Multiple structures of adeno-associated virus DNA: analysis of terminally labeled molecules with endonuclease R-Hae III.

(1)

Vol.18, No. 2 Printed inU.SA.

Multiple Structures of Adeno-Associated Virus DNA:

Analysis

of

Terminally

Labeled

Molecules with Endonuclease

RHaeIII

DAVID T. DENHARDT,* SHLOMO EISENBERG, KATALINA BARTOK, AND BARRIE J. CARTER

Department ofBiochemistry, McGill University, Montreal,Quebec, Canada H3G 1Y6,* and Laboratoryof

Experimental Pathology, National Institute of Arthritis, Metabolism andDigestive Diseases,National InstitutesofHealth, Bethesda, Maryland20014

Received forpublication25November1975

Thedouble-stranded form ofadeno-associatedvirus(AAV)DNAhasabout20

sitessensitivetoendonuclease R *HaeIII from Haemophilus aegyptius; the

frag-ments produced fallinto about13 size classes, 8 of which contain single

frag-ments.The location of theHaeIII-producedAAVfragmentsrelativetothethree

EcoRi

fragments was determined. Using revised figures for the molecular weights of the HaeIII cleavage products of+X174 replicative form DNA, we

calculatedthat AAV DNA contains about 4,000 nucleotides. After HaeIII

diges-tionofduplex DNA terminally labeled with32Pusingpolynucleotidekinase, the

majority offragments containing a 5' 32P label were about 40 nucleotides in

length,and fragments of similarsize weregenerated from each end, suggesting that theHae siteclosesttotheendiswithinthe terminal repetition. Two

more-slowly-migrating cleavage products also bore 5' 32p end label. These three

terminally labeled specieswerealsogenerated from single-stranded AAV DNA by digestion with HaeIII, and evidence thatonemayhave anonlinear

("rabbit-ear")structureispresented.Thepredominant5'terminal basewasidentifiedas

thymine for both the plus and minus strands of AAV. Single-stranded AAV molecules could not be efficiently covalently circularized by incubation with

polynucleotideligaseorligaseplus T4 DNA polymerase.

Adeno-associatedvirus(AAV) hasagenome

composed of approximately 4,200 nucleotides (4,000, accordingtoourdata), sufficienttocode for about 150,000 daltons of protein if all the nucleotides serve acoding function (see

refer-ence 2 for a review). However, because only

about 75% of thegenomeistranscribed (8) into

stable RNA, the aggregate amount ofprotein that canbecoded foris approximately 10°

dal-tons. The fact that the largest protein in the

virus has an apparent molecular weight of about 87,000to92,000(22,31)is consistent with the possibility that the AAV genome contains

only one gene and codes for only oneprotein

that is processed to yield the virion polypep-tides.

Forreplicationand for transcription, AAV is

normallydependentonadenovirus functions of

unknown nature (9); presumably, host cell

functionsarealsorequired. Becauseofthis

de-pendence on adenovirus and host functions, AAVshouldprovide auseful tool for investiga-tionof thereplicationofmammalian DNA. It is

likely thatthere are similarities between the

mechanisms of replication of the AAV genome

and the eucaryote chromosome. In particular,

AAV may yield clues as to how the ends of linear moleculesarereplicated; since DNA

po-lymerases are incapable of replicating the

ex-treme 3' ends ofDNA templates, some

provi-sion must be made so that information is not

lostatthese ends (11, 12).

Thegenomeof theAAV virioncomprisesone

single-stranded linear molecule of either plus

or minus polarity, and when the DNA is

ex-tracted from the virions the complementary strands will reassociatetoformaduplex, if

con-ditions permit (26, 30). Koczot etal. (23) and

Berns and Kelly (3) established that the

sin-gle-stranded viral molecules could form circles that could be seen in the electron microscope;

these studies suggested the presence of an

in-verted terminal repetitionofat least1.5%of the

genome. Studies of the properties of

double-stranded AAV DNA led Gerry et al. (16) to

conclude that AAV DNA contained a limited

number ofpermutations (perhaps only two) and aterminal repetition (notinverted)

represent-ing about 1% of thegenome.

It isevidentthat the structure of AAV DNA

is complex, perhaps so that the linear

single-stranded viral molecule can be replicated. We

672

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

ADENO-ASSOCIATED VIRUS DNA 673

have investigated aspects of the structure of

AAV DNA using acombinationof end-labeling

with polynucleotide kinase anddigestion with

the restriction enzyme fromHaemophilus

ae-gyptius, endonuclease R-HaeIII (27, 29), that cleaves the sequence 5' GGCC , at its center of

symmetry. Wehave used the terminology

sug-gested by Smith and Nathans (34) except that

wehave shortened it slightly for convenience. Endonuclease R-HaeIII will be referred to as

HaeHI; the fragments produced by HaeIII are designated A through M in this article,ordered

according to their mobility on our acrylamide-agarosegels. Further work is required before it can be concluded that this order also corre-sponds to their molecular weights (36).

Evi-dence is presented that AAV molecules have complementary sequences neareach terminus,

but that the terminal regions themselves, though self-complementary, are not always complementary to each other. This appearsto

lead to the formation of a terminal forked struc-turein aportion of thepopulation.

MATERIALS AND METHODS

Viral DNAs. AAV 2 DNA was prepared from AAV 2virionsproduced and purified as previously described (10). 3H-labeled and 33P-labeled kX174

replicative form (RF) DNA was purified (32) and wasgenerously provided by C. Hours.

Enzymes. The restrictionendonuclease R EcoRl wasobtained from MilesLaboratories. The restric-tion endonuclease R-HaeIII was purified as de-scribed (27), except thatthe nucleic acidswere

re-moved by precipitation with streptomycin sulfate rather than chromatography on Bio-Gel A 0.5 M. The spectrum ofOX fragments generated by this nuclease preparation and the salt concentration at which the enzyme eluted from phosphocellulose in-dicated that the activity was primarily that desig-nated HaeIII (29). Bacterial alkaline phosphatase and T4polynucleotide kinase wereprepared as de-scribed (14a). The Neurospora crassa single-strand-specific exo- and endonucleases were prepared and used as described (1).

Centrifugation. AAV DNAwaspurifiedfrom

en-zyme reactions, when necessary, by velocity sedi-mentation in neutral 5 to 20% sucrose gradients

containing1MNaCl,1mMEDTA,and50mM Tris-hydrochloride (pH 7.5). Samples (100 ,ul)were

lay-ered on 5-mlgradientsandcentrifugedfor4hat50 krpm in the SW50.1 rotor of the Beckman L265B

ultracentrifuge. Fractionswerecollectedthrougha

hole poked in the bottom of the tube, and those fractions containing the DNAwere pooledand di-luted with an equal volume of50 mM Tris-hydro-chloride (pH 8.0), and the DNA was precipitated

with0.1volume of3Msodiumacetate(pH5.5) and2

volumesofisopropanol.Precipitationwasallowedto occurfor12 h or more at -20C, and the DNAwas

collected by centrifugation at 10 krpmin the HB4 rotor of the Sorval RC2 for60 min. The DNA was

dissolved in 100 ,ul of 10 mM Tris-hydrochloride (pH 8).

