Modified model for the switch from Sendai virus transcription to replication.

(1)

0022-538X/89/051951-08$02.00/0

Modified Model for the Switch from

Sendai Virus

Transcription

to

Replication

SILVIA VIDALAND DANIELKOLAKOFSKY*

Department of Microbiology, Geneva MedicalSchool, CMU, CH1211 Geneva, Switzerland

Received 19August1988/Accepted 10 January 1989

RNasemappingwasusedtoestimate the levels of unencapsidatedSendai virusplus-strand RNAswhichcross

theleader-NPjunction relativetoNPmRNA.Significantamountsof leaderreadthroughRNAswerefound in

Zstrain-infected cells,similartothatdescribed forthe poiRmutantofvesicular stomatitisvirus, eventhough

this strain isconsidered wildtype.The levels ofthereadthroughRNAsdetected fellsharply when progressively longer probeswereused, unlike thatofNPmRNA. These studiessuggestthat polymerases which readthrough

the firstjunction terminate shortly afterwards in the absence of concurrent assembly ofthe nascentchain, whereas those which reinitiate at NP continue efficiently to the next junction. Reinitiation appears to be

necessary toconvert the polymerase toa mode inwhichelongation is independent ofconcurrentassembly. Concurrent assembly appears to be required not only for the polymerase to read

thro'ugh

the junction

efficiently, butalso for it tocontinueelongationbetweenjunctions. Sendai virus, amember of the Paramyxoviridae, contains

a nonsegmented minus-strand RNA genome of 15.3 kilo-bases (kb), from which a leader RNAand six mRNAs are

transcribed. The viral polymerase responsible for mRNA synthesis,referredtoasthetranscriptase,isthoughtto enter its nucleocapsid (NC) templateat or. nearthe 3' end andto sequentially synthesize the leader and the NP, P/C, M, F, HN, and L mRNAsby terminating and restartingateach of thegenejunctions. The mRNAsarecappedand

polyadeny-latedduring synthesis, whereas theleaderRNAis unmodi-fied at either end. The minus-strand genome is also a

templatefor thesynthesisofantigenomes,whichare unmod-ifiedfull-length complements ofthegenome whichserve as

intermediates in genome replication. The polymerase

re-sponsibleforantigenomesynthesis,whichwerefertoasthe

replicase, therefore ignores thejunctional stop-start signals that the transcriptase responds to (fora recent review, see

reference 16).

The manner in which this switch from transcription to replicationtakesplace for paramyxoviruses, acentralevent inthe lifecycle ofallminus-strandRNAviruses,isthought tobe similartothatproposed forthe rhabdovirus vesicular stomatitis virus (VSV) (1, 2, 5, 7, 15). Since, like VSV, Sendai virus genomes and antigenomes are found only encapsidated with the nucleocapsid protein (NP) and their synthesis is dependent on on-going protein synthesis

whereas that of mRNA is not, the switch may also be

effected here simply by the intracellular concentration of unassembled NP. In this minimal model, the template and the polymerase for mRNA and antigenome synthesis are

identical; the difference lies in whether the nascent RNA chainisbeing concurrentlyassembledinto NCs.Concurrent assemblywouldsomehow beresponsiblefor thepolymerase

ignoringthejunctional signals, all thatisrequired to switch itfromatranscriptase into areplicase.

The sitefor theinitiation of NC assemblyonVSVleader

RNAhas been mapped tothefirst 14 bases at the 5' end of the chain(7). Polymeraseswhosenascentleaderchainshad notassembled with NPbefore the firstjunctionwasreached

wouldterminate andthen restart atthe beginning ofNPto

* Correspondingauthor.

make all the mRNAs. All junctional signals would be ob-served,as noneof thenascentmRNAs could assembleNP. Polymeraseswhosenascentleader RNAs hadbegun

assem-bly wouldnotobserveanyofthejunctional signals,because

NC assemblywouldfollow thepolymerase, resultingin the synthesis ofa full-length and simultaneously encapsidated

plus-strand RNA. The frequency of genome replication would then bedetermined bythe level of unassembled NP, and the switch would. be part ofa self-regulatory system controllingthe relativeamountsoftranscriptionand replica-tion. We refertothisschemeasthe minimalmodel',since the

switch is effectedentir'elybyunassembledNP,theonlyviral proteinknowntobe requiredfor replicationthat is not also required for transcription.

Theorganizationof the Sendaivirus and VSVgenomesis

also quite similar, including the nature of the junctional signalsfortranscription (11-13). Wewould therefore expect that the minimalmodelwould alsoapplytoSendaivirus,at least in broad outline, even though several of the

experi-mentswhich ledtothemodel havesofaronlybeenreported for VSV. Paramyxoviruses such as Sendaivirus, however, have an overlapping gene (C) not found in rhabdoviruses, contained within thie P gene in an alternate reading frame (12). As the Cproteinhas beenlocalized with NCs intracel-lularly (4, 20),itmayalsoparticipate inRNAsynthesis. We have therefore further examined Sendai virus plus-strand RNAsynthesis to investigatehowclosely theabQve model applies.

