Overlapping retrovirus U5 sequence elements are required for efficient integration and initiation of reverse transcription.

(1)

JOURNAL OFVIROLOGY, JUIY1991, p. 3864-3872 0022-538X/91/073864-09$02.00/0

Overlapping

Retrovirus U5

Sequence

Elements

Are

Required for

Efficient

Integration

and

Initiation of

Reverse

Transcription

DAVID COBRINIK,t ASHOKAIYAR, ZHENG GE,MICHAEL

KATZMAN,t

HAN HUANG,§

ANDJONATHAN LEIS*

Department ofBiochemistry, Case Western Reserve University SchoolofMedicine, Cleveland, Ohio44106 Received 4February 1991/Accepted 9 April 1991

Asecondary structureinthe 5' noncoding regionof avianretrovirusRNA,calledtheU5-leader stem, was

shownpreviouslytohavearole in initiation ofreversetranscription (D. Cobrinik,L. Soskey, andJ. Leis, J. Virol.62:3622-3630, 1988). Wenowshow thatanadditional RNAsecondarystructureneartheU5 terminus, called the U5-IRstem,isalsoimportantforreversetranscription. Mutations thatdisruptthe U5-IR stemcause

areplicationdefect associated with bothadecreaseinsynthesis ofviral DNA in infected cells andadecrease

ininitiation ofreversetranscriptioninmelittin-permeabilized virions. Structure-compensatingbase substitu-tionsintheU5-IRrestorereversetranscriptionefficiency. In viralDNA, U5-IRsequencesareincludedinthe U5term'inalregionthat functionsas aviralintegrationdonor site. When basesubstitutionsareintroducedinto thesesequences,areducedefficiencyofintegrationinvitro andinvivois observed. Theseobservationsindicate thatU5-IRsequenceshaveastructural role inreversetranscriptionofviral RNAandasequence-specificrole in theintegration ofviral DNA.

Theearlysteps ofretrovirus replicationconsistofreverse

transcription ofviral RNAandintegrationof viral DNAinto cellular DNA. Reversetranscriptionis initiatedbyextension ofaspecifictRNAprimerannealedtosequencesnearthe 5' endofthe viralgenome.Anumber ofcomplexsteps result in theproduction of double-strandedlinear viral DNA contain-inglongterminalrepeat(LTR)sequencesin thecytoplasmof infectedcellsaswellaslinear and circular viral DNA forms incellnuclei (8). The linearform hasrecentlybeenshownto be the primary substrate for integration. This reaction in-volves cleavage of the U5 and U3LTRterminibythe viral INproteinfollowed byligationtocellularsequencesthatare

also likelyto becleaved byIN (5, 12, 15, 21, 26).

Avianretrovirusparticlesareinfectiousasearlyas10min after budding (30), yet there is no detectable reverse tran-scription priorto cellinfection (4, 13, 31). The mechanism that restricts reverse transcription in virions is notknown. However, we have recently found thatan RNA secondary

structurecalled the U5-leaderstem affects theefficiency of initiation andhave proposed that thisstructure servessucha

regulatory role (7).

TheU5-leader stem isapotential RNA secondary

struc-ture consisting ofsequences in the middle of U5 and in the leader region adjacent to the primer-binding site (7). Muta-tions which disrupt this structure impairreverse

transcrip-tion in infected cells, while mutations which alter the

se-quence but retain the stem structure have no effect (7).

Studies withpermeabilized virions have localized the defect to the initiation ofreverse transcription. Disruption of the

U5-leader stemstructurecauses adecrease in incorporation

* Correspondingauthor.

t Present address: WhiteheadInstituteforBiomedical Research, Cambridge, MA 02142.

tPresentaddress: DepartmentofMedicine, TheMilton S. Her-shey Medical Center, Pennsylvania State University College of

Medicine, Hershey, PA 17033.

§ Present address: Department ofPathology and Immunology,

Case WesternReserve University School of Medicine, Cleveland, OH44106.

of the firstdeoxynucleotides fromthe3'OH terminus ofthe tRNATrP primer, while the amount of thetRNATFP primer annealedtothe RNA primer-binding site is equivalenttothat found inwild-type virions (7).U5-leaderstemstructuresalso

appeartobe conserved among most retrovirus families (1, 7).

The U5-leaderstemof avianretrovirusesis in aregion of RNA whichcanpotentially formnumeroussecondary struc-tures (14, 17). In this studyweinvestigated whetherone of these other structures, termed the U5-inverted repeat

(U5-IR)stem(see Fig. 1A), could also affect initiation ofreverse

transcription. In viralRNA, thisstructureis composed ofan

invertedrepeatsequenceadjacenttotheprimer-binding site. In viral DNA, the U5-IR stemfalls within the U5 terminal

sequences that arerecognized by the IN protein and func-tionsas aviralintegrationsite(21, 22). Thus, mutationsthat

disrupt the U5-IR stem structure and alter its nucleotide

sequencecouldalso result indefectstointegration.Ourdata indicate that U5-IRsequences are indeed requiredfor both replication functions.

MATERIALSANDMETHODS

Reagents. The Klenow fragment of DNA polymerase I (4

U/4l),T4DNAligase(2,000U/pul),T4polynucleotide kinase (10U/pl),andvariousrestrictionenzymes werefrom

Boehr-ingerMannheimBiochemicalsorNewEngland Biolabs, Inc.

[cx-32P]dATP (800 Ci/mmol), [a-32P]dCTP (3,000 Ci/mmol), and [y-32P]ATP (6,000 Ci/mmol) were from New England

NuclearCorp. pTZ18 DNA wasobtained from Pharmacia.

Thermostable DNA polymerase (Pyrostase; 4 U/,l) was

fromMolecularGeneticsResources.