Labeling of the 5' terminus. The DNA was incu-bated with1U ofbacterial alkalinephosphatase per mlat 65C for15minin30mMTris-hydrochloride (pH 8). EGTA [ethylene-bis-(Q3-aminoethyl ether)N,

N'-tetraacetic acid] was added to 6.5 mM, and the incubationat 65 Cwascontinuedfor another15min (21). Thesolution was then made 1.5 mM in potas-sium phosphate (pH 6.8), 7 mM MgCl2, 14 mM in /-mercaptoethanol, and 60 mM in Tris-hydrochloride (pH 7.4). ATP labeled in the gamma position (17) was added in a 100- to 1,000-fold excess over the molar amount of 5' ends present (0.1 to 5pmol). The specificactivity ranged between 2 x 104 and 2 x 105 counts/min per pmol in the different experiments; the final concentration of ATP was maintained be-tween 5and 10 uM. The amount ofpolynucleotide kinase required to quantitatively label the 5' ends of

acontrol polynucleotide preparation in 60min under theseconditions was determined, and this amount was added at 0 min and again at 30 min.After 1 h at 36C, EDTA was added to terminate the reaction, andthe DNA waspurified by velocitysedimentation

inneutral sucrosegradients. This step resulted in the separation of the linearduplexmolecules, used inmost of this work, from the faster-sedimenting

circular andoligomeric forms (7).

Cleavage with restriction enzymes. Endonucle-ase REcoRl wasused as described (6). Duplex AAV DNA (0.01 to1pmol) wasdigested overnightat 37C in 10mMeach NaCl, MgCl2, mercaptoethanol, and Tris-hydrochloride (pH 8) with sufficientHaeIH to givecompletedigestionin 12h. EDTAwasaddedto 20mM,bromophenol blue was added to 0.004%, and solid sucrose was addedtoabout0.5M.

Electrophoresis. The composite acrylamide-aga-rose gels contained the designated percentage of acrylamide, N, N' methylene-bis-acrylamide at

0.05the concentration ofacrylamide, 0.5%agarose

(Seakem,product of MarineColloids,distributed by

Bauschand Lomb) and E buffer (14). Polymerization wasinitiated with 0.02% ammoniumpersulfate and 0.04% N,N,N',N'-tetramethylethylenediamine. Thegels were prerun for 30 min inE buffer, and then the DNA samples were injected onto the upper surface ofthegels. The gelswere 17 cmlong in glass tubes with an internaldiameter of6 mm.The DNA was subjected to electrophoresis for 12 h at room temperature with a current of 3 mAinducedby a 60-Vpotentialdifference across the electrodes. Under

theseconditions, thebromophenolbluedyemarker

migrated about 11 to12cmintothe17-cmgel. After

completionof theelectrophoresis,thegelswere

frac-tionated into 1- or 2-mm fractions using a Gilson Aliquogel gel fractionator. The fractions were

di-gested with alkaline hydrogenperoxideovernightat 55C andneutralized, and the radioactivitywas de-termined with 10 ml of a Triton X-100-toluene-PPO (2,5-diphenyloxazole) cocktail inan Intertechnique

scintillationspectrometer withappropriatesettings

and corrections forbackgroundand cross-talk where necessary. Allgelsare plotted with the cathodeon

the leftsothat themigration of the DNAfragments

isfrom lefttoright.

Determination of the 5'-mononucleotides. The VOL. 18, 1976

http://jvi.asm.org/

(3)

674 DENHARDT ET AL.

DNA (about 0.1

/ig)

was digested with 100 ,tg of DNase I per ml (Worthington Biochem. Corp., EC 3.1.4.5, code D) for 45 min at 37 C in 5 mM MgSO4, 50 mM Tris-hydrochloride (pH 8). Glycine buffer (pH 9.1) was added to a final concentration of 60 mM, and phosphodiesterase I (Worthington Bio-chem. Corp., EC 3.1.4.1, code VPH, treated [35]to

inactivate the 5' phosphatase) was added to 50

jig/

ml. Thedigestion was continued for another 60 min and then frozen. A 10- to

15-Aul

sample ofthe

di-gestedDNA wasapplied to a washed (distilled

wa-ter, 1 M LiCl, absolute methanol) polyethylene-imine-cellulose thin-layer plate (Macherey-Nagel

Co., Cel 300 PEI) together with20 ugeach of the

fourdeoxynucleotide monophosphates dissolved in

water. The plate wasdeveloped in the first dimen-sion with 1 M LiCl,washed with methanol,and then in the second dimension with 1 M aceticacid-3 M LiCl (9:1) (28). The location of the mononucleotide markers was determined by inspection under UV illumination. The entire plate was scanned for 32P radioactivity by autoradiography, and the amount ofradioactivity in the spots was quantitatedin 10-ml of X-100-toluene-PPO scintillation fluid after

A~~~~~

12 *A'

10

o~~~~~~~~~~~~~~~C

0 D

|g

i

F c

elution of the mononucleotides with 1 ml of 0.5 M ammoniumbicarbonate.

RESULTS AND DISCUSSION Characterization of the fragments of the AAV duplex produced byHaeIII. 32P-labeled double-stranded AAV DNA was digested,

to-gether with

[3HWOX

RF, with the restriction endonuclease R-HaeIH from H. aegyptius as

describedinMaterials and Methods. The

prod-ucts of the reaction wereseparated by electro-phoresis incomposite4%acrylamide-0.5% aga-rose gels, and the distribution of radioactivity throughout the gelwasdetermined by

scintilla-tion countingof thedissolved gel slices. Figure

1shows the distribution of the 32P-labeled AAV

DNA fragments (solid circles) and the

3H-la-beled

OX

RFfragments(opencircles)inthegel.

The 4X RF [3H]DNAfragments were used as

molecular-weight standardssince their

molecu-lar weightshave been determined (24); 11

OX

0 2 4 6 8

lo

12

14 16

DISTANCE

MIGRATED

(cm)

FIG. 1. Acrylamide-agarose gelelectrophoresis pattern of theHaeIIIcleavageproducts ofthe AAV duplex. AAV duplex [32P]DNA and 4X174 RF [3H]DNA were digested together with HaeIII and subjected to electrophoresis in a 4%acrylamide-0.5% agarose gel as described in Materials and Methods. Migration is from left to right;1-mm fractions were collected. Symbols: (0) 3H-labeled

OX1

74 RF cleavageproducts, (0)

32P-labeled AAVduplex cleavage products. Inset: Calibration curve of the size of the 3H-labeled

OX

RF

cleavage fragments (-) against the square root of the mobility. The size of the(X genome was taken as 5,500 base pairs (33).

