Virus-specific DNA in the cytoplasm of avian sarcoma virus-infected cells is a precursor to covalently closed circular viral DNA in the nucleus.

(1)

JOURNALOFVIROLOGY, Jan. 1978,p.104-114

Virus-Specific

DNA in

the Cytoplasm of

Avian

Sarcoma

Virus-Infected Cells Is a Precursor to

Covalently Closed

Circular Viral DNA in the Nucleus

PETER R. SHANK* AND HAROLD E. VARMUS

DepartmentofMicrobiology, University of California,San Francisco, California 94143

Received forpublication 9 August 1977

Threeprincipal forms of viral DNA have been identified in cells infected with aviansarcomavirus: (i) alinear duplexmolecule synthesizedin the cytoplasm, (ii) acovalently closed circularmolecule found in the nucleus,and (iii) proviral DNAcovalentlylinkedtohigh-molecular-weightcell DNA. To define

precursor-product relationships among these forms of viral DNA, we performed pulse-chase experiments using 5-bromodeoxyuridine to label by density the linear speciesof viral DNA in thecytoplasm duringthefirst4hafter infection. Aftera 4-to8-h chase withthymidine,aportion of thedensity-labeledviral DNA was

transported to the nucleus and converted to a covalently closed circular form. Weconclude that linear viralDNA, synthesizedinthecytoplasm,is theprecursor toclosed circular DNA observed in the nucleus.

Theability of RNAtumorvirusesto convert

theirsingle-strandedRNAgenomesinto double-strandedDNA isuniversallyrecognized (30, 37), butthemechanism bywhich this unusual form ofsynthesisoccurshasbeen

only partially

elu-cidated. Three major forms of

virus-specific

DNA have beenidentifiedinculturedcellsafter productiveinfectionbyRNAtumorviruses: (i) linear duplex DNA comprised of a

genome-lengthminus strand(8.5to10kilobases,

comple-mentary to viral

RNA)

and

segmented

plus

strands (0.3 to 1.5

kilobases)

(10, 23,

33); (ii)

covalentlyclosed circular

duplex

DNA

(form

I) of subunitlength (9, 12, 23);and

(iii)

viral DNA covalently integratedinto host cell DNA (pro-virus) (35). After infection by avian sarcoma

virus(ASV),the linear

duplex

DNA is observed inthe

cytoplasm

within4 h

(35);

form I DNA isfound

exclusively

inthe nucleusas

early

as5 h after infection (12); and integration ofviral DNAcan occur within9hafter infection (35). Basedupon thesefindings, inhibitor studies (11), andanalogieswithother viruses (1, 2), a hypo-thetical scheme for thesynthesisandintegration ofviral DNA canbe constructed (Fig. 1). The linear form of viralDNA issynthesized in the cytoplasm by virus-associated, RNA-directed

DNA polymerase, transported to the nucleus, and then circularized and integrated into host

cell DNA, presumably with the assistance of

host enzymes.

Tovalidatethishypothesis,it is necessaryto

demonstrate that thecytoplasmic linear DNA

is a precursortonuclear form IDNA and that

form I DNA isaprecursor tointegratedprovirus.

In this report we present evidence favoring a

precursor-product relationship between

cyto-plasmic linear DNA and nuclear form I DNA.

Viral DNA constitutes too small a fraction of

totalcellular DNAtopermitlabelingwith

radio-activedeoxyribonucleotides, as in atraditional pulse-chase experiment. We have previously

shown, however,that viral DNAsynthesized in

the presence of 5-bromodeoxyuridine (BUdR)

hasahigh density attributabletothe

substitu-tion of BUdR forthymidineinbothstrands (32,

34).Weused BUdRtodensity label the precur-sorform in the cytoplasm during the first4 h

of infection by ASV; after blocking further

in-corporationof BUdR into viralDNA with

thy-midine, we followed the conversion of

BUdR-substituted DNA into form I molecules in the

nucleus.

MATERIALS AND METHODS

Celsandvirus. Wehaveused the QT-6 ceH line

(derived fromamethylcholanthrene-induced fibrosar-coma ofJapanese quail [21]) to study synthesis of ASV DNA (12).Cellswere propagated as monolayers at41°Cinmedium199supplemented with10% tryp-tosephosphate broth,0.1%(wt/vol) sodium bicarbon-ate, 5% fetal calf serum, 1% heat-inactivated chick serum,and 1%dimethyl sulfoxide (growth medium). Monolayers of QT-6 cellswereinfected with the B77 strain ofASV(B77-ASV)at amultiplicity of infection

(MOI) of1focus-forming unit/cell in the presence of

4yg ofpolybreneperml. B77-ASVwasgrown in and titeredon chicken embryo fibroblasts as previously

reported (19). QT-6 cellular DNA was radioactively

104

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

CYTOPLASMIC ASV-SPECIFIC DNA 105

, s

3

i3'