Construction of U5-IR viral mutants. pRCAS-B and pRCAS-neo were generously provided by S. Hughes

(Na-tional Cancer Institute-Frederick Basic Research Facility). pRCAS-B is the pRCAS vector (20), except that it has a

central EcoRI fragment (spanning gag andpol genes)

de-rived from a cloned Bryan high-titer Rous sarcoma virus

(RSV) genome. pRCAS-neo contains the neomycin phos-photransferase gene inserted into the pRCAS ClaI cloning

3864

Vol.65,No. 7

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

site. pDC1O1B was derived from pRCAS-B by deletion of the EcoRV fragment between nucleotide 3655 (28) and pBR322 nucleotide 189. pDC102 was constructed by diges-tion of RCAS-neo with Sacl, partial digesdiges-tion with RsrII,

formation of blunt ends with the Klenow enzyme, and ligation toHindII-digested pTZ18. The gag-pol EcoRI frag-ment was then replaced with that of the Bryan high-titer clone to form pDC102B. U5 mutations were introduced into pDC1O1B by insertion of synthetic deoxynucleotides be-tweentheRsrII and BstEII sites. All clones were confirmed by sequencing.

Transfections were performed by the calcium phosphate

precipitation technique (27) with ligation products derived

from 7 ,ug ofdigested pDC102B and 5 ,ug of

HpaI-digested pDC1O1B (or its derivatives) per semiconfluent 60-mm-diameter dish of QT6 cells. The cells were fed 1 day after transfection and split on day 2, and selection was

initiated on day 2. Virus production was monitored by an exogenous reversetranscriptase assay (7) with the following

modifications: the final Nonidet P-40 concentration was

0.067%, and 15

RIJ

of virus was assayed per 30-,ul reaction mixture.

Analysisofreversetranscription products inpermeabilized virions. Virus preparations and melittin reactions were as previously described (7), except that the reaction volumes were 30

RI.

The amount of virus used in each reaction was that which incorporated 0.5 pmol of dCTP in a 30-min reverse transcriptase assay performed on a 1/10 dilution of the recovered virus and supplemented with excess dCTP. Thefinal concentrationand specificactivity of[a-32P]dATP

were 2 ,uMand 330 Ci/mmol, respectively, when dATP was present as the only deoxynucleotide, and 25 ,uM and 40

Ci/mmol, respectively,whenitwasused in combination with 1 mM (each) dCTP, dGTP, and dTTP. The dATP was not rate limiting at these concentrations. The final melittin concentration was 7.5 ,ug/ml.

Analysis of integrated viral DNA. High-molecular-weight cellular DNA waspreparedby the method of Hirt (19). The DNA was sheared by passing it through a 21-gauge needle severaltimes, heatedto 55°C, layered onto a gradient of 10 to40% sucrosein 0.1 MNaCl-10mMTris HCI (pH8.0)-1 mMEDTA, and spunfor16 h at 25,000 rpm in anSW28 rotor at4°C. Fractions containingDNAof>25kbwerecombined,

precipitatedwith 2volumesof ethanol, and used in polymer-asechain reactions (PCRs). Primers complementary to the gag gene (nucleotides 2251 to 2270) and to the pol gene

(nucleotides 2667 to2583) (28) were used in PCRs with the

following cycling conditions: 92°C for2 min, 55°Cfor 90 s, and then72°C for2 min.Amplification of pRCAS-BDNA,at concentrations between 0.01 to 1 pg, was linear up to 25

cycles.Thecell-derivedDNAbeinganalyzed forviral DNA wasthusamplified for24cyclesasdescribed elsewhere(23)

by using Pyrostase anda Hybaid thermal cycler. Amplified

DNAproductsweresubjectedtoagarosegel

electrophoresis

and transferred to nylon membranes (27), and viral DNA sequences were detected by hybridization to

[32P]dCTP-labeledprobes(2 x 109

cpm/Vig).

Theprobeswereprepared by DNA polymerase I (Klenow

fragment)

extension of primers complementary to the gagand

pol

genesand DNA

containingportions ofthese genes

amplified

from the viral clonepATV8 (6).

IN protein cleavage reactions. Reactions were

performed

asdescribed previously (22).

Integrative recombination invitro. Substrate

oligodeoxy-nucleotidesrepresenting wild-typeor mutantU5 sequences

were prepared as described previously (22) and are as

follows:

U5wild-type plus strand 5'TGAAGCAGAAGGCTTCA3' U5 wild-type minus strand 3'TCGTCTTCCGAAGTAA5'

U5 S2 plus strand U5 S2minus strand U5 S4plus strand U5 S4minus strand U5 S4C plus strand U5 S4C minus strand

5'AGCAGAAGGCAACA3'

3'TCGTCTTCCGITGTAA5'

5'AGCAGAAGGGAAGA3' 3'TCGTCTTCCCTTCTAA5'

5'TCCAGAAGGGAAGA3'

3'A&GTCTTCCCTTCTAA5'

The underlined sequences indicate base changes from the wild type for the S2, S4, and S4C mutations (see Fig. 1B and C). After the plus and minus strands are annealed, the resultant duplexes have a two-base 5' minus-strand over-hang.

Integrativerecombinationreactions were carried out iri 10-,u reaction volume containing 1 pmol of each of the duplexes shown above,5' 32plabeled only in the plusstrar.,

(105

cpm/pmol); 25 mM Tris HCl (pH 8.3); 10 mM 2-me,

captoethanol; 3 mM MgCl2; and 1 pmolof IN. The reaction mixtures were incubated at 37°C for 90 min. The reactions were stopped by the addition of 10 ,ul of sequencingloading buffer (22), and the mixtures were heated to90°C for 5 min andanalyzedon a20%polyacrylamide-7 M ureasequencing

gel (22).

RESULTS

The RSV(Schmidt-RuppinA)RNAsequencefrom

nudle-otides 55to 131, which includes the primer-binding site and thepreviouslystudiedU5-leader stem structure (7), is shown inFig. 1A. This region also includes an additional secondary structure termed the U5-IR stem. This structure comprises

the inverted repeat sequencebetweennucleotides 81 and 101

(U5-IR) andcanpotentially form a7-bp stemwitha 5-base loop. Analyses ofthe RSV 5' noncoding region have indi-cated that the U5-IRstemisthermodynamicallystableand is conserved among all members of the avian sarcoma and leukosis retrovirusfamily forwhich sequenceinformationis available(1, 3,7). This structuralconservationsuggests that theU5-IR stemmay havean importantrole in viral

replica-tion. We sought to understand this role by analyzing the effectsofmutations in the U5-IR sequenceon viral

replica-tion.