J. VIROL.

http://jvi.asm.org/

[image:3.503.77.459.302.587.2]

(4)

cleavage fragments were produced, two of

which had very similar mobilities and were

found inpeakno. 6at5.4cm.

The fragments produced from the AAV

du-plex fell into the 13 major mobility classes,

identifiedbythe letters Athrough M in order of

increasingmobilityasshown in Fig. 1.As

illus-trated in the inset to Fig. 1, an approximate

linearrelation exists between thelogarithmof

the size (e.g., the number of base pairs) of the

OX

markersand thesquarerootofthedistance

each fragment has migrated into the gel. This

allows anapparentmolecular weightto be

as-signedto each of the fragments A through M,

andthesearegiven in Table1;thesemolecular

weightswerecalculatedassumingthat the

cor-rect size of the

OX174

genome is 4,800

nucleo-tides (la).

As a first step towards locating these

frag-mentsonthe AAVgenome, theHaeIII

restric-tion enzyme pattern of the three AAV

frag-mentsproduced by digestion withendonuclease

R-EcoR1 was determined. This nuclease

cleaves the AAVduplexinto three fragments:

EcoR1 A,EcoRl1B, and EcoRlC, with

molecu-larweights of 1.6x 106, 1.1x 106, and 1.3x 105,

respectively (6).EcoRl A is theright-hand

ter-minus,EcoRliB is theleft-hand terminus, and

EcoRliC originates from themiddle of the

ge-nome (7); theycanbeseparated by

sedimenta-tion velocitycentrifugation on neutralsucrose

gradients. The HaeIII digestionpattern of the

intact duplex and the two largest fragments

producedbyEcoRi areillustrated in Fig. 2.

OX

RF DNA was added before the digestion with

R-HaeIIItoprovidemolecular-weightmarkers,

and thedigestsweresubjectedto

electrophore-sis on composite 3% acrylamide-0.5% agarose

gels. The three panels show, from top to

bot-tom,thedistribution ofradioactivityinthegels

of the fragmentsresulting from intact duplex,

the EcoRlDA segment, and the EcoRl B

seg-ment.The resolution in these gels isnotasgood

asinFig. 1because the acrylamide

concentra-tion was lower, and 2-mm rather than 1-mm

fractionswerecollected;nevertheless,the

simi-larity of the two patterns obtained from the

intact duplex (32P labeled) in Fig. 1 and the

intact duplex [radioactively labeled with

[3H]thymidine and density-labeled with

5-bromouraciltoabout 90%substitution]inFig.2

isapparent. Thetwopeaks labeled TA andTBin

the lower two panels contain the fragments

having as one terminus the end produced by

EcoRl. (Additionaldata in support ofthis

as-signmentare presentedbelow inFig. 5and6.)

The molecular weights of the various

frag-ments were determined from a calibration curve established using the known molecular

TABLE 1. Characteristicsofthecleavage productsof duplex AA V DNA produced by endonuclease

R HaeIII

HaeIII No. ofbase No. offrag- EcoRl

assign-sizeclass pairs" ments" ment

A 700 (640) 1

B 584 (550) 1 A

C 454(445) 2 A, B

D

253

(280) 1 A

F 210(214) 1 A

G 148(140) 2 A, A

H 113 (105) 2 A, B

I 92(83) 1 B

J 83(76) 1 B

K 71 (63) 1 '/2(?) B ('/2A?)

L 59 (51) 3 A, B, B

M 51(41) 4 A, A,B, B

TA 257 1

TB 183 1

The first number was determined from the square root relationship plotted in Fig. 1 (normal

duplex, 32P labeled); the numbers in parentheses

were derived using the linear relations shown in

Fig. 3 [bromouracil-substituted duplex, [3H]thy-mine-labeled]. The number of base pairs is calcu-lated assuming

4X174

RFhas 4,800 base pairs (la). bDependsontheassumption thatmigrationrate

isproportionaltothemolecularweight only.

weights of theHaeIIIfragmentsof AXRF

pres-ent inthegel.Inthiscase, asillustratedinFig.

3, agood linear relationwasobtainedbetween thelogarithm of the number of base pairsinthe fragment and themobility; whenplottedversus

the square rootof themobility, the fitwasnot

as good. We do not have any explanation for

why aplotagainstthesquare root in one case

(Fig. 1, 4% acrylamide) but not in the other

(Fig. 2, 3% acrylamide) allowsabetterfit,but Williamson (37) reported a similar

phenome-non. It may be a function of the acrylamide

concentration. The molecular weights esti-mated from the plots aretabulated inTable 1. The number offragments derived from one

genomeineachpeakcanbecalculatedfroman

analysis of the total radioactivity present in

each band; ifonly onefragment is present in

eachband, then theamountofradioactivityin

thatbandshouldbe proportionalto the size of

thefragment. The assumption implicit inthis

analysisisthat themobility of thefragmentis

proportionalto the sizeofthefragment andis

not affected by the base composition or

se-quence; this may be a poor assumption,

espe-ciallyforthe smallerfragments. Ananalysisof

the twogelsoftheintact AAVduplexshownin

Fig. 1and2ispresentedinFig. 4. Forthedata

in Fig. 1, a better linear relation is obtained

VOL. 18, 1976

http://jvi.asm.org/

(5)

676 DENHARDT ET AL. J. VIROL.

IntactA B C (D,E)F G H IUJ K L M

AAVsst99t999ttt

:3 4 5 6 7 8 9 10

§X

markers

0:

o . ... .

n,

EcoRlA B C TA

LT

G H t

10 F

s

₃ ₄ ₅₆

t

₇ ₈ ₉ ₁₀ _{IX markers}

5

LL * o

0

44

7 8 10

IXmarkers

0

z

W5

DISTANCEMIRTD(c)

0

4

8 ~~~2

1

6

FIG. 2. Acrylamide-agarosegelelectropherogramoftheHaeIIIcleavageproducts ofthe AAVduplexand theEcoR1 Aand EcoRl-B fragments. TheAAV DNApreparations were labeled with[3H]thymidine and substituted with 5-bromodeoxyuridine. 32P-labeled OX RF was digestedand subjected to electrophoresis

simultaneouslywith theAAVDNAina3%acrylamide-0.5%agarosegel.Thegelwasfractionatedinto2-mm

portions. A toMand3 to10indicate the distinctivecleavageproductsoftheAAVduplexandOX174RF

fragments, respectively. TA and TBaretheHaeIIIcleavageproductshavingone

EcoRi-produced

terminus.

Symbols:(O) 3H,(0)32P.

11 1. 11

http://jvi.asm.org/

[image:5.503.116.403.64.564.2]

(6)

when thelogarithm of thetotalradioactivityin

theband is plotted versus the squarerootof the distance migrated, whereas for the data of Fig.