MATERIALS ANDMETHODS

Isolation ofcytoplasmicRNAs. Confluent cultures of BHK cells in 10-cm dishes were infected with 20 PFU of Sendai virusHarris (H)orZ strainpercell and maintainedat33°C. Cycloheximidewasadded afteradsorption (1 h)ata

concen-tration of10 p.g/mlwhen indicated.The cellswereharvested at various times by scraping intophosphate-buffered saline andrecoveredby centrifugation. Cytoplasmicextractswere

preparedbyvortexing'107cells in0.5 ml oflysisbuffer(0.15 MNaCl,0.05 M Tris hydrochloride [pH7.4],0.6% Nonidet P-40[NP-40],followedby centrifugationfor 10minat4,000

xg.Theextractwaslayeredonapreformed20to40% CsCl

density gradientandcentrifugedfor 20 hat52,000rpminan

1951

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

1952 VIDAL AND KOLAKOFSKY

SW60rotor. TheRNApellet wastakenup in TNE (25mM

Tris chloride [pH 7.4], 50mMNaCl, 1 mM EDTA), precip-itated in ethanol, and dissolved inET (1 mM EDTA, 10 mM Tris chloride [pH 7.4]).

Invitropolymerasereaction.Purified Sendaivirus(50 ,1u,5 mg/ml)grownin LLC-MK2 cellswasused in250-,ul reaction

mixes containing 50 mM HEPES (N-2-hydroxethylpipera-zine-N'-2-ethanesulfonic acid, pH 8.1), 5 mMMgCl2, 5 mM spermidine,0.05% NP-40, 1 mM eachallfourNTPs, 100 U of RNasin. 48 mM Na2HPO4, and 1 mgofpoly-L-glutamic acid perml. The reaction mixwasincubatedat 30°C for 3h and thenbrought to0.5 M NaCl and 1%NP40, heated for2 minat37°C, and centrifugedon a20to40% CsClgradient to obtain the pelleted RNAs asabove.

RNase protection. Approximately 105 cpm ofriboprobe werecoprecipitatedwith 1to9,ug of CsCl pelletRNAand20 ,ug of tRNA, suspended in 9

RI

of ET, heated for 1 min at 90°C, quickchilled, andthenmade 0.3 MNaCl,20 mM Tris hydrochloride (pH 7.4), 2 mM EDTA (2x bufferA)inatotal

volume of10,ul. After annealing for 30 minat70'C, 50 pI of

0.1% sodium dodecyl sulfate (SDS)-40 ,ugof RNase A per

mi-5 U of RNase T1 per ml in 2.5x buffer A was added.

After 45 min atroom temperature, 600 ,ug ofproteinase K

permlandanadditional 0.1% SDSwereadded. After 20min

at30°C, the volumewasraisedto250,ul withTNEand 0.5% SDS, extracted twice with phenol-chloroform, and ethanol precipitated. The remaining RNAsweredissolved in 3 pI of 80% formamide plus dyes, heated for 2 min at 90°C, and separated on a6% polyacrylamide sequencing gel.

Preparation of riboprobes.TheH strain probe(nucleotides [nt] 22to 142)wasderived from pSL3, whichcontains the 3' endof thegenomestartingat nt22in the PstI siteofpBR322 (11). The first site for TaqI is at nt 142. This PstI-TaqI fragmentwasinserted intopSP65atthe PstI andAcclsites. After linearization with Hindlil, SP6 polymerase reactions

were carried outwith 0.5 mMATP, CTP, and GTP and50 ,uCi of[32P]UTP(10). Equalvolumesofformamide anddyes

were then added, and the radioactive transcript was

sepa-rateddirectly on a6% polyacrylamide sequencing gel. The

transcriptwas localized by autoradiography,andafter exci-sion of theband itwaselutedbyagitating overnightat4°Cih TNE.

The Z strain probes were prepared from clone B36 (21)

which contains the 3' end of thegenome startingatnt 1 plus 22 extra bases inserted into the BamHI site of pBR322. Three of theriboprobe plasmids containingtheentire leader sequence plus variable portions of the NP gene were ob-tainedby BamHI-NcoI (nt 358), RsaI (nt 234),and TaqI (nt 142) digestion, followed by insertion into the BamHI and AccI/HincII sitesofSP65 (nt1to142)orpGEM3 (nt 1to358 and 1 to 234). When the second enzyme was NcoI, it was

filled in with theKlenow enzyme before BamHI digestion.

The Sall site at nt58 was created by site-directed muta-genesis in clone 1-142 with EcoRI and HindIIIto create the gap and the synthetic oligodeoxynucleotide 3'-AATCCCA GCTGCATAGGTG-5' (the two base changes are

under-lined). The procedure used was essentially as described

before (10). The 81-bp BamHI-SalI fragment containingthe first 58ntofthe Sendai virusgenome wastheninserted into

the same sites of pGEM3. All riboprobes were prepared

simultaneouslywiththesamelabeled andunlabeled

nucleo-tide triphosphates by using DNAs linearized immediately after theinserts toensure thesame specific activity.

The plasmid containing the NP-PC junction (nt 1638 to

1812)wasderived from B36by digestionfirst with AccI and NcoI. The 459-bp fragment was isolated and digested with

TaqI, and the 174-bp NcoI-TaqI fragmentwas isolated,and the endswereblunted with the Klenow enzymeand inserted into the SmaIsite ofpGEM4.

RESULTS

Sendai virus genomesand

antigenomes

arefoundonly in

NCstructures, with abuoyant density of 1.31

g/ml

in CsCl

density gradients

(9). When

cytoplasmic

extractsof infected cells are

centrifuged

on such

gradients,

the

pellet

fraction

contains

only

the

unencapsidated

viral

RNAs,

i.e.,

the mRNAs and some of the leader RNAs (6), without their

complements.ThesepelletedRNAsaresuitable for

measur-ingthe amountsof thevarious viral

transcripts

whichcross a genejunction by RNase mapping, since

self-annealing

of

the viral RNAs or annealingof the minus-strand

riboprobe

withantigenome RNA should not occur.