Production ofU5-IRmutantvirus.Thevirusvectorsystem used for these studies was derived from the pRCAS-neo

vector described by Hughes et al. (20). pRCAS-neo is a

Schmidt-Ruppin Astrain of RSV in which the viralsrcgene isreplacedby the neomycinphosphotransferasegene. Two

plasmids were derived from pRCAS-neo:

pDC1O1B,

which

contains the 5' half ofpRCAS-neo, and

pDC102B,

which

contains the

majority

of the viral genome but lacks se-quences5' oftheleaderregionSacl site and those 3' ofthe U5regionRsrIIsite in the downstreamLTR

(Fig.

2). Both of these plasmids contain Bryan

high-titer

viral sequences, which confer the high-titer growth phenotype, within por-tions of the gag andpol genes.

To obtain virus-producer cells,

pDC102B

and

pDC1O1B

were digested with

HpaI,

ligated to assemble a

full-length

viraltranscriptionunit,andtransfected into

quail QT-6

cells. Cells that

expressed

the virus-encoded neo gene were then selected in the presence of G418. The

advantages

of this

on November 10, 2019 by guest

http://jvi.asm.org/

(3)

3866 COBRINIK ET AL.

A A

G . G .-..

. C

. U

U U5-IRSTEM . c * A UW U G U G C\ C C C PBS G A C/ G/

GU/

uA A * G U

* G US-LEADERSTEM

* G * A * A

* U

-A 131

B

A9 128 G A C G A A G U A C G u C C A C G A G C A U C A G C A G U C C C U U A G AA G * G . c .U * U . C * A . U U UGG U

GA

A

C

S6 C CC G A C G U G

uA

A * G * U U * A * G * G * G * A * A A U A

C

S2 S2C GA GAA

A G A G

C .G C. G

G .C G C

A

G .C G. C

U .A U A

A.*U A.*U

C U C U

S4 S4C

A A

GA GA

A G A G

C .G C. G

G

M.ICi1

A A U .IA

A . U A.U

C U C U

FIG. 1. Predicted RNAsecondarystructuresfor the 5'noncoding regionofwild-typeandmutantRSV(Schmidt-RuppinA)RNAnearthe

primer-bindingsite(PBS). (A) Wild-typestructure.Numbersindicate nucleotides from the viral RNA 5' end. ThePBS, U5-leader stem,and

U5-IRstemareindicated. (B) U5-IRstemmutationsA3 and A9. TheA6,A28,and S6 mutationsaredescribedinthe text.Theyareanalyzed elsewhere(1, 6, 7).Sequence deletionsareindicatedbyopenbars;thesequencesubstitution isindicatedbythestippledbar.(C)U5-IRstem substitution and compensating substitution mutationsS2, S2C, S4, andS4C. Sequence substitutions are boxed, and the effects ofthese substitutionsontheU5-IR stem structureareshown.

system are that US mutations can easily be cloned into pDC1O1B (see MaterialandMethods), thattransfectionscan

be performed without additional subcloning steps, and that largeamountsofdefectiveviruscanbeproducedfromQT-6

cells. In addition, truncation of the downstream LTR of pDC102B ensures that mutations introduced into the

pDC1O1B U5 region cannot undergoreversion by recombi-nation with wild-type U5sequences.

Threetypesof mutations wereintroduced into the U5-IR stem: deletionmutations A3 andA9, substitution mutations S2 and S4, and structure-compensating substitution muta-tionsS2C and S4C (Fig. 1B and C). The A3 mutation does not disrupt the U5-IR stem but decreases the spacing be-tweenthe U5-IRandU5-leaderstemstructures.The A9,S2, andS4mutationsdisrupttheU5-IRstem,whiletheS2Cand S4C mutations retain the U5-IR stem structurebut alterits

sequence. We found that release of virus, as measured by

reverse transcriptase activity, from pooled producer cells

was roughly the same for the wild type and each of the mutant viruses described above and showed no correlation

with viral growth rates (data not shown). This strongly suggests that late replication events such as transcription,

splicing, translation, particle assembly, and budding were

notaffected by the U5-IR mutations.

Analysis of viral growth rates. Viruses with each of the mutations described above, obtained from QT-6 producer cells,wereusedtomeasuretherateof viral growthfollowing infection of fresh QT-6 cells. Reverse transcriptase assays

usingexogenous template primers were performed directly

on cell supernatants both to equalize the amount ofvirus used for infection andtomonitor theappearanceofprogeny

virus. Results represent the average of several infections which had nearly identicalrates of viralgrowth.

The substitutionmutationsS2andS4resultedinimpaired viralgrowth compared with the wild type (Fig. 3). TheA9 and A3 deletion mutations also resulted in impaired and delayed growth of virus (datanotshown). The viruses with structure-compensating mutations, S2C and S4C, displayed improved growthkinetics comparedtotheiruncompensated S2 andS4counterparts, butwerestillsignificantly defective in growth compared with thewild type(Fig. 3). The

differ-ences between S2 and S2C and between S4 and S4C were

reproduciblyobserved infour separate kinetic virusgrowth experiments. This partialrestorationof growthefficiency for S2C and S4C suggests that the U5-IR stem structurehas a

role inreplicationthat isatleastpartiallysequence

indepen-dent. However, the delayed growth observed with viruses containing structure-compensating mutations indicates that there is a major sequence-dependent replication function

encoded in the U5-IR.

Each ofthe virus mutants analyzed in this study is only partially defective. The eventual appearance of virus is

unlikely to be the result of genetic reversions, since the

progeny virus exhibited the same impaired-growth

pheno-type as that exhibited by virus derived from the original

producercells (datanot shown).

Analysis ofreverse transcription in permeabilized virions.

