Chem.,

inpress)anddoesnotappear

supercoiled.

Ithas been

postulated

that the

di-mersandconcatenatesarise fromerrorsin

seg-regation of

daughter

molecules afteroneround of

replication (9).

We have found veryfew (2/

-1,000)

M13 molecules with a e structure like

those foundin the

replicating

intermediates of E.coli

(3)

orcolicin El

(14)

DNA. Themolecule shown inFig. 2D isacircle withtwoforks that

is 2

tum

when

only

one branch of the

double-stranded fork is included in the measurement. It appears possible that such structures, also observed by Ray (26),are intermediates in RF DNA

replication.

The light band from the CsCl gradient

por-trayed

in

Fig.

1 contains 70 to 80% RF type II

(RFII) DNA molecules (relaxed RF DNAwith one or more discontinuities in one or both

strands) (Fig. 3A),ssDNA circles (Fig. 3B),and a structureswithanssDNA tailattached to RF

8

-

6-4

21

0 10 20 30

FRACTIONNO.

FIG. 1. Equilibriumcentrifugationof intracellular M13 DNA in CsClcontainingethidium bromide. A 100-ml log-phase culture ofE. coli K37 grown in

supplementedM9 mediumat370Cwasinfectedwith

wild-typeM13for8min and thenlabeled with RH]-dThdfor90s.Afterisolationofthe totalphage DNA,

itwassubjectedtoequilibriumcentrifugationinCsCl containing ethidium bromide (both procedures as

describedin thetext).Fractionswerecollectedfrom

the bottom andmonitoredforradioactivity. Sedimen-tationwasfrom righttoleft.Thedensityoftheheavy

bandwas 1.60g/mlandthatofthelight bandwas

1.56g/ml.

DNA (Fig. 3C). The tailed molecules are

ex-pected intermediates for ssDNA synthesis on

circular, double-stranded DNA templates (10),

and the tails

generally

appear distinctly single stranded in ourelectron micrographs.

Dresser

(6) observed similar structures

during

4X174

DNA

synthesis,

and

Ray

(26) also detected such molecules in intracellular M13 DNA. Because

we never, in

examining

more than 200 a

mole-cules,

foundtails

longer

than2

Am

(unit length),

weconclude that the

single

strands of the

rolling

circlearecleaved as soon as

they

reachmature virus DNAlength. Figure 3Dshowsa molecule with two

single-stranded

tails attached to RF DNA; about 10% of the tailed molecules had twotails.

Figure 4 shows a a molecule spread at pH 11.1 in the presence of 50%

formamide,

condi-tions under which the double-strandedDNA is

partially denatured.That thefrequencyof such astructures didnotdecrease underthis condi-tion demonstrates thatthe ssDNA tails are in-deed

covalently

attachedtothedouble-stranded

RF DNA

molecules,

and the absence of any

675

on November 10, 2019 by guest

http://jvi.asm.org/

[image:3.500.303.395.74.297.2]

(4)

6 76 ALLISON ET AL.

"Z-:r

'.&N _V-6ft I, V' 0 _.0tv

-,-t IV a. 1%,.I -W, .%O%*4

AM ire.0 0 'k

-K.Z b, .16qq,

q;I Al: J,.^+d. I& W.a'.,its& 4V A

Aj V

10 , 4*., .1 P 'L, . ..

-W Oes. P

aIvv; la.

t,-Q,

"o.4% ₇_;,t %,e

v 4.

vp4A.c.r._Z" o6Aoi k W

-.4' % ,

dr- "a 4" a P.00

V %F- AI%

R

C4iDh- 19

y!"s 40 W'P

Ir U.

L",

is0. t.r

'B- lae

P f 'A VP

r r

A _VAOr. _S,

r-0.'aj

b' j4. -'.''. -*J-% qP1.*to

After

_I"

-161

-At'"".' #4 *..P.

.4 _V c.

L Ox _k ..v

IV e:. 4:

'.re

.3 'J t44S.,-. 01,

16n

P

..P & r

roll

cl t

OA kb

4? 4149g 06

V?4. V

V _j Ix

6,4

% _.4%

4, -J

If4 ₄₄

% % N v It 'fa r I.S190.11 It.

A Alr It 1,

1.VA

`V %1. N X%

6.4 4V.z r S

ZV

O

U, 4k A., I

0 A

f, Af rek

t

4_J.,4-' _rot _In ju .I, A

04

4W, If

_J4 'S

d jo P,

'J. _4.

IS_4. C' i

A

V 4 41 L. _. '%.,.F 'k,

P. A 400

.0 J A x.,I:

ile, 1.4- V'

?..-k.. :,;, I -11 '..,

..XJt 4

4, a

J A W'

k

T -v T J Qj. %

-6_A. TW A

5IA, 7_

`4 4.

Jk _A _-Jo, X*1.6 Le .6,

'k

41 S" 4 te

V _ok&;`

1k

V A.A q

A.010 F

x

V.

A.g

26t ft i

Al V', 4- 4

A.L U

_d A 'P .40

*il. K-W, 8 A%

_0IL

It _%

y/

.Ow,-J- IL

.0 _4- 4,

1.` 4.4? 44 exf'

.x Ak;.;.e. t

at

4;:

I--4 q,

4 1P

4t.

41

A y

A( rzf ₄

A V

%

..4.4.4 v

40N Z L 14

`X

-Z V

FIG. 2. Forms

of

intracellular M13DNA in the

heavy

band

represented

in Fig. 1. Supercoiled RFI

molecules

(A)

appearopenwhen

spread

with

formamide.