2 a better fitisobtained when the mobility is used directly.Fromtheseplots,aswellas simi-lar plotsof thedatainpanelsBand C of Fig.2

(datanotshown), itappearsthatfragments A,

B,D, E,F, I, and J are unique; onlyone frag-mentineach mobility class originates from the

AAVduplex. There are two copies each of frag-ments C and H, one each derived from the

EcoRlFA and EcoRl1B segments. Two

frag500

-00QF

.-e)

< 500

Lii

on

m

cr 100

m500

z

NO A

\XB

c

Intact

0cC

D, E

00F

"o G

B

Eco

RI.A

keKL

TA

,E

e

oF o G

\ H

'o

cEco

RI B

><M

bTo

°<< K

, .~

100l

0 4 8 12

MOBILITY

(cm)

FIG. 3. Sizeofthe HaeIIIcleavageproducts.From thedatapresentedinFig. 2,calibrationcurveswere

determined byplottingthe numberofbasepairs in the Xl74fragments (assumingatotalof5,500base

pairs) againstthe distanceeachfragmentmigrated.

Fromtheposition ofeach AAVcleavageproduct,the apparentnumberofbasepairsin thatfragmentwas

estimated.

(MOBILITY,

mm)/2

(o .o)

a

z

m

z

02

a.

4 8 12 16

(MOBILITY,

cm)

(°- )

FIG. 4. Numberof fragmentsfromonegenomein

eachmobility classofAAVHaeIII cleavageproducts.

The upper curve(0) isaplot of theradioactivity in

eachpeak against the squarerootofthemobility;the dataarefromFig.1. The lowercurve(L)isaplot of

the radioactivity in eachpeak against the mobility; the dataarefrom Fig.2.PeaksC, G, and H(andD +

E inthe lowercurvewheretheywere notresolved)are

replotted after division by 2; Lisreplottedafter

divi-sionby 3; and Misreplotted afterdivision by4.

mentsof the mobility class G are presentinthe

duplex, and both are found in the EcoRl-A segment. The two smallest fragments, L and M, are difficult to quantitate accurately

be-causeof the broadness of the bands and evident

heterogeneity. The best fit to all the gels is obtained, however, when it is assumed that thereare threefragments inL and fourinM, distributed as indicated in Table 1. Figure 1

also indicates that there may be onefragment running fasterthanM.Theonly fragment that is consistently ambiguous is K; possibly it is

migrating anomalously because ofan unusual

base composition orpossibly inthepopulation

ofAAVduplexesanonintegralnumber of

frag-mentsof thatsize areproducedbecause of

het-erogeneity (note that the terminally labeled component beta [see below] migrates in this position) in the population. We cannot distin-guish these alternatives atpresent.

VOL. 18, 1976

http://jvi.asm.org/

[image:6.503.254.450.69.334.2] [image:6.503.53.244.206.570.2]

(7)

678 DENHARDT ET AL.

Identification of the terminalfragments

us-ing polynucleotide kinase. The HaeIII

frag-mentsofAAV DNA possessingthe termini of

theintactmoleculewereidentifiedby labeling

those termini with 32P prior to the digestion

withHaeIII. The DNA wasfirst dephosphoryl-ated with alkaline phosphatase and then

re-phosphorylated using polynucleotide kinase

and [Ly-32P]ATP. The terminally labeled

mole-culesweredigested withHaellI,and the

result-ingfragmentswereseparated by electrophore-sis in the composite acrylamide-agarose gels.

The distribution ofradioactivityinthese gelsis

shown in Fig. 5. Most of the 32P-labeled

frag-ments derived from the intact molecules

mi-gratedonthe leadingedge of the peak

contain-ingthe fragments in the sizeclassM.

Invaria-bly asecond peak containing 10 to 25% of the

32Pwasobservedto migrate justahead of

frag-ment G. Often there wasalsoathirdpeak,very

closeto K, thatcontained5 to 10% of the total

a. 0161

Io

x

CL,

32P. In most gels, at least 70% ofthe total32P

labelrecoveredwasfoundinthetwopeaksnear

G andM. Theapparentmolecularweights ofa

duplex that would migrate at these positions

weredetermined using the AAV fragments as

standards;asshownintheinset(Fig. 5),aplot oflog molecularweightversusmobilitygave a

reasonablystraight line. Agel patternsimilar

(3Hdistributionapparently identical; 32P

pres-ent inall three peaks in theabove-mentioned

proportions) to that shown in Fig. 5 was

ob-tained from adigest of the circular and

oligo-meric DNA sedimenting ahead of the duplex

DNA(equivalenttofractions3 to 8ofFig. 5a in

reference 7). This suggests that the minor32P peaks are not derived from unusual, rapidly sedimentingmolecules and that the oligomeric molecules haveasimilarsetofterminal

config-urations.

Thefact that between 60 and80% of the:32P from the intact duplex migrated in an

appar-4 8 12

DISTANCE

MIGRATED

(cm)

4 .;a

3

1

N

2 -x

I 2

0a.

N

0aL,

cV

FIG. 5. Acrylamide-agarose gel electropherogram of the cleavage products produced by digestion with HaeIII of terminally labeled AAV duplex DNA. The3H-labeled AAV duplex was labeled with 32P at the

5'-terminiusingpolynucleotide kinase, digested withHaeIII,andsubjected to electrophoresis on 4%

acrylamide-0.5%agarose gels asdescribed in Materials and Methods. Symbols: (0) 3H, (0) 32P. Inset: Plot of the size of each AAVfragment versus its mobility. The arrows indicate the position and apparent size of the32P-labeled

fragments.

J. VIROL.

http://jvi.asm.org/

[image:7.503.122.404.294.603.2]

(8)

ently homogeneous peak suggested that the

two HaeHI terminal fragments were of the samesize, and thus within the terminal

repeti-tion. This was confirmed by examining the digestion patterns of terminally labeled

EcoRlA andEcoRlFBfragments.The data are

presented in Fig. 6; both EcoRl fragments

yielded 32P-labeled Hae fragments migrating

with theMmobility class,as well as thesecond

component migrating ahead of G. An

asymme-try in the labeling ofthe two ends is

appar-ent.TheterminiproducedbyEcoRl, labeled TA

and TB in the two panels, were labeled about

75% more efficiently than the "natural" end.

This could be forconformational reasons, such

as the presence of the protruding 5' end

pro-duced by EcoRl (19) or because the terminal A

is labeled more efficiently than the terminal

bromouracil(seebelowand reference25). Com-parison of the distribution of32P in these two

panels also reveals that each fragmentwas

con-taminated with about 10% of the other

frag-ment.

Location of fragmentA inthegenome.The

°B'

36 OC

EcoRI A

30

TA

24

° 18

RE

1I2 "FDY

1GM

....

...