1']_ii_ 1

-^^ ."..*

- **LI. . "I -,- CVIOPLASIA

,."...,.^d ~ ~ ~

~~~~~~~~~~~~~~~~~~~

i

iv

_~ ~ ~ ~~~~~~~~~~ILLt D "

FIG. 1. Modelforthe synthesis ofASVDNA. (i) The70SviralRNA molecule isadimercomposedof two35Ssubunitsjoinedattheir 5'ternini(18). Syn-thesisoftheminus strandofASVDNA bythe virion-associatedRNA-dependentDNApolymerase is ini-tiated near the 5' end of the viralgenome on a

tRNA"Pmolecule (5, 29). Using a terminal

redun-dancy (3, 14,24) in the viralRNA, thepolymerase "leaps"tothe 3' endofthesameor adifferentRNA subunit and completes the synthesis ofa

genome-lengthminus strand.Before completion ofthe minus

strand,plus-strand DNAsynthesis commences

(un-publisheddataoftheauthors)intheopposite

direc-tion, possibly using fragmented genome RNA as

primer. (ii)The linearviralDNAmolecule, synthe-sized in the cytoplasm and comprised of

genome-length minus strands andsegmented plus strands, istransported into the nucleus. (iii) Within the

nu-cleus, theplus-strand fragmentsarecompleted and

joined(possibly bycellularenzymes)immediately

be-forethemoleculeiscovalentlyclosed.Thisconversion

may occur via an asyet unobservedopen circular

intermediate. (iv) Covalently closed circular DNA

presumably integrates covalently into the cellular

genomeat unknown sitesbyarecombinational event

(1).Toallowsynthesis ofgenomicviral RNAdirectly fromatemplateofproviralDNA, atleastonecopy must beorientedsothatit is colinearwith the viral RNA.

labeledbythe inclusionof0.1 ytCi of[tH]- or

["C]-thymidinepermlinnormalgrowthmedium. To label

QT-6 cellular DNAradioactively inthe presence of BUdR, [3H]deoxyguanosine (0.1 to 0.4tiCi/ml) was

includedin the normalgrowthmedium.

BUdRlabeling and washprocedure.To label DNA withBUdR, QT-6 cellsweregrownin normal

growth medium supplemented with 10,g of BUdR

(Calbiochem) per ml. To stop BUdR incorporation,

the cells were washed twice with Tris-glucose (0.14 M NaCl, 5 mM KCI, 5.5 mM glucose, and 25 mM

Tris-hydrochloride, pH 7.4) containing 50 ,ug of

thy-midine perml and then incubated in normal growth medium containing 50,Ug ofthymidine per ml.

Cellfractionation and DNAextractions. Cells

were fractionated into nucleus and cytoplasm after disruption in 1% Nonidet P-40(ShellOil) aspreviously described (12). The nuclei were thendissolved at 1 x 107 per ml in 20 mM Tris (pH 7.2)-10 mM EDTA (TE buffer) andlysedby theaddition of sodium

do-decyl sulfate (SDS) to 1%, and chromosomal DNA

wasprecipitated by the addition of 5 M NaCl to 1 M according to the method ofHirt (15). Afterovernight

incubation at 4°C, the chromosomal DNA was pel-letedbycentrifugation of 13,000 x g for 30 min. DNA

waspreparedfromthecytoplasmandtheSDS-NaCl

supernatantbyconventional methods (32, 35). In brief,

thesampleswereincubated withself-digestedPronase

(500

ig/ml)

at37°Cfor 1 hfollowed byphenol extrac-tionand ethanol precipitation. The DNA was

resus-pended inTE.buffer and digestedfor 1 h with

pan-creatic RNase (100,ug/ml) at 37°C followedby Pro-nasedigestion (250

Ig/ml)

at 37°C for 30min, phenol

extraction,andethanolprecipitation.

Gradient sedimentation. Fractionation of DNA

according to the extent of substitutionof BUdR for

thytnidinewasaccomplished byequilibriumbanding

indensitygradientsof CsCl.Samplesof DNA in TE

buffer were sheared by three passages through a 19-guage needle and mixed with solid CsCl (Harshaw radiotracergrade)togiveafinaldensityof 1.75g/cm3. Centrifugationwasfor 60 hat200Cin either the type 40rotor(at33,000rpm) orthe type 42.1 rotor (Beck-man) (at31,000 rpm), dependingon the number of

cellsextracted. Densitiesweredeterminedboth from measurementsof refractive index (with a Bausch and Lomb refractometer) and from the bandingposition

ofradiolabeledcellularDNA.

Covalentlyclqsedcircular DNA(formI)was

sepa-rated from relaxed forms (linear or nicked circular forms IIIandII)by bandinginCsCl density gradients

containing intercalatingdyes. Each gradient contained

300ugof ethidium bromide (EB) (22) orpropidium diiodide (PI2) (16) per ml, and the sample was dis-solved in TE buffer to give final densities of 1.62

g/cm3. The gradients were centrifuged for 60 h at 33,000 rpm in a type 40 rotor at

200C.

Covalentlyclosed circular DNA was also separated

fromopen circular or linear forms bysedimentation in alkaline sucrosegradients (35). Gradients of 5 to 20% sucrosecontaining0.3NNaOH,0.7NNaCl,and 0.001 M EDTAwere prepared inpolyallomer tubes andcentrifugedat 17,000 rpm for 16 h at

200C

in an SW27.1 rotor (Beckman). 3H-labeled plasmid DNA, pML21 (formsIandII;molecularweight, 6.7 x 106) servedasanextemal sedimentation marker.

Hybridizationreagents. 32P-labeled

complemen-tary DNA(cDNA) waspreparedinvitro from

deter-gent-activatedB77-ASV in thepresence of

actinomy-cin D asdescribed (8). Theproduct synthesized was fractionated intosingle-anddouble-stranded fractions

on hydroxyapatite, and the single-stranded fraction

wasused forannealing.This32P-labeled cDNA, which

-moor,.--.-- . . . . .t

VOL. 25, 1978

http://jvi.asm.org/

[image:2.505.52.246.62.303.2]

(3)

SHANK VARMUS

containedsequencesprimarily complementarytothe 5' end of the viral genome (6, 13), had a specific

activity of1x 108 cpm/ug.

"2514abeledASV RNAwaspreparedaspreviously

describedby Commerford (4)and hadaspecific

activ-ity of1x108 cpm/uAg.

Hybridization assay for ASV specific DNA. Hybridization of labeled ASV cDNAtounlabeled viral DNA was performed as previously described (12).

After the addition of100,ug ofcalfthymus DNAas

carrier, sampleswerediluted to1.0ml in0.3 NNaOH,

heatedto80°C for2h(to fragment and denature the DNA), neutralized, ethanol precipitated, and

resus-pended in 20ptlof TE buffer.After addition of1,000 to 1,500 cpm of[nP]cDNA and adjustment of the

NaCl concentrationto0.6M, the sampleswere

over-layered with mineral oil and incubatedat680C for60 h.Annealing of the cDNAwasassayed by resistance

to the single-strand-specific nuclease Si (27). The amountof unlabeled viral DNA ineachsamplewas

then calculated fromacalibration curveconstructed

inparallel with each hybridization experiment by

an-nealing 32P-labeled cDNAto increasing amounts of DNA from XC cells (rat cells transformed by the