We have previously shown that reverse transcription is

A

_G A C G A A G U A C G U C C A C G A G C A U C A G C A G U C C C U U A G 55 J. VIROL. rli 9 16

on November 10, 2019 by guest

http://jvi.asm.org/

[image:3.612.103.534.72.365.2]

(4)

EcoRV

HpaI HpaI

Hpal

[image:4.612.88.530.79.336.2]

r

POL I EWN INEDI U3 IRIU5I pTZ18

FIG. 2. Strategy forproduction of virus containing U5 mutations. pDC1O1B (orU5 mutantderivatives) andpDC102B aredigested with

HpaI; theproductsareligated andtransfected into QT-6 cells.(A)Structure of pDC1O1BandpDC102B.Wide bars indicate viralsequences.

Thinlines indicateprokaryoticvectorsequences.TheU3, R, U5,and leader (L) noncodingregionsand thegag,pol,env,andneogenes are

indicated. The sizes of indicated regions of the viralgenome arenot toscale.ori,originof replication.(B)Structure of the ligation product

fromwhichboth theneogeneandfull-length viral RNAcanbe expressed.

substantiallyless efficient when theUS-leaderstem structure

is disrupted (7). This was established by analyzing the

synthesis of the initial intermediate inreversetranscription, a 101-nucleotide cDNA (cDNA101) extending from the

tRNATrp primertothe 5'terminus of viral RNA, in melittin-activated virions (Fig. 4E). In order to evaluate whether mutations in the U5-IR stem also affect reverse

transcrip-tion, equal amounts of wild-type and mutant virus, deter-mined by an exogenous reverse transcriptase assay, were

incubatedwith melittin and [t-32P]dATP in the presence of

dCTP, dGTP,anddTTP. cDNAswereisolated from virions

and treated with alkali to remove the covalently linked

tRNATrp, and the amount ofcDNA101 was determined by

polyacrylamide gel electrophoresis (Fig. 4B). The substitu-tionmutations S2 and S4 resulted in decreased amounts of

cDNA101synthesized,withalargerdecreasebeing observed for the S2 virions. Structure-compensating mutations

re-stored thereverse transcription efficiency,with that ofS4C restored essentially to the wild-type level. Only a partial

restoration was obtained with S2C (Fig. 4B). These

differ-ences arenotduetoapackagingdefect inS2CandS4C (see below).

ThedeletionmutationsA3 and A9 also resulted inreverse

transcription defects. Because of the deletion of U5

se-quences, the cDNA productsinthesecases werethree and nine bases shorter than that of the wild type, respectively (Fig. 4C). The A9 mutation resulted in barely detectable amounts of cDNA92, while the A3 mutation resulted in reduced levels of cDNA98 compared with the wild type. These results correlated with the growthcharacteristics of A3 and A9.

The first two deoxynucleotides incorporated onto the 3'

OH terminus of the tRNATrP primerare dAMPs (Fig. 4E).

Thus, initiation ofreverse transcription can be determined by monitoring the appearance ofa dA-tRNATrP formed in

virions when[ao-32P]dATP is included in the melittin reaction mixtures as the only deoxynucleoside triphosphate. The

relative amounts of dA-tRNATrP produced with the wild type and each of the mutant viruses (Fig. 4A and D) were

similar to the amounts of cDNA intermediate products obtained when all deoxynucleotides were present. These results indicate that disruption of the U5-IR stem impairs initiation of reverse transcription in melittin-activated

viri-ons.

Analysis offull-length double-stranded viral DNA in the cytoplasmofcells 12 h after infection with each of the U5 mutant viruses described above yielded results parallel to

thoseobserved inFig. 4; theamount of viral DNAdetected bySouthern blottingcorrelated with the amounts of strong-stop cDNA anddA-tRNATrPproduced in melittin-activated virions(data notshown). Theseresults cannotbeexplained by a RNA packaging defect, since equal amounts of wild-type and mutant virus (asjudged by reverse transcriptase

activity measured with exogenous template primers)

con-tainedequivalentamountsofviral RNA(as judged byadot blot assay using a labeled viral cDNA probe [data not

shown]). Taken together, these results indicate that the

U5-IRstem structure, like the U5-leader stem structure, is necessary for efficient initiation ofreverse transcription in

infected cells.

Effect of U5 mutationson integration in vivo. The in vitro

and in vivoanalyses described above revealed thatreverse

transcription was notdefective for the S4C mutantandwas

only partially defective for the S2C mutant. Since these

A

Rsai.

B 2

on November 10, 2019 by guest

http://jvi.asm.org/

(5)

E

0.

I-5;

c)

w

Cn

E-0.

cn

z

r

E

w

cn

w w UJ

C.

0

C.)

cn

:

cn

z :

w

cn

Cl)

w w

2000

1000

0

2000

1000

0

l

0

10 20

DAYS AFTERINFECTION

10 20

DAYS AFTERINFECTION

FIG. 3. Viral growthratesfollowing infection of QT-6 cellswith equal amounts of wild-type and U5-IR substitution mutant virus. Theviruswasdetectedbyassayofreversetranscriptase activityin

cell supernatants as described in Materials and Methods.O, wild-typevirus. (A) *,S2 virus;O,S2C virus. (B) A,S4 virus; A, S4C

virus.

viruses displayaconsiderable reductioningrowth efficiency

(Fig. 3), replication steps other than reverse transcription

seemed likelyto be affected. We felt that integration could be impairedin suchmutants, since theS2, R2C, S4,andS4C substitutions alterthe U5 terminal sequence recognized by

the viral INprotein (22). Toevaluate this possibility, the in

vivo integration efficiency formutantviruswascompared to

that ofthe wild type.

Equalamountsof eachmutant virus (measured byreverse

transcriptase activity) wereusedtoinfectQT-6 cells,andthe

high-molecular-weight Hirt pelletDNA (19), which contains

integrated provirus sequences, was extracted from cell

nu-clei after 12 h. In order tominimize the amount of

contam-inatingunintegrated viralDNA,high-molecular-weight DNA (length of>25 kb) was isolated by sucrose densitygradient

centrifugation as described in Materials and Methods. Viral

DNA was then detected by amplifying a 418-bp fragment fromthe pol region by the PCR technique and then

hybrid-izingthefragment by Southern blottingtoa32P-labeled viral

DNA probe. Various concentrations of pRCAS-B DNA (0.01 to 1 pg) wereamplified (Fig.5A) inparallel reactions.