L12V ISF

2

size of the EcoRlC fragmentwas obtained by subjecting it to electrophoresis together with

OX

RFmarkers. A value of 183 to 192 base pairs

(assuming 4,800 nucleotides in

OX

DNA) was

obtaineddependingonwhether thedatainFig.

7 wereplottedusingthe mobility orthe square

rootof the mobility (the datafitequallywell). AftertheEcoRl* C fragment was exposed tothe HaellI nuclease, no change in mobility was

detected, suggestingthat therewere noHaeIII sites withinthatfragment; however, asitevery

close to an end would nothave been detected. When the number ofbasepairs inEcoRlC is addedtothe number of basepairs in HaeIII-TA and HaeIII-TB (Table 1), an aggregate size of

627 basepairs isobtained. This isclose to the

sizeof the fragmentHaeHI -A thatwas not

pres-ent in either EcoRlA orEcoRlB. From this

resultandthe data given inTable 1wededuce

thearrangementoffragmentsindicatedinFig.

8a.

Digestion ofsingle-strandedAAV DNA with

HaeIII. The ability of single-stranded AAV DNA to form circular molecules (3) canbe

ex-16

ca

12

x

0-8 0

a-N

fn

12 16 4 8

DISTANCE MIGRATED (cm)

FIG. 6. Acrylamide-agarose gel electropherogram of the cleavage products produced by digestion with HaeIIIof terminally labeledEcoRi AandEcoRi*Bfragments. These fragmentswerepurified fromanEcoRi digestof[3H]thymidine-labeled, bromouracil-substitutedAAVduplex. TheEcoRi cleavagefragmentswere

labeled with 32Pusingpolynucleotide kinase, digestedwithHaeIII,andsubjectedtoelectrophoresis on4%

acrylamide-0.5% agarose gelsasdescribedinMaterialsand Methods.Symbols:(0) 3H, (0)32P. Thepeaks arelabeledasdescribedinFig.2.

VOL. 18, 1976

http://jvi.asm.org/

[image:8.503.53.445.324.605.2]

(9)

.103 plaind

by

theformation ofstructureslike those

l0 z drawn in Fig. 8b. If such structures areformed

and if

HaelIl-sensitive

sitesare presentin the

SWS. EcoRIC duplex portions, then terminally labeled

se-quencesshould be cleaved fromsingle-stranded

8~~~

.z,,,DNAby

the restriction enzyme and should

.

(BLTY"246 have the same mobility as fragments derived

1.(MOBILITY)'2 from the

duplex. Any

single-stranded

DNA

6.1 fragments derived by cleavage of the

single-5..6t9markersed DNArand byHaelIIwould be expectedto

x migratedifferentlyinthesegels. To testthese

ideas, theexperimentshown inFig. 9was

per-Ilb

4.

v formed.

a. In this experiment, 32P-end-labeled

single-N.i

@315'

stranded AAV DNA (the minus strand) was

2 l

ill°~~~~~~~

digested

with HaeIIIand

subjected

to

electro-phoresisongels usingthesameconditions

em-. \ ! |0 t 5 f ployed for the duplex DNA. As can be seen

M

!ikt\

li ..;

M6

_a

(Fig. 9),

essentiallyall of the 32Pwasfoundin

4 the same three peaks as were obtained when

DISTANCE MIGRATED (cm) end-labeled duplex DNA was digested. If the

FIG. 7. Size of the EcoRl C fragment of AAV DNA. The3H-labeled EcoRl -C cleavage productwas

subjectedto electrophoresis in 4% acrylamide-0.5%

15-agarosegelsasdescribedin Materials andMethods.

The HaeIII cleavage products of 32P-labeledOXRF

were includedas markers and the datawereplotted

assuming a genome size of 5,500 nucleotides for 0

OX174. 4

FIG. 8. (a) Partialordering of the cleavage prod-uctsofAA VduplexDNA produced byHaeIII. The conventional representation of the AAVgenome is shownatthe top, and underneath the threeEcoRi

cleavage productsaredrawnapproximatelytoscale.

Thenumberof basepairs(bp) iscalculatedfromthe

sumofthe HaeIIIfragmentsasgiven inTable1.The

location and relative sizes of A and the terminal

fragmentsTA, TB,and M*aregiven. M*appearsto

migrate slightly ahead ofM (Fig. 5). (b) Suggested structuresfor circularsingle-stranded and linear

du-plexAAVDNA.Forreasonsoutlined in thetext,the

single-stranded molecules are believed to form the

sameterminalduplexstructuresasthe linear duplex

bybase-pairing of thetermini.Proposed locations of HaeIII cleavage sites are indicated byarrows. The

variousstructuresarenotdrawntoscale althougha

is themost-slowly-migratingcomponentandyisthe

fastest.

0

0o

0

N

.

X-Io

0 4 8 12 16

[image:9.503.64.254.46.278.2]

DISTANCE MIGRATED (cm)

FIG. 9. Acrylamide-agarose gel electropherogram ofthe HaeIIIcleavage productsofsingle-stranded 3H-labeled AAV DNA (the minus strand) terminally

labeled with 32P using polynucleotide kinase. The

DNA was digestedandsubjected toelectrophoresis

on4%acrylamide-0.5%agarosegels usingthesame

conditionsasused inFig.5and 6. The insetshows the electropherogram ofthe DNA before digestion

withHaeIII.Symbols: (0) 3H, (0)32P.

(0)

5'(T) + strand (light)

- strand (heavy) 5'(T) EcoRI B,-1450bp EcoRI-C,-220bp EcoRIA,-2270bp

ML ,K, A

M J, I, H, D, C,) (B, C, E, F, G, G, H, L, M,) M*

Te TA

(b)

Q='-' <9

680 DENHARDT ET AL. J. VIROL.

http://jvi.asm.org/

[image:9.503.270.458.295.566.2] [image:9.503.63.257.373.478.2]

(10)

ADENO-ASSOCIATED VIRUS DNA 681 DNA was not digested with the restriction

en-donuclease, no radioactivity entered the gel

(Fig. 9, inset), thus providing strong evidence that the three peaks seen afterHae digestion were products of Hae action. We infer from these results that the same three types of

frag-ments are generated by HaeIII digestion of the

single-stranded DNAas aregenerated from the duplex. Several discrete peaks of[3H]DNAare

also evident, suggesting that single-stranded

AAVDNA, like fl DNA (19) and

OX

DNA (5),

can be cleaved at specific locations by HaeHI. Since underourconditions the single-stranded DNA fragments migrated differently from the duplex fragments, we cannot with certainty de-terminewhich peaks correspond. A gel pattern similartothatseen in Fig. 9 wasalso obtained when terminally labeled plus-strand DNA was digested.