Prague C strain of ASV) (28). Since the standard used (DNA from XCcells) is double stranded, the calibra-tion isaccurate only whenbothstrands of theviral DNAarepresent.Whenonlyonestrand of viral DNA

ispresent,i.e.,atthetopofanalkalinesucrosegradient where thelong minus strandsareseparated from the

shortplus strands, this calibrationcannotbe used.

["25I]RNA wasusedtodetect the minus strand of viral DNA. Annealings with [luI]RNA were

per-formedexactlyasdescribed abovefor[nP]cDNA

ex-cept that 100,ig ofyeast RNA carrier per ml was

addedtoeach reaction. Theannealingof[125I]RNA

wasassayed by resistance to digestion byamixture

ofpancreatic and Ti RNase (50,tg/ml,5 U/ml) in 2

xSSC (SSC=0.15M NaCl+0.015Msodium citrate)

at370C for 1 h. The nuclease resistance (ca. 5%) of the cDNA and RNAafterincubation with calfthymus DNAwasconsidered background for the assay and

wassubtracted from eachanalysis.

RESULTS

BUdRlabeling ofviralDNA. We have pre-viously shown (32, 34) thatviral DNA

synthe-sized in duckembryo fibroblasts infected with

ASVcanbedensitylabeled with BUdR. When

cells have been labeled foranappropriatetime,

viral DNAthat isfullysubstituted with BUdR

(HH DNA)canbeseparatedfrompartially

sub-stituted (HL) and unsubstituted (LL) cellular

DNAby banding in CsCl density gradients.

We usedQT-6cells in thisstudybecausethey

supportthree- tofivefoldmoreviral DNA

syn-thesis than duckembryofibroblasts after

infec-tion with B77-ASV (12). In addition, one to

three copiesof form I viralDNA are regularly

synthesizedpercellafterinfection ofQT-6cells

with B77-ASVathigh MOI (12). To document

our abffity to density label viral DNA in the

QT-6 line, cells were infected with B77-ASV

(MOI = 1) in the presence of 10 ,Ig of BUdR and0.1,uCiof

[3H]dG

per ml. After4h thecells were fractionated by the SDS-NaCl precipita-tion of Hirt (15), and DNAwasextracted from thesupernatantfraction and bandedto equilib-rium in aCsCl gradient. Thepositionof LL and HLcellularDNAin thegradientwasdetermined from the 3H radioactivity within a portion of each fraction, and viral DNAwas detected by annealing aportion of each fraction of the gra-dient with virus-specific [32P]cDNA, as de-scribed in Materials and Methods (Fig. 2). All the viral DNA banded at a density of ca. 1.78 g/cm3,consistent with the replacement of 80 to 100% ofthymidine residues by BUdR in both strandsduringinfection ofQT-6cells, as in duck embryo fibroblasts. Cellular DNA banded at densitiesof1.70and1.74g/cm3, appropriatefor LL and HL DNAs.

Efficacy of chasing BUdR from QT-6 cells. To use BUdR labeling in a pulse-chase experiment,weneededtodocument our ability toblockincorporation of BUdR into DNA dur-ing the chaseportion ofthe experiment. Wein-traub (38) has shown that incorporation of BUdR into chicken embryo fibroblast DNA is decreasedby90% within1to 2minby washing the cells in medium containing thymidine. We therefore designed a chase procedure in which QT-6 cells were washed twice in Tris-glucose buffercontaining50,ugofthymidineper ml and thenincubated in growth medium containing 50

,ug

ofthymidine per ml.

Wefirst assessed the effect of this chase pro-cedure on the labeling of cellular DNA under

ourexperimentalconditions.Cellswereinfected with B77-ASV in the presence of10,ugof BUdR perml. Onesetofcellswaslabeled from2to 4

hafter infection with[3H]dGand harvested4h postinfection; this procedure mightbeexpected

togenerate3H-labeledHL cellularDNA. Four hoursafterinfection,twoothersetsofcellswere washed and incubated with mediumcontaining 50 ug ofthymidine per ml as described above. One ofthese sets was labeled with

[3H]dG

for

15 min immediately after the wash, and the

other set waslabeled from 15 to 30 min after

the wash. After extraction and shearing, the DNA from all threesets wasbandedto equilib-rium in CsCl gradients

containing

an intemal "4C-labeled marker of LL QT-6 DNA. As ex-pected, the

[3H]DNA

radiolabeled in the

pres-ence of BUdR banded at the position of HL DNA (1.74

g/cm3)

(Fig. 3A), whereas the aver-age density of DNA synthesized 0 to 15 min after the chase was only

slightly

greater than that ofLL DNA(Fig. 3B);allthe DNA synthe-sized 15 to 30 min after the chase co-banded

withthe14C-labeledLL DNA(Fig. 3C),

http://jvi.asm.org/

(4)

.20

C

z

J

>l

.15 h .10 .05

Density-I

* (9/crn3) 1.80

_ ~ (g/cm') 11.75

1.70 elS CELL HLCELLLL

- ,0 \

I\

- I %O.&.-4.-4.O

5 10 15 20

FRACTION

FIG. 2. BUdR density labeling of viral DNA in QT-6 cells. 2 x10C QT-6 ceUs were infected with B77-ASV(MOI=1)in thepresenceof10pg ofBUdR, 4 pgofpolybrene,and0.1utCi of [3H]dGperml.Four hours after infection the cells were trypsinized,

washedtwice inThis-glucose, andlysed bythe

SDS-NaCiprocedure described by Hirt (15). DNA

ex-tractedfromtheSDS-NaCIsupernatant was

centri-fugedto equilibriumina CsCIdensity gradient at

33,000 rpm in atype40 rotorfor60h at200C. After

collection from below, 10% of each fraction was

counted directly todetermine the banding position

of 3H-labeled LL and HL cellular DNA. Densities

weredeterminedfromtherefractiveindex. After de-naturation andfragmentationby incubation at80fC

in 0.3 NNaOH, the sample waspreciitated with

ethanol, and viral DNA was detected by annealing

32P-labeled ASV-specific cDNA (1,000 cpm) to the

remainderofeachfractionas described in Materials and Methods.Annealingwasassessedby resistance

ofthecDNAtoSinucleasedigestion,andthe amount

ofviral DNA was determined by comparison to a calibration curvegenerated by annealing the/p]

cDNAtoincreasingamountsofXC DNA as described

inMaterials and Methods.

ing that the chase was completely effective within15min.

Although the chase procedure appeared highly effective with respect to cellular DNA, viral DNA, synthesized in the cytoplasm bya

viralpolymerase, mightpossiblyutilize different nucleotide pools. We therefore also tested our

abilitytopreventBUdRincorporationintoviral DNA.UninfectedQT-6cellswereincubated for

4 hwith10

jig

ofBUdRperml, washedtwice, and then infected with B77-ASV in thepresence

of50

Ag

ofthymidine perml. After 8 h, viral DNAextracted from theSDS-NaClsupernatant

wasanalyzedinaCsClequilibriumdensity

gra-dient (Fig. 4). All viral DNA detected by