Wefirstexamined the

transcripts

which crossthe leader-NPjunctionwithaminus-strand

riboprobe

containing

nt1to 142 (Fig. 1). When similar

preparations

of CsCl

pellet

RNA from H strain infectionswereexamined with 3'-end-labeled

genomeRNAas a

probe

(17),leader RNAs of31,

55,

and 65 ntwerefound, the

predominant

55-nt

species

having

termi-nated at the junction. Protected

fragments

consistent with

these species were also weakly seen with the

internally

labeled riboprobe

(Fig.

2, marked

1),

which detects NP mRNA (87-nt band) and leader-NP readthrough

transcripts

(142-ntband)aswell. (Inthis

experiment,

asmallamountof theintact 240-nt

riboprobe,

ofwhichonlythe central 142nt

arevirus

specific,

survived

digestion

in allcases).The

origin

ofthese bands isdefined

by

their

expected lengths,

by

their

ability

toannealto

riboprobes specific

toeither the leaderor NP mRNAregionswhen the

gel

is converted intoaNorthern blot (not

shown),

and whether their

mobility

is

unchanged

whenanestedsetofriboprobesis used

(Fig.

3A).The leader RNAs are minor

species

compared with the NP mRNA in thesesamples,

possibly

because theyareless stableorhave been

encapsidated

(6)

atnd

are removed

during

preparation.

Themost

striking

result, however, is therelativeamountof

leader-NP

readthroughtranscripts.

The

142-nt band

presumably

results from a

transcript

which initiatedatnt1and extendedtoatleastnt142andwas not encapsidated, as

juldged

by

CsCl

density

gradient

cen-trifugation.

These

transcripts

are not

antigenomes

whichhad somehow

escaped

the

separation protocol,

as

riboprobes

from the

opposite

end of the genome could not detect any

plus-strand

RNA

other than 3' end of the L

mRNA,

even

though the L mRNA was 20-fold less abundant than NP mRNA(notshown). Wealso examined therelativeamounts

ofthese

transcripts

during

the timecourse of the infection. RNAwas

prepared

from cells harvestedat4h aswellas 18 h

postinfection

and at 4 h from cultures incubated in the presenceof

cycloheximide.

Theinhibition of

protein

synthe-sis prevented genome replication but not

transcription,

so that under these conditions the viral RNAs were

predomi-nantly the result ofprimary

transcription.

When all these

samples were

compared

(Fig. 2)

and theresults were quan-titated by

densitometry

and normalized for the number of uridinesin eachfragment,thereadthroughRNA

represented

39% ofthe NP mRNA at 4 h when new genome

synthesis

waspreventedwith

cycloheximide,

43%at4h without

drug,

and 19% at18 h.

Unencapsidated

leader

readthrough

RNAs are thus present at

relatively

high

levels relative to NP mRNA throughout the infection.

The

readthrough

transcripts appeared

tobetooabundant and stable to be true

replication

intermediates,

which are

J. VIROL.

on November 10, 2019 by guest

http://jvi.asm.org/

(3)

1 358

1 234

1 142

1 58

/

1638 1812

NP

PC.

I

FIG. 1. Schematic diagramofriboprobes used and themodifiedmodel. The middle line represents the 3' orleft end of theminus-strand genome withtheleader. NP,and PCregionsseparatedbyverticalbarsrepresenting thejunctions.Mappositionsarenumbered from this end

orthe 5' endof theantigenome. Above areshown the Zstrain riboprobes used. Theextentsof the viral sequences areindicatedby heavy lines and mappositions;thethin lines at the endsindicate transcribed vector sequences. Below are shown thetwosituations ofplus-strand RNAsynthesisfoundin Z strain-infected cells in the absence of concurrentassembly. In each case thepolymeraseis shownjustafterhaving crossed the firstjunction, butwithorwithouthavingreinitiated, and the consequences for continued elongation.

expectedtobeverytransientspecies.Their levels relativeto NPmRNA were reduced by onlytwofold between 4 and 18

h, whereas NP mRNAs increased by more than 30-fold

between these times. J3ona fide replication intermediates describedforVSV(14)werefourd torepresent onlyasmall fractionof the RNAmade ina5-minpulse of[3H]uridine and were difficult to see at all after longer labeling. They were

also considerably larger (2,000 to 11,000 nt) and, more

importantly, wereassembled with N protein, astheirRNAs were resistant to RNase. The pelleted Sendai virus readthrough transcripts, onthe otherhand,wereonlyafew

hundred nucleotides in length, sensitive to RNase (not shown), and madeatthesamerate relative toNPmRNA in the

absencp

of viralproteinsynthesis. They therefore donot require assembly with NP for theirsynthesis either during primary transcription in vivo or in vitro (see below). True Sendaivirusreplication intermediates mustofcoursealso be

presentintheseinfections, but they would be expectedtobe retained in the CsCI gradient, along with mature genomes

and antigenomes.

Polymerases which read through the leader-NPjunction. Thislevel ofleaderreadthrough RNAs was unexpected, as

nascent chains which had not begun assembling with NP

shouldhaveterminatedefficientlyatthe firstjunction. Itwas

thereforeof interesttocharacterize further the polymerases which made readthrough RNAs; e.g., did they continue

alongthetemplate until the nextjunction, which mightnow

beobserved, ordid they somehow sensethe absenceof NC assembly on the nascent chain and abort prematurely? For this experiment, a nested set of minus-strand riboprobes

containing nt 1 to58, 142, 234, and 358were used (Fig. 1). Themappingswereagain carriedout atmorethanoneRNA concentration to ensure that the results were proportional,

asforFig. 2. Thegel from onesuchexperiment is shown in Fig. 3A. Riboprobe 1-58 only detected the leaderregion plus