This approach is similar to that previously described for

quantitating cellular mRNA levels by PCR (27).

The amount of viral DNA detected after 24 cycles of amplification for each of the U5mutantsis showninFig.5B. Comparison of the amount of amplified 418-bp fragment from the wildtype (Fig.5B, lane 4)with control DNA(Fig. SA,lane 2)indicatedthat approximately0.1pgof the former

was present before amplification. In the absence of virus infection, amplification of a 418-bp fragment was not de-tected (Fig. SB, lane 3). Theamounts ofS2, S2C, S4, S4C,

A9, andA3proviral DNAwereestimatedbydensitometryto

be 18, 25, 29, 44, 7, and 20% of the amount ofwild-type DNA, respectively (Fig. 5B).

A decrease in the amount of proviral DNA detected in cellsunder these conditions couldresult from eitheradirect effect upon integration or a decrease in the amount of the

linearviral DNAprecursortointegration synthesized during the early steps of viralreplication. Since the S4C virus does

not display a reverse transcription defect, we estimate that this mutationmustcausemorethanatwofold decreasein the efficiency ofintegration ofviral DNA. Since the S2, S4, and S2C mutants have defects in reverse transcription, the decreaseinintegrated proviruscannotbe attributedsolelyto

an integration defect. For example, the fivefold decrease in viral DNA detected for the S2C mutant might result from a

three- to fourfold decrease in integration efficiency and a

1.5-fold decreasein reversetranscription efficiency (Fig. 4).

Similar combined reverse transcription and integration

de-fects arelikelyto occurfor the S2and S4 viruses. Forthe A3

and A9 viruses, the reduction in proviral DNA is likely to result solely from defectsin reverse transcription, since the integration donor site sequences are intact in these mutants (21, 22).

Effect of US mutations on integration in vitro. To further define integration defects produced by U5-IR mutations, duplex oligodeoxynucleotides corresponding to the 16 ter-minal US bases of these mutants were evaluated for their abilitytobenicked by theviral INproteinin vitro(Fig. 6A). Previous studies have shown that purified avian retrovirus IN protein nicks double-stranded DNA at the in vivo viral integrationsite, twobasesfrom the U3 and U5LTR termini (22). In this assay, cleavage ofa 5'-32P-labeled plus-strand

16-meratasitetwobasesfrom the US terminus is monitored by conversion toa 14-mer.

In the presence of

Mg2",

the S2, S4, and S4C mutations nearly abolished the ability of the US terminal sequences to be cleaved (Fig. 6B, lanes b, e, h, and k). However, some

14-merproductcouldbedetected for theS4Csubstrate after

alongerexposure of theautoradiogram (Fig. 6B, lane n). In the presence of Mn2 , the endonuclease activity of IN was

20times higher than in the presence ofMg2+, but cleavage

occurred at manysites (lane c). When analyzed with Mn2 ,

cleavage of the S2 substrate was substantially reduced (lane

f). Cleavage of the S4 and S4C substrates was detected,

although thetotalamountofcleavage wasless than withthe wild-type substrate (lanes fand i). These data indicate that base changes in the US-IR stem decrease the efficiency of cleavage by IN and that the two-base substitution (S2) is

moredeleterious to cleavage thanthe four-base substitution (S4).

Oligodeoxynucleotides can also be used to demonstrate thejoiningreactioninvitro (12, 21). This is evidenced bythe

appearanceofoligodeoxynucleotides longer thanthestarting

substrate which arise by a self-integration reaction. The

substrates used in these experiments representLTR termini already nicked at the integration site. This allows one to examine primarily thejoining reaction, although some DNA

B

I

=13^

A A

A~~A

a &

A'-1

A

N *..A1 -.,~A I .

I

J. VIROL.

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.612.65.298.79.426.2]

(6)

A

N

3 U)enc en

M _ __m

B

0 0

NCM N T t

3: u) u cni

M _iuiat_

m

1*

_0).owl go-low_w _o.w101i

C

I-c) 0)

<

101F

i1-

98

9

D

0) CO

I-M-_

_1

:~

-- 92

tRNATRP_LdAl 4I

tRNA P-IdA1 4 _

4 AA

t I RF US I P L

FIG. 4. Analysis ofreverse transcription products for wild-type and U5-IR mutant virus in melittin-activated virions. Shown aretRNATm primer extensionproducts formed after 30minofincubation of virions with melittin and[32P]dATPin the absence (Aand D) or the presence (Band C)of the other three unlabeled deoxynucleotides. The viruses (wild type [WT] orU5mutantsA3,A9, S2, S2C, S4, andS4C)are indicatedabove the respective lanes. The migration distances of thedA-tRNAT"',cDNA101, cDNA98,andcDNA92products are as indicated. Mdenotesa32P-labeled125-bp DNA fragment, added to each reaction upon termination, that serves as a control for the recovery and loading of reaction products.(E)Schematic representation of the initiation of reverse transcription. AA, initial dAMP residues incorporated onto the 3' OHterminus of thetRNATrpprimer; closed arrow,cDNA101reversetranscription intermediate; open arrows, inverted repeat sequences; P, primer-bindingsite; L, leader region.

cleavage also occurs and results in oligodeoxynucleotides migrating faster than the starting substrate.