The small denaturedarea(arrow)isnearly always

presentwhen

supercoiled

moleculesare

spread

under these conditions.

Dimer-length, supercoiled

RF(B), concatenates(C),andafew Rstructures(D)arealsopresent. X57,560.

denaturedareasinthetailsisaclearindication adouble-strandeda

intermediate,

and

is,

in this

of their

single

strandedness.Nodouble-stranded

respect,

differentfrom that of

OX

174

(28).

tailswere seenamongafew hundredastructures AfterE. coli infection

by

wild-type

M13,

RF

Su

ejected

to denaturation. This

suggests

that

replication predominates

for the first 10 to 20

RF

replication

of M13 DNA doesnotoccurvia

min,

after which progeny ssDNA is the

major

on November 10, 2019 by guest

http://jvi.asm.org/

[image:4.500.75.448.77.551.2]

(5)

j44

4 4

'A 4~~~~~~~~~~~~~~~4

~ ~ ~A

d.~~ ~ ~ ~ ~ ~ ~

4h-~ ~ ~ ~ ~

I4

S5

JP~ ~ ~

-a4

pS,L4~,

W, '60a)

4&&t3

W44

4~~~~~~~A

10~~ ~ ~

'S~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.I

ir

~ ~ ~ ~ le

,6

al

^ssw.x.- - 4. M

FIG. 3. Formsof intracellular M13 DNAin thelight band represented in Fig. 1.(A) Relaxed RFII circles thatneverexhibit thesmalldenaturedareafound in the supercoiled circles. Unit-length, single-stranded

circles (B)are also present, as wellascircular RFmolecules withsingle-stranded linear tails (a). Most of

these molecules exhibit a single-stranded tail attached to a double-stranded circle (C). However, some

moleculesarefoundwith twosingle-strandedtailsissuing fromthe RF molecule(D). x57,560 formsynthesized (26). However, wesaw tailed

moleculesas early as8 min after wild-type in-fection(Fig. 3),and thefrequencyofthesetailed moleculesinthe total RF DNApopulation

ap-pearedtobeindependentof thetimeafter

infec-tion(Table 1).WethereforeagreewithForsheit etal.(7),who showedthatsomeprogenyssDNA

synthesisoccursveryearlyafterinfection. We found no tailed molecules in DNA ex-tracted from E. coli K38 10 to 15 min after

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.500.69.419.71.558.2]

(6)

678 ALLISON ET AL.

FIG. 4. Partially denatured RF DNA with tail. DNA isolatedfromthe lightbandrepresentedinFig. I was adjusted topH11.1 for 10min

before spreading

for electron microscopy. Two denatured sites are evidentonthe circular RF molecule. The single-strandednessofthe tailcan be inferred because the tail doesnotcontain any denatured sites and appears less thick than the RF molecule. xl10,000.

TABLE 1. PercentageofRF molecules with

single-stranded tailsearlyand latepostinfectiona

Expt Early Late

1 37/237(16)b 47/247(19)

2 4/38 (11) 5/41(12)

aLog-phase E. coli cultures(100ml)wereinfected withwild-type M13phage (multiplicityofinfection,

-100).Total intracellularDNA,after removal of the host DNA as described in the text, wasspread for electron microscopy after RNase treatment and

de-proteinization. "Early" and "late" refer to 8 and 45 minpostinfectionat

370C.

bNumber inparenthesesrepresents percent.

infectionwith M13 am5

phage

inthreeseparate

preparations, scoring approximately 500 RF

molecules. ThepresenceofM13 gene 5protein,

therefore,

appears to be essential for the exis-tenceof thea structure.

Ultracentrifugal

studies of M13 ssDNA

synthesis.

Ultracentrifugal analysis

and

detec-tion of

longer-than-unit-length, pulse-labeled

DNA of virus-strand type led Ray (25) and Tseng and Marvin (33) to propose that M13 ssDNAsynthesisoccursbya

rolling-circle

mech-anism.To

complement

ourelectron

microscopic

evidencefor the involvementofa structuresin M13 ssDNA

synthesis

(to be

presented

later),

we extendedtheearlier

ultracentrifugal

studies

by

investigating

DNAfrom bothwild-type and

am5

phages.

Band sedimentation analyses in

neutralsucrose ofboth

pulse-labeled

M13 wild

typeandam5DNA

early

andlate postinfection showed that most of the label was present in

RFI and

RFII,

withverylittle orno incorpora-tioninto maturessDNA, inagreementwithRay

(25) (datanot

shown).

After infection with

wild-type phage,a large fraction (70 to 80%) ofthis

label can,

however,

be transferred to ssDNA after a chase (4, 25). In contrast, inM13

am5-infected

cells,

inthe absence of ssDNA synthesis, theeffect ofasimilar

pulse

andchasewasonly

totransfer label from RFIItoRFI(S.Dasgupta

and S.

Mitra,

inpreparation).