3 nt, but the other three detected both NP mRNA and readthrough RNAs. Besides riboprobe 1-142, described be-fore, riboprobes 1-234 and 1-358 detected protected

frag-ments of 180 and 302 nt (NP mRNA) and 234 and 358 nt (leader-NP readthrough), respectively. In this experiment,

the bands were cut out and counted, and the results were plottedascountsperuridineperminuteineachfragment,so that therelativeamountsof eachspecies could becompared directly(Fig. 3B).Except forriboprobe 1-358, astraightline could be drawnthrough the origin with at least two of the threepoints,andtriplingtheprotectingRNAateach step led to three times as many counts protected. As the relative amountsofthevarious RNAsaregiven bytheslopesofthe lines, it is clear that there is a decreasing abundance of readthrough RNAsasafunction of how far their3' ends had extended. This can also be seen in Fig. 3C, in which the various readthrough RNAs from the lowest concentration point used for all four probes (from three separate

experi-ments) are plotted as the percentage of the NP mRNA

moll.--L

-- - - - -I

I I

4

on November 10, 2019 by guest

http://jvi.asm.org/

[image:3.612.51.557.56.385.2]

(4)

4h

INFECTION M 4h CHX 18h

pg RNA 8 4 8 4 8 48

642

240 PROBE

172

*. ..l

.0

_-.s.142 /NPI

-87

leader(I

T

_4

FIG. 2. Detection of Sendai virus plus-strand RNAs in CsCl pellet RNA by riboprobe 1-142. CsCl pellet RNA from

mock-infected(M)orvirus-infected cells harvestedat4 h withorwithout cycloheximide (CHX) added after virus adsorption or 18 h were

assayedforplus-strand RNAs witha minus-strandriboprobe

con-taining positionsnt1to142. TheamountoftheprotectingRNA used

in each lane is indicated above. The four right-hand lanes show riboprobesrun as lengthmarkers(small arrows). The large arrows

show the viral RNAs detected.Three leader RNAsareindicatedby the letter1.

detected by the same probe. There was no tendency for RNAs which had initiated at nt 56(NP mRNA)todecrease inabundance,asmeasuredby probeswhich extend

progres-sively downstream, whereas the readthrough RNAs de-creased from 19% of NP mRNA at nt 58 to 3% at nt 358. Furthermore, less than 10% of the polymerases which started the NP mRNA failed tofinish it (Fig. 4A), whereas

we have been unable to detect any complete leader-NP readthroughRNAsbyNorthernanalysiswithleader-specific riboprobes (not shown).

In theseexperiments, themappingswerecarried outwith

more than one concentration ofprotecting RNA to ensure

that the riboprobe was not limiting fortwo reasons. First,

under conditions of limiting riboprobe, formation of the longer hybrids mightoccurpreferentially.Moreimportantly, under these conditionsariboprobe mightbesimultaneously

hybridized to botha leader and an NP mRNA without any gapinbetween,asleader RNAmayextendto nt55 and NP begins at nt 56. Such a composite hybrid might not be sensitive to RNase attack and appear as a readthrough

transcriptin thisassay.Themoretheriboprobeis inexcess,

however,the less likely it is thatacomposite hybrid would

form. As shown in Fig. 3, the ratio ofreadthrough RNAto

NP mRNA did not

change

even when nine times less

protecting

RNA from a

given sample

was

used,

and we

measuredmorerather than less

readthrough

RNArelativeto NP mRNA when therewas 30times less viral RNA in the

sample

at 4 h relative to 18 h

postinfection (Fig. 2).

These results are inconsistentwith themeasurementof

composite

hybrids,

as is the

finding

that

readthrough

RNAs become less abundant relative to NP mRNA when

progressively

longer

riboprobes

areused.

Finally,

asthe sequence onthe

riboprobe

preceding

the

junction

is

AAAA,

in some

experi-ments we used RNase

T2

instead ofRNases A and

T1,

but

this did not

change

the results

(not shown).

As the

readthrough

RNAs do not appear to be artifacts of the

method,

these results indicatethat

polymerases

which start at nt 1and

readthrough

thefirst

junction

terminate

hetero-geneously

shortly

thereafter,

whereasthose whichreinitiate

at NPcontinue at

high frequency

tothe next

junction.

Zand Hstrains ofSendai virus. Several strains of Sendai

virusare

available,

and the onewe have described sofar is the Z

strain,

the

only

onewhichhas beencloned

completely

(21, 22).

The sequence of

the

H strain with whichwe have worked in the past has been cloned uptothe L gene,butour

clone closesttotheleft endstartsat nt22

(11). However,

as

these strainsare99% conserved atthenucleotide

level,

the Z strain

riboprobes

should also be able to measure the H

strain

transcripts.

When

equal

amountsof

CsCl

pellet

RNA

fromZand Hstrain-infectedcellswereusedtoprotectthe Z

strain

riboprobe 1-142,

these

samples

werefoundtocontain similaramountsofNPmRNAbut verydifferentamountsof

leader

readthrough

RNAs

(Fig. 4A). By

direct

counting

as

for

Fig.

3,

theZstrain

sample

wasfoundtocontain 12.2%as

much

readthrough

RNAsasNP

mRNAs,

in

good

agreement

with the

14%

averagein

Fig. 3C,

but the H strain

sample

was

found to contain 20 times less

(0.6%).

To ensure that this

differencewas not due tothe

slight

differences in sequence between the two strains in this

region,

the amount of

readthrough

RNA in H strain CsCI

pellet

RNA was

mea-sured withanHstrain

riboprobe

containing

nt22to142. As

shown in

Fig. 4B,

this

homologous

probe again

detectedless than 1%

readthrough

RNAinthe in vivo

sample

relativeto NP mRNA under conditionsinwhich the

products

ofan H

virion

polymerase

reaction werefoundtocontain consider-able amounts of

readthrough

RNA. The

ability

of the Z

strain

polymerase

to

readthrough

the first

junction

at an

unexpectedly high

frequency

in vivo is thus not shared

by

the H strain.