As shown in Fig. 6C, large integrative recombination

products were detected in the presence of

Mg2"

with the

wildtype (lane 8) but not with substrates containing the S2

mutation (lane 2). Such products, however, were detected

A B

(o 0

Cn CN4CM *

< cn aS 3~U) c) U U) <

418 _S _ _

~~~~~~~~418

12 3 1 2 3 4 5 6 7 8 9

FIG. 5. Quantitation of integrated wild-type and U5 mutant

provirus. (A) Amplification ofa418-bp fragment derived from the polgene,asdescribedinMaterials andMethods, by using pRCAS-B

DNA. Amplified DNA was analyzed on a 0.8% agarose gel and

transferredtoanylon membrane, and viralsequences weredetected

byhybridizationtorandomly primed32P-labeled cDNAasdescribed

inMaterialsand Methods. Theamountsof DNAamplifiedwere as

follows: lane 1, 1.0 pg; lane 2, 0.1 pg; and lane 3, 0.01 pg. (B) Chromosomal DNAs were isolated from cells 12 h after infection

with wild-type or U5 mutant viruses and used as templates for

amplification of the same 418-bp fragment. Equal amounts of

amplified DNAwereanalyzedon a0.8%agarosegelasdescribed for

panelA. The mutantviruses used for the initial infections of cells from whichthe DNA was isolatedare indicated above the lanes.

QT6indicatesDNAisolatedfrom uninfected cells.

with the S4 and S4C substrates (lanes 4 and 6), but in reduced amounts compared with the wild type. Densitomet-rictracing oftheautoradiogram indicates that the amount of

integrative recombinant product shown by the large arrow-head wasreduced42and63%for S4 and S4C, respectively.

Similar resultswereobtained in thepresence ofMn2+ (data notshown). Combiningthe invitro results displayed in Fig. 6B and C, the S2 mutation is predicted to cause a greater

integration defect than the S4 mutation because ofits se-verelycompromised nickingandjoining reactions.

DISCUSSION

Theexperiments described above support thenotionthat U5-IR sequences have roles in both reverse transcription

and integration. In viral RNA, U5-IR sequences form a

secondary structure necessary for efficient initiation of re-verse transcription, while in DNA these sequences are

specifically recognized by IN and are required for integra-tion. This represents an instance of a cis-acting element

possessing unrelated replication functions in its RNA and DNAforms.

The U5-IR stem RNA secondary structure, like the U5-leader stem, is a primary determinant for efficiency of initiation ofreverse transcription. There is a decrease in

initiationin melittin-activatedvirionswhen the U5-IR struc-tureisdisrupted andanimprovedreverse

transcription

and

growth efficiency when the U5-IR stem is restored by compensatingmutations. Wecannotruleoutthe possibility

that the

compensating

mutations activated initiation

inde-pendentlyof the

inhibitory

effect of the S2orS4substitution

on November 10, 2019 by guest

http://jvi.asm.org/

[image:6.612.92.534.73.321.2] [image:6.612.62.305.500.591.2]

(7)

U3 RU5 U3 RU5

AGCAGAAGGCTTCATT AGCAGAAGGCaaCATT AGCAGAAGGgaagATT t,

TCGTCTTCCGAAGTAA TCGTCTTCCGttGTAA TCGTCTTCCcttcTAA

WT S2

F

~~

IN IN IN IN

M Mg Mg Mn M MgMgMn

164 w416

15 14

-K

13 S1

12*

1051

9

8

7

a b c

13 12

11i

10

a

S4

- IN IN M Mg MgMn

16

15 '

14

f

13

1290 .1

119w

109..7

a

;cCAGAAGGgaagATT

igGTCTTCCcttcTAA

S4C

M IN IN - IN IN MgaMgMn MgMgMn

16j

15

14

13*

0

121

119

10*

810~8

81 ^

84

w

*

89

7

*-d e f g h

j

ik I mno

C

S2 S4 S4C WT

S

1 2 3 4 5 6 7 8

mutation. This appears unlikely, however, since the A9 deletiondisrupts the U5-IR stemstructurewithoutchanging the bases involved in the S2 and S4 mutations and also results in an initiation defect. Additional evidence that

supportsareplication role for the U5-IR stemisthe

conser-vation of this structure inseveral retrovirus families (1). Independent evidence supports a reverse transcription

rolefor U5-IRsequencesin murine leukemia virus (MuLV).

Murphy and Goff (24) have shown thatatleastonemutation

[image:7.612.69.490.69.571.2]

(dlAva2) constructed in the comparable region of MuLV RNAcauses areplication defect without affecting encapsi-dation of viral RNA. This mutation would disrupt the equivalent ofthe U5-IR stem. Another mutation, dlAva3, which wouldalso disrupttheU5-IRin MuLV, doesnothave

FIG. 6. EffectofUS-IRmutationsonintegrationandIN protein

cleavage. (A)U5-terminalsequencesforwild-type (WT)andU5-IR

mutant viruses are shown. Sequences represent double-stranded oligodeoxynucleotides of 16nucleotidesinlengthusedto assayIN protein cleavage. Each sequence is aligned over the appropriate column inpanelB.Lowercase letters indicatebasechangesfromthe WT. Asterisks indicate a 5'-end 32P-label on plus strands. (B)

Double-stranded oligodeoxynucleotides were incubated in the

ab-sence(lanesa,d,g,j,andm)orpresence(lanes b,c, e,f,h, i, k,1,

n,ando)ofIN,with eitherMg2+ (lanesa,b, d,e,g,h, i, k,m,and n) or Mn2+ (lanes c, f, i, 1, and o) as a divalent cation. The

arrowheads indicatethemajorinvitro and in vivocleavage site of WT virus(22). LaneslabeledMdisplaysnakevenom

phosphodies-terasecleavage productsfor each DNAsubstrate, and theproduct

sizes (in nucleotides) are shown. Lanes m through o represent a

longerexposure of lanesj through 1. (C) Analysis of integrative

recombinant products. Integrative recombination products were

analyzed as described in Materials and Methods by using 5' 32p_ labeledduplex U5oligodeoxynucleotide LTR substrates with

two-base 5' PoverhangsrepresentingtheWT(lanes7 and8)and the S2 (lanes1and2),S4(lanes3and4),and S4C(lanes5and6) mutations. Double-stranded oligodeoxynucleotides were incubated in the