We propose that the three terminally labeled structures(alpha, beta, and gamma,inorder of

increasing mobility) seen in the gels shown in

Fig. 5, 6, and 9 have the configurations alpha, beta, and gamma drawn in Fig. 8b. Hae sites located as indicated by the arrows would

gener-ate fragments of the various sizes found. We

think the real situation is still more complex, butintheabsence of more informationfurther speculation is not useful. To determine if there was any single-stranded DNA in the terminal regions, the DNA was exposed to the single-strand-specific endonuclease from N. crassa,

either before orafterdigestion with HaeIII. In

both experiments (datanotshown), theamount

of32P inall threepositions wasreducedtoabout

thesame extent.ThelossOf32P ismostly,ifnot

entirely, the result of the fact that the nuclease will attack duplex DNA slowly from the ends, presumably asthe result of"fraying." We

con-clude that atleast by this criteriontherewere

no single-stranded regionsinthe terminal

frag-ments.

There areanumber of experiments thatcan

bedonetocheckpredictions basedonthe hypo-thetical structures illustratedinFig. 8b.One of these is that themobility of alphashouldvary

with the pore sizeofagel indifferentfashion relative to beta, gamma, and normal linear duplex fragments. Because duplex DNA

frag-ments migratethrough the gel in an"end-on"

fashion (13) the apparent molecular weight of

alpha, relative to linear duplex markers,

should increase as the acrylamide

concentra-tion isincreased,since it willbe morestrongly

retardedbyvirtueof itsnonlinearstructure.To

test thisprediction,a seriesofgels ofdifferent

concentrations wererun,andtheapparent

mo-lecular weightsof the threeterminally labeled fragments were estimated relative to the

known AAV fragments. The data are

summa-rized in Fig. 10; in the 3.5% gel, alpha ran ahead of fragment

G,

whereas in the 6.5% gel it ran behind G. The other twofragments did not appear to change their mobility signifi-cantly. We therefore argue that a portion of bothsingle-stranded and double-stranded AAV molecules have the "rabbit-ear" structure showninFig. 8b.

Wededuce from these results that individual single-stranded AAV DNA molecules exist with two different terminal sequences, each of which is self-complementary and capable of forming "fold-back" structures, but which are

not complementary to each other. In other words, a portion of the population does not have

acomplete inverted terminal repetition. If the terminal sequences were perfectly

complemen-tary to each other they would be expected to

form the more stable single duplex (like

gamma) rather than the double "fold-back"

structure (alpha). It appears that about

one-quarterof the moleculesinthe population con-tain non-complementary terminal sequences, although this could be an underestimate if the kinase labeling was not 100% effective. These results are not incompatible with the

conclu-(I)

at

z

LbJ

(f)

cr-0li 0m

150 60 0.0-0" *~000

CC~0~

1001-. * . *_

50-ci-O

okI I , I

3 4 5 6

%

ACRYLAMIDE

FIG. 10. Apparent size of the terminally labeled fragmentsas afunction ofacrylamideconcentration.

3H-labeledAAVduplexDNAwas terminallylabeled

with32P,digested withHaeIII,andsubjectedto elec-trophoresisongels ofdifferentacrylamide

concentra-tions, butallwith0.5%agarose. Calibrationcurves were establishedfrom the positions of the 3Hpeaks (assuming an AAV genome size of 4,000 nucleo-tides), and the apparentsizeof the 32P-labeled

frag-ments wasthenestimated. a, 8, and yarebelievedto

have thestructuresillustrated inFig. 8b. VOL.18, 1976

http://jvi.asm.org/

[image:10.503.273.427.349.556.2]

(11)

682 DENHARDT ET AL.

sions(2,4, 16)that there are two or more types

of AAV molecules that differbyhaving a

lim-itedterminal permutation. However, itshould benoted that for technical reasonsithasnot so

farbeen possibletocloneAAV.

Identification of the5'terminal nucleotide.

Inthe initial attempt toidentify the5' endsof

AAV DNA, the plus and minus strands,

sepa-rated by centrifugation in CsCl equilibrium density gradients after substitution of the thy-mine with5-bromouracil, were individually

in-vestigated. The DNA was labeled with

[y-32P]ATP usingpolynucleotide kinase and then digestedwithDNaseIandvenom phosphodies-terase. Theresulting 5' mononucleotides were

resolved by two-dimensional thin-layer

chro-matography on polyethyleneimine-cellulose plates, and the32Pwaslocatedontheplates by

autoradiography. Both strands yieldedsimilar

autoradiograms. The spots from the minus-strandchromatogramwereexcised,andthe

ra-dioactivity was determined. As can be seen

from the data inTable 2, mostof the 32P was

recoveredinaregionof thechromatogram that didnotcoincide with any of the four markers.

As a control, Escherichia coli DNA

substi-tuted with5-bromouracil and uniformly labeled with 32P was similarly digested and quanti-tated. Figure lla shows an autoradiogram of

the 32P-labeled5-bromouracil-substituted DNA;

5' bromodeoxyuridylic acid migrated to the same location as the 5' terminal nucleotide of

the5-bromouracil-substituted AAV DNA. That the terminal nucleotide ofunsubstituted AAV

a )

A

®/

DNA was thymine was confirmed by labeling

the 5' end of AAV (the duplex this time) not

substituted with bromouracil; 50% of the

re-covered radioactivity was found in thymidylic

acid. Theautoradiogram is shown inFig.

lib,

and it can be seen that the onlyradioactivity

detected was contained in the four common

mononucleotides. The spots were excised, the nucleotides were eluted, and theradioactivity

presentwasmeasured.The resultsaregivenin

[image:11.503.269.459.208.286.2]

Table2.

TABLE 2. Determinationofthe5'endsofAAVDNA

%32precovered

Nucleotide

Expt1 Expt2b

5'dAMP 5.9 13.6

5'dTMP 4.4 50

5'dGMP 22 18.5

5'dCMP 8.8 16.9

5'dBrUMP 59

a The minus strand of 5-bromouracil-substituted AAV DNA was used. In this experiment, theends

were quantitatively labeled as calculated fromthe

3Hand32Pspecificactivities. Autoradiogramsof the

plus and minus strands were similar, with mostof

theradioactivityin5'-dBrUMPand alesseramount

in 5'dGMP.

bThe AAV duplex not substituted with

5-bro-mouracil. In this experiment, 50% of the input3H (theDNA was labeled with[3H]thymidine)and 46% of the input 32P (representing 5' terminal

phos-phates) was recovered in the fourmononucleotides. The autoradiogram is shown in Fig. lib.

Co

_CL

GC

0S..

G

FIG. 11. Autoradiograms of two-dimensional chromatograms of enzymatic hydrolysates of AAV DNA. Digestions with DNase I and phosphodiesterase I and two-dimensional chromatography on thin-layer polyethyleneimine-cellulose plateswereperformed as describedinMaterials and Methods. The reactions were completesince theonly detectableradioactivity was in mononucleotides; less than 5% ofthe input radioactivity remained atthe origin, and 50 to 75% of input radioactivity was accounted for as mononucleotides. (a)

32P-labeled,5-bromouracil-substitutedE.coliDNA.(b)AAV duplex DNA not substituted with5-bromouracil but

labeledinternally with[3H]thymidine and labeled at the 5' terminus with32P.(c)EcoRIArestriction enzyme fragment terminally labeled with 32p, substituted (about90%0o) with5-bromouracil, and labeled internally

with[3H~thymidine.