an-nealing with 32P-labeled cDNA banded in the

region of LL DNA, indicating that our wash procedure was effective in blocking BUdR

in-corporation into viral DNA. Since viral DNA

synthesisisdetectablewithin 1 h after infection

underthese conditions (unpublished data ofV.

Smith andauthors), the chase procedure rapidly

stops BUdRincorporationintoviralDNA.

How-ever, it was notpossibletoexamine thekinetics of the chase procedureforviralDNA as strin-gently asforcellularDNA.

Pulse-chaseexperiments with density-la-beled viral DNA. (i) Linear viral DNA is precursor to form I DNA. To test and illus-trate our experimental plan without the com-plexities introduced by cell fractionation, we

per-80

60

40

20

0

O 20

x

a-u 10

z a

I--30

20

10

18

b

x U

u

z a

u

[image:4.505.84.232.59.180.2]

5 10 1 5 20 25

FIG. 3. Efficiency of chase with QT-6 ceUular DNA. Three sets of QT-6 cells were infected with

B77-ASV(MOI=1)in thepresenceof10gofBUdR

and4pgofpolybreneperml. Twohoursafter

infec-tion,onesetofcells(A)waslabeled with 0.2,iCiof

pH]dGper ml in the presenceofBUdR. Four hours

afterinfection,setAwasharvested and total cellular

DNA was extracted At thesametime,theremaining

two sets ofcells (B and C) were washed twice in

Tris-glucose containing50pg ofthymidineand

in-cubated ingrowthmedium containing50pg of thy-midine perml.The cells insetBwerelabeledfrom 0to 15minafterthe wash with 0.4uCi of[3H]dG

per ml, and the other ceUs (C) were labeled with

[3H]dG from 15 to 30min afterthe wash. Celular

DNA extractedfromeach setofcells was sheared and bandedtoequilibriuminCsCIdensity gradients,

asdescribedin thelegend ofFig.2,aftertheaddition

of14C-labeledunsubstitutedQT-6ceU DNA. The

gra-dientfractionswerecounteddirectlyin anaqueous

scintiUationfluortodetermine thebanding position of3H(0)-and14C()_-labeledcelularDNA. VOL. 25, 1978

http://jvi.asm.org/

[image:4.505.266.458.167.454.2]

(5)

~0.8-

<0.6-z

a

_J

0.4-

>0.2-5 10 15 20

FRACTION

FIG. 4. Efficiencyof chase with viral DNA. QT-6

cellswereincubatedfor4hwith10pgof BUdR and 0.1,uCiofPH]dGper ml. The cellswerethenwashed

twice withTris-glucose containing

50jg

of thymidine

permlandinfected with B77-ASV (MOI=1) in the

presenceof4pgofpolybrene and50pgof thymidine

perml. Eight hours after infection, the cells were

harvested, andthedensity of viral DNA in the

SDS-NaCisupernatantwasdetermined bybandingina

CsCldensity gradientasdescribed in thelegend of

Fig. 2.

formed the preliminary experiment illustrated inFig.5.Ourpreviousresults(12) demonstrated

that little or no form I viral DNA would be

present4hafterinfection. Ourexperimentwas

therefore designed to density label the linear form of viral DNA during the fit 4 h after

infection, wash out the BUdR, and determine

whether density-labeled DNA was converted

into form I molecules.

QT-6 cellswereinfected with B77-ASV in the presence of BUdR and 0.1 ,uCi of [3H]dG per

ml;after 4h,aportionofthe cells(pulse sample)

washarvested,andasecondportion (chase sam-ple) wassubjected tothethymidinechase

pro-cedure and allowedtoincubate foranadditional

4hbeforecollection. ViralDNA,extractedfrom

theSDS-NaCl supernatantofthepulsesample,

banded in the position expected for HH DNA

inaCsClequilibriumdensity gradient (Fig. 5A)

and in the position expected for linear DNA

when rebanded inaCsCldensity gradient

con-tainingpropidiumdiiodide (CsCl-P12) (Fig.50). (Binding ofP12, abuoyantintercalating dye, is

restricted in superhelical [form I] DNA

com-pared with linear ornicked circular molecules,

causingformI DNA to exhibitagreaterdensity

than do the other forms inCsCl-PI2gradients.)

In contrast, viral DNA extracted from the

SDS-NaCl supernatantofthe chasesample

ex-hibitedabroad spectrum of densities after

equi-librium centrifugation in CsCl gradients (Fig.

5B).The DNArangedfrom HH molecules

pre-sumably completed duringthe 4-hlabelingwith

BUdRtoLLmolecules synthesized onlyduring the chase period; the species of intermediate densitywerepresumablymolecules of different "ages," dating from the time of the thymidine chase. Theheterogeneous densityof viral DNA observed after the thymidine chase contrasted sharplywith thehomogeneity of cellular DNA synthesized after a similar chase (Fig. 3B and C). Although it is possible that the chase pro-cedure is lessefficient with respect to viral DNA, the observed data are also consistent with the hypothesis that viral DNA is elongated at a relatively slow rate. This would mean that many incomplete molecules were present 4 h after infection and completed during the chase; this idea issupportedby kinetic analysis of the syn-thesis of viral DNA (seeDiscussion).

When the HH DNA recovered from thechase sample was centrifuged in CsCl-PI2 gradients (Fig. 5D), approximately 25% of the viral DNA banded in the position of form I DNA, demon-strating that linear DNAsynthesized during the first4h ofinfection could be found in a

cova-lently closed circular form4h later.

(ii) Linear viral DNA in the cytoplasm is precursor to form I DNA in the nucleus. To determine whether the linear precursor to form I DNA was present initially in the cyto-plasm of infected cells and then transported into the nucleus, a pulse-chaseexperiment including cell fractionation was performed. The DNAs from the nuclear and cytoplasmic fractions of the pulse and chase cells werebanded to equilib-rium in CsCldensitygradients,andthe position of cellular DNA and viral DNA was detected asdescribedpreviously (legend toFig. 2). All of the viral DNApresent in the 4-hpulsesamples banded at the position of HH DNA (Fig. 6A andC). A small amount of viral DNA(less than 2% of the total) was detected in the nuclear fraction 4 h after infection and probably

re-flectedcontamination of the nuclei with cyto-plasmic material, since the nuclei were not washed after cellfractionation.

Both the nucleus and cytoplasm from the chase cells contained viral DNA banding in a

broadspectrumof densities(Fig.6BandD),in

agreement with the previous experiment (Fig. 5B). Thedensity of each molecule presumably reflects thedegreeof itscompletionatthe time of the chase procedure; therefore thegradient

ofdensitiescorrespondstoagradient of agesof