ab-sence(lanes 1, 3, 5,and7)orpresence(lanes 2, 4, 6,and8)of IN. S denotesthe respective startingsubstrates. DNAproducts larger

than thestartingsubstratesareindicated bythearrowheads.

anyeffecton replication. However, closer inspection of the sequence in dlAva3 reveals that the potential to form an

alternative but smaller U5-IRstemexists.Thus, MuLVstem structures mayhave similar roles tothose demonstrated for avian sarcoma andleukosis virus.

In viral DNA, U5-IRsequences were found to be

neces-A

B

J. VIROL.

on November 10, 2019 by guest

http://jvi.asm.org/

(8)

sary for efficient

integration

as well as for specific in vitro

nicking

at the viral

integration

site by purified IN protein

(Fig.

5 and 6). These results were not unexpected, both

because mutations near the U5 terminus of other retrovi-ruses have been shown to cause integration defects (9, 10,

25, 26)

andbecausemutations inthisregionwerepreviously

found to inhibit in vitro

nicking

by theavian retrovirus IN

protein (22). Furthermore,

Roth etal. (26) have shown that

U5 mutations which block

integration

also blockcleavage at theU5 terminusinvivo.Thecurrentstudy indicatesthat the

U5-IR mutationswhich affect

integration

interfere withboth

IN-catalyzed

cleavageandjoining reactions in vitro, but to

different extents. The S2 substitution appears to induce a stronger defect than the S4 substitution. In the case ofthe

S2C

virus,

in which thereis no defect in reverse

transcrip-tion,

we

speculate

that the ability of this virus to replicate

and

integrate

in vivo

(Fig.

3A and 5) might be due to

cooperative

interactions between the IN protein bound to

the

wild-type

U3 LTR terminus and that bound to the

mutated U5 LTR terminus. Alternatively, other viral or

cellular

proteins

may contribute to the S2 integration

effi-ciency

in

vivo,

or

integration

mediatedby

oligomeric

forms

of viral DNA asdescribed

by Hagino-Yamagishi

etal. (16)

mayoccur.

We have found that the S2 mutation caused greater

initiation and

integration

defects than the S4 mutation.

Furthermore,

S4C

fully

restored the reverse transcription

efficiency,

whereas S2C led to

only

a

partial

restoration.

These

findings

wereunexpected, since S2C has fewer

nucle-otide substitutions than S4C and forms a U5-IR stem of

similar

stability. Thus,

the

partial

initiation defect of S2C suggests that both the U5-IR stem structure and sequence affect initiation.

Apparently,

the CTTC/GAAG U5-IR se-quence is favored over

GTTG/CAAC,

which more

closely

resembles the

wild-type

sequence, GAAG/CTTC (Fig. 1C).

We

speculate

that this couldreflecttherole ofaproteinthat

recognizes

both the U5-IR stem structure and the GAAG

sequence in either orientation in the stem. Although the same sequence

preferences

are

displayed by

the IN

protein

in in vitro

cleavage

and

integration reactions,

IN is

unlikely

to be involved in initiation since it canbe mutated without

affecting

reverse

transcription (29).

The

biological

consequencesofthe

S2, S4, S2C, S4C,

and A9mutationscanbe

explained

onthe basis of either

disrupt-ing

the U5-IR stem structure,

thereby affecting

initiation of reverse

transcription,

or

altering

DNA sequences

recognized

by

the IN

protein

for

integration.

The A3 mutation ismore

enigmatic.

The sequences that are deleted

by

this mutation are not

required by

INfor

specific

cleavage

of the U5 LTR terminus

(22),

nordo

they

appeartodisturb the base

pairing

orsequenceoftheU5-IRstem.This suggests that features in addition tothe U5-IR stem and U5-leader stem structures

are

required

for maximalinitiation

efficiency.

One

intriguing

possibility

is that U5 sequences between the U5-leaderand

U5-IR

stems,

namely,

those substituted in the S6 mutation

(Fig. 1B),

form base

pairs

with

complementary

sequencesin the

TqlC

loop

of the

tRNATrP

primer,

as

initially proposed by

Haseltineetal.

(17,

18)

and

supported

by

the

experiments

of Cordelletal.

(11).

This appearstobe the case,since in vitro andinvivo studies indicate that this

U5-primer

interactionis

important

forinitiationofreverse

transcription

(1). Thus,

the A3 mutation

might impair

initiation ofreverse

transcription

by altering

the

spacing

between the U5-leader and U5-IR

structures,

thereby

affecting

the interactions between the

TtlC

loop

of the

tRNATrp

primer andU5.Thisis discussed in

moredetail elsewhere

(1).

Our previous analyses of a mutation that disrupts the US-leader stem (A6 [Fig. 1B]) revealed a defect in the initiation of reverse transcription in infected cells and in reactions ofmelittin-permeabilized virions but not in reac-tions of purified viral RNA and reverse transcriptase (7).

Thisraises the possibility that the US-leader and possibly the U5-IR structures are important for signalling the initiation of reversetranscription following infection in the context of the highly structured virion core. The role of viral proteins in such signalling is unclear, although it may be relevant that initiation ofpoliovirus replication also is likely to require a RNA secondary structure near the initiation site. In this case, poliovirus genes whose products could interact with the secondary structure in trans have been identified by analysis of second-site revertants (2). We have isolated second-site revertants for severalof the US mutations (Sa), and a similar analysis might identify a retrovirus gene product that interacts with the US-leader and U5-IR stem structures to regulate initiation of reverse transcription.

ACKNOWLEDGMENTS