A more revealing setof experimentalresults wasobtained from extendedband sedimentation ofpulse-labeledDNAs inalkaline sucrose (Fig. 5) whereRFI DNAsediments into the cushion atthe bottom whileRFII DNA separates into

unit-length linear single strands, circularsingle

strands, and longer-than-unit-length

single

strands. It is evident by comparing with the

markerphage DNA that inboth

early

(A)

and

late(B) postinfectionwithwild-type phage,the

label is distributed

bimodally

into

unit-length

linearDNA(16S) anda

faster-sedimenting

DNA with a sedimentationcoefficientof ca. 20S. The

faster-sedimentingDNAdidnotcorrespond

ex-actly with, but contained material that moved

on November 10, 2019 by guest

http://jvi.asm.org/

[image:6.500.118.403.73.310.2]

(7)

ORIGIN OF M13 SINGLE-STRANDED 679

1)20

30

1020

30 10 2030 40

FRACTIONNO

FG5.Bandsedimentation in alkaline sucrose of pulse-labeled wild-type andam5M13 DNA. Twenty-milliliter log-phasecultures of E. coli K37 (A and B) and K38 (C) infected with wild type (A and B) and

am5phage (C) werepulse-labeled with 20 ,uCi of

[3Hlthymidineper ml as described in the text. (A) 30-spulse-label at37°C,8minpostinfection; (B) 30-spulse-label at37°C,45 minpostinfection; (C)1-mmn pulse-label at25°C, 3mi after shift-down to25°C from37°C,13 min postinfection. After termination of labeling, the infected cells were washed and lysed, and the bacterial DNA was removed by centrifuga-tion as described in the text. Portions ofthe DNA,

alongwith '4C-labeled M3phageDNA,were

centri-fuged in 5 to 20% alkaline sucrose gradients in a Beckman SW41 swinging-bucket rotor at 5°C at 39,000rpm for 11 h (A, B) or38,000)rpm for 12 h (C).

Symbols:*, 3H; C. The positions of circular and

unit-length, linearM13DNAareindicatedby C and L,respectively.RFI DNAsedimentinginto the

cush-ionis indicated by a dashed line.

fasterthan, the circular phage DNA.Incontrast, DNA extractedfrom M13 am5-infected culture

(C) early postinfection sedimented mainly as

unit-length linear DNA in alkaline sucrose, with

little labeled material sedimenting faster than circular DNA. Asignificant amount of label is

also in the RFI DNA. Figure 6shows that

fol-lowing the chase of pulse-labeled DNA after

wild-type M13 infection, the label from both

unit-length linear DNA and the faster-moving DNA(20Speak) isin the DNA that sediments

exactlywith circular DNAmarker.

These results can easily be explained by the

situation depicted in our electron micrographs, where longer-than-unit-length strands (i.e.,

cir-cles with tails) are clearly evident. If these are

indeed intermediates insSDNA synthesis, any

labelthey contain should be chased into mature

circular DNA during continued synthesis with unlabeled thymidine. This proposition was

testedintwoways:First, thepulse-labeled

frac-tions containing longer-than-unit-length linear

DNA (Fig. 6A) and the corresponding chased

fractions (Fig. 6B) wereseparately pooled. The

DNAs werethen tested for susceptibilitytoE.

coli exonuclease I, which attacks only ssDNA

from the 3' end. Circular phage DNA is

com-pletely resistant to this exonuclease (20). The majority of the label from pulse-labeled DNA issusceptible to the exonuclease (Table 2), in-dicating that most ofthe DNA sedimenting in theposition of circular DNA is, in fact, longer-than-unit-length linear DNA. The majority of the DNA isolated afterchasing is, asexpected,

resistanttotheexonuclease and, therefore,

cir-u

Il

.

18

-A

16 C

8-CL

L12

6-J~~~~~~~~~~~~

04 1

FRACTIO NO.

E)

Cli

Q

a.

FIG. 6. Bandsedimentation in alkaline sucrose of wild-type M13 DNA afterpulse and chase. E. coli K38infected withphagewaslabeled with

f3HjdThd

for30 s 45minpostinfection at370C.One half of the culture(A)wasstoppedwithethanol-phenolmix (22), while the other(B)wastreated with500,agof unla-beled dThd per mlfor10minbefore additionof the stopper. Furtherprocessing and centrifugation are described in thelegendtoFig. 5C, except M13virion

[image:7.500.50.235.53.215.2]

/32P]DNAwasusedasthe internal marker.

TABLE 2. Susceptibilityof alkali-denatured,

pulse-labeledM13DNAtoexonuclease Ia

DNAmade acidsoluble by exonuclease I(%) Type of DNA

[3H]DNA Control [32p]_

DNA After pulse-labeling (Fig.

7A) 74.4 91.2

After chase(Fig. 7B) 35.6 89.8

aThe fractions 8 to19and 10 to 17inFig.6Aand B,respectively,werepooled, quickly neutralized,and alcoholprecipitatedinthe presence of0.3Msodium acetate. Portions (-4 x 103 cpm) of the dissolved precipitate were mixed with 75 pmol ofpancreatic

DNase-treated and denatured 32P-labeled M13 RF DNA (4.7x 103cpm) (G.LavelleandS.Mitra,inP. Tattersall andD. Ward, ed., Parvoviruses, inpress)

and thendigestedwith0.6U of E.coliexonucleaseI in 0.3 ml at37°Cfor30min(20).

on November 10, 2019 by guest

http://jvi.asm.org/

[image:7.500.255.442.205.358.2]

(8)

12-- 8-I

I

4-0

0 20 40 20 40 20 40

FRACTION NO,

'4

6

k8

FIG. 7. Equilibrium centrifugation in alkaline CsCl. Pooled fractions of unit-length, linear and longer-than-unit-length M13 DNApooledfrom the alkalinesucrose gradientof Fig. 5were banded in

alkaline CsCl asdescribed in the text. (A) Longer-than-unit-length ssDNA from fractions 13 to 16of Fig.5B; (B) unit-length,linear ssDNAfromfractions 17to20ofFig. 5B;and(C)unit-length, linear ssDNA from fractions 18to22ofFig. 5C. Symbols:*, 3H; 0, `4C marker phage DNA. (v) and (c) denote the position ofviralandcomplementary strands,

respec-tively.

cular. Second, equilibrium centrifugation in

al-kaline CsCl (Fig. 7) was used to determine the

strandednessof(B) unit-lengthand (A)

longer-than-unit-length linear DNA samples from pulse-labeled DNA late after wild-type phage infection (pooled from fractions in Fig. 5B) and of (C) pulse-labeled, unit-length linear DNA after M13 am5 infection (Fig. 50). The pulse-labeled linear DNAfromM13am5-infected cells contained bothviralandcomplementary strands inanequimolaramountaftercorrecting for dif-ferences in thymine content (26), whereas the label in the DNA fromwild-type M13-infected cells late postinfection contains predominantly viral strands, which is in agreement with the results ofSuggs and Ray (31).