viral molecules. Most if not all of the oldest (HH) viralDNAinthenucleus must have been fully synthesized in the cytoplasm during the

pulseofBUdR andtransportedto thenucleus duringthe chaseperiod,since therewas atleast

five times more HH DNA inthe nuclei at 8h

thanat 4h after infection.

Similarly,

the

mole-J.VIROL.

http://jvi.asm.org/

[image:5.505.72.223.47.210.2]

(6)

CYTOPLASMIC ASV-SPECIFIC DNA

.15 .10 m).05

z a

< .15

.10 .05

Pulse Chose

0.8 0.6 0.4 0.2 .25 .20

.15 .10

.05

z a

5:i

5 10 15 20 5 10 15 20

FRACTION

FiG. 5.Analysisofpulse-labeled viral DNA in CsCl and CsCl-P1 gradients. Twosetsof QT-6cells (7x 101 each)wereinfectedwith B77-ASV(MOI=1)inthepresenceof10 pgof BUdR,4ugofpolybrene, and 0.1

itCiof[3H]dGperml.Fourhoursafter infection,thepulsecellsweretrypsinizedandsubjectedtoan SDS-NaCifractionation; thechase cells werewashed as described in the legend toFig. 3 and incubated in growth mediumcontaining50pgofthymidineperml. Eighthoursafterinfection, the chasesample was

harvestedasdescribedforthe 4-hsample. AfterDNAextraction,thepulseand chasesampleswerecentrifuged

toequilibrium in CsCldensity gradients asdescribedin the legend of Fig.2. Viral DNA wasdetectedby annealing20%ofeachgradientfractionwith3P-labeledASVcDNA. Theposition ofHLand LL cellular

DNA, indicatedbythearrows,wasdetermined bydirectly counting1%ofeachgradientfraction.Pools of HH viralDNA,indicatedbythebars,weremade,and the DNAwasrecoveredbyprecipitationwith ethanol. Thesecondarystructureofthe viral DNA wasthen assayed by bandingtheDNAfrom each poolin CsCl densitygradients containingtheintercalating buoyant dyePI2.One-thirdofthe HH DNA recoveredfrom thepulseand chasesampleswasbanded toequilibriuminaCsClgradient (startingdensityof1.62g/cm3) containing300(pgofPI2perml. Thedensityatwhichopencircular(formI)orlinear(form III)HH DNA would bandisindicatedbythearrow. TheCsCl-PI2gradientswerecentrifuged for60 h at33,000rpm ina

type40 rotor at20°C. Viral DNA wasdetected in the CsCl-P1 gradients by annealing with 3P-labeled cDNAafterdenaturation and ethanolprecipitationasdescribedin thelegendtoFig.2.

cules of intermediatedensitywereatleast

initi-ated in thecytoplasm during the pulse period.

However, the siteatwhich thesemoleculeswere

completed and the site of synthesis of the

young-est (LL) molecules found in the nucleus after

the chasecannotbedetermined fromthis

exper-iment. Assuming thatall viral DNA synthesis

is initiated in thecytoplasm, thepresenceof LL

viral DNA molecules in the nucleus at times

when HH and HL moleculesare presentinthe

cytoplasm (Fig. 6B and D) implies thattransport

of viral DNA into thenucleus isnotdependent

ontheageof themolecules.

Fractions from the indicated regions of the

CsClgradientsshown inFig.6AthroughDwere

pooled, and thestructure of DNAineach pool

wasanalyzed by centrifugation inCsCl-EB

gra-dients (Fig. 7). Inaccord with ourprevious

re--ilts (12),noform Iviral DNAwasdetectable in thecytoplasmateither4or8hafter infection

(Fig. 7A). In addition,noneof thesmallamount

ofviralDNAassociated with thenucleusat4h

postinfectionwasformI DNA(Fig. 7B);asnoted

earlier, the small amount of DNA present in

the nuclear fractionat4hprobablyrepresented

cytoplasmic contamination. By 8 h after

infec-tion, when therewasfivetimesmoreHHDNA

in the nucleus than at 4 h (Fig. 6C and D),

approximately 50% of the HH DNA in the

nu-cleus consisted of form I molecules (Fig. 70).

Similar analysis in CsCl-EB gradients of DNA

recovered from the HLregion of theCsCl

gra-dient shown in Fig. 6D indicated that 5to 10%

consisted of form I molecules (datanotshown).

Theseresults indicate that the majority of form

I viral DNA molecules present in the nucleus

musthavecomefromthepool of

BUdR-substi-tutedcytoplasmic viral DNA. Sincewe

encoun-teredsignificant variability in recovering

BUdR-substituted DNA fromthepreparativeCsCl

gra-dients (Fig. 6), ourconclusions arebased upon

theproportionsofforn I DNA intherecovered

DNA(Fig. 7) andnotupontheabsoluteamounts

ofform IDNA.

Since the density of DNA used in these

ex-perimentswasalteredbyBUdRsubstitutionas

wellasby the binding ofintercalating dyes,we

confirmedourobservationsby sedimentation in

I~~~~~ '

C

E L HL CEL LL

/CL'L(EL

D

's

1

CeCI-Pl, CiCI-Pp,

A

A I

.1 \\

i

_11A4

109

VOL. 25,1978

http://jvi.asm.org/

[image:6.505.122.360.69.251.2]

(7)

110 SHANK

alkaline sucrose gradients, a method independ-ent of the density of DNA. In alkaline sucrose gradients, form I molecules partially denature and collapse into a fast-sedimenting form (36), permitting both the detection of form I mole-cules and an estimation of their size (26). Ge-nomelength strands derived from linear mole-cules should sediment as 18 to 20S in these gradients (26); since there is no evidence for multimeric linearmolecules,viralDNA migrat-ing faster than 20S was presumed to be conva-lently closed circles. In addition, these gradients permitmeasurementof the size of each strand present in the linear viral DNA molecules (see below).

HH viral DNA from the various pools shown

in Fig. 6 was sedimented in alkaline sucrose

gradients and assayed by annealing with

[32P]-cDNA (Fig. 8). These gradients confirmed the observationsmade by equilibrium density cen-trifugation inCsCl-EB gradients (Fig. 7). There was no evidence of rapidly sedimenting (form I) DNA in the cytoplasm at either 4 or 8 h.

(Insufficient viral DNA was recovered from the 4-hsample of nuclear DNA to allow analysis in an alkaline sucrose gradient). The HH sample from the nuclear fraction after 8 h of infection contained twopopulations of forn I viral DNA. The large population sedimented at 65S, con-sistentwith amolecularweight of about 6.5 x 106.The other componentsedimented at around 45S, consistentwithamolecular weight of about

2 x 106 to 3 x 106. This smaller population

presumably represents the highly defectiveform I viral DNA we have previously observed in QT-6 cells infected with B77-ASV (12). Al-thoughcDNAannealed much moreextensively tolinearDNAatthetop ofthese gradients than tothe closedcircularDNA, therelative propor-tionsofform Iandlinear DNAcannot be esti-mated from these analyses. The efficiency of annealing of cDNA to plus strands fromlinear DNA was exaggerated because the minus strands, which normally compete with the cDNA, wereseparatedfrom the plus strands on thebasis of size. This accounts for the apparent