SincetheZand Hstrain

polymerases

readthrough

thefirst

junction

at such different

frequencies,

we examined the

readthrough

frequency

at the next

junction (NP-PC),

by

using

a Z strain

riboprobe

containing

positions

nt 1638 to 1812

(the

NP mRNA ends at nt

1736,

and PC

begins

at nt

1740).

As

shown

in

Fig.

4A, this

probe

detected

roughly

equal

amountsof NP

(100 nt)

and PC

(71

nt) mRNAs,

aswell asNP-PC

readthrough

RNAs

(174

nt)

in

equal

amountsofZ and HCsCl

pellet

RNA.

However,

the level of

readthrough

atthis

junction by

theZand H

polymerases

was

identical

to thatofthe H

polymerase

at the first

junction (0.6

and

0.5%,

respectively).

The Z strain

polymerase

therefore reads

through

the first and secondjunctions at different

frequen-cies,

whereas the Hstrain

polymerase

reads

through

bothat the samelow

frequency.

Invitro studies. The level of

transcripts

from the leader-NP

region

thatwe measured in vivoare

steady-state levels,

determined

by

turnoveraswellas

synthesis.

Moreover,

for

transcripts

whichcontain theleader sequences, turnoverin thiscontext includes

encapsidation

as wellas

degradation.

J. VIROL.

on November 10, 2019 by guest

http://jvi.asm.org/

[image:4.612.86.283.71.382.2]

(5)

A PROBE 1-358 1-234 1-142 1-58 pgRNAM.7 2 6 M .7 2 6 M.726 M 1 3 1

INP 358_ NP 302 1,NP

234-NP 180

'NP 142 ~d- 9

NP 87

leaderi)

_

56-142 X

NP mRNA

FIG. 3. Quantitation of leaderreadthrough and NP mRNAswith 9PROBES a nested set of riboprobes. (A) Three different amounts of CsCl 431 pellet RNA (18 h) eachathreefold increase,were used toprotect riboprobescontaining positions 1to58, 142, 234, and358(Fig. 1),

307 and theRNAsremainingafterdigestion wereseparated by electro-phoresis. Theautoradiogram fromone suchexperiment is shown; * - 240 the lengths (innucleotides) ofthe various probes used andRNAs detected are indicated on the sides. The probes used in each experimentareindicatedabove,asisthe amountofprotectingRNA for each lane. M refersto 6 or9 ,ugofRNA from mock-infected cells. The firsteight lanes containing the larger probeswereexposed for shorter times than those containing the smaller probes. (B) * - 104 Bandsrepresenting the leader readthroughRNAs

(1/NP)

andthe NP mRNAs (NP) were excised from a parallel gel which contained blank lanes in between, and their Cerenkov radiation was deter-mined in a beta counter. After background subtraction (30 cpm total), thecountspresent ineachband weredividedbythenumber of uridines and plotted as afunction of the amountofprotecting RNAforthe threesmallestriboprobes. Onlythe NP mRNAbands from riboprobe 1-142 are shown. (C) Results obtained with the lowestamountofprotectingRNAfromthree separateexperiments, P - 58 NP one of which is shown in panels A and B, are plotted according to the 3' end of the plus-strand fragment protected in the various hybrids.Tonormalizethevaluesforthe NPmRNA,the averageof the valuesfrom thethree riboprobes in each experiment (115, 34, and 189cpm/Uprotected)wasset to100, and the results from each riboprobeareplotted relativetothis value. Thereadthrough RNAs, ontheotherhand,areplotted relativetothe NP mRNAdetectedby the sameprobein the sameexperiment,exceptforthat detectedby riboprobe 1-58, whichisplotted relativetotheaverage value for the NP mRNA. Large and small crossesand circles indicatethemean and individual values forNP mRNA and the leader readthrough RNAs,respectively.

1-58

z

a

cc

z cc

FE

0.

0

z

1-142 5

cc

a-0u

C

120-100

80

60-40

20

READ -THR U

20F

n

u0

:i o

E-v

CL

uJ 10

z

a

0.

u)

1-234

.71 2 3

pg CsCI PELLET RNA

Wetherefore alsoexaminedreadthrough in vitro with puri-fied Sendai virions, as degradation of the uncapped RNAs containing the leader region might be less important here than in vivo, and encapsidation of these transcripts should not take place. RNA levels may therefore more closely

approximate their rates of synthesis in vitro. Polymerase reactionmixes containing equal amounts of Z and H strain virionsweremade,and theproductswereexamined with the

Z strain riboprobe 1-142 asbefore. To distinguish between

MAPPOSITION (NUCLEOTIDE) 3 END OF FRAGMENT

RNAmadedenovoand thatpresentasvirioncontaminants,

mock reaction mixes containing excess EDTA were

proc-essedinparallel. As showninFig. 5, the results in vitrowere

basically similar to those in vivo, but there were some

notable differences. The leader RNAs, which were very

minorspeciesinvivo,weremoreprominentinvitro.Z strain

virions made two leader RNAs which migrated at

approxi-mately 55 and 40 nt in length, and H virions made an

additional leader band at 25 nt (which was also present in

vivo [not shown]). When the results were quantitated by

densitometry and normalized for uridine content and the

signaldue tocontaminatingNP mRNA(EDTA reaction)was

subtracted,therewas62 + 4%asmuch leaderasNP mRNA

in the Z reaction whentheleader bands were summed, and

B

0

LC

U

:L

z

10

C:

CL

0

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.612.285.554.381.561.2]

(6)

B:

TI Ti TI

RNase

A A A A A A I/iNP NP/PC JNP

3 1 3 1 3 1

_4I_P 120 /NP

NP 100-_-o

71.