Origin of ssDNA synthesis on RF DNA

template.Alltheexperimentsdescribedsofar

indicate that progeny M13 ssDNA synthesis,

like that of 4X174 DNA, occurs bya a

mecha-nism where the 3' end of the growing virus strand displaces the 5' end of the same strand on the complementary strand template of RF

DNA. If the synthesis starts at a fixed site on

theRFtemplate,itshouldbepossibletolocate the sitebyvisually"folding"thesingle-stranded

tail backonthecircularpart oftheastructure

shownin electronmicrographs. The basic prob-lemofsuchamethodlies inestablishingapoint

of reference in the circular RF DNA, but we

circumvented thisdifficultybytakingadvantage of the site of cleavage by HindII restriction endonuclease on M13 RF (34). This enzyme

does not attack M13 ssDNA (2). Total phage DNA wasextracted from 100mlof E. coli K37

culture in supplemented M9 medium, 45 min

after infection withwild-type M13. After

treat-ment ofthe DNA solution with Hind ertdonu-clease (4), the DNA was spread for electron microscopyinthepresenceof formamideatpH

8.7 asdescribed in Materials and Methods. We were thus able to observe linear RF DNA (2

,Am long) with or without single-stranded tails

of various lengths attached to the double-stranded DNA. Thelengthof the tailwas

mea-sured in both directions from the three-point junction, giving two possible locations for the origin ofssDNAsynthesis.Assuming there isa

unique origin of synthesis, only oneof the two

sites from each molecule, inanarrayofdifferent molecules, would becommontoall.Becausethe "right" and "left" ends of the double-stranded RF DNA could not be distinguished, we could

only establish the distance of the origin from the Hind cleavage site. Figure 8 represents a

collection of molecules with the two possible origins,oneindicated foreachmoleculeoneither

side of thethree-point junction. It isquite clear thatonlyoneregion, correspondingtoabout 0.2 ,um from the Hindcleavage site, iscommonto mostof themolecules. The direction of replica-tion can be established in a relative sense in

H1DCUT SITE

0.2

b 140.14

:1 0.16

*|z~ ~ ~ ~ ~ ~ ~ ~ ~~~~~-30.2

.@ z 0~~~~~~~~~~~~~~~~.34

I$&Ie 0.45

* I6*:~0.49

**.- m4058E

a- -z 0is.63

a- '0.65

I.00

1.09

130 1.32

1.36

1.45 1.49

1.54

163 1.80

2.0

1.0

[image:8.500.65.254.52.175.2]

MOLECULARLENGTH(I.m)

FIG. 8. Mapping ofsingle-stranded tailsonlinear

RF DNA. The total viral DNA waspurified from

K37cultures,maintainedat370C,45minafter infec-tion withwild-typeM13asdescribed inthetext.After digestion ofthe DNAwithHindendonuclease,itwas

spreadforelectron microscopy asdescribed in the text.The"Y"-shapedmoleculesarising from a

struc-tureswerephotographed. The double-strandedarms were measured and normalized to 2 wun, and the length ofthe single-stranded tailwas interpolated

onto the double-stranded arms. The two possible origins ofssDNA synthesis thusobtainedare

indi-cated by arrows on the lines representing the RF DNAmolecules. The moleculesareorientedsothat,

ofthe twopossible origins, the one closest to the Hindcutislocatedontheleft. Thenumbersatthe endofthelinesindicate taillength.

A B

8-2 4'