Ts

.28,-4hr Cytoplasm

A CELL CELI

f\ HL LL

-

0 \

0

I**I I

4hrNuclei C

.20k _CELL _CELL HL LL

.12k

.04

;-9

5 10 15

FRACTION

8hr Cytoplasm

B CELL CELL

4'

H.L

LL

0 0

I I

4,0ot

8hrNuclei CELL CELL

D ILL

*urii@4' I I

5 10 15

[image:7.505.148.390.331.547.2]

FRACTION

FIG. 6. CsCI gradient analysis ofpulse-labeled viral DNA in thenucleus andcytoplasm. QT-6cells(2x

1(P')wereprelabeled for24 h with0.I#CiofPHJthymidineperml. CeUswerethen infectedwithB77-ASV

(MOI= 1) in thepresenceof10 pgofBUdR per mlasdescribedinMaterials and Methods. Four hours

after infection, halfthecellswereharvested and theotherhalfweresubjectedto thechaseprocedureand incubatedforanadditional4 h ingrowthmedium containing50 pgofthymidineperml. Thecells were fractionatedintonucleiandcytoplasm;DNAwasprepared fromthecytoplasm,andanSDS-NaClsupernatant wasprepared from the nuclei as described in Materials and Methods. The DNA was then banded to equilibriuminCsCI density gradientsinatype42.1rotor,and viral DNAwasdetectedby annealing10%of eachfractionwith32P-labeledcDNAasdescribed in thelegendtoFig.2. Thepositions ofthecellularHL

andLL DNA, indicated by thearrow, were determined by directly counting 3Hin 1%ofeachgradient

fraction. Thepools ofHHDNAindicatedbythe barsweremade,and DNA wasrecoveredby precipitation withethanol andanalyzedinCsCI-EB gradients (Fig. 7)oralkalinesucrosegradients (Fig. 8).

1.6

0)

C

< 1.2 z

-J 0.8

> nA

z

a

5:

V.4

I

http://jvi.asm.org/

(8)

CYTOPLASMIC ASV-SPECIFIC DNA

a

LU)

z z z

u

z

u

CL

.n

0 250

200

150 [

100 50 1

[image:8.505.54.246.61.239.2]

5 10 15 20 25 5 10 15 2025 FRACTION

FIG. 7. Analysis of pulse-labeled viral DNA in CsCI-EBgradients. HH DNA from the pools indi-cated in Fig. 6 was concentrated byprecipitation with ethanol and banded in CsCl-EB density gra-dients(startingdensity=1.62g/cm). The 4- and 8-h cytoplasmicsampleswerepooled to reduce the

num-ber of analyses, sinceprevious data (12) indicated

the absenceof formIDNAfromthe cytoplasm. The arrowindicates the bandingposition oflinear (III) oropen circular(I) DNA. Fivepercentof the cyto-plasmicpoolwas analyzed, whereas allof the 4-h nuclear and one-thirdofthe 8-h nuclearpoolswere

analyzed.Viral DNA wasdetected byannealingeach

fractionofthegradients with

132P]cDNA

asdescribed

in Materials and Methods.

a

-I

z z

z

a-u

zK

0a-1200

1000

800

600

400

200

discrepancy betweenFig. 7C and 8B.

Size ofthe strands ofviralDNApresent in linear forms. We have reported that the linear viral DNA present in the cytoplasm of ASV-infected duckcells contains a long minus strand(complementarytothe viralgenome) and ashortplus strand (same polarity as the viral genome) (31). The same form of viral DNA is present inASV-infectedQT-6cells (H. E. Var-mus, S. Heasley, H. J. Kung, H. 0. Oppermann, V. C. Smith, J. M. Bishop, and P. R. Shank, submittedforpublication),andsimilar forms of viral DNA have been found inmurineleukemia virus-infectedcells(10)and inmousemammary tumorvirus-infectedcells(23).This unusual lin-earduplex DNA appears to be thecharacteristic form of DNA transcribed in vivo from the ge-nomeof RNAtumorviruses.

We therefore designed a pulse-chase experi-mentwith the BUdR density-labelingprotocol to determine whether elongation of the plus-strand fragments was detectable during the chase. Viral DNA recovered from the HH region of CsCl density gradients from the cytoplasm aftera4-hpulsewithBUdR and from both the nucleus and cytoplasm after an additional 8-h chase withthymidine (cf. Fig. 6) wassedimented in alkaline sucrose gradients (Fig. 9). Minus-strand viral DNA extracted from both the 4-h and 12-hcytoplasmand the 12-hnuclei, detected by annealingwith

"MI-labeled

viral RNA, sedi-mented at 10 to 20S (Fig. 9). Theplus strand,

1200 a 1000 < z

z

800 <

600

u

400 a

200 r,

z~~~~~~~~~~~~~~~~~~~~~~~~~~~~I"

5 10 15 20 25 5 10 15 20 25

FRACTION

FIG. 8. Analysis of pulse-labeledviralDNA inalkalinesucrosegradients. DNArecovered from the HH poolsshown inFig. 6wassedimentedin alkalinesucrosegradientsat17,000rpmfor 16h inanSW27.1

rotor at20°C. Two-thirds ofthenuclearpoolandone-twentiethofthecytoplasmic poolwereanalyzed. A

parallel gradient contained3H-labeled plasmid DNA (pML21) as asedimentation marker. (pML-21 isa

bacterialplasmid that exists as a form IDNA molecule of6.7x 106 daltons andhas a sedimentation

coefficient of70S in alkaline sucrose.) Sedimentation wasfrom right to left. ViralDNApresent in each gradient fractionwasdetectedby annealing with32P-labeled cDNAasdescribed in Materials and Methods.

B Il/ll

4hr. Nuclei I

A.

I

-IIrl

8 hr. Nuclei

11/111

qD

I

0id

'

4-+8 hr. 11/111 Sytoplasmf

i

ll

0

-

.!I

I

-_-J.

VOL. 25,1978 _ill

http://jvi.asm.org/

[image:8.505.133.374.401.580.2]

(9)

0

" 800

z

z 600 400

u

Z 200

l.

10 20 30 10 20 30 10 20 30

-S

0I

a 800 6

w

z 600 Z 4:

400 Q-200 Z a

U1

FRACTION

FIG. 9. Analysisof both strands ofpulse-labeledviral DNA inalkaline sucrose gradients. HH viral DNA

wasrecoveredfromCsCldensitygradients ofthe nucleus andcytoplasmof QT-6cellsinfectedanddensity

pulselabeled as described inFig. 6, except that the chase wasextended until 12 h after infected to increase

the amountofHH DNAtransportedintothe nucleus. The HH DNA samplesfromthecytoplasmofIx 10, cells andfromtheSDS-NaClsupernatantof the nucleifrom5 x 108 cells were sedimented in an alkaline sucrosegradient for16hat17,000rpm in an SW27.1 rotor at20°C.Minus-strand DNA was detected by

annealing to125I-labeledviralRNA, and plus-strand DNA was detected byannealing to 32P-labeled cDNA. To reduce the numberof analyses (sinceanyrapidlysedimenting[form I]DNAmustcontain both strands ofviralDNA), fractions1 to 20wereanalyzedwith '32P]cDNA, whereasfractions21 to 40 weredivided in

halfandanalyzedwith each reagent. Aparallel gradientcontained the3H-labeledpML21DNA (formI)as

asedimentation marker. Sedimentation wasfrom righttoleft.