J. VIROL.

I-1 i

http://jvi.asm.org/

[image:11.503.66.460.419.561.2]

(12)

[image:12.503.54.450.79.148.2]

TABLE 3. Formation ofprotected 5' ends by incubation with DNA polymerase and DNA ligase"

OXts7DNAb Random linearDNAN AAVDNAd

Prepn

lig lig+pol lig lig+pol lig lig +pol

Control 2.5 4.9 2 1.6 3.35 6.8

Incubated 7 27.5 3.3 8.7 3.35 11.3

Nophosphatase 205 206 91 89 43.5 59

%Protection 2.2 11 1.4 8.0 0 7.6

aThe DNA, internally labeled with

[3H]thymidine

and the 5' end labeledwith32P,

was

incubated with

polynucleotide ligase (lig) or polynucleotide ligase plus T4 DNA polymerase (lig + pol) and thentreated withphosphatase (20). The numbers shown in the first three lines are the ratio of 32P counts per minute to 3H counts per minute after acid precipitation. The "control" was not incubated with ligase or ligase plus polymerase. The "No phosphatase" preparation was not reacted with phosphatase. "% Protection" is [(Incubated-Control)/NoPhosphatase] x 100.

b TheOXlinear DNA was purified from phage grown under conditions of ligase deficiency in E. coli ts7 (20).The specific activity was 8.1 x 104counts/min per,ug, and the P/H ratios given in the table have been multiplied by 103.

cThe random linearOXDNAmolecules were prepared by nicking circular single-strandedOXDNA with

DNase I (15). Lessthan one nick per molecule wasintroduced, and the linear molecules were purified by sedimentation on alkaline sucrose gradients. The specific activity of the DNA was 2.5x 103counts/min per

,gg, and the P/H ratios given in the table have beenmultiplied by 10.

d The AAV DNA in this experiment was the 5-bromouracil-substituted Hstrand. Its specific activity was

4.4 x 104counts/min per ,g,and the P/H ratiosinthetable have been multiplied by 102.

In another experiment, the separated

EcoRlA and EcoRlB fragments were

termi-nally labeled with 32p and examined. In both

cases, two spots wereobserved on the

autoradi-ograms: one in the position of

5'-deoxyadeno-sine monophosphate, as expected, since EcoRl

leavesa 5'A(18),and the secondintheposition of 5-bromodeoxyuridylic acid. An

autoradi-ogram of the EcoRlAchromatogram is shown

inFig. lic. Thisexperimentconfirms thatthe

5' ends ofboth the plus and minus strands of

AAV DNA are thymine. Consistent with the

observation thatTA is labeled more efficiently

than the natural end (Fig. 6a) is the

observa-tionthat dAMP is somewhatmoreheavily

la-beled than deoxybromouridine monophosphate. Circularization of AAV DNA. Iwaya et al.

(20) described an experiment that suggested

that linearsingle-stranded kX DNA molecules could beconvertedtocircular molecules under appropriate conditions. The experiment

in-volvedisolating linearDNAfromvirionsmade under conditions ofpolynucleotide ligase defi-ciency,labeling the5'ends with32Pusing

poly-nucleotidekinase,and thendemonstratingthat

someof the 32pcould be made resistantto

bacte-rial alkaline phosphatase by incubation with

T4DNA polymerase and T4 DNAligase. This

could have occurred if a structurelike beta in

Fig. 8b wereformed. Tosee ifAAVcould form

suchastructure, thesameexperimentwas

per-formed with single-stranded AAV DNA. The

details ofthe experimentaregiven inTable 3. Theresultwasthat AAV DNAcouldbe

"circu-larized" toabout thesame extent (5to 10%)as

the linear

OX

DNA molecule. The criteria for

circularization included resistance to alkaline phosphatase and velocity sedimentation in al-kalinesucrosegradients to the positionof

circu-lar DNA; this is not rigorous, however, since

anydimers thatmightbeformedbythe

polym-erase-ligase reaction would also exhibit these

properties. As acontrol, the same experiment

wasperformed with linear

OX

DNAmolecules having randomly located ends generated by nicking circular single-stranded DNA with DNaseI;unexpectedly,protectionof the5'ends

toabout thesamelevelwasachieved. Thusit is notpossibletoconclude that the smallamount

of"circularization"observed for AAV DNA and the linear

OX

DNA from the ligase-defective

hostisevidence foranyparticularstructure at

theends. This result alsosuggeststhata major-ity of themolecules cannothave their termini

inthe (3 conformation.

ACKNOWLEDGMENTS

Wethank H. E. David Lane for the32P-labeled, bromo-uracil-substitutedE. coli DNA,Christian Hours for the32p_ labeled4X174RFDNA, Carol Kerr for the[y-32P]ATP,and Linda Pallett for typing the manuscript.

Thisresearch was supported by the National Cancer Institute ofCanada, the Medical Research Council of Can-ada, andtheUnitedStates Public Health Service.

LITERATURE CITED

1. Bartok, K., B. Harbers, and D. T. Denhardt. 1975. Isolation and characterization of self complementary sequences from 4X174 viral DNA. J. Mol. Biol. 99:93-105.

la. Berkowitz, S. A., and L. A. Day. 1975. Molecular weight of single-stranded bacteriophage fd DNA. High speed equilibrium sedimentation andlight scat-teringmeasurements. Biochemistry13:4825-4831. 2. Berns, K. I. 1974. Molecularbiology ofthe

adeno-associ-VOL. 18, 1976

http://jvi.asm.org/

(13)

684 DENHARDT ET AL.

ated viruses. Curr.Top.Microbiol. Immunol. 65:1-20. 3. Berns, K. I., and T. J.Kelly, Jr. 1974. Visualization of the inverted terminal repetition inadeno-associated virus DNA. J.Mol. Biol. 82:267-271.

4. Berns, K. I., J. Kort, K. H.Fife, E. W. Grogan, and I. Spear.1975.Studyof the finestructureof adeno-asso-ciatedvirusDNA withbacterial restriction endonu-cleases.J.Virol. 16:712-719.

5. Blakesley, R. W., and R. D. Wells. 1975. Single-stranded DNA from 4X174 and M13 is cleavedby certain restriction endonucleases. Nature (London) 257:421-422.

6. Carter, B. J., and G. Khoury. 1975.Specific cleavages of adenovirus-associatedvirusDNAbyarestriction en-donuclease REcoRl-characterization of cleavage products.Virology 63:523-538.