:I2

2c

I

IC

I ~

~~~~~.

--I

0

_ * a0.31

_ b W&Os

b~~~~~~~~~~~~~~II-OS

I- 0.54

_ * 0. 56

_. _ 0.6 "

4 I V0.7X)Z

** I0.74'

*4 * 0.75

P * 0.81

SI II .0.91

1

4 0

0 4

4 b

-6 b

4, b

4 0

4 p

on November 10, 2019 by guest

http://jvi.asm.org/

[image:8.500.270.455.348.497.2]

(9)

681

that the DNA chain grows either toward or away from the center of the double-stranded template. However,oneproblem thatmay com-plicate such a determination is branch migration

occurring in vitro (17) after replicating DNA

has been freed ofgene 5 protein, which in vivo covers up the peeled-off 5' end of the strand (24) and preventsbranch migration. Branch

mi-grationin DNAwould,ineffect,apparently re-verse the direction of synthesis. In fact, we can see both directions ofchain growth in Fig. 8. However, we may assume that the possibility of branch migration is significantly reduced in

moleculeswithlonger single-strandedtails, and thesemolecules (showninthelower half of the array) clearlyindicate that the direction ofchain growthisaway from the center of themolecules. To establish the absolute direction ofchain

growth and relate it tothe genetic map ofthe

phage,weutilized thepartial denaturation map-pingtechnique ofInman and Schnos (13). The

partial denaturation map (Dasgupta et al., J.

Biol.Chem., inpress) oflinear RFDNA

gener-ated by Hind endonuclease cleavage was then

related to the HaeII restriction endonuclease

map (35) that had already been related to the

genetic map of thephage (18). We utilized the

fact that glyoxal predominantly fixes a region in supercoiled M13 RF DNA that corresponds

to the major denaturable region and digested glyoxal-treated M13 RFI DNA with HaeII, whichcutsthe DNA into threepieces of3,500,

2,600, and320base

pairs,

denotedasA,

B,

and

C, respectively

(35).A

comparison

of theHaeII

restriction map and the

partial

denaturation map of Hind endonuclease-generated linear DNA(Fig. 10) shows that the

glyoxal-fixed

bub-ble

corresponding

tothe

major

A-T-rich

region

wouldbe located in the B

fragment

if the

ori-entation ofthedenaturation mapwerethesame

asthat of the conventional restrictionmap

(i.e.,

with theintragenic region locatedneartheHind site inaclockwise direction

[35]

or onthe

right

side of the Hind site in a

similarly

oriented linear

representation). Alternatively,

thebubble wouldbe in the A

fragment

ifthe molecules in the

partially

denatured map were oriented in

the

opposite

way. Out of 43 DNA

fragments

scored, 12 contained

bubbles,

and 11 of these (>90%) fell in the size range

corresponding

to

2,300 to2,500basepairs.The

bubble-containing

fragments, therefore, clearly correspond to

HaeIIB

fragments.

This establishes the orienta-tionof thepartial denaturation maprelativeto

theHaeIIrestriction map

and, by extension,

to

the

genetic

map.

Toestablishtheoriginand direction of ssDNA

synthesis, we

partially

denatured circular RFII

molecules

(Fig.

4)

by

raising

the

pH

to11.1 for

10minbefore spreading forelectron microscopy.

Molecules with single-stranded tailswere

mea-sured so that both the position and length of the denatured areas and the single-stranded tails

could be relatedto a convenient, but arbitrary,

starting point on the circular molecules. The data were fed into the computer, where the molecules were normalized to 2

pm

and aligned

for maximum overlap of denatured regions as described in Materials and Methods. We then

asked the computer to translate these data to

correspond to the partial denaturation profile of RFImoleculesthathadbeencleaved by the

site-specific HindII endonuclease before

dena-turation. TheHindII-cut moleculesrepresented

in Fig. 9A are plotted so that the site of the

predominant denatured area is in the left half

of themoleules; thearrayof partially denatured RFII molecules with single-stranded tails

ar-ranged by thecomputer arerepresented inFig. 9B. When the single-stranded tails ofvarious

lengths are superimposed onto these partially denatured RF molecules, they indicate a com-mon origin of replication. If the best-fit data plotted by thecomputer areused, outofthe39

moleculesshown, 31 have a common origin lo-cated 0.16,m (standard deviation = 0.12

tum

out of a total length of 2.0 ,um) or about 8%

from the left endofthemap withthedirection

of chaingrowthaway fromthecenter.Weshould point out that eight molecules (indicated by dotted circles around the two possible origins

andthe tail inFig. 9B) do notfitourresultsas

plotted. However, sinceweasked thecomputer to align the molecules by the best fit of the denatured areas without

considering

where the

origin of

replication

would belocated and since we weredealing withasmall molecule with few

denatured sites, we feel that these results are

quite good. If the 8 molecules that do not fit

wererotated 180° aboutthe major denaturation

area located in theleft half of the

molecules,

5 more molecules

(asterisk,

Fig. 9B) would have

an origin

coinciding

with that of the other 31

molecules.

Acomparisonwith thegeneticmap

(Fig.

10)

shows that this

origin

of ssDNA

synthesis

is

between gene IV andgene IIofthe

phage

and

that

replication proceeds

ina counterclockwise