detected by annealing with [32P]cDNA, sedi-mentedat 4 to8S in both cytoplasmic samples (Fig.9Aand B). Furthermore, the plus strands present inthe nucleus 12hafter infection sedi-mentedat4 to 8Sunlessthey were present in form I molecules (Fig.

90).

Therefore we are unable to detect plus-strand intermediates be-tweenthe shortplusstrands in thelinear mole-cule and thefull-length plus strands present in form I molecules.

DISCUSSION

We wish to drawonecentralconclusion from the experiments presented in this report: the linear form of viral DNA, synthesized in the cytoplasm ofcellsinfected withASV, is a

pre-cursor tothe closed circular form found in the

nucleus afew hours later. We haveestablished this relationship by the use of the traditional pulse-chase protocol, which has been used in

many contexts to illustrate precursor-product

relationships.Although pulse-chaseexperiments

most commonlyuse labeling ofprecursor can-didates with radioactive isotopes, the small quantities of viral DNA inRNA tumor virus-infected cellsobligedus touse densitylabeling with BUdRto identifyprecursor DNA, which

wasthen detected by molecularhybridization. Usingthisprotocol, wehave shown that a sig-nificantproportionofthedensity-labeledlinear viralDNA canbe chasedinto formImolecules (Fig. 5-9) and that thisDNA isalsotransported from thenucleus into the

cytoplasn

during the chaseperiod (Fig.6).

We encounteredsignificant variabilityin the recovery of the small amounts of viral DNA from the preparative CsCl gradients, thereby impairing our ability to account for all of the viralDNA insubsequentanalyses.On the other hand,since therewasfive times more HH DNA in thenucleus after the chase (Fig. 6) and over 50% of the HH DNA in the nucleus after the chasewasinthe form Iconfiguration (Fig.

70),

we conclude that the majority of the form I DNA in the nucleus was synthesized as linear molecules in the cytoplasm. Taken together, these observations support steps ito iii in the proposedscheme for thesynthesisofviral DNA (Fig. 1).

The resultspresentedhere donotexclude the possibility that the cytoplasmic form of viral DNA is an open circle rather than a linear molecule. Usingagarose gel electrophoresisand analysiswith restrictionendonucleases,wehave establishedthat the

cytoplasmic

viral DNA mol-ecule is linear (Shank et al., in preparation). Similar results have been observed with the

DNA ofmouse mammarytumorvirus (Shank

et al., in preparation); in addition, the open forms ofMoloneymurineleukemiavirus,mouse

sarcoma virus, andspleen necrosis virusDNAs

are in linear rather than open circular

confor-mation (3, 7, 23).

Inview of the evidence for the efficiency of

our chase procedure (Fig. 3 and 4), we were

surprisedtofind thatalarge proportionof viral DNAexhibited the densityofpartially labeled molecules after the chase period (Fig. 5C and

http://jvi.asm.org/

[image:9.505.116.402.71.193.2]

(10)

D; Fig 6C and D). One obvious interpretation of this result is that many of the viral DNA molecules initiated during the first 4 h after infection were not complete at that time, and that their completion during the chase period conferred an intermediate density upon them. We have recently obtained corroborating evi-dence for this view. Studies of the kinetics of synthesis of viral DNA (underconditions iden-tical to those used in this report) have revealed that minus strandsareslowlyelongated, requir-ing 3 to 6 h for completion, and synthesis of plus strands is retarded about 1 to 2 hrelative

to synthesis of minus strands (Varmus et al., submitted). Thus,atthe time of our chase pro-cedure,onlyaminority of DNA molecules were likely to be completed duplexes which would retain their HH density during the chase pro-cedure. It was, however, impossible for us to lengthenourpulse period, since effective dem-onstrationof the precursor-productrelationship demanded that little or no nuclear DNA be detectableattheconclusion of the BUdR label-ingperiod.

Althoughit is apparent from our results that atleast aportion of the cytoplasmic DNA mi-grates to thenucleus and enters a closed circular form, we do not know whether all of the ob-served cytoplasmic DNA is capable of these transitions. Thelargeamountof HH DNAthat persistedin thecytoplasm duringthe chase(Fig. 6B) might have represented incomplete mole-cules that could not be transported into the nucleus. Previous results from our laboratory and others indicate that viral DNA is synthe-sized inconsiderableexcess overtheintegrated species (12, 17) and that some of this excess

appears to remain in the cytoplasm (36). In addition, weand others have shown thata sig-nificantportion of the unintegrated linear DNA isless than genomelength(12, 25).These short species could represent abortive transcription products,precursorstofull-length forms, and/or precursorstothesmallspeciesofclosed circular viral DNA we have previously described (12). Althoughourexperiments provide direct evi-dence for theprecursor-productrelationship

be-tweenthecytoplasmiclinear and thecovalently

closed nuclear forms of viral DNA,theydonot

address the questionof whether the covalently closed form is the precursor to the integrated provirus (step iv, Fig. 1). In addition, they do notrevealthemolecular natureofthe

circulari-zationprocess. Although thesimplestmodelfor

circularization of theviral DNA wouldinvolve single-stranded regions ("sticky ends") at the

termini of the linearmolecules, suchendshave

notyet been demonstrated, and recombinational

mechanisms of circularization remain equally plausible.

ACKNOWLEDGMENTS

Wethank R. V.Guntaka,L.Levintow,and J. M.Bishop forhelpful discuionof theseexperimentsandmanuscript. ThisworkwassupportedbyPublicHealth Servicegrants CA 12705and CA 19287, awardedbythe National Cancer Institute.H.E.V. istherecipientofResearch Career Devel-opment Award CA 70193 from the National Cancer Institute. P.RS. acknowledges support from Public Health Service

btining grant 5T01 CA 05303 fromthe National Cancer Institute.

LITERATURE CITED

1.Campbell,A. 1962.Episomes.Adv.Genet. 11:101-145. 2.Canaani, E.,P.Duesberg,and D.Dina.1977.Cleavage mapof linearmouse sarcomavirus DNA. Proc.Natl. Acad.Sci.U.S.A.74:29-33.