PC 71 _-4

40

STRAIN H

_." _l .~

AR

4,..T..

*

0

--- 87 NP

P

;;-z

qbf Im

45,I o ---87

[image:6.612.99.530.72.355.2]

MOCK INF POL

FIG. 4. Estimation of readthrough frequencyat the leader-NP and NP-PCjunctions in strain Z- and H-infected cells. (A) Equal and increasingamountsof CsCl pellet RNA from strain Z and H infectionswereusedtoprotectriboprobes whichspanthe leader-NP and NP-PC junctions(Fig. 1). The length (in nucleotides) of the protected fragments and their originareshownonthe sides. (B) CsCl pellet RNA from

mock- andstrainH-infected cells (20 ,ug)aswellastheproducts from 10,ulofavirionpolymerase reactionwereusedtoprotectanHstrain riboprobe containingnt22to142. Thehybridsweredigested with RNase A aloneorplus RNaseTl.Thelengths and origins of theprotected

bandsare indicatedonthe side.

120± 13%intheHreaction. LeaderRNAswerethus made

inroughly equimolaramountsrelativetoNP mRNAinvitro,

asexpected, althoughmore leaderRNAwas made intheH

reaction. The equimolar levels of leaders andNPmRNAs in vitro suggest that the scarcity of leader RNA in vivo is predominantly due to turnover. Another difference is that the Z and H polymerases read through more frequently in vitro than in vivo. In this experiment, leader readthrough RNAswhichextendedto nt 142 orbeyondwere 106 ± 12%

asabundant asNPmRNA in the Z reactions andonly38 ±

4%asabundantintheHreaction.Thus,similartothe results in vivo, the Z polymerase reads through more frequently

than H, but the difference is only 3-fold, as opposed to 20-fold invivo. Inaddition, similar results wereobtained in

vitro when the reaction products were radioactive and the

riboprobe was not, and here the readthroughRNAs cannot be the result ofcomposite hybrids.

The apparentdifference in readthrough frequencyinvivo and invitro of both strains isunlikely to be dueentirely to more turnover in vivo, as we measured only 39%

read-through with the Z strain in vivo at 4 h even when protein

synthesis was inhibited to prevent encapsidation.

Further-more it is difficult to accept that the H strain difference of 63-fold (0.6 versus 38%) is only a question ofturnover. It appearsmorelikelythatthis differencereflects thepeculiar conditions of the in vitro reaction. Forexample, the virion reactions take place within envelopes containing the M protein, which may be a transcriptional inhibitor. Further-more, the optimum conditions for the in vitro reaction are

Z H

_ -w__142I/NP

-_-a -W ---87 NP

.._,,,,, .. - 5 e

-40e

2

^,,_ 5 e"

E DTA+ -+-+

+±+-FIG. 5. Polymerase readthrough in Z and H virion reactions.

Triplicate nonradioactiveinvitropolymerase reactionswerecarried

outwith bothstrainsasdescribed in Materials and Methods. Mock

reactionswerealsodone inparallel,whichcontainedanadditional

20 mM EDTA as indicated, to prevent de novo synthesis. The

products from 10 of each reaction were analyzed by RNase

mapping with riboprobe 1-142. The lengths of the leader RNAs

(script l) wereestimated relativetothose ofriboprobes of known

length (not shown) andaretherefore approximate. A:

PROBE NP/PC jgRNA 3 1

NP/PC174-_

NP

J. VIROL.

dO--- 142 I/N P

*

o

9

:sl;:.

0 .0.

on November 10, 2019 by guest

http://jvi.asm.org/

[image:6.612.370.513.425.638.2]

(7)

quite different from those of VSV. For Sendai virus, the replacement of Tris with HEPES buffer, chloride or acetate with phosphate as counter ions, and the addition of poly-L-glutamic acid, led to a 500-fold stimulation of activity. All or any of the above could account for the differences between readthrough in vivo and in vitro. We also note that whereas parallel reactions were remarkably reproducible (Fig. 5), readthrough levels varied significantly from one set of reactions to the next, even with the same preparation of virus, for reasons that are unclear (not shown).

DISCUSSION

RNase mapping is well suited to examine transcripts which cross gene junctions, as all three possibletranscripts can be measured simultaneously with a single probe. How-ever, as it is unreliable when thecomplementary RNA is also present and leader readthrough RNAs cannot be distin-guished from antigenomes by this method, we have re-stricted ourselves to characterizing the CsCl pellet RNAs. We therefore do not know what fraction of the leader readthrough RNAs are represented by this material. How-ever, by comparing their levels to those of the NP mRNA, whichquantitatively pellets in CsCl gradients and is thought to be the most stable of these RNAs, we can get a minimum estimate of their rates of synthesis. The time course of their accumulation during Z strain infections suggests that their levels at 18 h, when most of ourmeasurements were made, do in fact underestimate their rate ofsynthesis relative to NP mRNA, but only by about twofold.

We were surprised to find such large amounts of leader readthrough RNAs in Z-infected cells which had not been encapsidated but not in H straininfections. The presence of easily measurable amounts of the readthrough RNAs al-lowed us to measure how far their 3' ends had extended. According to the minimal model, polymerases which read through the firstjunction would be expected to continue on to the next. Concurrent assembly is supposedly only re-quired for the polymerase to disregard the junctional signals andshould have no effectbetween junctions. This test of the model has clearly not held up for the Z strain, as these replicases aborted synthesis shortly after having crossed the first junction, even though they are traversing the same template sequences as the transcriptases. Z strain tran-scriptases and replicases can thus be distinguished even in the absence of concurrent assembly.