direction. Because both DNA and RNA chains grow in the 5'-+ 3' directionand because both

M13mRNAandprogeny ssDNA

syntheses

oc-cur on the complementary

strand,

one would expect the direction of DNA

synthesis

to be

counterclockwise, asis RNA

synthesis

(21).

DISCUSSION

The precise mechanism of

synthesis

of prog-eny M13 RF and ssDNA isnotyet

completely

on November 10, 2019 by guest

http://jvi.asm.org/

(10)

682

ALLISON ET AL.

)ERIVED HIND I SITE B

MOLECULARLENGTH (elm)

TAILLENGTH(Im) 0431 0 36

*--- 043

.053 05 0 56 0530

0601 0666 065

-.-075

-0

0-

786

66 081 09 91

--94

9

096

.02'

108

- i-4₁₁₄₁111

116

26 1.27

-1-35- 1.7

171

81-l881.

2 00 2

inprogeny RF DNA

synthesis (26),

and the lack ofrequirement for many of the dnafunctions inprogenyssDNA

synthesis

(4, 20,

26)

suggest that different

replication

complexes

are respon-sible for the three different stages ofintracellular M13 DNA

replication.

.4X174,

which also

con-tainscircular ssDNA asthe genome,

produces

intermediatesinDNAreplication thatare

anal-ogous to thoseofM13 (26). Dressler(6) showed that progeny ssDNA in

4)X174

is

synthesized

via the a intermediate

by

displacement

of the 5' end of the viral strand in RF with a

newly

synthesized 3' end of the same strand. Since then,

Schroder

and Kaerner (28) have shown that progeny RF DNA

synthesis

occurs

by

rep-lication of the

displaced single

strand in the a

intermediates.It isobvious thatprogeny ssDNA

is synthesized most economically by a strand displacementmechanism,

i.e.,

one

involving

aa

structure. On the other

hand,

RF DNA could besynthesizedinthesameway orviaacircular replicating intermediate, such as the structure

observedinothersystems(3, 14).Although this manuscript deals with theorigin of ssDNA syn-thesis, our

negative

findings

bring

out some

in-teresting

possibilities

regarding RF synthesis, which we will discuss first. Evidence based on

studies with wild-type phage has been used to

IHINDICLEAVAGE SITE

FIG. 9. Denaturation maps of M13 RF molecules and astructures. (A) M13 RFI DNA isolatedfrom

theheavy band described inFig.1wascutwith Hind endonuclease andpartially denaturedforelectron microscopy asdescribed in thetext. Themolecules, indicatedbyhorizontal lines(thedenaturedregions are shown by open blocks),are aligned so that the major asymmetrically located, denatured region is in the left half ofthe molecules. (B) Circular RF molecules with tails(astructures),presentinthe total phage DNA isolatedfromE. coli K37,45minafter infectionwithwild-typeM13 andpartiallydenatured asdescribed in the. text, were aligned by computer for the bestfitofthedenatured areas, and transposed by computertofit theprofile of the Hind-cut mole-cules. The location and directionof the arrows indi-catetheoriginanddirectionof replication, whereas the small vertical lines indicate theposition of the tail. Molecules that have bothorigins and tails cir-cled donothave anorigin that fitsourconclusion. Theasterisk indicates moleculeswhoseorigins would fitif the moleculeswererotated1800aboutthemajor denaturedarea.

understood. The lack ofrequirement for host

dna functions other than that of dnaE (DNA polymeraseIII) inparentalRFDNAsynthesis (27), therequirementforseveral dna functions

HAE n CLEAVAGE MAP

6 9 | . 3

A B C A

GENETIC MAP

IR. z

v3' ' i . a .... ..

1 I s I I ."

EARLYMELTINGREGIONS

ORIGIN OF ss DNA SYNTHESIS v3' a

[image:10.500.64.251.60.395.2]

c5'

FIG. 10. Origin ofssDNAsynthesisonM13phage

genetic map. The HaeII restriction endonuclease cleavagemap ofRF DNAwasusedtocomparethe

genetic map and thepartial denaturation map as

described in the text. The origin and direction of ssDNAsynthesisindicatedbythearrow wasobtained

from Fig.9. (v)and(c) correspondtothe virus and thecomplementary strand,respectively. Thegenetic map and HaeII restriction map are redrawn from reference35.

t L

> -=_

K _:__

* C= _-1 I ah

h -L_

__n

.oC

2S.=

_ * _ _.

_ g

*__

An==

_=D {

£ -_ L_

=E he

C

5,.

3

:5

3s

l-

J. VIROL.

1.

v

c .j

on November 10, 2019 by guest

http://jvi.asm.org/

(11)

indicatethat RF replication in M13 also occurs

by a a intermediate (29, 33), but our data do not support this model. We followed progeny RF DNA synthesis in the absence of progeny

ssDNAsynthesisbyinfectingsuppressor-freeE.

coliK38withM13am5,in whichRFreplication continues for at least 50 min at

340C

in the

complete absence ofprogeny ssDNA synthesis

(S. Dasgupta andS. Mitra, unpublished data).

Under these circumstances,we found no a struc-turesof RF among several hundred DNA

mol-ecules scored by electron microscopy and no

pulse-labeled, longer-than-unit-length viral ssDNA afterbandsedimentationinalkaline su-crose(Fig.5C). Thesefindingssuggest that M13

RF isnotreplicatedin the same way asprogeny

ssDNA. Furthermore, the presence of DNA from both strands of RF DNA in both

unit-length linear and smaller-than-unit-unit-length frag-mentsin nascentRFDNA afterinfection with am5 phage (Dasgupta and Mitra, unpublished data) supportsamodel of discontinuous

synthe-sisof both strandsin RFsynthesis. In fact, we

did see a few Cairns-type 0 structures, which may be replicating intermediates of RF DNA.

Their paucity in the DNA population could result from the factthat the replication time for RF is expected to be in the order of seconds,iftherateofDNAchain growthinRF