Wethank H.-J. Kungforcritical comments on the manuscript. This workwassupported inpartby Public Health Service grant CA 38046 from the National Cancer Institute and by Cancer ResearchCentergrantP30CA43703.D.C.and H.H. aresupported inpartbyPublic HealthServicetraininggrantT32GM07250from the NationalInstitute of GeneralMedical Sciences.

REFERENCES

1. Aiyar, A., D. Cobrinik, Z. Ge, H.-J. Kung, and J. Leis. Submit-tedfor publication.

2. Andino, R., G. E. Rieckhof, and D. Baltimore. 1990. A func-tional ribonucleoprotein complex forms around the 5' end of poliovirusRNA.Cell63:369-380.

3. Bilofosky, H. S., C. Burks, J. W. Fickett, W. B. Goad, F. I. Lewitter, W. P. Rindone, C. D. Swindell, and C. S. Tung. 1986. The GenBankgeneticsequencedatabank. Nucleic Acids Res. 14:1-4.

4. Boone,L.R., and A. M. Skalka. 1981. Viral DNA synthesized in vitroby avian retrovirus particles permeabilized with melittin. I. Kinetics ofsynthesis andsize of minus- andplus-strand tran-scripts.J. Virol. 37:109-116.

5. Brown, B., B. Bowerman, H. Varmus, and J. M. Bishop. 1987. Correctintegrationof retroviral DNA in vitro. Cell 49:347-356. 5a.Cobrinik,D. Unpublisheddata.

6. Cobrinik, D.,R.Katz,R.Terry, A. M.Skalka,andJ. Leis. 1987. Aviansarcomaandleukosis viruspol-endonucleaserecognition of the tandem long terminal repeat junction: minimum site required for cleavage isalsorequired forviralgrowth.J.Virol. 61:1999-2008.

7. Cobrinik, D., L. Soskey, andJ. Leis. 1988. A retroviral RNA secondary structure required for efficient initiation of reverse transcription. J. Virol. 62:3622-3630.

8. Coffin, J. M. 1990. Retroviridae and theirreplication, p. 1437-1500. In B. N. Fields, D. M. Knipe, R. M. Chanock, M. S. Hirsch, J. L. Melnick, T. P. Monath, and B. Roizman (ed.), Virology, 2nd ed. RavenPress,Ltd., NewYork.

9. Colicelli, J.,andS. Goff. 1985. Mutants andpseudorevertantsof Moloney murine leukemiaviruswith alterationsatthe integra-tion site. Cell 42:573-580.

10. Colicelli, J.,and S. Goff. 1988. Sequence andspacing

require-mentsofaretrovirusintegrationsite. J. Mol. Biol. 199:47-59. 11. Cordell, B., R. Swanstrom, H. Goodman, and J. M. Bishop.

1979. tRNATmPas primerfor RNA-directed DNApolymerase: structural determinants offunction. J. Biol. Chem. 254:1866-1874.

12. Craigie, R., A.-T. Fujiwara, and F. Bushman. 1990. The IN protein of Moloneymurine leukemia virus processes the viral DNA ends and accomplishes their integration in vitro. Cell 64:829-37.

on November 10, 2019 by guest

http://jvi.asm.org/

(9)

13. Dahlberg,J. E., R. C.Sawyer,J. M. Taylor, A. J. Faras, W. E.

Levinson,H. M.Goodman, and J. M.Bishop. 1974. Transcrip-tion ofDNA from the 70S RNA of Rous sarcoma virus. I.

Identification ofaspecific 4SRNAwhich serves asprimer. J.

Virol. 13:1126-1133.

14. Darlix, J.-L., M. Zuker, and P.-F. Spahr. 1982. Structure-functionrelationship ofRous sarcomavirusleaderRNA. Nu-cleic AcidsRes. 10:5183-5196.

15. Fujiwara, T., and R. Craigie.1989.Integration ofmini-retroviral DNA: acell-freereaction for biochemical analysis of retroviral integration. Proc. Natl. Acad. Sci. USA 86:3056-3069. 16. Hagino-Yamagishi, K., L. A. Donehower, and H. E. Varmus.

1987.RetroviralDNAintegratedduringinfection byan integra-tion-deficientmutantof murine leukemiavirusisoligomeric.J. Virol.61:1964-1971.

17. Haseltine, W. A.,D.G. Kleid,A. Panet, E. Rothenberg, and D. Baltimore. 1976. Ordered transcription of RNA tumor virus

genomes. J.Mol. Biol. 106:109-131.

18. Haseltine, W. A., A. M. Maxam, and W. Gilbert. 1977. Rous

sarcomavirusgenomeisterminallyredundant:the 5'sequence.

Proc.Natl. Acad. Sci. USA 74:989-993.

19. Hirt, B. 1967. Selective extraction of polyoma DNA from infectedmousecellcultures. J.Mol. Biol. 26:365-371. 20. Hughes, S. H., J. J. Greenhouse, C. J. Petropoulos, and P.

Sutrave.1987. Adaptorplasmids simplify the insertion of foreign DNA intohelper-independent retroviral vectors. J. Virol. 61: 3004-3012.

21. Katz, R., G. Merkel, J. Kulkosky, J. Leis, and A. M. Skalka. 1990. The avian retroviral IN protein is both necessary and

sufficientforintegrative recombinationin vitro.Cell63:87-95. 22. Katzman, M., R. A. Katz, A. M.Skalka, and J. Leis. 1989.The

avian retroviral integration protein cleaves the terminal se-quences of linear viralDNA attheinvivo sitesofintegration.J. Virol.63:5319-5327.

23. Mullis, K. B., and F. A. Faloona. 1987. Specific synthesis of DNA in vitro viaapolymerase-catalyzed chain reaction. Meth-ods Enzymol. 155:335-350.

24. Murphy, J. E.,and S.Goff. 1989. Construction and analysis of deletion mutations in the U5regionof Moloney murine leuke-mia virus:effectson RNApackaging and reversetranscription. J. Virol. 63:319-327.

25. Panganiban, A.,and H. Temin. 1983.The terminalnucleotides of retrovirusDNA arerequired for integration butnotfor virus production. Nature(London) 306:155-160.

26. Roth, M., P.Schwartzberg, andS. Goff. 1989. Structureof the termini ofDNA intermediates in the integration of retroviral DNA:dependence ofINfunction andterminalDNAsequence. Cell58:47-54.

27. Sambrook, J.,E. F. Fritsch,and T. Maniatis. 1989.Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, ColdSpring Harbor, N.Y.

28. Schwartz, D.E., R. Tizard,and W. Gilbert. 1983. Nucleotide sequence ofRous sarcomavirus. Cell 22:853-869.

29. Schwartzberg, P., J.Colicelli, and S. Goff. 1984. Construction andanalysisofdeletion mutations in the pol gene of Moloney murine leukemia virus: a new viral function required for pro-ductive infection. Cell 37:1043-1052.

30. Smith, R.E. 1974. High specific infectivity avianRNA tumor viruses. Virology60:534-547.

31. Temin, H. M., and D. Baltimore. 1972. RNA-directed DNA synthesis andRNAtumorviruses. Adv. VirusRes.17:129-186. J. VIROL.