3. Coffin, J. M., and W. A. Haseltine. 1977. Terminal redundancyand theoriginofreplicationof Rous sar-coma virus RNA. Proc. Natl. Acad. Sci. U.S.A. 74:1908-1912.

4. Commerford,S. L 1971.Iodinationofnucleicacids in vitro.Biochemistry10:1993-1999.

5. Dahlberg, J.E., R.C.Sawyer,J.M. Taylor, A. J. Faras,W. E.Levinson,H. M.Goodman,and J.M. Bishop. 1974. Transcription of DNA from the 70S RNA of Rous sarcoma virus. I. Identification of a

speciesof 4S RNAwhich serves asprimer.J.Virol. 13:1126-1133.

6. Friedrich, R., H. J.King,B.Baker, H. E.Varmus, H.M.Goodman, andJ. M.Bishop.1977. Character-izationof DNAcomplementarytonucleotidesequences atthe 5'-terminus of the aviansarcomavirus genome. Virology79:198-215.

7.Fritsch, E., and H. M. Temin. 1977. Formation and structureof infectious DNA ofspleennecrosisvirus. J. Virol. 21:119-130.

8. Garapin,A. C., H. E. Varmus, A. J. Faras,W. E. Levinson, and J. M. Bishop. 1973. RNA-directed DNAsynthesis byvirions ofRoussarcoma virus: fur-ther characterization of thetemplatesand theextent of theirtranscription.Virology52:264-274.

9. Gianni,A.M.,D.Smotkin,and R. A.Weinberg.1975. Murineleukemia virus: detection ofunintegrated dou-ble-stranded DNA forms of theprovirus. Proc.Natl. Acad. Sci. U.S.A.76:447-451.

10.Gianni,A. M., and R.A. Weinberg. 1975. Partially single-strandedformof freeMoloneyviral DNA. Na-ture(London)256:646-648.

11.Guntaka,R.V.,B.W.J.Mahy,J. M.Bishop, andH. E.Varmus.1975.Ethidiumbromide inhibits appear-anceofclosed circular viral DNA andintegrationof virus-specific DNA in duckcellsinfected by avian sar-comavirus. Nature(London)253:507-511.

12.Guntakca,R.V.,0. C. Richards,P.R.Shank,H. J. Kung,N.Davidson,E.Fritsch,J. M.Bishop, and H. E.Varmus. 1976.Covalentlyclosedcircular DNA ofaviansarcomavirus:purification from nuclei of in-fectedquailtumorcelLsandmeasurementby electron microscopy and gel electrophoresis. J. Mol. Biol. 106:337-357.

13.Haseline,W.A.,D.G.Kleid,A.Panet, E. Rothen-berg,andD.Baltimore.1976.Orderedtranscription of RNAtumorvirusgenomes. J. Mol. Biol.106:109-131. 14. Haseltine,W.A.,A.Maxam,and W.Gilbert. 1977. Rous sarcomavirusgenome isterminallyredundant: the 5' sequence. Proc. Natl. Acad. Sci. U.S.A. 74:989-993.

15.Hirt,B.1967.Selective extraction ofpolyoma DNA from VOL. 25,1978

http://jvi.asm.org/

(11)

infected mouse cell cultures. J. Mol. Biol. 26:365-369. 16. Hudson, B., W. B. Upholt, J. Devinny, and J. Vino-grad. 1969. The useof an ethidium analogue in the dye-buoyant density procedure for the isolation of closed circular DNA: the variation of the superhelix density ofmitochondrialDNA. Proc. Natl. Acad. Sci. U.S.A.62:813-830.

17. Khoury,A. T.,andH.Hanafusa. 1976. Synthesis and integration ofviral DNA inchickencellsatdifferent times after infection with variousmultiplicities of avian oncornaviruses. J. Virol. 18:383-400.

18. Kung,H. J., J. M. Bailey, N. Davidson, M.0. Nicol-son,and R.M.McAllister.1975.Structure,subunit composition, andmolecular weight of RD-114 RNA. J. Virol. 16:397-411.

19. Levinson,W. E. 1967.Fragmentation of the nucleus in Rous sarcoma virus-infected chickembryo cells. Virol-ogy 32:74-83.

20.Mizuuchi, K., and H. A. Nash. 1976. Restriction assay forintegrativerecombinationofbacteriophageXDNA in vitro: requirementforclosedcircular DNA substrate. Proc. Natl. Acad. Sci. U.S.A. 73:3524-3528.

21.Moscovici,C.,M.G. Moscovici,H.Jimenez,M. M. C. Lai,M.J.Hayman, andP. K.Vogt.1977.Continuous tissueculture celllinesderived fromchemically induced tumorsofJapanesequail. Cell11:95-104.

22.Radloff,R., W.Bauer, and J.Vinograd.1967. A dye-buoyant-density methodforthe detection and isolation ofclosedcircularduplexDNA:the closed circular DNA inHeLacells.Proc.Natl.Acad.Sci. U.S.A. 57:1514-1521. 23. Ringold, G.M., K. R. Yamamoto, P. R. Shank, and H. E. Varmus. 1977. Mousemammary tumor virus DNA in infected ratcells:characterization of uninte-gratedforms.Cell10:19-26.

24. Schwartz,D.E.,P.C.Zamecnik, andH. L.Weith. 1977.Rous sarcoma virus genome isterminally redun-dant: the 3' sequence. Proc. Natl. Acad. Sci. U.S.A. 74:994-998.

25. Smotkin, D.,F. K.Yoshimura,and R. A.Weinberg. 1976.Infectious, linear, unintegratedDNA ofMoloney murineleukemiavirus. J.Virol.20:621-629.

26. Studier, F. W. 1965.Sedimentation studies of the size andshapeof DNA. J. Mol. Biol. 11:373-390.

27. Sutton,W. D. 1971. Acrude nuclease preparationsuitable for use inDNA reassociation experiments. Biochim. Biophys. Acta 240:522-531.

28. Svoboda,J. 1960. Presence of chicken tumorvirus in the sarcomaof theadultratincubated after birth with Rous sarcomatissue. Nature(London)186:980-981. 29. Taylor,J.M.,and R.Illmensee. 1975. SiteontheRNA

of aviansarcoma virusatwhichprimeris bound. J. Virol. 16:553-558.

30. Temin,H. M. 1971. Mechanism of celltransformationby RNA tumor viruses. Annu. Rev. Microbiol. 25:609-648. 31. Varmus, H. E., R. V.Guntaka,C. T.Deng,and J. M. Bishop. 1974.Synthesis, structure and function of avian sarcomavirus-specificDNA inpermissive and non-per-missive cells. ColdSpringHarborSymp.Quant. Biol. 39:987-996.

32.Varmus,H.E.,R. V.Guntaka,W.J.Fan, S.Heasley, and J.M. Bishop. 1974. Synthesisof viral DNA in thecytoplasmof duckembryofibroblasts and in enu-cleated cells after infectionby aviansarcoma virus. Proc. Natl. Acad. Sci. U.S.A. 71:3874-3878.

33.Varmus, H.E., S.Heasley,J.Linn, and K. Wheeler. 1976.Use of alkalinesucrosegradientsinazonalrotor todetect integrated andunintegratedavian sarcoma virus-specific DNA in cells. J. Virol. 18:574-585. 34. Varmus, H.E.,and P. R. Shank. 1976. Unintegrated

viral DNA is synthesized in thecytoplasm of avian sarcoma virus-transformed duck cells by viral DNA polymerase.J. Virol. 18:567-573.

35. Varmus, H. E., P. K.Vogt,andJ. M. Bishop.1973. Integrationofdeoxyribonucleicacidspecificfor Rous sarcoma virusafter infection ofpermissiveand non-permissive hosts. Proc. Natl. Acad. Sci. U.S.A. 70:3067-3071.

36. Weil, R.,and J.Vinograd. 1963.Thecyclic helixand cycliccoil formsofpolyoma viralDNA. Proc. Natl. Acad.Sci.U.S.A.50:730-738.

37. Weinberg, R. A 1977. Structure of the intermediates leadingtotheintegratedprovirus.Biochim.Biophys. Acta 473:39-55.

38.Weintraub, H. 1972. Bi-directional initiation ofDNA synthesis in developing chick erythroblasts. Nature (London)New.Biol. 236:195-197.

http://jvi.asm.org/