is comparable to thatin the host. Hence, it is

quite possible thatmost of the molecules

com-pletearound ofreplicationevenwithin thetime

thepoison (usedtostopthepulse label) inhibits DNA

synthesis.

Alternatively, the replicating RFmoleculesmaybeselectively lost either by bindingvery

tightly

tothemembranouscomplex (29) orby existingin a structurethatcannotbe

spread

properly

forelectronmicroscopic visual-ization.

Finally,

the lack of a structures with double-stranded tails after infectionwith

wild-typephagemakesit

unlikely

that the a structure isinvolvedin M13RFreplicationinthemanner

postulated for 4X174 (28).

On the other

hand,

our

ultracentrifugal

anal-yses of

pulse-labeled

M13 DNA confirm the a

modeofprogenyssDNA

synthesis

suggested by Ray (25). His

experiments

and those ofTseng and Marvin

(33)

indicated the presence of

longer-than-unit-length, pulse-labeled

strandsin thevirusstrand ofRF DNA after infectionwith wild-type phage.We showedbyultracentrifugal and

enzymatic techniques

that both

longer-than-unit-length and unit-length DNA derived

from pulse-labeled RFII DNA are of the viral-strand type andareindeedlinearbythe criterion of theirsusceptibilitytoexonuclease I. Further-more, afterachase,the labelcanbetransferred

intocircularDNA.These

results,

inagreement with the

rolling-circle

model of M13

ssDNA synthesis, are further supported by elec-tronmicroscopic studies.Many astructureswith

single-stranded

tails covalently attachedto RF

moleculeswerereadily observed in the electron microscopeand constitutedasignificant

propor-tion ofthe total intracellularRF pool. The

oc-casional presence of RF molecules with two

single-stranded tails was puzzling at first, but

couldeasilybeexplained by incompletebranch

migration in vitro (17). That the two tails are

always attached to the RF in close proximity and that there isalwaysasingle-stranded region

on the RF molecules between these points of attachment support such apossibility. It is ob-vious that branch migration is prevented in vivo,

wherethedisplaced singlestrand of a structures

iscovered bygene 5protein (24).

Wehave also observed severalinteresting as-pectsof the control ofssDNAsynthesis. The a

modelpostulatesacontinuouselongationof the

tail witha5'end viadisplacement bythe grow-ing3'end of thevirusstrandonthe

complemen-tary strand template (10). After one round of replication, the single-stranded tail is cleaved

and circularized into mature progeny ssDNA. We have firstof all shown here that the tailin the a structures is neverlongerthan unitlength, which indicates an efficient cleavage reaction

independent

of the DNA chain growth.

Sec-ondly,thefraction of RFDNAparticipatingin

ssDNA

synthesis

as

rolling

circles does not

change

significantly

from early posinfection, when a very small amount ofprogeny ssDNA

is

detected,

tolate

postinfection,

whenprogeny

ssDNAconstitutes the

bulk

of thephageDNA

synthesized.

The amountof gene5

protein

also

increases

significantly

late

postinfection

(26); therefore, ourdata suggest that gene 5

protein

stimulates the

synthesis

of viral DNA via a

structuresandthuscontrolssuch

synthesis

ina

positive manner

originally

postulated

by

Stau-denbauer and Hofschneider (30). Mazur and

Model(19) showedthat gene5

protein

controls M13 ssDNA

synthesis

in a

negative

fashion

by

preventing the ssDNA,

complexed

with gene 5

protein, from

acting

asa

template

for RF

syn-thesis.

We have established that the

unique origin

of M13 ssDNA

synthesis

is located about 8%

from theHind

cleavage

site,

which

corresponds

to a location in

HpaII

fragment

F (34) and

HaeIIIfragment G

(12),

and that the direction of itsreplicationiscounterclockwiseonthe

con-ventional

genetic

map. Tabak et al.

(32)

have

already established that the

origin

of

parental

RF DNA

synthesis

in vitro is also located in

HpaIIfragment

F;

this

region

isthe location of the

intragenic

space between genes IV and II (12).

on November 10, 2019 by guest

http://jvi.asm.org/

(12)

During the preparation of this manuscript, the establishment ofthe originof ssDNA

syn-thesis in M13phagewasreportedbytwo labo-ratories thatrelied uponthegradientofpulsed label in different regions of RF during

ssDNA synthesis (12, 31). Our results agree

re-markably well with thesereports,although our investigationwasentirelydifferent inapproach,

being based on physicalmapping of the origin andrelying on thea structure. The agreement

therefore provides positive evidence for the

in-volvement of a structures as intermediates in

ssDNA synthesis.

ACKNOWLEDGMENTS

This researchwassupported by theEnergyResearch and DevelopmentAdministration undercontractwiththe Union CarbideCorporation.A. T.Ganesanwassupportedby Public Health Service grant GM108andResearch Career Develop-mentAward 6M 50199 fromtheNationalInstituteofGeneral Medical Sciences.

Wethank Letha Oggs for her competent technical assist-anceandR.diLaurofor thegiftofHaeII restriction endonu-clease.

LITERATURE CITED

1. Bauer, W.,andJ.Vinograd.1971.Theuseof interca-lativedyesinthestudy of closed circular DNA. Prog. Subcell. Mol. Biol.2:181-215.