Very few, if any, of the polymerases which read through thefirstjunction continue on to the next. Readthrough of the secondjunction can then only take place with polymerases which had reinitiated at the first, i.e., transcriptases. As the polymerases of the Z and H strains are indistinguishable by their frequency ofreadthrough at the second junction (0.6 and 0.5%, respectively) but vary 20-fold in the frequency with which they read through the firstjunction, this again suggests that thesepolymerases must somehow be different, independent of assembly.

The reason for the difference in these readthrough fre-quencies at the first junction remains unknown. However, a remarkably similar situation has previously been described forVSV. Perrault and co-workers, using a selection protocol forheat resistance, isolated mutants (polR) with enhanced ability to read through the leader-N junction but not subse-quent junctions (18, 19). The enhanced ability to read through the junction is reproducible in vitro, and here it maps to the N protein of the template rather than the exchangeable polymerase components (L and NS). The

VSVpolRmutants stressthepoint that the assembled Nor NPprotein, which isanintegralpartofthetemplate, is thus an integral part of the polymerase complex as well. The differences between the Z and H polymerases could then also lie in the template component ofthecomplex, but any component of this complex could of course be equally

responsible for the readthrough phenotype. This might now also include theSendai virus equivalent of the simian virus 5 and measles virus V proteins (23; R. Cattaneo, K. Kaelin,

and K. Baczko, Cell, in press), which does in fact existas

predicted (unpublished).

Moreimportantly, whatdifferencesbetween thereplicase

andtranscriptase can accountforthefindingthat the former requires concurrent assembly tocontinuedown thetemplate when it has crossed the first junction, whereas the latterdoes not? And to what extent does the minimal model still apply to Sendai virus? A key element of this model is that the switch from transcription to replication is controlled not at the level of chain initiation, but during elongation. This would also appear to be true for Sendai virus, as leader readthrough RNA synthesis continues similarly to NP mRNA synthesis in the presence ofcycloheximide (Fig. 2).

In that case, transcriptases and replicases cannot be distin-guished until the polymerase has crossed the firstjunction. It would appear that the act of crossing thejunction is respon-sible for this difference. Before the first junction, only one form of the polymerase can be distinguished, that which has initiated at nt 1. After this junction there are two forms; transcriptases which have reinitiated at NP and whose nascentchains no longer carry the leader sequences and are capped, and replicaseswhich have not reinitiated and whose nascent chains still contain the leader, with a 5' triphos-phate. Any or all of these may account for the difference in how they continue elongation without assembly.

Exchange or modification of polymerase subunits during reinitiation at the junction may also be involved. In the case of VSV polymerase reactions in vitro (3, 19), wild-type (wt) N mRNA synthesis requires high ATP concentrations for half-maximal synthesis, whereas leader synthesis takes place efficiently at lower ATP concentrations. In the polR mutants, however, N mRNA synthesis behaves similarly to leader synthesis with respect to ATP. These results are consistent with different phosphorylation states of the poly-merases which initiate leader and N mRNA synthesis, and these differences could take place during reinitiation at the first junction. In the case of paramyxoviruses, it has been found that in a persistent measles virus infection of human brain, a single gene (M) could be highly and specifically mutated whereas these mutations stopped precisely at the flanking PC junction (R. Cattaneo, A. Schmid, K. Baczko, V. ter Meulen, and M. A. Billeter, Cell, in press). One explanation they advance is that during replication a highly mutagenic polymerase copied the M but not the flanking genes. This suggests that polymerases could also be ex-changed at genejunctions during replication.

The minimal model clearly requires some modification to account for the Z straininfections. Amodified model which more adequatelydescribes both Sendai virus Z and H as well as VSV polR and wt is one in which the polymerase, in the absence of concurrent assembly, cannot continue for very long on its template unless it has reinitiated. Howfrequently it would stop at or read through the junction without assembly could of course becontrolled genetically. Concur-rent assembly would be required here not only for the polymerase to read through junctions, but also for itto read throughcontinuously. The leader-NP junctionwouldonly be

on November 10, 2019 by guest

http://jvi.asm.org/

(8)

the first ofmany places where the polymerase would termi-natewithout assembly if it had not reinitiated. The intracel-lular level of unassembled NP would still be the key element in the switch and would remain the rate-determining one. It would still act by initiating NC assembly on the nascent leader sequences, thereby controlling the rate with which polymerases continue without interruption to the end of the template. Initiation of nascent-chain assembly before the

polymerasehad reached the junction would result in efficient readthrough regardless of the basal rate of readthrough in the absence of assembly. It should be noted that our data do not rule out the possibility that the polymerases (and/or tem-plates) are committed to either the transcription or replica-tion mode before initiareplica-tion, but these differences only be-come apparent atthe first junction. This possibility appears less likely, however, as virions would be expected to contain predominantly those in the transcription mode, whereas in vitro they read through the first junction more frequently than is found in vivo (Fig. 5).

Finally, in Z versus H strain infections, termination at the first junction is relatively inefficient, yet this appears to have only amodest effect, as the Z strain grows only a little less

quickly than H (again similar to polR and wt VSV). As a

majority ofthe polymerases which read through terminate within 100 nt, some of them may move back to the junction and reinitiate, as suggested for the respiratory syncytial virus polymerase at the 22K-L junction (8). However, the modest effect seen in one-step growth curves in culture couldof course have been important in the natural evolution of thisstrain, which is considered to beas wt asthe H strain.