7. Carter, B. J., G. Khoury,and D. T. Denhardt. 1975. Physical map and strand polarity of specific frag-ments ofadeno-associated virus DNA produced by endonucleaseREcoRl. J. Virol. 16:559-568. 8. Carter,B.J.,G. Khoury, andJ. A. Rose. 1972.

Adeno-associated virusmultiplication. IX. Extentof tran-scription of the viral genome in vivo. J. Virol. 10:1118-1125.

9. Carter,B.J.,F.J.Koczot,J.Garrison,J. A.Rose,and R. Dolin. 1973. Separatefunction providedby adeno-virusforadenovirus-associatedvirusmultiplication. Nature (London) New Biol. 244:71-73.

10. Carter, B. J., and J. A. Rose. 1974. Transcription in vivo of a defective parvovirus: sedimentation and electrophoretic analysis of RNA synthesized by ad-eno-associated virus and itshelper adenovirus. Virol-ogy 61:182-199.

11. Cavalier-Smith, T. 1974. Palindromic basesequences andreplication of eucaryote chromosomeends. Na-ture(London) 250:467-470.

12. Denhardt, D. T. 1972. A theory of DNA replication. J. Theor.Biol. 34:487-508.

13. Dingman, C. W., M. P.Fisher,and T.Kakefuda.1972. Role ofmolecular conformation in determiningthe electrophoreticproperties ofpolynucleotides in aga-rose-acrylamide gels. Biochemistry 11:1242-1250. 14. Edgell,M.H., C. A. HutchisonIII, and M. Sclair. 1972.

Specific endonuclease Rfragments ofbacteriophage kX174 DNA. J. Virol. 9:575-582.

14a. Eisenberg,S.,B. Harbers,C.Hours,and D. T. Den. hardt. 1975.The mechanism ofreplicationofkX174 DNA. XII. Non-random location of gaps in nascent

OX174RF IIDNA. J.Mol.Biol. 99:107-123. 15. Fiers, W., and R. L. Sinsheimer. 1962. The structure of

the DNA ofbacteriophage4X174.III. Ultracentrifu-galevidence for a ringstructure. J.Mol. Biol. 5:424-434.

16. Gerry, H. W., T. J.Kelly and K. I. Berns. 1973. The arrangementof nucleotide sequences in adeno-associ-ated virus DNA. J. Mol.Biol. 79:207-226.

17. Glynn, I. M., and J. B. Chappell. 1964. A simple method for thepreparation of32P-labeledadenosine triphosphate of high specific activity. Biochem. J. 90:147-149.

18. Hedgpeth, J., H. M.Goodman, and H. W. Boyer. 1972. DNA nucleotide sequence restricted by the R1 endo-nuclease. Proc. Natl. Acad.Sci.U.S.A. 69:3448-3452. 19. Horiuchi, K., and N. D. Zinder. 1975. Site specific cleavage ofsingle-stranded DNA by aHemophilus restriction endonuclease. Proc. Natl. Acad. Sci. U.S.A. 72:2555-2558.

20. Iwaya, M., S. Eisenberg, K. Bartok, and D. T.

Den-J. VIROL. hardt. 1973. Mechanism of replication of single-strandedOX174DNA. VII.Circularization of the pro-genyviral strand. J. Virol. 12:808-818.

21. Johnson, P. H., A. S. Lee,and R. L. Sinsheimer. 1973. Production of specific fragments of 4X174 replicative form DNAby arestriction enzyme from Haemophilus parainfluenzae, endonuclease HP. J. Virol. 11:596-599.

22. Johnson, F.B., H. L.Ozer, and M. D. Hoggan. 1971. Structural proteins of adeno-associated virus type 3. J.Virol. 8:860-863.

23. Koczot, F. J., B. J. Carter,C. F. Garon, and J. A. Rose. 1973. Self-complementarity of terminal sequences withinplus or minusstrands of adeno-associated vi-rus DNA. Proc.Natl. Acad. Sci. U.S.A. 70:215-219. 24. Lee, A. S., and R. L. Sinsheimer. 1974. Acleavage map

ofbacteriophage 4X174 genome. Proc. Natl. Acad. Sci. U.S.A. 71:2882-2886.

25. Lillehaug, J. R., and K. Kleppe. 1975. Kinetics and specificity of T4 polynucleotide kinase. Biochemistry 14:1221-1225.

26. Mayor, H. D., K. Torikai, J. L.Melnick, and M. Man-del. 1969. Plus and minussingle-strandedDNA sepa-rately encapsidated inadeno-associatedsatellite viri-ons.Science166:1280-1282.

27. Middleton, J. H., M. H. Edgell, and C. A. Hutchison III. 1972.Specific fragments ofOX174DNAproduced byarestriction enzymefromHaemophilus aegyptius, endonuclease Z. J. Virol. 10:42-50.

28. Randerath, K., and E. Randerath. 1967. Thin-layer separation methods for nucleic acid derivatives, p. 323-347. In L.Grossmanand K. Moldave (ed.), Meth-ods in enzymology, vol. 12A. Academic PressInc., New York.

29. Roberts, R. J., J. B.Breitmeyer,N. F.Tabachnik, and P. A. Myers. 1975. A second specific endonuclease fromHemophilus aegyptius. J. Mol. Biol. 91:121-123. 30. Rose,J.A.,K. I.Berns, M. D.Hoggan,and F. Koczot. 1969.Evidence for asingle-stranded adeno-associated virus genome:formation of a DNAdensityhybrid on release of viral DNA. Proc. Natl. Acad. Sci. U.S.A. 64:863-869.

31. Rose, J. A., J. V. Maizel, Jr., K. Inman, and A. J. Shatkin. 1971. Structural proteinsof adenovirus-as-sociated viruses. J. Virol. 8:766-770.

32. Schekman,R.W.,M.Iwaya,K.Bromstrup,andD. T. Denhardt. 1972. Mechanism ofreplication of4X174 DNA. III. An enzymicstudy of the structure of the replicative form II DNA. J. Mol. Biol. 57:177-199. 33. Sinsheimer, R. L. 1968.Bacteriophage 4X174and

re-lated viruses. Prog. Nucleic Acid Res. Mol. Biol. 8:115-170.

34. Smith,H.O., and D. Nathans. 1973.Asuggested no-menclature for bacterial host modification and re-striction systemsand their enzymes. J. Mol. Biol. 81:419-423.

35. Sulkowski,E., andM.Laskowski.1971.Inactivationof 5'nucleotidaseincommercial preparations of venom exonuclease (phosphodiesterase). Biochim. Biophys. Acta 240:443-447.

36. Thomas, M., and R. W. Davis. 1975. Studies on the cleavageofbacteriophage lambda DNA with EcoRl restrictionendonuclease.J. Mol. Biol. 91:315-329. 37. Williamson,R. 1970. Properties ofrapidly labeled DNA

fragments isolated from the cytoplasm of primary cultures ofembryonic mouse liver cells. J. Mol. Biol. 51:157-168.

http://jvi.asm.org/