2. Blakesley, R. W., and R. D. Wells. 1975. Single-stranded DNA from OX174 and M13 is cleaved by certain restriction endonucleases. Nature (London) 257:421-422.

3. Cairns,J. 1963. The chromosome ofEscherichia coli. ColdSpring Harbor Symp.Quant. Biol. 28:43-46. 4. Dasgupta,S., and S. Mitra. 1976. The role of

Esche-richia coli dnaG function incoliphageM13DNA syn-thesis.Eur. J. Biochem.67:47-51.

5. Davis,R.W.,and N. Davidson. 1968.Electron micro-scopevisualization ofdeletion mutations.Proc. Natl. Acad. Sci. U.S.A.60:243-250.

6.Dressler,D. 1970. Therollingcircle forOX174replication.

II.Synthesisofsingle-stranded circles.Proc.Natl. Acad.

Sci.U.S.A.67:1934-1942.

7. Forsheit,A.P., D. S.Ray,and L.Lica.1971. Replica-tionofbacteriophageM13. V.Single-strand synthesis duringM13 infection. J.Mol. Biol.57:117-127. 8. Fujimura, R. K., and E. Volkin. 1968. Biochemical

analysis of the naturally repaired sections of bacterio-phage.T5 deoxyribonucleicacid.I.Bromodeoxyuridine incorporationintoparentaldeoxyribonucleicacid in the absence ofdeoxyribonucleicacidreplication. Biochem-istry7:3488-3498.

9. Gefter, M. L. 1975. DNAreplication. Annu. Rev. Bio-chem.44:45-78.

10. Gilbert, W.,and D.Dresser. 1968.DNA replications: the rolling circle model. Cold Spring Harbor Symp. Quant. Biol. 33:473-484.

11. Godson,G. N. 1973. Asimplemethod of preparing large

amountsofOX RFIsupercoiled DNA. Biochim. Bio-phys.Acta299:516-520.

12. Horiuchi, K.,and N. D.Zinder.1976.Origin and direc-tionofsynthesis ofbacteriophageF1 DNA. Proc. Natl. Acad.Sci.U.S.A.73:2341-2345.

13. Inman,R.B.,and M.Schnos.1970.Partialdenaturation ofthymine and 5-bromouracil-containing A DNA in alkali. J.Mol.Biol.49:93-98.

14. Inselburg,J., and M. Fuke. 1971.Isolationof catenated and replicating DNA molecules ofcolicin factor El from minicells. Proc. Natl. Acad. Sci. U.S.A. 68:2839-2842.

15. Jacobson,M.K.,and K.G. Lark. 1973.DNAreplication

inEscherichiacoli:evidencefor twoclassesof small deoxyribonucleotidechains. J. Mol. Biol. 73:371-396. 16.Kleinschmidt, A. K., D. Lang, D. Jacherts, andR. K.Zahn. 1962.Preparationandlengthmeasurements ofthetotaldeoxyribonucleicacid content ofT2 bacte-riophage. Biochim. Biophys.Acta61:857-864. 17. Lee,C. S., R. W.Davis, and N. Davidson. 1970. A

physical study by electron microscopy oftheterminally

repetitious, circularly permuted DNA from the

coli-phage particles ofEscherichia coli 15. J. Mol. Biol. 48:1-22.

18. Lyons, L. B.,and N. D. Zinder. 1972. Thegeneticmap of thefilamentousbacteriophage Fl. Virology 49:45-60. 19. Mazur, B.J., and P. Model. 1976. Regulationof coli-phage F1 single-stranded DNAsynthesisbya DNA-binding protein. J. Mol. Biol. 78:285-300.

20. Mitra, S.,andD.R.Stallions. 1976. Therole of Esche-richiacolidnaAgeneanditsintegrativesuppression in M13 coliphage DNA synthesis. Eur. J. Biochem. 67:37-45.

21. Model, P.,and N. D. Zinder. 1974. In vitrosynthesisof bacteriophageF1 proteins. J. Mol. Biol.83:231-251. 22. Okazaki, R. 1974. Short chain intermediates in DNA

replication,p. 1-32. In R. B. Wickner(ed.), Methods inmolecularbiology,vol. 7.Marcel Dekker,Inc.,New York.

23. Pratt, D., andW. S. Erdahl. 1968. Geneticcontrolof bacteriophage M13 DNA synthesis. J. Mol. Biol. 37:181-200.

24. Pratt, D.,P.Laws,and J.Griffith. 1974.Complexof bacteriophage M13 single-stranded DNA and gene 5 protein. J.Mol. Biol. 82:425-439.

25. Ray,D. S. 1969. Replicationofbacteriophage M13. II. Therole ofreplicativeforms in single-strand synthesis. J.Mol. Biol.43:631-643.

26. Ray, D. S. 1977. Replication offilamentous bacterio-phages,p. 105-178. In H.Fraenkel-Conrat and R.R. Wagner (ed.), Comprehensive virology,vol. 7.Plenum Publishing Corp.,NewYork.

27. Schekman, R., A. Weiner, and A.Kornberg. 1974. Multienzyme systems of DNA replication. Science 186:987-993.

28. Schroder,C.H.,and H.C.Kaerner.1972.Replication ofbacteriophage 4X174 replicativeformDNAin vivo. J. Mol. Biol.71:351-362.

29. Staudenbauer,W. L., and P. H.Hofschneider.1971. Membrane attachment of replicating parental DNA molecules of bacteriophage M13. Biochem. Biophys. Res.Commun. 42:1035-1041.

30. Staudenbauer,W.L., and P. H. Hofschneider. 1973. Replicationofbacteriophage M13: positive role of gene 5protein in single-stranded DNA synthesis. Eur. J. Biochem.34:569-576.

31. Suggs,S.V., and D. S. Ray. 1977.Replication of bac-teriophage M13.XI.Localization of the origin for M13 single-strand synthesis. J. Mol. Biol. 110:147-163. 32. Tabak,H.F.,J.Griffith,K.Geider,H.Schaller,and

A.Kornberg.1974.Initiationofdeoxyribonucleic acid synthesis. VII. Auniquelocation ofthe gapsinM13 replicative duplex synthesized in vitro. J. Biol.Chem. 249:3049-3054.

33. Tseng, B. Y., and D. A.Marvin. 1972. Filamentous bacterialviruses. V.Asymmetric replicationoffdduplex deoxyribonucleicacid. J. Virol. 10:371-383.

34. vandenHondel, C. A., andJ.G.G.Schoenmakers. 1975.Studiesonbacteriophage M13 DNA.I. Acleavage map of the M13genome.Eur. J. Biochem.53:547-558. 35. vanden Hondel, C. A., and J. G. Schoenmakers. 1976.Cleavage mapsofthefilamentous bacteriophages M13, fd,fl andZJ/2.J. Virol.18:1024-1039. 36. Young,I.T.,D.Levinstone,M.Eden, B.-K. Tye,and

D. Botstein. 1974.Alignmentofpartial denaturation mapsofcircularly permutedDNAby computer.J. Mol. Biol. 85:528-532.(Appendixin J. Mol. Biol.85:501-532, 1974.)