ACKNOWLEDGMENTS

Wethank HiroshiShibuta (Tokyo) forthe Zstrain of Sendai virus and cloneB36andJosephCurran (Geneva) for experimentaladvice. This work was supported by a grant from the Swiss National Science Fund.

LITERATURE CITED

1. Arnheiter, H., N. L. Davis, G. Wertz, M. Schubert, and R. A. Lazzarini. 1985. Role of nucleocapsid protein in regulating vesicularstomatitis virus RNAsynthesis. Cell41:259-267. 2. Banerjee, A. K. 1987. Transcription andreplication of

rhabdo-viruses. Microbiol. Rev.51:66-87.

3. Beckes, J. D., A. A. Haller, and J. Perrault. 1987. Differential effect ofATPconcentrationonsynthesis ofvesicularstomatitis virus leaderRNAs andmRNAs. J. Virol. 61:3470-3478. 4. Bellini, W. J., G. Englund, S. Rozenblatt, H. Arnheiter, and

C. D. Richardson. 1985. Measles virus P gene codes for two proteins. J.Virol. 53:908-919.

5. Blumberg, B. M., M. Leppert, andD. Kolakofsky. 1981. Inter-action of VSV leader RNA and nucleocapsid protein may control VSV genome replication. Cell23:837-845.

6. Blumberg, B. M., and D. Kolakofsky. 1981. Intracellular vesic-ular stomatitis virus leader RNAs are found in nucleocapsid structures. J. Virol. 40:568-576.

7. Blumberg, B. M., C. Giorgi, and D. Kolakofsky. 1983. Nprotein of vesicular stomatitis virus selectively encapsidates leader

RNA in vitro. Cell 32:559-567.

8. Collins, P. L., R. A. Olmsted, M. K. Spriggs, P. R. Johnson, and A.J.Buckler-White. 1987.Geneoverlap and site-specific atten-uationoftranscriptionof the viralpolymerase L gene of human respiratory syncytial virus. Proc. Natl. Acad. Sci. USA 84: 5134-5138.

9. Compans, R. W., and P. W. Choppin. 1967. Isolation and properties of the helical nucleocapsid of theparainfluenza virus SV5. Proc. Natl. Acad. Sci. USA57:949-956.

10. Curran, J. A., and D. Kolakofsky. 1988. Ribosomal initiation fromanACG codon intheSendai virus P/CmRNA. EMBO J. 7:245-251.

11. Dowling, P. C., C. Giorgi, L. Roux, L. A. Dethlefsen, M. E. Galantowicz, B.M.Blumberg, and D. Kolakofsky. 1983. Molec-ular cloning of the 3'-proximal third of Sendai virus genome. Proc. Natl. Acad. Sci. USA 80:5213-5216.

12. Giorgi, C., B. M. Blumberg, and D. Kolakofsky. 1983. Sendai virus contains overlapping genes expressed from a single mRNA.Cell35:829-836.

13. Gupta, K. C., and D. W. Kingsbury. 1984.Completesequences ofthe intergenic and mRNA start signals in the Sendai virus genome: homologies with the genome of vesicular stomatitis virus.Nucleic AcidsRes. 12:3829-3841.

14. Hill, V. M., C. C.Simonsen, and D. F. Summers. 1979. Charac-terization of vesicular stomatitis virus replicating complexes isolated inrenografin gradients. Virology99:75-83.

15. Kolakofsky, D., and B. M. Blumberg. 1982. A modelfor the control ofnon-segmented negative strandvirusesgenome rep-lication,p. 203-213. In B. W. J.Mahy,A.C.Minson,and G. K. Darby (ed.), Virus Persistence Symposium 33. Society for GeneralMicrobiology Ltd., Cambridge UniversityPress, Cam-bridge.

16. Kolakofsky, D., and L. Roux. 1987. The molecularbiology of paramyxoviruses,p. 277-279. In P.Bercoff(ed.),The molecular basis ofviralreplication. Plenum Publishing Corp.,NewYork. 17. Leppert, M., L.Rittenhouse,J. Perrault, D. F.Summers,and D. Kolakofsky. 1979. Plus and minus strand leaderRNAsin nega-tive strandvirus-infected cells. Cell 18:735-747.

18. Perrault, J., G. M. Clinton, and M. A. McClure. 1983. RNP template of vesicularstomatitisvirusregulatestranscription and replication functions. Cell35:175-185.

19. Perrault, J., and P. W. McClear. 1984. ATP dependence of vesicularstomatitisvirustranscription initiation andmodulation by mutationin the nucleocapsid protein.J. Virol. 51:635-642. 20. Portner, A., K. C. Gupta, J. M.Seyer, E. H. Beachey, and D. W.

Kingsbury. 1986. Localization and characterization of Sendai virus nonstructural C and C' proteins by antibodies against syntheticpeptides.Virus Res. 6:109-121.

21. Shioda, T., Y. Hidaka, T. Kanda, H. Shibuta, A. Nomoto, andK.

Iwasaki. 1983.Sequence of3,687 nucleotides fromthe 3'endof Sendai virus genome RNA and the predicted amino acid

se-quence of viral NP, P and C proteins. Nucleic Acids Res. 11:7317-7338.

22. Shioda, T., K. Iwasaki, and H. Shibuta. 1986. Determination of the complete nucleotidesequenceoftheSendai virus genome RNAand thepredicted amino acidsequencesoftheF, HN,and Lproteins. Nucleic Acids Res. 14:1545-1563.

23. Thomas, S. M., R. A. Lamb, and R. G. Paterson. 1988. Two mRNAs thatdifferbytwonontemplatednucleotides encode the aminocoterminalproteinsPand V of theparamyxovirus SV5. Cell 54:891-902.

J. VIROL.